llvm.org GIT mirror llvm / release_90 lib / Target / AMDGPU / SIOptimizeExecMasking.cpp
release_90

Tree @release_90 (Download .tar.gz)

SIOptimizeExecMasking.cpp @release_90raw · history · blame

  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
//===-- SIOptimizeExecMasking.cpp -----------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//

#include "AMDGPU.h"
#include "AMDGPUSubtarget.h"
#include "SIInstrInfo.h"
#include "MCTargetDesc/AMDGPUMCTargetDesc.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/Support/Debug.h"

using namespace llvm;

#define DEBUG_TYPE "si-optimize-exec-masking"

namespace {

class SIOptimizeExecMasking : public MachineFunctionPass {
public:
  static char ID;

public:
  SIOptimizeExecMasking() : MachineFunctionPass(ID) {
    initializeSIOptimizeExecMaskingPass(*PassRegistry::getPassRegistry());
  }

  bool runOnMachineFunction(MachineFunction &MF) override;

  StringRef getPassName() const override {
    return "SI optimize exec mask operations";
  }

  void getAnalysisUsage(AnalysisUsage &AU) const override {
    AU.setPreservesCFG();
    MachineFunctionPass::getAnalysisUsage(AU);
  }
};

} // End anonymous namespace.

INITIALIZE_PASS_BEGIN(SIOptimizeExecMasking, DEBUG_TYPE,
                      "SI optimize exec mask operations", false, false)
INITIALIZE_PASS_DEPENDENCY(LiveIntervals)
INITIALIZE_PASS_END(SIOptimizeExecMasking, DEBUG_TYPE,
                    "SI optimize exec mask operations", false, false)

char SIOptimizeExecMasking::ID = 0;

char &llvm::SIOptimizeExecMaskingID = SIOptimizeExecMasking::ID;

/// If \p MI is a copy from exec, return the register copied to.
static unsigned isCopyFromExec(const MachineInstr &MI, const GCNSubtarget &ST) {
  switch (MI.getOpcode()) {
  case AMDGPU::COPY:
  case AMDGPU::S_MOV_B64:
  case AMDGPU::S_MOV_B64_term:
  case AMDGPU::S_MOV_B32:
  case AMDGPU::S_MOV_B32_term: {
    const MachineOperand &Src = MI.getOperand(1);
    if (Src.isReg() &&
        Src.getReg() == (ST.isWave32() ? AMDGPU::EXEC_LO : AMDGPU::EXEC))
      return MI.getOperand(0).getReg();
  }
  }

  return AMDGPU::NoRegister;
}

/// If \p MI is a copy to exec, return the register copied from.
static unsigned isCopyToExec(const MachineInstr &MI, const GCNSubtarget &ST) {
  switch (MI.getOpcode()) {
  case AMDGPU::COPY:
  case AMDGPU::S_MOV_B64:
  case AMDGPU::S_MOV_B32: {
    const MachineOperand &Dst = MI.getOperand(0);
    if (Dst.isReg() &&
        Dst.getReg() == (ST.isWave32() ? AMDGPU::EXEC_LO : AMDGPU::EXEC) &&
        MI.getOperand(1).isReg())
      return MI.getOperand(1).getReg();
    break;
  }
  case AMDGPU::S_MOV_B64_term:
  case AMDGPU::S_MOV_B32_term:
    llvm_unreachable("should have been replaced");
  }

  return AMDGPU::NoRegister;
}

/// If \p MI is a logical operation on an exec value,
/// return the register copied to.
static unsigned isLogicalOpOnExec(const MachineInstr &MI) {
  switch (MI.getOpcode()) {
  case AMDGPU::S_AND_B64:
  case AMDGPU::S_OR_B64:
  case AMDGPU::S_XOR_B64:
  case AMDGPU::S_ANDN2_B64:
  case AMDGPU::S_ORN2_B64:
  case AMDGPU::S_NAND_B64:
  case AMDGPU::S_NOR_B64:
  case AMDGPU::S_XNOR_B64: {
    const MachineOperand &Src1 = MI.getOperand(1);
    if (Src1.isReg() && Src1.getReg() == AMDGPU::EXEC)
      return MI.getOperand(0).getReg();
    const MachineOperand &Src2 = MI.getOperand(2);
    if (Src2.isReg() && Src2.getReg() == AMDGPU::EXEC)
      return MI.getOperand(0).getReg();
    break;
  }
  case AMDGPU::S_AND_B32:
  case AMDGPU::S_OR_B32:
  case AMDGPU::S_XOR_B32:
  case AMDGPU::S_ANDN2_B32:
  case AMDGPU::S_ORN2_B32:
  case AMDGPU::S_NAND_B32:
  case AMDGPU::S_NOR_B32:
  case AMDGPU::S_XNOR_B32: {
    const MachineOperand &Src1 = MI.getOperand(1);
    if (Src1.isReg() && Src1.getReg() == AMDGPU::EXEC_LO)
      return MI.getOperand(0).getReg();
    const MachineOperand &Src2 = MI.getOperand(2);
    if (Src2.isReg() && Src2.getReg() == AMDGPU::EXEC_LO)
      return MI.getOperand(0).getReg();
    break;
  }
  }

  return AMDGPU::NoRegister;
}

static unsigned getSaveExecOp(unsigned Opc) {
  switch (Opc) {
  case AMDGPU::S_AND_B64:
    return AMDGPU::S_AND_SAVEEXEC_B64;
  case AMDGPU::S_OR_B64:
    return AMDGPU::S_OR_SAVEEXEC_B64;
  case AMDGPU::S_XOR_B64:
    return AMDGPU::S_XOR_SAVEEXEC_B64;
  case AMDGPU::S_ANDN2_B64:
    return AMDGPU::S_ANDN2_SAVEEXEC_B64;
  case AMDGPU::S_ORN2_B64:
    return AMDGPU::S_ORN2_SAVEEXEC_B64;
  case AMDGPU::S_NAND_B64:
    return AMDGPU::S_NAND_SAVEEXEC_B64;
  case AMDGPU::S_NOR_B64:
    return AMDGPU::S_NOR_SAVEEXEC_B64;
  case AMDGPU::S_XNOR_B64:
    return AMDGPU::S_XNOR_SAVEEXEC_B64;
  case AMDGPU::S_AND_B32:
    return AMDGPU::S_AND_SAVEEXEC_B32;
  case AMDGPU::S_OR_B32:
    return AMDGPU::S_OR_SAVEEXEC_B32;
  case AMDGPU::S_XOR_B32:
    return AMDGPU::S_XOR_SAVEEXEC_B32;
  case AMDGPU::S_ANDN2_B32:
    return AMDGPU::S_ANDN2_SAVEEXEC_B32;
  case AMDGPU::S_ORN2_B32:
    return AMDGPU::S_ORN2_SAVEEXEC_B32;
  case AMDGPU::S_NAND_B32:
    return AMDGPU::S_NAND_SAVEEXEC_B32;
  case AMDGPU::S_NOR_B32:
    return AMDGPU::S_NOR_SAVEEXEC_B32;
  case AMDGPU::S_XNOR_B32:
    return AMDGPU::S_XNOR_SAVEEXEC_B32;
  default:
    return AMDGPU::INSTRUCTION_LIST_END;
  }
}

// These are only terminators to get correct spill code placement during
// register allocation, so turn them back into normal instructions. Only one of
// these is expected per block.
static bool removeTerminatorBit(const SIInstrInfo &TII, MachineInstr &MI) {
  switch (MI.getOpcode()) {
  case AMDGPU::S_MOV_B64_term:
  case AMDGPU::S_MOV_B32_term: {
    MI.setDesc(TII.get(AMDGPU::COPY));
    return true;
  }
  case AMDGPU::S_XOR_B64_term: {
    // This is only a terminator to get the correct spill code placement during
    // register allocation.
    MI.setDesc(TII.get(AMDGPU::S_XOR_B64));
    return true;
  }
  case AMDGPU::S_XOR_B32_term: {
    // This is only a terminator to get the correct spill code placement during
    // register allocation.
    MI.setDesc(TII.get(AMDGPU::S_XOR_B32));
    return true;
  }
  case AMDGPU::S_OR_B32_term: {
    // This is only a terminator to get the correct spill code placement during
    // register allocation.
    MI.setDesc(TII.get(AMDGPU::S_OR_B32));
    return true;
  }
  case AMDGPU::S_ANDN2_B64_term: {
    // This is only a terminator to get the correct spill code placement during
    // register allocation.
    MI.setDesc(TII.get(AMDGPU::S_ANDN2_B64));
    return true;
  }
  case AMDGPU::S_ANDN2_B32_term: {
    // This is only a terminator to get the correct spill code placement during
    // register allocation.
    MI.setDesc(TII.get(AMDGPU::S_ANDN2_B32));
    return true;
  }
  default:
    return false;
  }
}

static MachineBasicBlock::reverse_iterator fixTerminators(
  const SIInstrInfo &TII,
  MachineBasicBlock &MBB) {
  MachineBasicBlock::reverse_iterator I = MBB.rbegin(), E = MBB.rend();
  for (; I != E; ++I) {
    if (!I->isTerminator())
      return I;

    if (removeTerminatorBit(TII, *I))
      return I;
  }

  return E;
}

static MachineBasicBlock::reverse_iterator findExecCopy(
  const SIInstrInfo &TII,
  const GCNSubtarget &ST,
  MachineBasicBlock &MBB,
  MachineBasicBlock::reverse_iterator I,
  unsigned CopyToExec) {
  const unsigned InstLimit = 25;

  auto E = MBB.rend();
  for (unsigned N = 0; N <= InstLimit && I != E; ++I, ++N) {
    unsigned CopyFromExec = isCopyFromExec(*I, ST);
    if (CopyFromExec != AMDGPU::NoRegister)
      return I;
  }

  return E;
}

// XXX - Seems LivePhysRegs doesn't work correctly since it will incorrectly
// report the register as unavailable because a super-register with a lane mask
// is unavailable.
static bool isLiveOut(const MachineBasicBlock &MBB, unsigned Reg) {
  for (MachineBasicBlock *Succ : MBB.successors()) {
    if (Succ->isLiveIn(Reg))
      return true;
  }

  return false;
}

bool SIOptimizeExecMasking::runOnMachineFunction(MachineFunction &MF) {
  if (skipFunction(MF.getFunction()))
    return false;

  const GCNSubtarget &ST = MF.getSubtarget<GCNSubtarget>();
  const SIRegisterInfo *TRI = ST.getRegisterInfo();
  const SIInstrInfo *TII = ST.getInstrInfo();
  unsigned Exec = ST.isWave32() ? AMDGPU::EXEC_LO : AMDGPU::EXEC;

  // Optimize sequences emitted for control flow lowering. They are originally
  // emitted as the separate operations because spill code may need to be
  // inserted for the saved copy of exec.
  //
  //     x = copy exec
  //     z = s_<op>_b64 x, y
  //     exec = copy z
  // =>
  //     x = s_<op>_saveexec_b64 y
  //

  for (MachineBasicBlock &MBB : MF) {
    MachineBasicBlock::reverse_iterator I = fixTerminators(*TII, MBB);
    MachineBasicBlock::reverse_iterator E = MBB.rend();
    if (I == E)
      continue;

    unsigned CopyToExec = isCopyToExec(*I, ST);
    if (CopyToExec == AMDGPU::NoRegister)
      continue;

    // Scan backwards to find the def.
    auto CopyToExecInst = &*I;
    auto CopyFromExecInst = findExecCopy(*TII, ST, MBB, I, CopyToExec);
    if (CopyFromExecInst == E) {
      auto PrepareExecInst = std::next(I);
      if (PrepareExecInst == E)
        continue;
      // Fold exec = COPY (S_AND_B64 reg, exec) -> exec = S_AND_B64 reg, exec
      if (CopyToExecInst->getOperand(1).isKill() &&
          isLogicalOpOnExec(*PrepareExecInst) == CopyToExec) {
        LLVM_DEBUG(dbgs() << "Fold exec copy: " << *PrepareExecInst);

        PrepareExecInst->getOperand(0).setReg(Exec);

        LLVM_DEBUG(dbgs() << "into: " << *PrepareExecInst << '\n');

        CopyToExecInst->eraseFromParent();
      }

      continue;
    }

    if (isLiveOut(MBB, CopyToExec)) {
      // The copied register is live out and has a second use in another block.
      LLVM_DEBUG(dbgs() << "Exec copy source register is live out\n");
      continue;
    }

    unsigned CopyFromExec = CopyFromExecInst->getOperand(0).getReg();
    MachineInstr *SaveExecInst = nullptr;
    SmallVector<MachineInstr *, 4> OtherUseInsts;

    for (MachineBasicBlock::iterator J
           = std::next(CopyFromExecInst->getIterator()), JE = I->getIterator();
         J != JE; ++J) {
      if (SaveExecInst && J->readsRegister(Exec, TRI)) {
        LLVM_DEBUG(dbgs() << "exec read prevents saveexec: " << *J << '\n');
        // Make sure this is inserted after any VALU ops that may have been
        // scheduled in between.
        SaveExecInst = nullptr;
        break;
      }

      bool ReadsCopyFromExec = J->readsRegister(CopyFromExec, TRI);

      if (J->modifiesRegister(CopyToExec, TRI)) {
        if (SaveExecInst) {
          LLVM_DEBUG(dbgs() << "Multiple instructions modify "
                            << printReg(CopyToExec, TRI) << '\n');
          SaveExecInst = nullptr;
          break;
        }

        unsigned SaveExecOp = getSaveExecOp(J->getOpcode());
        if (SaveExecOp == AMDGPU::INSTRUCTION_LIST_END)
          break;

        if (ReadsCopyFromExec) {
          SaveExecInst = &*J;
          LLVM_DEBUG(dbgs() << "Found save exec op: " << *SaveExecInst << '\n');
          continue;
        } else {
          LLVM_DEBUG(dbgs()
                     << "Instruction does not read exec copy: " << *J << '\n');
          break;
        }
      } else if (ReadsCopyFromExec && !SaveExecInst) {
        // Make sure no other instruction is trying to use this copy, before it
        // will be rewritten by the saveexec, i.e. hasOneUse. There may have
        // been another use, such as an inserted spill. For example:
        //
        // %sgpr0_sgpr1 = COPY %exec
        // spill %sgpr0_sgpr1
        // %sgpr2_sgpr3 = S_AND_B64 %sgpr0_sgpr1
        //
        LLVM_DEBUG(dbgs() << "Found second use of save inst candidate: " << *J
                          << '\n');
        break;
      }

      if (SaveExecInst && J->readsRegister(CopyToExec, TRI)) {
        assert(SaveExecInst != &*J);
        OtherUseInsts.push_back(&*J);
      }
    }

    if (!SaveExecInst)
      continue;

    LLVM_DEBUG(dbgs() << "Insert save exec op: " << *SaveExecInst << '\n');

    MachineOperand &Src0 = SaveExecInst->getOperand(1);
    MachineOperand &Src1 = SaveExecInst->getOperand(2);

    MachineOperand *OtherOp = nullptr;

    if (Src0.isReg() && Src0.getReg() == CopyFromExec) {
      OtherOp = &Src1;
    } else if (Src1.isReg() && Src1.getReg() == CopyFromExec) {
      if (!SaveExecInst->isCommutable())
        break;

      OtherOp = &Src0;
    } else
      llvm_unreachable("unexpected");

    CopyFromExecInst->eraseFromParent();

    auto InsPt = SaveExecInst->getIterator();
    const DebugLoc &DL = SaveExecInst->getDebugLoc();

    BuildMI(MBB, InsPt, DL, TII->get(getSaveExecOp(SaveExecInst->getOpcode())),
            CopyFromExec)
      .addReg(OtherOp->getReg());
    SaveExecInst->eraseFromParent();

    CopyToExecInst->eraseFromParent();

    for (MachineInstr *OtherInst : OtherUseInsts) {
      OtherInst->substituteRegister(CopyToExec, Exec,
                                    AMDGPU::NoSubRegister, *TRI);
    }
  }

  return true;

}