aboutsummaryrefslogtreecommitdiffstats
path: root/lib/Target/Mips/MipsSEFrameLowering.cpp
blob: 7c79c4c564a7d592239f755b3ef56af02dab70b3 (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
//===-- MipsSEFrameLowering.cpp - Mips32/64 Frame Information -------------===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file contains the Mips32/64 implementation of TargetFrameLowering class.
//
//===----------------------------------------------------------------------===//

#include "MipsSEFrameLowering.h"
#include "MCTargetDesc/MipsBaseInfo.h"
#include "MipsAnalyzeImmediate.h"
#include "MipsMachineFunction.h"
#include "MipsSEInstrInfo.h"
#include "MipsSubtarget.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineModuleInfo.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/RegisterScavenging.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/Function.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Target/TargetOptions.h"

using namespace llvm;

namespace {
typedef MachineBasicBlock::iterator Iter;

static std::pair<unsigned, unsigned> getMFHiLoOpc(unsigned Src) {
  if (Mips::ACC64RegClass.contains(Src))
    return std::make_pair((unsigned)Mips::PseudoMFHI,
                          (unsigned)Mips::PseudoMFLO);

  if (Mips::ACC64DSPRegClass.contains(Src))
    return std::make_pair((unsigned)Mips::MFHI_DSP, (unsigned)Mips::MFLO_DSP);

  if (Mips::ACC128RegClass.contains(Src))
    return std::make_pair((unsigned)Mips::PseudoMFHI64,
                          (unsigned)Mips::PseudoMFLO64);

  return std::make_pair(0, 0);
}

/// Helper class to expand pseudos.
class ExpandPseudo {
public:
  ExpandPseudo(MachineFunction &MF);
  bool expand();

private:
  bool expandInstr(MachineBasicBlock &MBB, Iter I);
  void expandLoadCCond(MachineBasicBlock &MBB, Iter I);
  void expandStoreCCond(MachineBasicBlock &MBB, Iter I);
  void expandLoadACC(MachineBasicBlock &MBB, Iter I, unsigned RegSize);
  void expandStoreACC(MachineBasicBlock &MBB, Iter I, unsigned MFHiOpc,
                      unsigned MFLoOpc, unsigned RegSize);
  bool expandCopy(MachineBasicBlock &MBB, Iter I);
  bool expandCopyACC(MachineBasicBlock &MBB, Iter I, unsigned MFHiOpc,
                     unsigned MFLoOpc);
  bool expandBuildPairF64(MachineBasicBlock &MBB,
                          MachineBasicBlock::iterator I, bool FP64) const;
  bool expandExtractElementF64(MachineBasicBlock &MBB,
                               MachineBasicBlock::iterator I, bool FP64) const;

  MachineFunction &MF;
  MachineRegisterInfo &MRI;
  const MipsSubtarget &Subtarget;
  const MipsSEInstrInfo &TII;
  const MipsRegisterInfo &RegInfo;
};
}

ExpandPseudo::ExpandPseudo(MachineFunction &MF_)
    : MF(MF_), MRI(MF.getRegInfo()),
      Subtarget(static_cast<const MipsSubtarget &>(MF.getSubtarget())),
      TII(*static_cast<const MipsSEInstrInfo *>(Subtarget.getInstrInfo())),
      RegInfo(*Subtarget.getRegisterInfo()) {}

bool ExpandPseudo::expand() {
  bool Expanded = false;

  for (MachineFunction::iterator BB = MF.begin(), BBEnd = MF.end();
       BB != BBEnd; ++BB)
    for (Iter I = BB->begin(), End = BB->end(); I != End;)
      Expanded |= expandInstr(*BB, I++);

  return Expanded;
}

bool ExpandPseudo::expandInstr(MachineBasicBlock &MBB, Iter I) {
  switch(I->getOpcode()) {
  case Mips::LOAD_CCOND_DSP:
    expandLoadCCond(MBB, I);
    break;
  case Mips::STORE_CCOND_DSP:
    expandStoreCCond(MBB, I);
    break;
  case Mips::LOAD_ACC64:
  case Mips::LOAD_ACC64DSP:
    expandLoadACC(MBB, I, 4);
    break;
  case Mips::LOAD_ACC128:
    expandLoadACC(MBB, I, 8);
    break;
  case Mips::STORE_ACC64:
    expandStoreACC(MBB, I, Mips::PseudoMFHI, Mips::PseudoMFLO, 4);
    break;
  case Mips::STORE_ACC64DSP:
    expandStoreACC(MBB, I, Mips::MFHI_DSP, Mips::MFLO_DSP, 4);
    break;
  case Mips::STORE_ACC128:
    expandStoreACC(MBB, I, Mips::PseudoMFHI64, Mips::PseudoMFLO64, 8);
    break;
  case Mips::BuildPairF64:
    if (expandBuildPairF64(MBB, I, false))
      MBB.erase(I);
    return false;
  case Mips::BuildPairF64_64:
    if (expandBuildPairF64(MBB, I, true))
      MBB.erase(I);
    return false;
  case Mips::ExtractElementF64:
    if (expandExtractElementF64(MBB, I, false))
      MBB.erase(I);
    return false;
  case Mips::ExtractElementF64_64:
    if (expandExtractElementF64(MBB, I, true))
      MBB.erase(I);
    return false;
  case TargetOpcode::COPY:
    if (!expandCopy(MBB, I))
      return false;
    break;
  default:
    return false;
  }

  MBB.erase(I);
  return true;
}

void ExpandPseudo::expandLoadCCond(MachineBasicBlock &MBB, Iter I) {
  //  load $vr, FI
  //  copy ccond, $vr

  assert(I->getOperand(0).isReg() && I->getOperand(1).isFI());

  const TargetRegisterClass *RC = RegInfo.intRegClass(4);
  unsigned VR = MRI.createVirtualRegister(RC);
  unsigned Dst = I->getOperand(0).getReg(), FI = I->getOperand(1).getIndex();

  TII.loadRegFromStack(MBB, I, VR, FI, RC, &RegInfo, 0);
  BuildMI(MBB, I, I->getDebugLoc(), TII.get(TargetOpcode::COPY), Dst)
    .addReg(VR, RegState::Kill);
}

void ExpandPseudo::expandStoreCCond(MachineBasicBlock &MBB, Iter I) {
  //  copy $vr, ccond
  //  store $vr, FI

  assert(I->getOperand(0).isReg() && I->getOperand(1).isFI());

  const TargetRegisterClass *RC = RegInfo.intRegClass(4);
  unsigned VR = MRI.createVirtualRegister(RC);
  unsigned Src = I->getOperand(0).getReg(), FI = I->getOperand(1).getIndex();

  BuildMI(MBB, I, I->getDebugLoc(), TII.get(TargetOpcode::COPY), VR)
    .addReg(Src, getKillRegState(I->getOperand(0).isKill()));
  TII.storeRegToStack(MBB, I, VR, true, FI, RC, &RegInfo, 0);
}

void ExpandPseudo::expandLoadACC(MachineBasicBlock &MBB, Iter I,
                                 unsigned RegSize) {
  //  load $vr0, FI
  //  copy lo, $vr0
  //  load $vr1, FI + 4
  //  copy hi, $vr1

  assert(I->getOperand(0).isReg() && I->getOperand(1).isFI());

  const TargetRegisterClass *RC = RegInfo.intRegClass(RegSize);
  unsigned VR0 = MRI.createVirtualRegister(RC);
  unsigned VR1 = MRI.createVirtualRegister(RC);
  unsigned Dst = I->getOperand(0).getReg(), FI = I->getOperand(1).getIndex();
  unsigned Lo = RegInfo.getSubReg(Dst, Mips::sub_lo);
  unsigned Hi = RegInfo.getSubReg(Dst, Mips::sub_hi);
  DebugLoc DL = I->getDebugLoc();
  const MCInstrDesc &Desc = TII.get(TargetOpcode::COPY);

  TII.loadRegFromStack(MBB, I, VR0, FI, RC, &RegInfo, 0);
  BuildMI(MBB, I, DL, Desc, Lo).addReg(VR0, RegState::Kill);
  TII.loadRegFromStack(MBB, I, VR1, FI, RC, &RegInfo, RegSize);
  BuildMI(MBB, I, DL, Desc, Hi).addReg(VR1, RegState::Kill);
}

void ExpandPseudo::expandStoreACC(MachineBasicBlock &MBB, Iter I,
                                  unsigned MFHiOpc, unsigned MFLoOpc,
                                  unsigned RegSize) {
  //  mflo $vr0, src
  //  store $vr0, FI
  //  mfhi $vr1, src
  //  store $vr1, FI + 4

  assert(I->getOperand(0).isReg() && I->getOperand(1).isFI());

  const TargetRegisterClass *RC = RegInfo.intRegClass(RegSize);
  unsigned VR0 = MRI.createVirtualRegister(RC);
  unsigned VR1 = MRI.createVirtualRegister(RC);
  unsigned Src = I->getOperand(0).getReg(), FI = I->getOperand(1).getIndex();
  unsigned SrcKill = getKillRegState(I->getOperand(0).isKill());
  DebugLoc DL = I->getDebugLoc();

  BuildMI(MBB, I, DL, TII.get(MFLoOpc), VR0).addReg(Src);
  TII.storeRegToStack(MBB, I, VR0, true, FI, RC, &RegInfo, 0);
  BuildMI(MBB, I, DL, TII.get(MFHiOpc), VR1).addReg(Src, SrcKill);
  TII.storeRegToStack(MBB, I, VR1, true, FI, RC, &RegInfo, RegSize);
}

bool ExpandPseudo::expandCopy(MachineBasicBlock &MBB, Iter I) {
  unsigned Src = I->getOperand(1).getReg();
  std::pair<unsigned, unsigned> Opcodes = getMFHiLoOpc(Src);

  if (!Opcodes.first)
    return false;

  return expandCopyACC(MBB, I, Opcodes.first, Opcodes.second);
}

bool ExpandPseudo::expandCopyACC(MachineBasicBlock &MBB, Iter I,
                                 unsigned MFHiOpc, unsigned MFLoOpc) {
  //  mflo $vr0, src
  //  copy dst_lo, $vr0
  //  mfhi $vr1, src
  //  copy dst_hi, $vr1

  unsigned Dst = I->getOperand(0).getReg(), Src = I->getOperand(1).getReg();
  unsigned VRegSize = RegInfo.getMinimalPhysRegClass(Dst)->getSize() / 2;
  const TargetRegisterClass *RC = RegInfo.intRegClass(VRegSize);
  unsigned VR0 = MRI.createVirtualRegister(RC);
  unsigned VR1 = MRI.createVirtualRegister(RC);
  unsigned SrcKill = getKillRegState(I->getOperand(1).isKill());
  unsigned DstLo = RegInfo.getSubReg(Dst, Mips::sub_lo);
  unsigned DstHi = RegInfo.getSubReg(Dst, Mips::sub_hi);
  DebugLoc DL = I->getDebugLoc();

  BuildMI(MBB, I, DL, TII.get(MFLoOpc), VR0).addReg(Src);
  BuildMI(MBB, I, DL, TII.get(TargetOpcode::COPY), DstLo)
    .addReg(VR0, RegState::Kill);
  BuildMI(MBB, I, DL, TII.get(MFHiOpc), VR1).addReg(Src, SrcKill);
  BuildMI(MBB, I, DL, TII.get(TargetOpcode::COPY), DstHi)
    .addReg(VR1, RegState::Kill);
  return true;
}

/// This method expands the same instruction that MipsSEInstrInfo::
/// expandBuildPairF64 does, for the case when ABI is fpxx and mthc1 is not
/// available and the case where the ABI is FP64A. It is implemented here
/// because frame indexes are eliminated before MipsSEInstrInfo::
/// expandBuildPairF64 is called.
bool ExpandPseudo::expandBuildPairF64(MachineBasicBlock &MBB,
                                      MachineBasicBlock::iterator I,
                                      bool FP64) const {
  // For fpxx and when mthc1 is not available, use:
  //   spill + reload via ldc1
  //
  // The case where dmtc1 is available doesn't need to be handled here
  // because it never creates a BuildPairF64 node.
  //
  // The FP64A ABI (fp64 with nooddspreg) must also use a spill/reload sequence
  // for odd-numbered double precision values (because the lower 32-bits is
  // transferred with mtc1 which is redirected to the upper half of the even
  // register). Unfortunately, we have to make this decision before register
  // allocation so for now we use a spill/reload sequence for all
  // double-precision values in regardless of being an odd/even register.
  if ((Subtarget.isABI_FPXX() && !Subtarget.hasMTHC1()) ||
      (FP64 && !Subtarget.useOddSPReg())) {
    unsigned DstReg = I->getOperand(0).getReg();
    unsigned LoReg = I->getOperand(1).getReg();
    unsigned HiReg = I->getOperand(2).getReg();

    // It should be impossible to have FGR64 on MIPS-II or MIPS32r1 (which are
    // the cases where mthc1 is not available). 64-bit architectures and
    // MIPS32r2 or later can use FGR64 though.
    assert(Subtarget.isGP64bit() || Subtarget.hasMTHC1() ||
           !Subtarget.isFP64bit());

    const TargetRegisterClass *RC = &Mips::GPR32RegClass;
    const TargetRegisterClass *RC2 =
        FP64 ? &Mips::FGR64RegClass : &Mips::AFGR64RegClass;

    // We re-use the same spill slot each time so that the stack frame doesn't
    // grow too much in functions with a large number of moves.
    int FI = MF.getInfo<MipsFunctionInfo>()->getMoveF64ViaSpillFI(RC2);
    if (!Subtarget.isLittle())
      std::swap(LoReg, HiReg);
    TII.storeRegToStack(MBB, I, LoReg, I->getOperand(1).isKill(), FI, RC,
                        &RegInfo, 0);
    TII.storeRegToStack(MBB, I, HiReg, I->getOperand(2).isKill(), FI, RC,
                        &RegInfo, 4);
    TII.loadRegFromStack(MBB, I, DstReg, FI, RC2, &RegInfo, 0);
    return true;
  }

  return false;
}

/// This method expands the same instruction that MipsSEInstrInfo::
/// expandExtractElementF64 does, for the case when ABI is fpxx and mfhc1 is not
/// available and the case where the ABI is FP64A. It is implemented here
/// because frame indexes are eliminated before MipsSEInstrInfo::
/// expandExtractElementF64 is called.
bool ExpandPseudo::expandExtractElementF64(MachineBasicBlock &MBB,
                                           MachineBasicBlock::iterator I,
                                           bool FP64) const {
  // For fpxx and when mfhc1 is not available, use:
  //   spill + reload via ldc1
  //
  // The case where dmfc1 is available doesn't need to be handled here
  // because it never creates a ExtractElementF64 node.
  //
  // The FP64A ABI (fp64 with nooddspreg) must also use a spill/reload sequence
  // for odd-numbered double precision values (because the lower 32-bits is
  // transferred with mfc1 which is redirected to the upper half of the even
  // register). Unfortunately, we have to make this decision before register
  // allocation so for now we use a spill/reload sequence for all
  // double-precision values in regardless of being an odd/even register.

  if ((Subtarget.isABI_FPXX() && !Subtarget.hasMTHC1()) ||
      (FP64 && !Subtarget.useOddSPReg())) {
    unsigned DstReg = I->getOperand(0).getReg();
    unsigned SrcReg = I->getOperand(1).getReg();
    unsigned N = I->getOperand(2).getImm();
    int64_t Offset = 4 * (Subtarget.isLittle() ? N : (1 - N));

    // It should be impossible to have FGR64 on MIPS-II or MIPS32r1 (which are
    // the cases where mfhc1 is not available). 64-bit architectures and
    // MIPS32r2 or later can use FGR64 though.
    assert(Subtarget.isGP64bit() || Subtarget.hasMTHC1() ||
           !Subtarget.isFP64bit());

    const TargetRegisterClass *RC =
        FP64 ? &Mips::FGR64RegClass : &Mips::AFGR64RegClass;
    const TargetRegisterClass *RC2 = &Mips::GPR32RegClass;

    // We re-use the same spill slot each time so that the stack frame doesn't
    // grow too much in functions with a large number of moves.
    int FI = MF.getInfo<MipsFunctionInfo>()->getMoveF64ViaSpillFI(RC);
    TII.storeRegToStack(MBB, I, SrcReg, I->getOperand(1).isKill(), FI, RC,
                        &RegInfo, 0);
    TII.loadRegFromStack(MBB, I, DstReg, FI, RC2, &RegInfo, Offset);
    return true;
  }

  return false;
}

MipsSEFrameLowering::MipsSEFrameLowering(const MipsSubtarget &STI)
    : MipsFrameLowering(STI, STI.stackAlignment()) {}

unsigned MipsSEFrameLowering::ehDataReg(unsigned I) const {
  static const unsigned EhDataReg[] = {
    Mips::A0, Mips::A1, Mips::A2, Mips::A3
  };
  static const unsigned EhDataReg64[] = {
    Mips::A0_64, Mips::A1_64, Mips::A2_64, Mips::A3_64
  };

  return STI.isABI_N64() ? EhDataReg64[I] : EhDataReg[I];
}

void MipsSEFrameLowering::emitPrologue(MachineFunction &MF) const {
  MachineBasicBlock &MBB   = MF.front();
  MachineFrameInfo *MFI    = MF.getFrameInfo();
  MipsFunctionInfo *MipsFI = MF.getInfo<MipsFunctionInfo>();

  const MipsSEInstrInfo &TII =
      *static_cast<const MipsSEInstrInfo *>(STI.getInstrInfo());
  const MipsRegisterInfo &RegInfo =
      *static_cast<const MipsRegisterInfo *>(STI.getRegisterInfo());

  MachineBasicBlock::iterator MBBI = MBB.begin();
  DebugLoc dl = MBBI != MBB.end() ? MBBI->getDebugLoc() : DebugLoc();
  unsigned SP = STI.isABI_N64() ? Mips::SP_64 : Mips::SP;
  unsigned FP = STI.isABI_N64() ? Mips::FP_64 : Mips::FP;
  unsigned ZERO = STI.isABI_N64() ? Mips::ZERO_64 : Mips::ZERO;
  unsigned ADDu = STI.isABI_N64() ? Mips::DADDu : Mips::ADDu;

  // First, compute final stack size.
  uint64_t StackSize = MFI->getStackSize();

  // No need to allocate space on the stack.
  if (StackSize == 0 && !MFI->adjustsStack()) return;

  MachineModuleInfo &MMI = MF.getMMI();
  const MCRegisterInfo *MRI = MMI.getContext().getRegisterInfo();
  MachineLocation DstML, SrcML;

  // Adjust stack.
  TII.adjustStackPtr(SP, -StackSize, MBB, MBBI);

  // emit ".cfi_def_cfa_offset StackSize"
  unsigned CFIIndex = MMI.addFrameInst(
      MCCFIInstruction::createDefCfaOffset(nullptr, -StackSize));
  BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
      .addCFIIndex(CFIIndex);

  const std::vector<CalleeSavedInfo> &CSI = MFI->getCalleeSavedInfo();

  if (CSI.size()) {
    // Find the instruction past the last instruction that saves a callee-saved
    // register to the stack.
    for (unsigned i = 0; i < CSI.size(); ++i)
      ++MBBI;

    // Iterate over list of callee-saved registers and emit .cfi_offset
    // directives.
    for (std::vector<CalleeSavedInfo>::const_iterator I = CSI.begin(),
           E = CSI.end(); I != E; ++I) {
      int64_t Offset = MFI->getObjectOffset(I->getFrameIdx());
      unsigned Reg = I->getReg();

      // If Reg is a double precision register, emit two cfa_offsets,
      // one for each of the paired single precision registers.
      if (Mips::AFGR64RegClass.contains(Reg)) {
        unsigned Reg0 =
            MRI->getDwarfRegNum(RegInfo.getSubReg(Reg, Mips::sub_lo), true);
        unsigned Reg1 =
            MRI->getDwarfRegNum(RegInfo.getSubReg(Reg, Mips::sub_hi), true);

        if (!STI.isLittle())
          std::swap(Reg0, Reg1);

        unsigned CFIIndex = MMI.addFrameInst(
            MCCFIInstruction::createOffset(nullptr, Reg0, Offset));
        BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
            .addCFIIndex(CFIIndex);

        CFIIndex = MMI.addFrameInst(
            MCCFIInstruction::createOffset(nullptr, Reg1, Offset + 4));
        BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
            .addCFIIndex(CFIIndex);
      } else if (Mips::FGR64RegClass.contains(Reg)) {
        unsigned Reg0 = MRI->getDwarfRegNum(Reg, true);
        unsigned Reg1 = MRI->getDwarfRegNum(Reg, true) + 1;

        if (!STI.isLittle())
          std::swap(Reg0, Reg1);

        unsigned CFIIndex = MMI.addFrameInst(
          MCCFIInstruction::createOffset(nullptr, Reg0, Offset));
        BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
            .addCFIIndex(CFIIndex);

        CFIIndex = MMI.addFrameInst(
          MCCFIInstruction::createOffset(nullptr, Reg1, Offset + 4));
        BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
            .addCFIIndex(CFIIndex);
      } else {
        // Reg is either in GPR32 or FGR32.
        unsigned CFIIndex = MMI.addFrameInst(MCCFIInstruction::createOffset(
            nullptr, MRI->getDwarfRegNum(Reg, 1), Offset));
        BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
            .addCFIIndex(CFIIndex);
      }
    }
  }

  if (MipsFI->callsEhReturn()) {
    const TargetRegisterClass *RC = STI.isABI_N64() ?
        &Mips::GPR64RegClass : &Mips::GPR32RegClass;

    // Insert instructions that spill eh data registers.
    for (int I = 0; I < 4; ++I) {
      if (!MBB.isLiveIn(ehDataReg(I)))
        MBB.addLiveIn(ehDataReg(I));
      TII.storeRegToStackSlot(MBB, MBBI, ehDataReg(I), false,
                              MipsFI->getEhDataRegFI(I), RC, &RegInfo);
    }

    // Emit .cfi_offset directives for eh data registers.
    for (int I = 0; I < 4; ++I) {
      int64_t Offset = MFI->getObjectOffset(MipsFI->getEhDataRegFI(I));
      unsigned Reg = MRI->getDwarfRegNum(ehDataReg(I), true);
      unsigned CFIIndex = MMI.addFrameInst(
          MCCFIInstruction::createOffset(nullptr, Reg, Offset));
      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
          .addCFIIndex(CFIIndex);
    }
  }

  // if framepointer enabled, set it to point to the stack pointer.
  if (hasFP(MF)) {
    // Insert instruction "move $fp, $sp" at this location.
    BuildMI(MBB, MBBI, dl, TII.get(ADDu), FP).addReg(SP).addReg(ZERO)
      .setMIFlag(MachineInstr::FrameSetup);

    // emit ".cfi_def_cfa_register $fp"
    unsigned CFIIndex = MMI.addFrameInst(MCCFIInstruction::createDefCfaRegister(
        nullptr, MRI->getDwarfRegNum(FP, true)));
    BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
        .addCFIIndex(CFIIndex);
  }
}

void MipsSEFrameLowering::emitEpilogue(MachineFunction &MF,
                                       MachineBasicBlock &MBB) const {
  MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
  MachineFrameInfo *MFI            = MF.getFrameInfo();
  MipsFunctionInfo *MipsFI = MF.getInfo<MipsFunctionInfo>();

  const MipsSEInstrInfo &TII =
      *static_cast<const MipsSEInstrInfo *>(STI.getInstrInfo());
  const MipsRegisterInfo &RegInfo =
      *static_cast<const MipsRegisterInfo *>(STI.getRegisterInfo());

  DebugLoc dl = MBBI->getDebugLoc();
  unsigned SP = STI.isABI_N64() ? Mips::SP_64 : Mips::SP;
  unsigned FP = STI.isABI_N64() ? Mips::FP_64 : Mips::FP;
  unsigned ZERO = STI.isABI_N64() ? Mips::ZERO_64 : Mips::ZERO;
  unsigned ADDu = STI.isABI_N64() ? Mips::DADDu : Mips::ADDu;

  // if framepointer enabled, restore the stack pointer.
  if (hasFP(MF)) {
    // Find the first instruction that restores a callee-saved register.
    MachineBasicBlock::iterator I = MBBI;

    for (unsigned i = 0; i < MFI->getCalleeSavedInfo().size(); ++i)
      --I;

    // Insert instruction "move $sp, $fp" at this location.
    BuildMI(MBB, I, dl, TII.get(ADDu), SP).addReg(FP).addReg(ZERO);
  }

  if (MipsFI->callsEhReturn()) {
    const TargetRegisterClass *RC = STI.isABI_N64() ?
        &Mips::GPR64RegClass : &Mips::GPR32RegClass;

    // Find first instruction that restores a callee-saved register.
    MachineBasicBlock::iterator I = MBBI;
    for (unsigned i = 0; i < MFI->getCalleeSavedInfo().size(); ++i)
      --I;

    // Insert instructions that restore eh data registers.
    for (int J = 0; J < 4; ++J) {
      TII.loadRegFromStackSlot(MBB, I, ehDataReg(J), MipsFI->getEhDataRegFI(J),
                               RC, &RegInfo);
    }
  }

  // Get the number of bytes from FrameInfo
  uint64_t StackSize = MFI->getStackSize();

  if (!StackSize)
    return;

  // Adjust stack.
  TII.adjustStackPtr(SP, StackSize, MBB, MBBI);
}

bool MipsSEFrameLowering::
spillCalleeSavedRegisters(MachineBasicBlock &MBB,
                          MachineBasicBlock::iterator MI,
                          const std::vector<CalleeSavedInfo> &CSI,
                          const TargetRegisterInfo *TRI) const {
  MachineFunction *MF = MBB.getParent();
  MachineBasicBlock *EntryBlock = MF->begin();
  const TargetInstrInfo &TII = *STI.getInstrInfo();

  for (unsigned i = 0, e = CSI.size(); i != e; ++i) {
    // Add the callee-saved register as live-in. Do not add if the register is
    // RA and return address is taken, because it has already been added in
    // method MipsTargetLowering::LowerRETURNADDR.
    // It's killed at the spill, unless the register is RA and return address
    // is taken.
    unsigned Reg = CSI[i].getReg();
    bool IsRAAndRetAddrIsTaken = (Reg == Mips::RA || Reg == Mips::RA_64)
        && MF->getFrameInfo()->isReturnAddressTaken();
    if (!IsRAAndRetAddrIsTaken)
      EntryBlock->addLiveIn(Reg);

    // Insert the spill to the stack frame.
    bool IsKill = !IsRAAndRetAddrIsTaken;
    const TargetRegisterClass *RC = TRI->getMinimalPhysRegClass(Reg);
    TII.storeRegToStackSlot(*EntryBlock, MI, Reg, IsKill,
                            CSI[i].getFrameIdx(), RC, TRI);
  }

  return true;
}

bool
MipsSEFrameLowering::hasReservedCallFrame(const MachineFunction &MF) const {
  const MachineFrameInfo *MFI = MF.getFrameInfo();

  // Reserve call frame if the size of the maximum call frame fits into 16-bit
  // immediate field and there are no variable sized objects on the stack.
  // Make sure the second register scavenger spill slot can be accessed with one
  // instruction.
  return isInt<16>(MFI->getMaxCallFrameSize() + getStackAlignment()) &&
    !MFI->hasVarSizedObjects();
}

// Eliminate ADJCALLSTACKDOWN, ADJCALLSTACKUP pseudo instructions
void MipsSEFrameLowering::
eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
                              MachineBasicBlock::iterator I) const {
  const MipsSEInstrInfo &TII =
      *static_cast<const MipsSEInstrInfo *>(STI.getInstrInfo());

  if (!hasReservedCallFrame(MF)) {
    int64_t Amount = I->getOperand(0).getImm();

    if (I->getOpcode() == Mips::ADJCALLSTACKDOWN)
      Amount = -Amount;

    unsigned SP = STI.isABI_N64() ? Mips::SP_64 : Mips::SP;
    TII.adjustStackPtr(SP, Amount, MBB, I);
  }

  MBB.erase(I);
}

void MipsSEFrameLowering::
processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
                                     RegScavenger *RS) const {
  MachineRegisterInfo &MRI = MF.getRegInfo();
  MipsFunctionInfo *MipsFI = MF.getInfo<MipsFunctionInfo>();
  unsigned FP = STI.isABI_N64() ? Mips::FP_64 : Mips::FP;

  // Mark $fp as used if function has dedicated frame pointer.
  if (hasFP(MF))
    MRI.setPhysRegUsed(FP);

  // Create spill slots for eh data registers if function calls eh_return.
  if (MipsFI->callsEhReturn())
    MipsFI->createEhDataRegsFI();

  // Expand pseudo instructions which load, store or copy accumulators.
  // Add an emergency spill slot if a pseudo was expanded.
  if (ExpandPseudo(MF).expand()) {
    // The spill slot should be half the size of the accumulator. If target is
    // mips64, it should be 64-bit, otherwise it should be 32-bt.
    const TargetRegisterClass *RC = STI.hasMips64() ?
      &Mips::GPR64RegClass : &Mips::GPR32RegClass;
    int FI = MF.getFrameInfo()->CreateStackObject(RC->getSize(),
                                                  RC->getAlignment(), false);
    RS->addScavengingFrameIndex(FI);
  }

  // Set scavenging frame index if necessary.
  uint64_t MaxSPOffset = MF.getInfo<MipsFunctionInfo>()->getIncomingArgSize() +
    estimateStackSize(MF);

  if (isInt<16>(MaxSPOffset))
    return;

  const TargetRegisterClass *RC = STI.isABI_N64() ?
    &Mips::GPR64RegClass : &Mips::GPR32RegClass;
  int FI = MF.getFrameInfo()->CreateStackObject(RC->getSize(),
                                                RC->getAlignment(), false);
  RS->addScavengingFrameIndex(FI);
}

const MipsFrameLowering *
llvm::createMipsSEFrameLowering(const MipsSubtarget &ST) {
  return new MipsSEFrameLowering(ST);
}