| 1 | //===- LowerSwitch.cpp - Eliminate Switch instructions --------------------===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | // |
| 9 | // The LowerSwitch transformation rewrites switch instructions with a sequence |
| 10 | // of branches, which allows targets to get away with not implementing the |
| 11 | // switch instruction until it is convenient. |
| 12 | // |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | #include "llvm/Transforms/Utils/LowerSwitch.h" |
| 16 | #include "llvm/ADT/DenseMap.h" |
| 17 | #include "llvm/ADT/STLExtras.h" |
| 18 | #include "llvm/ADT/SmallPtrSet.h" |
| 19 | #include "llvm/ADT/SmallVector.h" |
| 20 | #include "llvm/Analysis/AssumptionCache.h" |
| 21 | #include "llvm/Analysis/LazyValueInfo.h" |
| 22 | #include "llvm/Analysis/ValueTracking.h" |
| 23 | #include "llvm/IR/BasicBlock.h" |
| 24 | #include "llvm/IR/CFG.h" |
| 25 | #include "llvm/IR/ConstantRange.h" |
| 26 | #include "llvm/IR/Constants.h" |
| 27 | #include "llvm/IR/Function.h" |
| 28 | #include "llvm/IR/InstrTypes.h" |
| 29 | #include "llvm/IR/Instructions.h" |
| 30 | #include "llvm/IR/PassManager.h" |
| 31 | #include "llvm/IR/Value.h" |
| 32 | #include "llvm/InitializePasses.h" |
| 33 | #include "llvm/Pass.h" |
| 34 | #include "llvm/Support/Casting.h" |
| 35 | #include "llvm/Support/Compiler.h" |
| 36 | #include "llvm/Support/Debug.h" |
| 37 | #include "llvm/Support/KnownBits.h" |
| 38 | #include "llvm/Support/raw_ostream.h" |
| 39 | #include "llvm/Transforms/Utils.h" |
| 40 | #include "llvm/Transforms/Utils/BasicBlockUtils.h" |
| 41 | #include <cassert> |
| 42 | #include <iterator> |
| 43 | #include <vector> |
| 44 | |
| 45 | using namespace llvm; |
| 46 | |
| 47 | #define DEBUG_TYPE "lower-switch" |
| 48 | |
| 49 | namespace { |
| 50 | |
| 51 | struct IntRange { |
| 52 | APInt Low, High; |
| 53 | }; |
| 54 | |
| 55 | } // end anonymous namespace |
| 56 | |
| 57 | namespace { |
| 58 | // Return true iff R is covered by Ranges. |
| 59 | bool IsInRanges(const IntRange &R, const std::vector<IntRange> &Ranges) { |
| 60 | // Note: Ranges must be sorted, non-overlapping and non-adjacent. |
| 61 | |
| 62 | // Find the first range whose High field is >= R.High, |
| 63 | // then check if the Low field is <= R.Low. If so, we |
| 64 | // have a Range that covers R. |
| 65 | auto I = llvm::lower_bound( |
| 66 | Range: Ranges, Value: R, C: [](IntRange A, IntRange B) { return A.High.slt(RHS: B.High); }); |
| 67 | return I != Ranges.end() && I->Low.sle(RHS: R.Low); |
| 68 | } |
| 69 | |
| 70 | struct CaseRange { |
| 71 | ConstantInt *Low; |
| 72 | ConstantInt *High; |
| 73 | BasicBlock *BB; |
| 74 | |
| 75 | CaseRange(ConstantInt *low, ConstantInt *high, BasicBlock *bb) |
| 76 | : Low(low), High(high), BB(bb) {} |
| 77 | }; |
| 78 | |
| 79 | using CaseVector = std::vector<CaseRange>; |
| 80 | using CaseItr = std::vector<CaseRange>::iterator; |
| 81 | |
| 82 | /// The comparison function for sorting the switch case values in the vector. |
| 83 | /// WARNING: Case ranges should be disjoint! |
| 84 | struct CaseCmp { |
| 85 | bool operator()(const CaseRange &C1, const CaseRange &C2) { |
| 86 | const ConstantInt *CI1 = cast<const ConstantInt>(Val: C1.Low); |
| 87 | const ConstantInt *CI2 = cast<const ConstantInt>(Val: C2.High); |
| 88 | return CI1->getValue().slt(RHS: CI2->getValue()); |
| 89 | } |
| 90 | }; |
| 91 | |
| 92 | /// Used for debugging purposes. |
| 93 | LLVM_ATTRIBUTE_USED |
| 94 | raw_ostream &operator<<(raw_ostream &O, const CaseVector &C) { |
| 95 | O << "[" ; |
| 96 | |
| 97 | for (CaseVector::const_iterator B = C.begin(), E = C.end(); B != E;) { |
| 98 | O << "[" << B->Low->getValue() << ", " << B->High->getValue() << "]" ; |
| 99 | if (++B != E) |
| 100 | O << ", " ; |
| 101 | } |
| 102 | |
| 103 | return O << "]" ; |
| 104 | } |
| 105 | |
| 106 | /// Update the first occurrence of the "switch statement" BB in the PHI |
| 107 | /// node with the "new" BB. The other occurrences will: |
| 108 | /// |
| 109 | /// 1) Be updated by subsequent calls to this function. Switch statements may |
| 110 | /// have more than one outcoming edge into the same BB if they all have the same |
| 111 | /// value. When the switch statement is converted these incoming edges are now |
| 112 | /// coming from multiple BBs. |
| 113 | /// 2) Removed if subsequent incoming values now share the same case, i.e., |
| 114 | /// multiple outcome edges are condensed into one. This is necessary to keep the |
| 115 | /// number of phi values equal to the number of branches to SuccBB. |
| 116 | void FixPhis(BasicBlock *SuccBB, BasicBlock *OrigBB, BasicBlock *NewBB, |
| 117 | const APInt &NumMergedCases) { |
| 118 | for (auto &I : SuccBB->phis()) { |
| 119 | PHINode *PN = cast<PHINode>(Val: &I); |
| 120 | |
| 121 | // Only update the first occurrence if NewBB exists. |
| 122 | unsigned Idx = 0, E = PN->getNumIncomingValues(); |
| 123 | APInt LocalNumMergedCases = NumMergedCases; |
| 124 | for (; Idx != E && NewBB; ++Idx) { |
| 125 | if (PN->getIncomingBlock(i: Idx) == OrigBB) { |
| 126 | PN->setIncomingBlock(i: Idx, BB: NewBB); |
| 127 | break; |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | // Skip the updated incoming block so that it will not be removed. |
| 132 | if (NewBB) |
| 133 | ++Idx; |
| 134 | |
| 135 | // Remove additional occurrences coming from condensed cases and keep the |
| 136 | // number of incoming values equal to the number of branches to SuccBB. |
| 137 | SmallVector<unsigned, 8> Indices; |
| 138 | for (; LocalNumMergedCases.ugt(RHS: 0) && Idx < E; ++Idx) |
| 139 | if (PN->getIncomingBlock(i: Idx) == OrigBB) { |
| 140 | Indices.push_back(Elt: Idx); |
| 141 | LocalNumMergedCases -= 1; |
| 142 | } |
| 143 | // Remove incoming values in the reverse order to prevent invalidating |
| 144 | // *successive* index. |
| 145 | for (unsigned III : llvm::reverse(C&: Indices)) |
| 146 | PN->removeIncomingValue(Idx: III); |
| 147 | } |
| 148 | } |
| 149 | |
| 150 | /// Create a new leaf block for the binary lookup tree. It checks if the |
| 151 | /// switch's value == the case's value. If not, then it jumps to the default |
| 152 | /// branch. At this point in the tree, the value can't be another valid case |
| 153 | /// value, so the jump to the "default" branch is warranted. |
| 154 | BasicBlock *NewLeafBlock(CaseRange &Leaf, Value *Val, ConstantInt *LowerBound, |
| 155 | ConstantInt *UpperBound, BasicBlock *OrigBlock, |
| 156 | BasicBlock *Default) { |
| 157 | Function *F = OrigBlock->getParent(); |
| 158 | BasicBlock *NewLeaf = BasicBlock::Create(Context&: Val->getContext(), Name: "LeafBlock" ); |
| 159 | F->insert(Position: ++OrigBlock->getIterator(), BB: NewLeaf); |
| 160 | |
| 161 | // Emit comparison |
| 162 | ICmpInst *Comp = nullptr; |
| 163 | if (Leaf.Low == Leaf.High) { |
| 164 | // Make the seteq instruction... |
| 165 | Comp = |
| 166 | new ICmpInst(NewLeaf, ICmpInst::ICMP_EQ, Val, Leaf.Low, "SwitchLeaf" ); |
| 167 | } else { |
| 168 | // Make range comparison |
| 169 | if (Leaf.Low == LowerBound) { |
| 170 | // Val >= Min && Val <= Hi --> Val <= Hi |
| 171 | Comp = new ICmpInst(NewLeaf, ICmpInst::ICMP_SLE, Val, Leaf.High, |
| 172 | "SwitchLeaf" ); |
| 173 | } else if (Leaf.High == UpperBound) { |
| 174 | // Val <= Max && Val >= Lo --> Val >= Lo |
| 175 | Comp = new ICmpInst(NewLeaf, ICmpInst::ICMP_SGE, Val, Leaf.Low, |
| 176 | "SwitchLeaf" ); |
| 177 | } else if (Leaf.Low->isZero()) { |
| 178 | // Val >= 0 && Val <= Hi --> Val <=u Hi |
| 179 | Comp = new ICmpInst(NewLeaf, ICmpInst::ICMP_ULE, Val, Leaf.High, |
| 180 | "SwitchLeaf" ); |
| 181 | } else { |
| 182 | // Emit V-Lo <=u Hi-Lo |
| 183 | Constant *NegLo = ConstantExpr::getNeg(C: Leaf.Low); |
| 184 | Instruction *Add = BinaryOperator::CreateAdd( |
| 185 | V1: Val, V2: NegLo, Name: Val->getName() + ".off" , InsertBefore: NewLeaf); |
| 186 | Constant *UpperBound = ConstantExpr::getAdd(C1: NegLo, C2: Leaf.High); |
| 187 | Comp = new ICmpInst(NewLeaf, ICmpInst::ICMP_ULE, Add, UpperBound, |
| 188 | "SwitchLeaf" ); |
| 189 | } |
| 190 | } |
| 191 | |
| 192 | // Make the conditional branch... |
| 193 | BasicBlock *Succ = Leaf.BB; |
| 194 | BranchInst::Create(IfTrue: Succ, IfFalse: Default, Cond: Comp, InsertBefore: NewLeaf); |
| 195 | |
| 196 | // Update the PHI incoming value/block for the default. |
| 197 | for (auto &I : Default->phis()) { |
| 198 | PHINode *PN = cast<PHINode>(Val: &I); |
| 199 | auto *V = PN->getIncomingValueForBlock(BB: OrigBlock); |
| 200 | PN->addIncoming(V, BB: NewLeaf); |
| 201 | } |
| 202 | |
| 203 | // If there were any PHI nodes in this successor, rewrite one entry |
| 204 | // from OrigBlock to come from NewLeaf. |
| 205 | for (BasicBlock::iterator I = Succ->begin(); isa<PHINode>(Val: I); ++I) { |
| 206 | PHINode *PN = cast<PHINode>(Val&: I); |
| 207 | // Remove all but one incoming entries from the cluster |
| 208 | APInt Range = Leaf.High->getValue() - Leaf.Low->getValue(); |
| 209 | for (APInt j(Range.getBitWidth(), 0, false); j.ult(RHS: Range); ++j) { |
| 210 | PN->removeIncomingValue(BB: OrigBlock); |
| 211 | } |
| 212 | |
| 213 | int BlockIdx = PN->getBasicBlockIndex(BB: OrigBlock); |
| 214 | assert(BlockIdx != -1 && "Switch didn't go to this successor??" ); |
| 215 | PN->setIncomingBlock(i: (unsigned)BlockIdx, BB: NewLeaf); |
| 216 | } |
| 217 | |
| 218 | return NewLeaf; |
| 219 | } |
| 220 | |
| 221 | /// Convert the switch statement into a binary lookup of the case values. |
| 222 | /// The function recursively builds this tree. LowerBound and UpperBound are |
| 223 | /// used to keep track of the bounds for Val that have already been checked by |
| 224 | /// a block emitted by one of the previous calls to switchConvert in the call |
| 225 | /// stack. |
| 226 | BasicBlock *SwitchConvert(CaseItr Begin, CaseItr End, ConstantInt *LowerBound, |
| 227 | ConstantInt *UpperBound, Value *Val, |
| 228 | BasicBlock *Predecessor, BasicBlock *OrigBlock, |
| 229 | BasicBlock *Default, |
| 230 | const std::vector<IntRange> &UnreachableRanges) { |
| 231 | assert(LowerBound && UpperBound && "Bounds must be initialized" ); |
| 232 | unsigned Size = End - Begin; |
| 233 | |
| 234 | if (Size == 1) { |
| 235 | // Check if the Case Range is perfectly squeezed in between |
| 236 | // already checked Upper and Lower bounds. If it is then we can avoid |
| 237 | // emitting the code that checks if the value actually falls in the range |
| 238 | // because the bounds already tell us so. |
| 239 | if (Begin->Low == LowerBound && Begin->High == UpperBound) { |
| 240 | APInt NumMergedCases = UpperBound->getValue() - LowerBound->getValue(); |
| 241 | FixPhis(SuccBB: Begin->BB, OrigBB: OrigBlock, NewBB: Predecessor, NumMergedCases); |
| 242 | return Begin->BB; |
| 243 | } |
| 244 | return NewLeafBlock(Leaf&: *Begin, Val, LowerBound, UpperBound, OrigBlock, |
| 245 | Default); |
| 246 | } |
| 247 | |
| 248 | unsigned Mid = Size / 2; |
| 249 | std::vector<CaseRange> LHS(Begin, Begin + Mid); |
| 250 | LLVM_DEBUG(dbgs() << "LHS: " << LHS << "\n" ); |
| 251 | std::vector<CaseRange> RHS(Begin + Mid, End); |
| 252 | LLVM_DEBUG(dbgs() << "RHS: " << RHS << "\n" ); |
| 253 | |
| 254 | CaseRange &Pivot = *(Begin + Mid); |
| 255 | LLVM_DEBUG(dbgs() << "Pivot ==> [" << Pivot.Low->getValue() << ", " |
| 256 | << Pivot.High->getValue() << "]\n" ); |
| 257 | |
| 258 | // NewLowerBound here should never be the integer minimal value. |
| 259 | // This is because it is computed from a case range that is never |
| 260 | // the smallest, so there is always a case range that has at least |
| 261 | // a smaller value. |
| 262 | ConstantInt *NewLowerBound = Pivot.Low; |
| 263 | |
| 264 | // Because NewLowerBound is never the smallest representable integer |
| 265 | // it is safe here to subtract one. |
| 266 | ConstantInt *NewUpperBound = ConstantInt::get(Context&: NewLowerBound->getContext(), |
| 267 | V: NewLowerBound->getValue() - 1); |
| 268 | |
| 269 | if (!UnreachableRanges.empty()) { |
| 270 | // Check if the gap between LHS's highest and NewLowerBound is unreachable. |
| 271 | APInt GapLow = LHS.back().High->getValue() + 1; |
| 272 | APInt GapHigh = NewLowerBound->getValue() - 1; |
| 273 | IntRange Gap = {.Low: GapLow, .High: GapHigh}; |
| 274 | if (GapHigh.sge(RHS: GapLow) && IsInRanges(R: Gap, Ranges: UnreachableRanges)) |
| 275 | NewUpperBound = LHS.back().High; |
| 276 | } |
| 277 | |
| 278 | LLVM_DEBUG(dbgs() << "LHS Bounds ==> [" << LowerBound->getValue() << ", " |
| 279 | << NewUpperBound->getValue() << "]\n" |
| 280 | << "RHS Bounds ==> [" << NewLowerBound->getValue() << ", " |
| 281 | << UpperBound->getValue() << "]\n" ); |
| 282 | |
| 283 | // Create a new node that checks if the value is < pivot. Go to the |
| 284 | // left branch if it is and right branch if not. |
| 285 | Function *F = OrigBlock->getParent(); |
| 286 | BasicBlock *NewNode = BasicBlock::Create(Context&: Val->getContext(), Name: "NodeBlock" ); |
| 287 | |
| 288 | ICmpInst *Comp = new ICmpInst(ICmpInst::ICMP_SLT, Val, Pivot.Low, "Pivot" ); |
| 289 | |
| 290 | BasicBlock *LBranch = |
| 291 | SwitchConvert(Begin: LHS.begin(), End: LHS.end(), LowerBound, UpperBound: NewUpperBound, Val, |
| 292 | Predecessor: NewNode, OrigBlock, Default, UnreachableRanges); |
| 293 | BasicBlock *RBranch = |
| 294 | SwitchConvert(Begin: RHS.begin(), End: RHS.end(), LowerBound: NewLowerBound, UpperBound, Val, |
| 295 | Predecessor: NewNode, OrigBlock, Default, UnreachableRanges); |
| 296 | |
| 297 | F->insert(Position: ++OrigBlock->getIterator(), BB: NewNode); |
| 298 | Comp->insertInto(ParentBB: NewNode, It: NewNode->end()); |
| 299 | |
| 300 | BranchInst::Create(IfTrue: LBranch, IfFalse: RBranch, Cond: Comp, InsertBefore: NewNode); |
| 301 | return NewNode; |
| 302 | } |
| 303 | |
| 304 | /// Transform simple list of \p SI's cases into list of CaseRange's \p Cases. |
| 305 | /// \post \p Cases wouldn't contain references to \p SI's default BB. |
| 306 | /// \returns Number of \p SI's cases that do not reference \p SI's default BB. |
| 307 | unsigned Clusterify(CaseVector &Cases, SwitchInst *SI) { |
| 308 | unsigned NumSimpleCases = 0; |
| 309 | |
| 310 | // Start with "simple" cases |
| 311 | for (auto Case : SI->cases()) { |
| 312 | if (Case.getCaseSuccessor() == SI->getDefaultDest()) |
| 313 | continue; |
| 314 | Cases.push_back(x: CaseRange(Case.getCaseValue(), Case.getCaseValue(), |
| 315 | Case.getCaseSuccessor())); |
| 316 | ++NumSimpleCases; |
| 317 | } |
| 318 | |
| 319 | llvm::sort(C&: Cases, Comp: CaseCmp()); |
| 320 | |
| 321 | // Merge case into clusters |
| 322 | if (Cases.size() >= 2) { |
| 323 | CaseItr I = Cases.begin(); |
| 324 | for (CaseItr J = std::next(x: I), E = Cases.end(); J != E; ++J) { |
| 325 | const APInt &nextValue = J->Low->getValue(); |
| 326 | const APInt ¤tValue = I->High->getValue(); |
| 327 | BasicBlock *nextBB = J->BB; |
| 328 | BasicBlock *currentBB = I->BB; |
| 329 | |
| 330 | // If the two neighboring cases go to the same destination, merge them |
| 331 | // into a single case. |
| 332 | assert(nextValue.sgt(currentValue) && |
| 333 | "Cases should be strictly ascending" ); |
| 334 | if ((nextValue == currentValue + 1) && (currentBB == nextBB)) { |
| 335 | I->High = J->High; |
| 336 | // FIXME: Combine branch weights. |
| 337 | } else if (++I != J) { |
| 338 | *I = *J; |
| 339 | } |
| 340 | } |
| 341 | Cases.erase(first: std::next(x: I), last: Cases.end()); |
| 342 | } |
| 343 | |
| 344 | return NumSimpleCases; |
| 345 | } |
| 346 | |
| 347 | /// Replace the specified switch instruction with a sequence of chained if-then |
| 348 | /// insts in a balanced binary search. |
| 349 | void ProcessSwitchInst(SwitchInst *SI, |
| 350 | SmallPtrSetImpl<BasicBlock *> &DeleteList, |
| 351 | AssumptionCache *AC, LazyValueInfo *LVI) { |
| 352 | BasicBlock *OrigBlock = SI->getParent(); |
| 353 | Function *F = OrigBlock->getParent(); |
| 354 | Value *Val = SI->getCondition(); // The value we are switching on... |
| 355 | BasicBlock *Default = SI->getDefaultDest(); |
| 356 | |
| 357 | // Don't handle unreachable blocks. If there are successors with phis, this |
| 358 | // would leave them behind with missing predecessors. |
| 359 | if ((OrigBlock != &F->getEntryBlock() && pred_empty(BB: OrigBlock)) || |
| 360 | OrigBlock->getSinglePredecessor() == OrigBlock) { |
| 361 | DeleteList.insert(Ptr: OrigBlock); |
| 362 | return; |
| 363 | } |
| 364 | |
| 365 | // Prepare cases vector. |
| 366 | CaseVector Cases; |
| 367 | const unsigned NumSimpleCases = Clusterify(Cases, SI); |
| 368 | IntegerType *IT = cast<IntegerType>(Val: SI->getCondition()->getType()); |
| 369 | const unsigned BitWidth = IT->getBitWidth(); |
| 370 | // Explicitly use higher precision to prevent unsigned overflow where |
| 371 | // `UnsignedMax - 0 + 1 == 0` |
| 372 | APInt UnsignedZero(BitWidth + 1, 0); |
| 373 | APInt UnsignedMax = APInt::getMaxValue(numBits: BitWidth); |
| 374 | LLVM_DEBUG(dbgs() << "Clusterify finished. Total clusters: " << Cases.size() |
| 375 | << ". Total non-default cases: " << NumSimpleCases |
| 376 | << "\nCase clusters: " << Cases << "\n" ); |
| 377 | |
| 378 | // If there is only the default destination, just branch. |
| 379 | if (Cases.empty()) { |
| 380 | BranchInst::Create(IfTrue: Default, InsertBefore: OrigBlock); |
| 381 | // Remove all the references from Default's PHIs to OrigBlock, but one. |
| 382 | FixPhis(SuccBB: Default, OrigBB: OrigBlock, NewBB: OrigBlock, NumMergedCases: UnsignedMax); |
| 383 | SI->eraseFromParent(); |
| 384 | return; |
| 385 | } |
| 386 | |
| 387 | ConstantInt *LowerBound = nullptr; |
| 388 | ConstantInt *UpperBound = nullptr; |
| 389 | bool DefaultIsUnreachableFromSwitch = false; |
| 390 | |
| 391 | if (SI->defaultDestUnreachable()) { |
| 392 | // Make the bounds tightly fitted around the case value range, because we |
| 393 | // know that the value passed to the switch must be exactly one of the case |
| 394 | // values. |
| 395 | LowerBound = Cases.front().Low; |
| 396 | UpperBound = Cases.back().High; |
| 397 | DefaultIsUnreachableFromSwitch = true; |
| 398 | } else { |
| 399 | // Constraining the range of the value being switched over helps eliminating |
| 400 | // unreachable BBs and minimizing the number of `add` instructions |
| 401 | // newLeafBlock ends up emitting. Running CorrelatedValuePropagation after |
| 402 | // LowerSwitch isn't as good, and also much more expensive in terms of |
| 403 | // compile time for the following reasons: |
| 404 | // 1. it processes many kinds of instructions, not just switches; |
| 405 | // 2. even if limited to icmp instructions only, it will have to process |
| 406 | // roughly C icmp's per switch, where C is the number of cases in the |
| 407 | // switch, while LowerSwitch only needs to call LVI once per switch. |
| 408 | const DataLayout &DL = F->getDataLayout(); |
| 409 | KnownBits Known = computeKnownBits(V: Val, DL, AC, CxtI: SI); |
| 410 | // TODO Shouldn't this create a signed range? |
| 411 | ConstantRange KnownBitsRange = |
| 412 | ConstantRange::fromKnownBits(Known, /*IsSigned=*/false); |
| 413 | const ConstantRange LVIRange = |
| 414 | LVI->getConstantRange(V: Val, CxtI: SI, /*UndefAllowed*/ false); |
| 415 | ConstantRange ValRange = KnownBitsRange.intersectWith(CR: LVIRange); |
| 416 | // We delegate removal of unreachable non-default cases to other passes. In |
| 417 | // the unlikely event that some of them survived, we just conservatively |
| 418 | // maintain the invariant that all the cases lie between the bounds. This |
| 419 | // may, however, still render the default case effectively unreachable. |
| 420 | const APInt &Low = Cases.front().Low->getValue(); |
| 421 | const APInt &High = Cases.back().High->getValue(); |
| 422 | APInt Min = APIntOps::smin(A: ValRange.getSignedMin(), B: Low); |
| 423 | APInt Max = APIntOps::smax(A: ValRange.getSignedMax(), B: High); |
| 424 | |
| 425 | LowerBound = ConstantInt::get(Context&: SI->getContext(), V: Min); |
| 426 | UpperBound = ConstantInt::get(Context&: SI->getContext(), V: Max); |
| 427 | DefaultIsUnreachableFromSwitch = (Min + (NumSimpleCases - 1) == Max); |
| 428 | } |
| 429 | |
| 430 | std::vector<IntRange> UnreachableRanges; |
| 431 | |
| 432 | if (DefaultIsUnreachableFromSwitch) { |
| 433 | DenseMap<BasicBlock *, APInt> Popularity; |
| 434 | APInt MaxPop(UnsignedZero); |
| 435 | BasicBlock *PopSucc = nullptr; |
| 436 | |
| 437 | APInt SignedMax = APInt::getSignedMaxValue(numBits: BitWidth); |
| 438 | APInt SignedMin = APInt::getSignedMinValue(numBits: BitWidth); |
| 439 | IntRange R = {.Low: SignedMin, .High: SignedMax}; |
| 440 | UnreachableRanges.push_back(x: R); |
| 441 | for (const auto &I : Cases) { |
| 442 | const APInt &Low = I.Low->getValue(); |
| 443 | const APInt &High = I.High->getValue(); |
| 444 | |
| 445 | IntRange &LastRange = UnreachableRanges.back(); |
| 446 | if (LastRange.Low.eq(RHS: Low)) { |
| 447 | // There is nothing left of the previous range. |
| 448 | UnreachableRanges.pop_back(); |
| 449 | } else { |
| 450 | // Terminate the previous range. |
| 451 | assert(Low.sgt(LastRange.Low)); |
| 452 | LastRange.High = Low - 1; |
| 453 | } |
| 454 | if (High.ne(RHS: SignedMax)) { |
| 455 | IntRange R = {.Low: High + 1, .High: SignedMax}; |
| 456 | UnreachableRanges.push_back(x: R); |
| 457 | } |
| 458 | |
| 459 | // Count popularity. |
| 460 | assert(High.sge(Low) && "Popularity shouldn't be negative." ); |
| 461 | APInt N = High.sext(width: BitWidth + 1) - Low.sext(width: BitWidth + 1) + 1; |
| 462 | // Explict insert to make sure the bitwidth of APInts match |
| 463 | APInt &Pop = Popularity.insert(KV: {I.BB, APInt(UnsignedZero)}).first->second; |
| 464 | if ((Pop += N).ugt(RHS: MaxPop)) { |
| 465 | MaxPop = Pop; |
| 466 | PopSucc = I.BB; |
| 467 | } |
| 468 | } |
| 469 | #ifndef NDEBUG |
| 470 | /* UnreachableRanges should be sorted and the ranges non-adjacent. */ |
| 471 | for (auto I = UnreachableRanges.begin(), E = UnreachableRanges.end(); |
| 472 | I != E; ++I) { |
| 473 | assert(I->Low.sle(I->High)); |
| 474 | auto Next = I + 1; |
| 475 | if (Next != E) { |
| 476 | assert(Next->Low.sgt(I->High)); |
| 477 | } |
| 478 | } |
| 479 | #endif |
| 480 | |
| 481 | // As the default block in the switch is unreachable, update the PHI nodes |
| 482 | // (remove all of the references to the default block) to reflect this. |
| 483 | const unsigned NumDefaultEdges = SI->getNumCases() + 1 - NumSimpleCases; |
| 484 | for (unsigned I = 0; I < NumDefaultEdges; ++I) |
| 485 | Default->removePredecessor(Pred: OrigBlock); |
| 486 | |
| 487 | // Use the most popular block as the new default, reducing the number of |
| 488 | // cases. |
| 489 | Default = PopSucc; |
| 490 | llvm::erase_if(C&: Cases, |
| 491 | P: [PopSucc](const CaseRange &R) { return R.BB == PopSucc; }); |
| 492 | |
| 493 | // If there are no cases left, just branch. |
| 494 | if (Cases.empty()) { |
| 495 | BranchInst::Create(IfTrue: Default, InsertBefore: OrigBlock); |
| 496 | SI->eraseFromParent(); |
| 497 | // As all the cases have been replaced with a single branch, only keep |
| 498 | // one entry in the PHI nodes. |
| 499 | if (!MaxPop.isZero()) |
| 500 | for (APInt I(UnsignedZero); I.ult(RHS: MaxPop - 1); ++I) |
| 501 | PopSucc->removePredecessor(Pred: OrigBlock); |
| 502 | return; |
| 503 | } |
| 504 | |
| 505 | // If the condition was a PHI node with the switch block as a predecessor |
| 506 | // removing predecessors may have caused the condition to be erased. |
| 507 | // Getting the condition value again here protects against that. |
| 508 | Val = SI->getCondition(); |
| 509 | } |
| 510 | |
| 511 | BasicBlock *SwitchBlock = |
| 512 | SwitchConvert(Begin: Cases.begin(), End: Cases.end(), LowerBound, UpperBound, Val, |
| 513 | Predecessor: OrigBlock, OrigBlock, Default, UnreachableRanges); |
| 514 | |
| 515 | // We have added incoming values for newly-created predecessors in |
| 516 | // NewLeafBlock(). The only meaningful work we offload to FixPhis() is to |
| 517 | // remove the incoming values from OrigBlock. There might be a special case |
| 518 | // that SwitchBlock is the same as Default, under which the PHIs in Default |
| 519 | // are fixed inside SwitchConvert(). |
| 520 | if (SwitchBlock != Default) |
| 521 | FixPhis(SuccBB: Default, OrigBB: OrigBlock, NewBB: nullptr, NumMergedCases: UnsignedMax); |
| 522 | |
| 523 | // Branch to our shiny new if-then stuff... |
| 524 | BranchInst::Create(IfTrue: SwitchBlock, InsertBefore: OrigBlock); |
| 525 | |
| 526 | // We are now done with the switch instruction, delete it. |
| 527 | BasicBlock *OldDefault = SI->getDefaultDest(); |
| 528 | SI->eraseFromParent(); |
| 529 | |
| 530 | // If the Default block has no more predecessors just add it to DeleteList. |
| 531 | if (pred_empty(BB: OldDefault)) |
| 532 | DeleteList.insert(Ptr: OldDefault); |
| 533 | } |
| 534 | |
| 535 | bool LowerSwitch(Function &F, LazyValueInfo *LVI, AssumptionCache *AC) { |
| 536 | bool Changed = false; |
| 537 | SmallPtrSet<BasicBlock *, 8> DeleteList; |
| 538 | |
| 539 | // We use make_early_inc_range here so that we don't traverse new blocks. |
| 540 | for (BasicBlock &Cur : llvm::make_early_inc_range(Range&: F)) { |
| 541 | // If the block is a dead Default block that will be deleted later, don't |
| 542 | // waste time processing it. |
| 543 | if (DeleteList.count(Ptr: &Cur)) |
| 544 | continue; |
| 545 | |
| 546 | if (SwitchInst *SI = dyn_cast<SwitchInst>(Val: Cur.getTerminator())) { |
| 547 | Changed = true; |
| 548 | ProcessSwitchInst(SI, DeleteList, AC, LVI); |
| 549 | } |
| 550 | } |
| 551 | |
| 552 | for (BasicBlock *BB : DeleteList) { |
| 553 | LVI->eraseBlock(BB); |
| 554 | DeleteDeadBlock(BB); |
| 555 | } |
| 556 | |
| 557 | return Changed; |
| 558 | } |
| 559 | |
| 560 | /// Replace all SwitchInst instructions with chained branch instructions. |
| 561 | class LowerSwitchLegacyPass : public FunctionPass { |
| 562 | public: |
| 563 | // Pass identification, replacement for typeid |
| 564 | static char ID; |
| 565 | |
| 566 | LowerSwitchLegacyPass() : FunctionPass(ID) { |
| 567 | initializeLowerSwitchLegacyPassPass(*PassRegistry::getPassRegistry()); |
| 568 | } |
| 569 | |
| 570 | bool runOnFunction(Function &F) override; |
| 571 | |
| 572 | void getAnalysisUsage(AnalysisUsage &AU) const override { |
| 573 | AU.addRequired<LazyValueInfoWrapperPass>(); |
| 574 | } |
| 575 | }; |
| 576 | |
| 577 | } // end anonymous namespace |
| 578 | |
| 579 | char LowerSwitchLegacyPass::ID = 0; |
| 580 | |
| 581 | // Publicly exposed interface to pass... |
| 582 | char &llvm::LowerSwitchID = LowerSwitchLegacyPass::ID; |
| 583 | |
| 584 | INITIALIZE_PASS_BEGIN(LowerSwitchLegacyPass, "lowerswitch" , |
| 585 | "Lower SwitchInst's to branches" , false, false) |
| 586 | INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker) |
| 587 | INITIALIZE_PASS_DEPENDENCY(LazyValueInfoWrapperPass) |
| 588 | INITIALIZE_PASS_END(LowerSwitchLegacyPass, "lowerswitch" , |
| 589 | "Lower SwitchInst's to branches" , false, false) |
| 590 | |
| 591 | // createLowerSwitchPass - Interface to this file... |
| 592 | FunctionPass *llvm::createLowerSwitchPass() { |
| 593 | return new LowerSwitchLegacyPass(); |
| 594 | } |
| 595 | |
| 596 | bool LowerSwitchLegacyPass::runOnFunction(Function &F) { |
| 597 | LazyValueInfo *LVI = &getAnalysis<LazyValueInfoWrapperPass>().getLVI(); |
| 598 | auto *ACT = getAnalysisIfAvailable<AssumptionCacheTracker>(); |
| 599 | AssumptionCache *AC = ACT ? &ACT->getAssumptionCache(F) : nullptr; |
| 600 | return LowerSwitch(F, LVI, AC); |
| 601 | } |
| 602 | |
| 603 | PreservedAnalyses LowerSwitchPass::run(Function &F, |
| 604 | FunctionAnalysisManager &AM) { |
| 605 | LazyValueInfo *LVI = &AM.getResult<LazyValueAnalysis>(IR&: F); |
| 606 | AssumptionCache *AC = AM.getCachedResult<AssumptionAnalysis>(IR&: F); |
| 607 | return LowerSwitch(F, LVI, AC) ? PreservedAnalyses::none() |
| 608 | : PreservedAnalyses::all(); |
| 609 | } |
| 610 | |