1 //===----- ARMCodeGenPrepare.cpp ------------------------------------------===//
2 //
3 // The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 /// \file
11 /// This pass inserts intrinsics to handle small types that would otherwise be
12 /// promoted during legalization. Here we can manually promote types or insert
13 /// intrinsics which can handle narrow types that aren't supported by the
14 /// register classes.
15 //
16 //===----------------------------------------------------------------------===//
17
18 #include "ARM.h"
19 #include "ARMSubtarget.h"
20 #include "ARMTargetMachine.h"
21 #include "llvm/ADT/StringRef.h"
22 #include "llvm/CodeGen/Passes.h"
23 #include "llvm/CodeGen/TargetPassConfig.h"
24 #include "llvm/IR/Attributes.h"
25 #include "llvm/IR/BasicBlock.h"
26 #include "llvm/IR/IRBuilder.h"
27 #include "llvm/IR/Constants.h"
28 #include "llvm/IR/InstrTypes.h"
29 #include "llvm/IR/Instruction.h"
30 #include "llvm/IR/Instructions.h"
31 #include "llvm/IR/IntrinsicInst.h"
32 #include "llvm/IR/Intrinsics.h"
33 #include "llvm/IR/Type.h"
34 #include "llvm/IR/Value.h"
35 #include "llvm/IR/Verifier.h"
36 #include "llvm/Pass.h"
37 #include "llvm/Support/Casting.h"
38 #include "llvm/Support/CommandLine.h"
39
40 #define DEBUG_TYPE "arm-codegenprepare"
41
42 using namespace llvm;
43
44 static cl::opt<bool>
45 DisableCGP("arm-disable-cgp", cl::Hidden, cl::init(true),
46 cl::desc("Disable ARM specific CodeGenPrepare pass"));
47
48 static cl::opt<bool>
49 EnableDSP("arm-enable-scalar-dsp", cl::Hidden, cl::init(false),
50 cl::desc("Use DSP instructions for scalar operations"));
51
52 static cl::opt<bool>
53 EnableDSPWithImms("arm-enable-scalar-dsp-imms", cl::Hidden, cl::init(false),
54 cl::desc("Use DSP instructions for scalar operations\
55 with immediate operands"));
56
57 namespace {
58
59 class IRPromoter {
60 SmallPtrSet<Value*, 8> NewInsts;
61 SmallVector<Instruction*, 4> InstsToRemove;
62 Module *M = nullptr;
63 LLVMContext &Ctx;
64
65 public:
IRPromoter(Module * M)66 IRPromoter(Module *M) : M(M), Ctx(M->getContext()) { }
67
Cleanup()68 void Cleanup() {
69 for (auto *I : InstsToRemove) {
70 LLVM_DEBUG(dbgs() << "ARM CGP: Removing " << *I << "\n");
71 I->dropAllReferences();
72 I->eraseFromParent();
73 }
74 InstsToRemove.clear();
75 NewInsts.clear();
76 }
77
78 void Mutate(Type *OrigTy,
79 SmallPtrSetImpl<Value*> &Visited,
80 SmallPtrSetImpl<Value*> &Leaves,
81 SmallPtrSetImpl<Instruction*> &Roots);
82 };
83
84 class ARMCodeGenPrepare : public FunctionPass {
85 const ARMSubtarget *ST = nullptr;
86 IRPromoter *Promoter = nullptr;
87 std::set<Value*> AllVisited;
88 Type *OrigTy = nullptr;
89 unsigned TypeSize = 0;
90
91 bool isNarrowInstSupported(Instruction *I);
92 bool isSupportedValue(Value *V);
93 bool isLegalToPromote(Value *V);
94 bool TryToPromote(Value *V);
95
96 public:
97 static char ID;
98
ARMCodeGenPrepare()99 ARMCodeGenPrepare() : FunctionPass(ID) {}
100
getAnalysisUsage(AnalysisUsage & AU) const101 void getAnalysisUsage(AnalysisUsage &AU) const override {
102 AU.addRequired<TargetPassConfig>();
103 }
104
getPassName() const105 StringRef getPassName() const override { return "ARM IR optimizations"; }
106
107 bool doInitialization(Module &M) override;
108 bool runOnFunction(Function &F) override;
109 bool doFinalization(Module &M) override;
110 };
111
112 }
113
114 /// Can the given value generate sign bits.
isSigned(Value * V)115 static bool isSigned(Value *V) {
116 if (!isa<Instruction>(V))
117 return false;
118
119 unsigned Opc = cast<Instruction>(V)->getOpcode();
120 return Opc == Instruction::AShr || Opc == Instruction::SDiv ||
121 Opc == Instruction::SRem;
122 }
123
124 /// Some instructions can use 8- and 16-bit operands, and we don't need to
125 /// promote anything larger. We disallow booleans to make life easier when
126 /// dealing with icmps but allow any other integer that is <= 16 bits. Void
127 /// types are accepted so we can handle switches.
isSupportedType(Value * V)128 static bool isSupportedType(Value *V) {
129 if (V->getType()->isVoidTy())
130 return true;
131
132 const IntegerType *IntTy = dyn_cast<IntegerType>(V->getType());
133 if (!IntTy)
134 return false;
135
136 // Don't try to promote boolean values.
137 if (IntTy->getBitWidth() == 1)
138 return false;
139
140 if (auto *ZExt = dyn_cast<ZExtInst>(V))
141 return isSupportedType(ZExt->getOperand(0));
142
143 return IntTy->getBitWidth() <= 16;
144 }
145
146 /// Return true if V will require any promoted values to be truncated for the
147 /// use to be valid.
isSink(Value * V)148 static bool isSink(Value *V) {
149 auto UsesNarrowValue = [](Value *V) {
150 return V->getType()->getScalarSizeInBits() <= 32;
151 };
152
153 if (auto *Store = dyn_cast<StoreInst>(V))
154 return UsesNarrowValue(Store->getValueOperand());
155 if (auto *Return = dyn_cast<ReturnInst>(V))
156 return UsesNarrowValue(Return->getReturnValue());
157
158 return isa<CallInst>(V);
159 }
160
161 /// Return true if the given value is a leaf that will need to be zext'd.
isSource(Value * V)162 static bool isSource(Value *V) {
163 if (isa<Argument>(V) && isSupportedType(V))
164 return true;
165 else if (isa<TruncInst>(V))
166 return true;
167 else if (auto *ZExt = dyn_cast<ZExtInst>(V))
168 // ZExt can be a leaf if its the only user of a load.
169 return isa<LoadInst>(ZExt->getOperand(0)) &&
170 ZExt->getOperand(0)->hasOneUse();
171 else if (auto *Call = dyn_cast<CallInst>(V))
172 return Call->hasRetAttr(Attribute::AttrKind::ZExt);
173 else if (auto *Load = dyn_cast<LoadInst>(V)) {
174 if (!isa<IntegerType>(Load->getType()))
175 return false;
176 // A load is a leaf, unless its already just being zext'd.
177 if (Load->hasOneUse() && isa<ZExtInst>(*Load->use_begin()))
178 return false;
179
180 return true;
181 }
182 return false;
183 }
184
185 /// Return whether the instruction can be promoted within any modifications to
186 /// it's operands or result.
isSafeOverflow(Instruction * I)187 static bool isSafeOverflow(Instruction *I) {
188 if (isa<OverflowingBinaryOperator>(I) && I->hasNoUnsignedWrap())
189 return true;
190
191 unsigned Opc = I->getOpcode();
192 if (Opc == Instruction::Add || Opc == Instruction::Sub) {
193 // We don't care if the add or sub could wrap if the value is decreasing
194 // and is only being used by an unsigned compare.
195 if (!I->hasOneUse() ||
196 !isa<ICmpInst>(*I->user_begin()) ||
197 !isa<ConstantInt>(I->getOperand(1)))
198 return false;
199
200 auto *CI = cast<ICmpInst>(*I->user_begin());
201 if (CI->isSigned())
202 return false;
203
204 bool NegImm = cast<ConstantInt>(I->getOperand(1))->isNegative();
205 bool IsDecreasing = ((Opc == Instruction::Sub) && !NegImm) ||
206 ((Opc == Instruction::Add) && NegImm);
207 if (!IsDecreasing)
208 return false;
209
210 LLVM_DEBUG(dbgs() << "ARM CGP: Allowing safe overflow for " << *I << "\n");
211 return true;
212 }
213
214 // Otherwise, if an instruction is using a negative immediate we will need
215 // to fix it up during the promotion.
216 for (auto &Op : I->operands()) {
217 if (auto *Const = dyn_cast<ConstantInt>(Op))
218 if (Const->isNegative())
219 return false;
220 }
221 return false;
222 }
223
shouldPromote(Value * V)224 static bool shouldPromote(Value *V) {
225 auto *I = dyn_cast<Instruction>(V);
226 if (!I)
227 return false;
228
229 if (!isa<IntegerType>(V->getType()))
230 return false;
231
232 if (isa<StoreInst>(I) || isa<TerminatorInst>(I) || isa<TruncInst>(I) ||
233 isa<ICmpInst>(I))
234 return false;
235
236 if (auto *ZExt = dyn_cast<ZExtInst>(I))
237 return !ZExt->getDestTy()->isIntegerTy(32);
238
239 return true;
240 }
241
242 /// Return whether we can safely mutate V's type to ExtTy without having to be
243 /// concerned with zero extending or truncation.
isPromotedResultSafe(Value * V)244 static bool isPromotedResultSafe(Value *V) {
245 if (!isa<Instruction>(V))
246 return true;
247
248 if (isSigned(V))
249 return false;
250
251 // If I is only being used by something that will require its value to be
252 // truncated, then we don't care about the promoted result.
253 auto *I = cast<Instruction>(V);
254 if (I->hasOneUse() && isSink(*I->use_begin()))
255 return true;
256
257 if (isa<OverflowingBinaryOperator>(I))
258 return isSafeOverflow(I);
259 return true;
260 }
261
262 /// Return the intrinsic for the instruction that can perform the same
263 /// operation but on a narrow type. This is using the parallel dsp intrinsics
264 /// on scalar values.
getNarrowIntrinsic(Instruction * I,unsigned TypeSize)265 static Intrinsic::ID getNarrowIntrinsic(Instruction *I, unsigned TypeSize) {
266 // Whether we use the signed or unsigned versions of these intrinsics
267 // doesn't matter because we're not using the GE bits that they set in
268 // the APSR.
269 switch(I->getOpcode()) {
270 default:
271 break;
272 case Instruction::Add:
273 return TypeSize == 16 ? Intrinsic::arm_uadd16 :
274 Intrinsic::arm_uadd8;
275 case Instruction::Sub:
276 return TypeSize == 16 ? Intrinsic::arm_usub16 :
277 Intrinsic::arm_usub8;
278 }
279 llvm_unreachable("unhandled opcode for narrow intrinsic");
280 }
281
Mutate(Type * OrigTy,SmallPtrSetImpl<Value * > & Visited,SmallPtrSetImpl<Value * > & Leaves,SmallPtrSetImpl<Instruction * > & Roots)282 void IRPromoter::Mutate(Type *OrigTy,
283 SmallPtrSetImpl<Value*> &Visited,
284 SmallPtrSetImpl<Value*> &Leaves,
285 SmallPtrSetImpl<Instruction*> &Roots) {
286 IRBuilder<> Builder{Ctx};
287 Type *ExtTy = Type::getInt32Ty(M->getContext());
288 unsigned TypeSize = OrigTy->getPrimitiveSizeInBits();
289 SmallPtrSet<Value*, 8> Promoted;
290 LLVM_DEBUG(dbgs() << "ARM CGP: Promoting use-def chains to from " << TypeSize
291 << " to 32-bits\n");
292
293 auto ReplaceAllUsersOfWith = [&](Value *From, Value *To) {
294 SmallVector<Instruction*, 4> Users;
295 Instruction *InstTo = dyn_cast<Instruction>(To);
296 for (Use &U : From->uses()) {
297 auto *User = cast<Instruction>(U.getUser());
298 if (InstTo && User->isIdenticalTo(InstTo))
299 continue;
300 Users.push_back(User);
301 }
302
303 for (auto &U : Users)
304 U->replaceUsesOfWith(From, To);
305 };
306
307 auto FixConst = [&](ConstantInt *Const, Instruction *I) {
308 Constant *NewConst = nullptr;
309 if (isSafeOverflow(I)) {
310 NewConst = (Const->isNegative()) ?
311 ConstantExpr::getSExt(Const, ExtTy) :
312 ConstantExpr::getZExt(Const, ExtTy);
313 } else {
314 uint64_t NewVal = *Const->getValue().getRawData();
315 if (Const->getType() == Type::getInt16Ty(Ctx))
316 NewVal &= 0xFFFF;
317 else
318 NewVal &= 0xFF;
319 NewConst = ConstantInt::get(ExtTy, NewVal);
320 }
321 I->replaceUsesOfWith(Const, NewConst);
322 };
323
324 auto InsertDSPIntrinsic = [&](Instruction *I) {
325 LLVM_DEBUG(dbgs() << "ARM CGP: Inserting DSP intrinsic for "
326 << *I << "\n");
327 Function *DSPInst =
328 Intrinsic::getDeclaration(M, getNarrowIntrinsic(I, TypeSize));
329 Builder.SetInsertPoint(I);
330 Builder.SetCurrentDebugLocation(I->getDebugLoc());
331 Value *Args[] = { I->getOperand(0), I->getOperand(1) };
332 CallInst *Call = Builder.CreateCall(DSPInst, Args);
333 ReplaceAllUsersOfWith(I, Call);
334 InstsToRemove.push_back(I);
335 NewInsts.insert(Call);
336 };
337
338 auto InsertZExt = [&](Value *V, Instruction *InsertPt) {
339 LLVM_DEBUG(dbgs() << "ARM CGP: Inserting ZExt for " << *V << "\n");
340 Builder.SetInsertPoint(InsertPt);
341 if (auto *I = dyn_cast<Instruction>(V))
342 Builder.SetCurrentDebugLocation(I->getDebugLoc());
343 auto *ZExt = cast<Instruction>(Builder.CreateZExt(V, ExtTy));
344 if (isa<Argument>(V))
345 ZExt->moveBefore(InsertPt);
346 else
347 ZExt->moveAfter(InsertPt);
348 ReplaceAllUsersOfWith(V, ZExt);
349 NewInsts.insert(ZExt);
350 };
351
352 // First, insert extending instructions between the leaves and their users.
353 LLVM_DEBUG(dbgs() << "ARM CGP: Promoting leaves:\n");
354 for (auto V : Leaves) {
355 LLVM_DEBUG(dbgs() << " - " << *V << "\n");
356 if (auto *ZExt = dyn_cast<ZExtInst>(V))
357 ZExt->mutateType(ExtTy);
358 else if (auto *I = dyn_cast<Instruction>(V))
359 InsertZExt(I, I);
360 else if (auto *Arg = dyn_cast<Argument>(V)) {
361 BasicBlock &BB = Arg->getParent()->front();
362 InsertZExt(Arg, &*BB.getFirstInsertionPt());
363 } else {
364 llvm_unreachable("unhandled leaf that needs extending");
365 }
366 Promoted.insert(V);
367 }
368
369 LLVM_DEBUG(dbgs() << "ARM CGP: Mutating the tree..\n");
370 // Then mutate the types of the instructions within the tree. Here we handle
371 // constant operands.
372 for (auto *V : Visited) {
373 if (Leaves.count(V))
374 continue;
375
376 if (!isa<Instruction>(V))
377 continue;
378
379 auto *I = cast<Instruction>(V);
380 if (Roots.count(I))
381 continue;
382
383 for (auto &U : I->operands()) {
384 if ((U->getType() == ExtTy) || !isSupportedType(&*U))
385 continue;
386
387 if (auto *Const = dyn_cast<ConstantInt>(&*U))
388 FixConst(Const, I);
389 else if (isa<UndefValue>(&*U))
390 U->mutateType(ExtTy);
391 }
392
393 if (shouldPromote(I)) {
394 I->mutateType(ExtTy);
395 Promoted.insert(I);
396 }
397 }
398
399 // Now we need to remove any zexts that have become unnecessary, as well
400 // as insert any intrinsics.
401 for (auto *V : Visited) {
402 if (Leaves.count(V))
403 continue;
404 if (auto *ZExt = dyn_cast<ZExtInst>(V)) {
405 if (ZExt->getDestTy() != ExtTy) {
406 ZExt->mutateType(ExtTy);
407 Promoted.insert(ZExt);
408 }
409 else if (ZExt->getSrcTy() == ExtTy) {
410 ReplaceAllUsersOfWith(V, ZExt->getOperand(0));
411 InstsToRemove.push_back(ZExt);
412 }
413 continue;
414 }
415
416 if (!shouldPromote(V) || isPromotedResultSafe(V))
417 continue;
418
419 // Replace unsafe instructions with appropriate intrinsic calls.
420 InsertDSPIntrinsic(cast<Instruction>(V));
421 }
422
423 LLVM_DEBUG(dbgs() << "ARM CGP: Fixing up the roots:\n");
424 // Fix up any stores or returns that use the results of the promoted
425 // chain.
426 for (auto I : Roots) {
427 LLVM_DEBUG(dbgs() << " - " << *I << "\n");
428 Type *TruncTy = OrigTy;
429 if (auto *Store = dyn_cast<StoreInst>(I)) {
430 auto *PtrTy = cast<PointerType>(Store->getPointerOperandType());
431 TruncTy = PtrTy->getElementType();
432 } else if (isa<ReturnInst>(I)) {
433 Function *F = I->getParent()->getParent();
434 TruncTy = F->getFunctionType()->getReturnType();
435 }
436
437 for (unsigned i = 0; i < I->getNumOperands(); ++i) {
438 Value *V = I->getOperand(i);
439 if (Promoted.count(V) || NewInsts.count(V)) {
440 if (auto *Op = dyn_cast<Instruction>(V)) {
441
442 if (auto *Call = dyn_cast<CallInst>(I))
443 TruncTy = Call->getFunctionType()->getParamType(i);
444
445 if (TruncTy == ExtTy)
446 continue;
447
448 LLVM_DEBUG(dbgs() << "ARM CGP: Creating " << *TruncTy
449 << " Trunc for " << *Op << "\n");
450 Builder.SetInsertPoint(Op);
451 auto *Trunc = cast<Instruction>(Builder.CreateTrunc(Op, TruncTy));
452 Trunc->moveBefore(I);
453 I->setOperand(i, Trunc);
454 NewInsts.insert(Trunc);
455 }
456 }
457 }
458 }
459 LLVM_DEBUG(dbgs() << "ARM CGP: Mutation complete.\n");
460 }
461
isNarrowInstSupported(Instruction * I)462 bool ARMCodeGenPrepare::isNarrowInstSupported(Instruction *I) {
463 if (!ST->hasDSP() || !EnableDSP || !isSupportedType(I))
464 return false;
465
466 if (ST->isThumb() && !ST->hasThumb2())
467 return false;
468
469 if (I->getOpcode() != Instruction::Add && I->getOpcode() != Instruction::Sub)
470 return false;
471
472 // TODO
473 // Would it be profitable? For Thumb code, these parallel DSP instructions
474 // are only Thumb-2, so we wouldn't be able to dual issue on Cortex-M33. For
475 // Cortex-A, specifically Cortex-A72, the latency is double and throughput is
476 // halved. They also do not take immediates as operands.
477 for (auto &Op : I->operands()) {
478 if (isa<Constant>(Op)) {
479 if (!EnableDSPWithImms)
480 return false;
481 }
482 }
483 return true;
484 }
485
486 /// We accept most instructions, as well as Arguments and ConstantInsts. We
487 /// Disallow casts other than zext and truncs and only allow calls if their
488 /// return value is zeroext. We don't allow opcodes that can introduce sign
489 /// bits.
isSupportedValue(Value * V)490 bool ARMCodeGenPrepare::isSupportedValue(Value *V) {
491 LLVM_DEBUG(dbgs() << "ARM CGP: Is " << *V << " supported?\n");
492
493 // Non-instruction values that we can handle.
494 if (isa<ConstantInt>(V) || isa<Argument>(V))
495 return true;
496
497 // Memory instructions
498 if (isa<StoreInst>(V) || isa<LoadInst>(V) || isa<GetElementPtrInst>(V))
499 return true;
500
501 // Branches and targets.
502 if (auto *ICmp = dyn_cast<ICmpInst>(V))
503 return ICmp->isEquality() || !ICmp->isSigned();
504
505 if( isa<BranchInst>(V) || isa<SwitchInst>(V) || isa<BasicBlock>(V))
506 return true;
507
508 if (isa<PHINode>(V) || isa<SelectInst>(V) || isa<ReturnInst>(V))
509 return true;
510
511 // Special cases for calls as we need to check for zeroext
512 // TODO We should accept calls even if they don't have zeroext, as they can
513 // still be roots.
514 if (auto *Call = dyn_cast<CallInst>(V))
515 return Call->hasRetAttr(Attribute::AttrKind::ZExt);
516 else if (auto *Cast = dyn_cast<CastInst>(V)) {
517 if (isa<ZExtInst>(Cast))
518 return Cast->getDestTy()->getScalarSizeInBits() <= 32;
519 else if (auto *Trunc = dyn_cast<TruncInst>(V))
520 return Trunc->getDestTy()->getScalarSizeInBits() <= TypeSize;
521 else {
522 LLVM_DEBUG(dbgs() << "ARM CGP: No, unsupported cast.\n");
523 return false;
524 }
525 } else if (!isa<BinaryOperator>(V)) {
526 LLVM_DEBUG(dbgs() << "ARM CGP: No, not a binary operator.\n");
527 return false;
528 }
529
530 bool res = !isSigned(V);
531 if (!res)
532 LLVM_DEBUG(dbgs() << "ARM CGP: No, it's a signed instruction.\n");
533 return res;
534 }
535
536 /// Check that the type of V would be promoted and that the original type is
537 /// smaller than the targeted promoted type. Check that we're not trying to
538 /// promote something larger than our base 'TypeSize' type.
isLegalToPromote(Value * V)539 bool ARMCodeGenPrepare::isLegalToPromote(Value *V) {
540 if (!isSupportedType(V))
541 return false;
542
543 unsigned VSize = 0;
544 if (auto *Ld = dyn_cast<LoadInst>(V)) {
545 auto *PtrTy = cast<PointerType>(Ld->getPointerOperandType());
546 VSize = PtrTy->getElementType()->getPrimitiveSizeInBits();
547 } else if (auto *ZExt = dyn_cast<ZExtInst>(V)) {
548 VSize = ZExt->getOperand(0)->getType()->getPrimitiveSizeInBits();
549 } else {
550 VSize = V->getType()->getPrimitiveSizeInBits();
551 }
552
553 if (VSize > TypeSize)
554 return false;
555
556 if (isPromotedResultSafe(V))
557 return true;
558
559 if (auto *I = dyn_cast<Instruction>(V))
560 return isNarrowInstSupported(I);
561
562 return false;
563 }
564
TryToPromote(Value * V)565 bool ARMCodeGenPrepare::TryToPromote(Value *V) {
566 OrigTy = V->getType();
567 TypeSize = OrigTy->getPrimitiveSizeInBits();
568
569 if (!isSupportedValue(V) || !shouldPromote(V) || !isLegalToPromote(V))
570 return false;
571
572 LLVM_DEBUG(dbgs() << "ARM CGP: TryToPromote: " << *V << "\n");
573
574 SetVector<Value*> WorkList;
575 SmallPtrSet<Value*, 8> Leaves;
576 SmallPtrSet<Instruction*, 4> Roots;
577 WorkList.insert(V);
578 SmallPtrSet<Value*, 16> CurrentVisited;
579 CurrentVisited.clear();
580
581 // Return true if the given value can, or has been, visited. Add V to the
582 // worklist if needed.
583 auto AddLegalInst = [&](Value *V) {
584 if (CurrentVisited.count(V))
585 return true;
586
587 if (!isSupportedValue(V) || (shouldPromote(V) && !isLegalToPromote(V))) {
588 LLVM_DEBUG(dbgs() << "ARM CGP: Can't handle: " << *V << "\n");
589 return false;
590 }
591
592 WorkList.insert(V);
593 return true;
594 };
595
596 // Iterate through, and add to, a tree of operands and users in the use-def.
597 while (!WorkList.empty()) {
598 Value *V = WorkList.back();
599 WorkList.pop_back();
600 if (CurrentVisited.count(V))
601 continue;
602
603 if (!isa<Instruction>(V) && !isSource(V))
604 continue;
605
606 // If we've already visited this value from somewhere, bail now because
607 // the tree has already been explored.
608 // TODO: This could limit the transform, ie if we try to promote something
609 // from an i8 and fail first, before trying an i16.
610 if (AllVisited.count(V)) {
611 LLVM_DEBUG(dbgs() << "ARM CGP: Already visited this: " << *V << "\n");
612 return false;
613 }
614
615 CurrentVisited.insert(V);
616 AllVisited.insert(V);
617
618 // Calls can be both sources and sinks.
619 if (isSink(V))
620 Roots.insert(cast<Instruction>(V));
621 if (isSource(V))
622 Leaves.insert(V);
623 else if (auto *I = dyn_cast<Instruction>(V)) {
624 // Visit operands of any instruction visited.
625 for (auto &U : I->operands()) {
626 if (!AddLegalInst(U))
627 return false;
628 }
629 }
630
631 // Don't visit users of a node which isn't going to be mutated unless its a
632 // source.
633 if (isSource(V) || shouldPromote(V)) {
634 for (Use &U : V->uses()) {
635 if (!AddLegalInst(U.getUser()))
636 return false;
637 }
638 }
639 }
640
641 unsigned NumToPromote = 0;
642 unsigned Cost = 0;
643 for (auto *V : CurrentVisited) {
644 // Truncs will cause a uxt and no zeroext arguments will often require
645 // a uxt somewhere.
646 if (isa<TruncInst>(V))
647 ++Cost;
648 else if (auto *Arg = dyn_cast<Argument>(V)) {
649 if (!Arg->hasZExtAttr())
650 ++Cost;
651 }
652
653 // Mem ops can automatically be extended/truncated and non-instructions
654 // don't need anything done.
655 if (Leaves.count(V) || isa<StoreInst>(V) || !isa<Instruction>(V))
656 continue;
657
658 // Will need to truncate calls args and returns.
659 if (Roots.count(cast<Instruction>(V))) {
660 ++Cost;
661 continue;
662 }
663
664 if (shouldPromote(V))
665 ++NumToPromote;
666 }
667
668 LLVM_DEBUG(dbgs() << "ARM CGP: Visited nodes:\n";
669 for (auto *I : CurrentVisited)
670 I->dump();
671 );
672 LLVM_DEBUG(dbgs() << "ARM CGP: Cost of promoting " << NumToPromote
673 << " instructions = " << Cost << "\n");
674 if (Cost > NumToPromote || (NumToPromote == 0))
675 return false;
676
677 Promoter->Mutate(OrigTy, CurrentVisited, Leaves, Roots);
678 return true;
679 }
680
doInitialization(Module & M)681 bool ARMCodeGenPrepare::doInitialization(Module &M) {
682 Promoter = new IRPromoter(&M);
683 return false;
684 }
685
runOnFunction(Function & F)686 bool ARMCodeGenPrepare::runOnFunction(Function &F) {
687 if (skipFunction(F) || DisableCGP)
688 return false;
689
690 auto *TPC = &getAnalysis<TargetPassConfig>();
691 if (!TPC)
692 return false;
693
694 const TargetMachine &TM = TPC->getTM<TargetMachine>();
695 ST = &TM.getSubtarget<ARMSubtarget>(F);
696 bool MadeChange = false;
697 LLVM_DEBUG(dbgs() << "ARM CGP: Running on " << F.getName() << "\n");
698
699 // Search up from icmps to try to promote their operands.
700 for (BasicBlock &BB : F) {
701 auto &Insts = BB.getInstList();
702 for (auto &I : Insts) {
703 if (AllVisited.count(&I))
704 continue;
705
706 if (isa<ICmpInst>(I)) {
707 auto &CI = cast<ICmpInst>(I);
708
709 // Skip signed or pointer compares
710 if (CI.isSigned() || !isa<IntegerType>(CI.getOperand(0)->getType()))
711 continue;
712
713 LLVM_DEBUG(dbgs() << "ARM CGP: Searching from: " << CI << "\n");
714 for (auto &Op : CI.operands()) {
715 if (auto *I = dyn_cast<Instruction>(Op)) {
716 if (isa<ZExtInst>(I))
717 MadeChange |= TryToPromote(I->getOperand(0));
718 else
719 MadeChange |= TryToPromote(I);
720 }
721 }
722 }
723 }
724 Promoter->Cleanup();
725 LLVM_DEBUG(if (verifyFunction(F, &dbgs())) {
726 dbgs();
727 report_fatal_error("Broken function after type promotion");
728 });
729 }
730 if (MadeChange)
731 LLVM_DEBUG(dbgs() << "After ARMCodeGenPrepare: " << F << "\n");
732
733 return MadeChange;
734 }
735
doFinalization(Module & M)736 bool ARMCodeGenPrepare::doFinalization(Module &M) {
737 delete Promoter;
738 return false;
739 }
740
741 INITIALIZE_PASS_BEGIN(ARMCodeGenPrepare, DEBUG_TYPE,
742 "ARM IR optimizations", false, false)
743 INITIALIZE_PASS_END(ARMCodeGenPrepare, DEBUG_TYPE, "ARM IR optimizations",
744 false, false)
745
746 char ARMCodeGenPrepare::ID = 0;
747
createARMCodeGenPreparePass()748 FunctionPass *llvm::createARMCodeGenPreparePass() {
749 return new ARMCodeGenPrepare();
750 }
751