| 1 | //===-- PredicateInfo.cpp - PredicateInfo Builder--------------------===// |
| 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 | // This file implements the PredicateInfo class. |
| 10 | // |
| 11 | //===----------------------------------------------------------------===// |
| 12 | |
| 13 | #include "llvm/Transforms/Utils/PredicateInfo.h" |
| 14 | #include "llvm/ADT/DenseMap.h" |
| 15 | #include "llvm/ADT/STLExtras.h" |
| 16 | #include "llvm/ADT/SmallPtrSet.h" |
| 17 | #include "llvm/Analysis/AssumeBundleQueries.h" |
| 18 | #include "llvm/Analysis/AssumptionCache.h" |
| 19 | #include "llvm/IR/AssemblyAnnotationWriter.h" |
| 20 | #include "llvm/IR/Dominators.h" |
| 21 | #include "llvm/IR/IRBuilder.h" |
| 22 | #include "llvm/IR/InstIterator.h" |
| 23 | #include "llvm/IR/IntrinsicInst.h" |
| 24 | #include "llvm/IR/PatternMatch.h" |
| 25 | #include "llvm/Support/CommandLine.h" |
| 26 | #include "llvm/Support/Debug.h" |
| 27 | #include "llvm/Support/DebugCounter.h" |
| 28 | #include "llvm/Support/FormattedStream.h" |
| 29 | #define DEBUG_TYPE "predicateinfo" |
| 30 | using namespace llvm; |
| 31 | using namespace PatternMatch; |
| 32 | |
| 33 | static cl::opt<bool> VerifyPredicateInfo( |
| 34 | "verify-predicateinfo" , cl::init(Val: false), cl::Hidden, |
| 35 | cl::desc("Verify PredicateInfo in legacy printer pass." )); |
| 36 | DEBUG_COUNTER(RenameCounter, "predicateinfo-rename" , |
| 37 | "Controls which variables are renamed with predicateinfo" ); |
| 38 | |
| 39 | // Maximum number of conditions considered for renaming for each branch/assume. |
| 40 | // This limits renaming of deep and/or chains. |
| 41 | static const unsigned MaxCondsPerBranch = 8; |
| 42 | |
| 43 | namespace { |
| 44 | // Given a predicate info that is a type of branching terminator, get the |
| 45 | // branching block. |
| 46 | const BasicBlock *getBranchBlock(const PredicateBase *PB) { |
| 47 | assert(isa<PredicateWithEdge>(PB) && |
| 48 | "Only branches and switches should have PHIOnly defs that " |
| 49 | "require branch blocks." ); |
| 50 | return cast<PredicateWithEdge>(Val: PB)->From; |
| 51 | } |
| 52 | |
| 53 | // Given a predicate info that is a type of branching terminator, get the |
| 54 | // branching terminator. |
| 55 | static Instruction *getBranchTerminator(const PredicateBase *PB) { |
| 56 | assert(isa<PredicateWithEdge>(PB) && |
| 57 | "Not a predicate info type we know how to get a terminator from." ); |
| 58 | return cast<PredicateWithEdge>(Val: PB)->From->getTerminator(); |
| 59 | } |
| 60 | |
| 61 | // Given a predicate info that is a type of branching terminator, get the |
| 62 | // edge this predicate info represents |
| 63 | std::pair<BasicBlock *, BasicBlock *> getBlockEdge(const PredicateBase *PB) { |
| 64 | assert(isa<PredicateWithEdge>(PB) && |
| 65 | "Not a predicate info type we know how to get an edge from." ); |
| 66 | const auto *PEdge = cast<PredicateWithEdge>(Val: PB); |
| 67 | return std::make_pair(x: PEdge->From, y: PEdge->To); |
| 68 | } |
| 69 | } |
| 70 | |
| 71 | namespace llvm { |
| 72 | enum LocalNum { |
| 73 | // Operations that must appear first in the block. |
| 74 | LN_First, |
| 75 | // Operations that are somewhere in the middle of the block, and are sorted on |
| 76 | // demand. |
| 77 | LN_Middle, |
| 78 | // Operations that must appear last in a block, like successor phi node uses. |
| 79 | LN_Last |
| 80 | }; |
| 81 | |
| 82 | // Associate global and local DFS info with defs (PInfo set) and uses (U set), |
| 83 | // so we can sort them into a global domination ordering. |
| 84 | struct ValueDFS { |
| 85 | int DFSIn = 0; |
| 86 | int DFSOut = 0; |
| 87 | unsigned int LocalNum = LN_Middle; |
| 88 | // Only one of U or PInfo will be set. |
| 89 | Use *U = nullptr; |
| 90 | PredicateBase *PInfo = nullptr; |
| 91 | }; |
| 92 | |
| 93 | // This compares ValueDFS structures. Doing so allows us to walk the minimum |
| 94 | // number of instructions necessary to compute our def/use ordering. |
| 95 | struct ValueDFS_Compare { |
| 96 | DominatorTree &DT; |
| 97 | ValueDFS_Compare(DominatorTree &DT) : DT(DT) {} |
| 98 | |
| 99 | bool operator()(const ValueDFS &A, const ValueDFS &B) const { |
| 100 | if (&A == &B) |
| 101 | return false; |
| 102 | |
| 103 | // Order by block first. |
| 104 | if (A.DFSIn != B.DFSIn) |
| 105 | return A.DFSIn < B.DFSIn; |
| 106 | assert(A.DFSOut == B.DFSOut && |
| 107 | "Equal DFS-in numbers imply equal out numbers" ); |
| 108 | |
| 109 | // Then order by first/middle/last. |
| 110 | if (A.LocalNum != B.LocalNum) |
| 111 | return A.LocalNum < B.LocalNum; |
| 112 | |
| 113 | // We want to put the def that will get used for a given set of phi uses, |
| 114 | // before those phi uses. |
| 115 | // So we sort by edge, then by def. |
| 116 | // Note that only phi nodes uses and defs can come last. |
| 117 | if (A.LocalNum == LN_Last) |
| 118 | return comparePHIRelated(A, B); |
| 119 | |
| 120 | // Use block-local ordering for instructions in the middle. |
| 121 | if (A.LocalNum == LN_Middle) |
| 122 | return localComesBefore(A, B); |
| 123 | |
| 124 | // The order of PredicateInfo definitions at the start of the block does not |
| 125 | // matter. |
| 126 | assert(A.LocalNum == LN_First); |
| 127 | assert(A.PInfo && B.PInfo && "Must be predicate info def" ); |
| 128 | return false; |
| 129 | } |
| 130 | |
| 131 | // For a phi use, or a non-materialized def, return the edge it represents. |
| 132 | std::pair<BasicBlock *, BasicBlock *> getBlockEdge(const ValueDFS &VD) const { |
| 133 | if (VD.U) { |
| 134 | auto *PHI = cast<PHINode>(Val: VD.U->getUser()); |
| 135 | return std::make_pair(x: PHI->getIncomingBlock(U: *VD.U), y: PHI->getParent()); |
| 136 | } |
| 137 | // This is really a non-materialized def. |
| 138 | return ::getBlockEdge(PB: VD.PInfo); |
| 139 | } |
| 140 | |
| 141 | // For two phi related values, return the ordering. |
| 142 | bool comparePHIRelated(const ValueDFS &A, const ValueDFS &B) const { |
| 143 | BasicBlock *ASrc, *ADest, *BSrc, *BDest; |
| 144 | std::tie(args&: ASrc, args&: ADest) = getBlockEdge(VD: A); |
| 145 | std::tie(args&: BSrc, args&: BDest) = getBlockEdge(VD: B); |
| 146 | |
| 147 | #ifndef NDEBUG |
| 148 | // This function should only be used for values in the same BB, check that. |
| 149 | DomTreeNode *DomASrc = DT.getNode(ASrc); |
| 150 | DomTreeNode *DomBSrc = DT.getNode(BSrc); |
| 151 | assert(DomASrc->getDFSNumIn() == (unsigned)A.DFSIn && |
| 152 | "DFS numbers for A should match the ones of the source block" ); |
| 153 | assert(DomBSrc->getDFSNumIn() == (unsigned)B.DFSIn && |
| 154 | "DFS numbers for B should match the ones of the source block" ); |
| 155 | assert(A.DFSIn == B.DFSIn && "Values must be in the same block" ); |
| 156 | #endif |
| 157 | (void)ASrc; |
| 158 | (void)BSrc; |
| 159 | |
| 160 | // Use DFS numbers to compare destination blocks, to guarantee a |
| 161 | // deterministic order. |
| 162 | DomTreeNode *DomADest = DT.getNode(BB: ADest); |
| 163 | DomTreeNode *DomBDest = DT.getNode(BB: BDest); |
| 164 | unsigned AIn = DomADest->getDFSNumIn(); |
| 165 | unsigned BIn = DomBDest->getDFSNumIn(); |
| 166 | bool isAUse = A.U; |
| 167 | bool isBUse = B.U; |
| 168 | assert((!A.PInfo || !A.U) && (!B.PInfo || !B.U) && |
| 169 | "Def and U cannot be set at the same time" ); |
| 170 | // Now sort by edge destination and then defs before uses. |
| 171 | return std::tie(args&: AIn, args&: isAUse) < std::tie(args&: BIn, args&: isBUse); |
| 172 | } |
| 173 | |
| 174 | const Instruction *getDefOrUser(const ValueDFS &VD) const { |
| 175 | if (VD.U) |
| 176 | return cast<Instruction>(Val: VD.U->getUser()); |
| 177 | |
| 178 | // For the purpose of ordering, we pretend the def is right after the |
| 179 | // assume, because that is where we will insert the info. |
| 180 | assert(VD.PInfo && "No use, and no predicateinfo should not occur" ); |
| 181 | assert(isa<PredicateAssume>(VD.PInfo) && |
| 182 | "Middle of block should only occur for assumes" ); |
| 183 | return cast<PredicateAssume>(Val: VD.PInfo)->AssumeInst->getNextNode(); |
| 184 | } |
| 185 | |
| 186 | // This performs the necessary local basic block ordering checks to tell |
| 187 | // whether A comes before B, where both are in the same basic block. |
| 188 | bool localComesBefore(const ValueDFS &A, const ValueDFS &B) const { |
| 189 | const Instruction *AInst = getDefOrUser(VD: A); |
| 190 | const Instruction *BInst = getDefOrUser(VD: B); |
| 191 | return AInst->comesBefore(Other: BInst); |
| 192 | } |
| 193 | }; |
| 194 | |
| 195 | class PredicateInfoBuilder { |
| 196 | // Used to store information about each value we might rename. |
| 197 | struct ValueInfo { |
| 198 | SmallVector<PredicateBase *, 4> Infos; |
| 199 | }; |
| 200 | |
| 201 | PredicateInfo &PI; |
| 202 | Function &F; |
| 203 | DominatorTree &DT; |
| 204 | AssumptionCache &AC; |
| 205 | |
| 206 | // This stores info about each operand or comparison result we make copies |
| 207 | // of. The real ValueInfos start at index 1, index 0 is unused so that we |
| 208 | // can more easily detect invalid indexing. |
| 209 | SmallVector<ValueInfo, 32> ValueInfos; |
| 210 | |
| 211 | // This gives the index into the ValueInfos array for a given Value. Because |
| 212 | // 0 is not a valid Value Info index, you can use DenseMap::lookup and tell |
| 213 | // whether it returned a valid result. |
| 214 | DenseMap<Value *, unsigned int> ValueInfoNums; |
| 215 | |
| 216 | BumpPtrAllocator &Allocator; |
| 217 | |
| 218 | ValueInfo &getOrCreateValueInfo(Value *); |
| 219 | const ValueInfo &getValueInfo(Value *) const; |
| 220 | |
| 221 | void processAssume(AssumeInst *, BasicBlock *, |
| 222 | SmallVectorImpl<Value *> &OpsToRename); |
| 223 | void processBranch(BranchInst *, BasicBlock *, |
| 224 | SmallVectorImpl<Value *> &OpsToRename); |
| 225 | void processSwitch(SwitchInst *, BasicBlock *, |
| 226 | SmallVectorImpl<Value *> &OpsToRename); |
| 227 | void renameUses(SmallVectorImpl<Value *> &OpsToRename); |
| 228 | void addInfoFor(SmallVectorImpl<Value *> &OpsToRename, Value *Op, |
| 229 | PredicateBase *PB); |
| 230 | |
| 231 | struct StackEntry { |
| 232 | const ValueDFS *V; |
| 233 | Value *Def = nullptr; |
| 234 | |
| 235 | StackEntry(const ValueDFS *V) : V(V) {} |
| 236 | }; |
| 237 | |
| 238 | using ValueDFSStack = SmallVectorImpl<StackEntry>; |
| 239 | void convertUsesToDFSOrdered(Value *, SmallVectorImpl<ValueDFS> &); |
| 240 | Value *materializeStack(unsigned int &, ValueDFSStack &, Value *); |
| 241 | bool stackIsInScope(const ValueDFSStack &, const ValueDFS &) const; |
| 242 | void popStackUntilDFSScope(ValueDFSStack &, const ValueDFS &); |
| 243 | |
| 244 | public: |
| 245 | PredicateInfoBuilder(PredicateInfo &PI, Function &F, DominatorTree &DT, |
| 246 | AssumptionCache &AC, BumpPtrAllocator &Allocator) |
| 247 | : PI(PI), F(F), DT(DT), AC(AC), Allocator(Allocator) { |
| 248 | // Push an empty operand info so that we can detect 0 as not finding one |
| 249 | ValueInfos.resize(N: 1); |
| 250 | } |
| 251 | |
| 252 | void buildPredicateInfo(); |
| 253 | }; |
| 254 | |
| 255 | bool PredicateInfoBuilder::stackIsInScope(const ValueDFSStack &Stack, |
| 256 | const ValueDFS &VDUse) const { |
| 257 | assert(!Stack.empty() && "Should not be called with empty stack" ); |
| 258 | // If it's a phi only use, make sure it's for this phi node edge, and that the |
| 259 | // use is in a phi node. If it's anything else, and the top of the stack is |
| 260 | // a LN_Last def, we need to pop the stack. We deliberately sort phi uses |
| 261 | // next to the defs they must go with so that we can know it's time to pop |
| 262 | // the stack when we hit the end of the phi uses for a given def. |
| 263 | const ValueDFS &Top = *Stack.back().V; |
| 264 | assert(Top.PInfo && "RenameStack should only contain predicate infos (defs)" ); |
| 265 | if (Top.LocalNum == LN_Last) { |
| 266 | if (!VDUse.U) { |
| 267 | assert(VDUse.PInfo && "A non-use VDUse should have a predicate info" ); |
| 268 | // We should reserve adjacent LN_Last defs for the same phi use. |
| 269 | return VDUse.LocalNum == LN_Last && |
| 270 | // If the two phi defs have the same edge, they must be designated |
| 271 | // for the same succ BB. |
| 272 | getBlockEdge(PB: Top.PInfo) == getBlockEdge(PB: VDUse.PInfo); |
| 273 | } |
| 274 | auto *PHI = dyn_cast<PHINode>(Val: VDUse.U->getUser()); |
| 275 | if (!PHI) |
| 276 | return false; |
| 277 | // Check edge |
| 278 | BasicBlock *EdgePred = PHI->getIncomingBlock(U: *VDUse.U); |
| 279 | if (EdgePred != getBranchBlock(PB: Top.PInfo)) |
| 280 | return false; |
| 281 | |
| 282 | // Use dominates, which knows how to handle edge dominance. |
| 283 | return DT.dominates(BBE: getBlockEdge(PB: Top.PInfo), U: *VDUse.U); |
| 284 | } |
| 285 | |
| 286 | return VDUse.DFSIn >= Top.DFSIn && VDUse.DFSOut <= Top.DFSOut; |
| 287 | } |
| 288 | |
| 289 | void PredicateInfoBuilder::popStackUntilDFSScope(ValueDFSStack &Stack, |
| 290 | const ValueDFS &VD) { |
| 291 | while (!Stack.empty() && !stackIsInScope(Stack, VDUse: VD)) |
| 292 | Stack.pop_back(); |
| 293 | } |
| 294 | |
| 295 | // Convert the uses of Op into a vector of uses, associating global and local |
| 296 | // DFS info with each one. |
| 297 | void PredicateInfoBuilder::convertUsesToDFSOrdered( |
| 298 | Value *Op, SmallVectorImpl<ValueDFS> &DFSOrderedSet) { |
| 299 | for (auto &U : Op->uses()) { |
| 300 | if (auto *I = dyn_cast<Instruction>(Val: U.getUser())) { |
| 301 | // Lifetime intrinsics must work directly on alloca, do not replace them |
| 302 | // with a predicated copy. |
| 303 | if (I->isLifetimeStartOrEnd()) |
| 304 | continue; |
| 305 | |
| 306 | ValueDFS VD; |
| 307 | // Put the phi node uses in the incoming block. |
| 308 | BasicBlock *IBlock; |
| 309 | if (auto *PN = dyn_cast<PHINode>(Val: I)) { |
| 310 | IBlock = PN->getIncomingBlock(U); |
| 311 | // Make phi node users appear last in the incoming block |
| 312 | // they are from. |
| 313 | VD.LocalNum = LN_Last; |
| 314 | } else { |
| 315 | // If it's not a phi node use, it is somewhere in the middle of the |
| 316 | // block. |
| 317 | IBlock = I->getParent(); |
| 318 | VD.LocalNum = LN_Middle; |
| 319 | } |
| 320 | DomTreeNode *DomNode = DT.getNode(BB: IBlock); |
| 321 | // It's possible our use is in an unreachable block. Skip it if so. |
| 322 | if (!DomNode) |
| 323 | continue; |
| 324 | VD.DFSIn = DomNode->getDFSNumIn(); |
| 325 | VD.DFSOut = DomNode->getDFSNumOut(); |
| 326 | VD.U = &U; |
| 327 | DFSOrderedSet.push_back(Elt: VD); |
| 328 | } |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | bool shouldRename(Value *V) { |
| 333 | // Only want real values, not constants. Additionally, operands with one use |
| 334 | // are only being used in the comparison, which means they will not be useful |
| 335 | // for us to consider for predicateinfo. |
| 336 | return (isa<Instruction>(Val: V) || isa<Argument>(Val: V)) && !V->hasOneUse(); |
| 337 | } |
| 338 | |
| 339 | // Collect relevant operations from Comparison that we may want to insert copies |
| 340 | // for. |
| 341 | void collectCmpOps(CmpInst *Comparison, SmallVectorImpl<Value *> &CmpOperands) { |
| 342 | auto *Op0 = Comparison->getOperand(i_nocapture: 0); |
| 343 | auto *Op1 = Comparison->getOperand(i_nocapture: 1); |
| 344 | if (Op0 == Op1) |
| 345 | return; |
| 346 | |
| 347 | CmpOperands.push_back(Elt: Op0); |
| 348 | CmpOperands.push_back(Elt: Op1); |
| 349 | } |
| 350 | |
| 351 | // Add Op, PB to the list of value infos for Op, and mark Op to be renamed. |
| 352 | void PredicateInfoBuilder::addInfoFor(SmallVectorImpl<Value *> &OpsToRename, |
| 353 | Value *Op, PredicateBase *PB) { |
| 354 | auto &OperandInfo = getOrCreateValueInfo(Op); |
| 355 | if (OperandInfo.Infos.empty()) |
| 356 | OpsToRename.push_back(Elt: Op); |
| 357 | OperandInfo.Infos.push_back(Elt: PB); |
| 358 | } |
| 359 | |
| 360 | // Process an assume instruction and place relevant operations we want to rename |
| 361 | // into OpsToRename. |
| 362 | void PredicateInfoBuilder::processAssume( |
| 363 | AssumeInst *II, BasicBlock *AssumeBB, |
| 364 | SmallVectorImpl<Value *> &OpsToRename) { |
| 365 | if (II->hasOperandBundles()) { |
| 366 | for (auto BOI : II->bundle_op_infos()) { |
| 367 | if (RetainedKnowledge RK = getKnowledgeFromBundle(Assume&: *II, BOI)) { |
| 368 | if (RK.AttrKind == Attribute::NonNull) |
| 369 | addInfoFor(OpsToRename, Op: RK.WasOn, |
| 370 | PB: new (Allocator) PredicateBundleAssume(RK.WasOn, II, |
| 371 | Attribute::NonNull)); |
| 372 | } |
| 373 | } |
| 374 | return; |
| 375 | } |
| 376 | |
| 377 | SmallVector<Value *, 4> Worklist; |
| 378 | SmallPtrSet<Value *, 4> Visited; |
| 379 | Worklist.push_back(Elt: II->getOperand(i_nocapture: 0)); |
| 380 | while (!Worklist.empty()) { |
| 381 | Value *Cond = Worklist.pop_back_val(); |
| 382 | if (!Visited.insert(Ptr: Cond).second) |
| 383 | continue; |
| 384 | if (Visited.size() > MaxCondsPerBranch) |
| 385 | break; |
| 386 | |
| 387 | Value *Op0, *Op1; |
| 388 | if (match(V: Cond, P: m_LogicalAnd(L: m_Value(V&: Op0), R: m_Value(V&: Op1)))) { |
| 389 | Worklist.push_back(Elt: Op1); |
| 390 | Worklist.push_back(Elt: Op0); |
| 391 | } |
| 392 | |
| 393 | SmallVector<Value *, 4> Values; |
| 394 | Values.push_back(Elt: Cond); |
| 395 | if (auto *Cmp = dyn_cast<CmpInst>(Val: Cond)) |
| 396 | collectCmpOps(Comparison: Cmp, CmpOperands&: Values); |
| 397 | else if (match(V: Cond, P: m_NUWTrunc(Op: m_Value(V&: Op0)))) |
| 398 | Values.push_back(Elt: Op0); |
| 399 | |
| 400 | for (Value *V : Values) { |
| 401 | if (shouldRename(V)) { |
| 402 | auto *PA = new (Allocator) PredicateConditionAssume(V, II, Cond); |
| 403 | addInfoFor(OpsToRename, Op: V, PB: PA); |
| 404 | } |
| 405 | } |
| 406 | } |
| 407 | } |
| 408 | |
| 409 | // Process a block terminating branch, and place relevant operations to be |
| 410 | // renamed into OpsToRename. |
| 411 | void PredicateInfoBuilder::processBranch( |
| 412 | BranchInst *BI, BasicBlock *BranchBB, |
| 413 | SmallVectorImpl<Value *> &OpsToRename) { |
| 414 | BasicBlock *FirstBB = BI->getSuccessor(i: 0); |
| 415 | BasicBlock *SecondBB = BI->getSuccessor(i: 1); |
| 416 | |
| 417 | for (BasicBlock *Succ : {FirstBB, SecondBB}) { |
| 418 | bool TakenEdge = Succ == FirstBB; |
| 419 | // Don't try to insert on a self-edge. This is mainly because we will |
| 420 | // eliminate during renaming anyway. |
| 421 | if (Succ == BranchBB) |
| 422 | continue; |
| 423 | |
| 424 | SmallVector<Value *, 4> Worklist; |
| 425 | SmallPtrSet<Value *, 4> Visited; |
| 426 | Worklist.push_back(Elt: BI->getCondition()); |
| 427 | while (!Worklist.empty()) { |
| 428 | Value *Cond = Worklist.pop_back_val(); |
| 429 | if (!Visited.insert(Ptr: Cond).second) |
| 430 | continue; |
| 431 | if (Visited.size() > MaxCondsPerBranch) |
| 432 | break; |
| 433 | |
| 434 | Value *Op0, *Op1; |
| 435 | if (TakenEdge ? match(V: Cond, P: m_LogicalAnd(L: m_Value(V&: Op0), R: m_Value(V&: Op1))) |
| 436 | : match(V: Cond, P: m_LogicalOr(L: m_Value(V&: Op0), R: m_Value(V&: Op1)))) { |
| 437 | Worklist.push_back(Elt: Op1); |
| 438 | Worklist.push_back(Elt: Op0); |
| 439 | } |
| 440 | |
| 441 | SmallVector<Value *, 4> Values; |
| 442 | Values.push_back(Elt: Cond); |
| 443 | if (auto *Cmp = dyn_cast<CmpInst>(Val: Cond)) |
| 444 | collectCmpOps(Comparison: Cmp, CmpOperands&: Values); |
| 445 | else if (match(V: Cond, P: m_NUWTrunc(Op: m_Value(V&: Op0)))) |
| 446 | Values.push_back(Elt: Op0); |
| 447 | |
| 448 | for (Value *V : Values) { |
| 449 | if (shouldRename(V)) { |
| 450 | PredicateBase *PB = new (Allocator) |
| 451 | PredicateBranch(V, BranchBB, Succ, Cond, TakenEdge); |
| 452 | addInfoFor(OpsToRename, Op: V, PB); |
| 453 | } |
| 454 | } |
| 455 | } |
| 456 | } |
| 457 | } |
| 458 | // Process a block terminating switch, and place relevant operations to be |
| 459 | // renamed into OpsToRename. |
| 460 | void PredicateInfoBuilder::processSwitch( |
| 461 | SwitchInst *SI, BasicBlock *BranchBB, |
| 462 | SmallVectorImpl<Value *> &OpsToRename) { |
| 463 | Value *Op = SI->getCondition(); |
| 464 | if ((!isa<Instruction>(Val: Op) && !isa<Argument>(Val: Op)) || Op->hasOneUse()) |
| 465 | return; |
| 466 | |
| 467 | // Remember how many outgoing edges there are to every successor. |
| 468 | SmallDenseMap<BasicBlock *, unsigned, 16> SwitchEdges; |
| 469 | for (BasicBlock *TargetBlock : successors(BB: BranchBB)) |
| 470 | ++SwitchEdges[TargetBlock]; |
| 471 | |
| 472 | // Now propagate info for each case value |
| 473 | for (auto C : SI->cases()) { |
| 474 | BasicBlock *TargetBlock = C.getCaseSuccessor(); |
| 475 | if (SwitchEdges.lookup(Val: TargetBlock) == 1) { |
| 476 | PredicateSwitch *PS = new (Allocator) PredicateSwitch( |
| 477 | Op, SI->getParent(), TargetBlock, C.getCaseValue(), SI); |
| 478 | addInfoFor(OpsToRename, Op, PB: PS); |
| 479 | } |
| 480 | } |
| 481 | } |
| 482 | |
| 483 | // Build predicate info for our function |
| 484 | void PredicateInfoBuilder::buildPredicateInfo() { |
| 485 | DT.updateDFSNumbers(); |
| 486 | // Collect operands to rename from all conditional branch terminators, as well |
| 487 | // as assume statements. |
| 488 | SmallVector<Value *, 8> OpsToRename; |
| 489 | for (BasicBlock &BB : F) { |
| 490 | if (!DT.isReachableFromEntry(A: &BB)) |
| 491 | continue; |
| 492 | |
| 493 | if (auto *BI = dyn_cast<BranchInst>(Val: BB.getTerminator())) { |
| 494 | if (!BI->isConditional()) |
| 495 | continue; |
| 496 | // Can't insert conditional information if they all go to the same place. |
| 497 | if (BI->getSuccessor(i: 0) == BI->getSuccessor(i: 1)) |
| 498 | continue; |
| 499 | processBranch(BI, BranchBB: &BB, OpsToRename); |
| 500 | } else if (auto *SI = dyn_cast<SwitchInst>(Val: BB.getTerminator())) { |
| 501 | processSwitch(SI, BranchBB: &BB, OpsToRename); |
| 502 | } |
| 503 | } |
| 504 | for (auto &Assume : AC.assumptions()) { |
| 505 | if (auto *II = cast_or_null<AssumeInst>(Val&: Assume)) |
| 506 | if (DT.isReachableFromEntry(A: II->getParent())) |
| 507 | processAssume(II, AssumeBB: II->getParent(), OpsToRename); |
| 508 | } |
| 509 | // Now rename all our operations. |
| 510 | renameUses(OpsToRename); |
| 511 | } |
| 512 | |
| 513 | // Given the renaming stack, make all the operands currently on the stack real |
| 514 | // by inserting them into the IR. Return the last operation's value. |
| 515 | Value *PredicateInfoBuilder::materializeStack(unsigned int &Counter, |
| 516 | ValueDFSStack &RenameStack, |
| 517 | Value *OrigOp) { |
| 518 | // Find the first thing we have to materialize |
| 519 | auto RevIter = RenameStack.rbegin(); |
| 520 | for (; RevIter != RenameStack.rend(); ++RevIter) |
| 521 | if (RevIter->Def) |
| 522 | break; |
| 523 | |
| 524 | size_t Start = RevIter - RenameStack.rbegin(); |
| 525 | // The maximum number of things we should be trying to materialize at once |
| 526 | // right now is 4, depending on if we had an assume, a branch, and both used |
| 527 | // and of conditions. |
| 528 | for (auto RenameIter = RenameStack.end() - Start; |
| 529 | RenameIter != RenameStack.end(); ++RenameIter) { |
| 530 | auto *Op = |
| 531 | RenameIter == RenameStack.begin() ? OrigOp : (RenameIter - 1)->Def; |
| 532 | StackEntry &Result = *RenameIter; |
| 533 | auto *ValInfo = Result.V->PInfo; |
| 534 | ValInfo->RenamedOp = (RenameStack.end() - Start) == RenameStack.begin() |
| 535 | ? OrigOp |
| 536 | : (RenameStack.end() - Start - 1)->Def; |
| 537 | auto CreateSSACopy = [](Instruction *InsertPt, Value *Op, |
| 538 | const Twine &Name = "" ) { |
| 539 | // Use a no-op bitcast to represent ssa copy. |
| 540 | return new BitCastInst(Op, Op->getType(), Name, InsertPt->getIterator()); |
| 541 | }; |
| 542 | // For edge predicates, we can just place the operand in the block before |
| 543 | // the terminator. For assume, we have to place it right after the assume |
| 544 | // to ensure we dominate all uses except assume itself. Always insert |
| 545 | // right before the terminator or after the assume, so that we insert in |
| 546 | // proper order in the case of multiple predicateinfo in the same block. |
| 547 | if (isa<PredicateWithEdge>(Val: ValInfo)) { |
| 548 | BitCastInst *PIC = CreateSSACopy(getBranchTerminator(PB: ValInfo), Op, |
| 549 | Op->getName() + "." + Twine(Counter++)); |
| 550 | PI.PredicateMap.insert(KV: {PIC, ValInfo}); |
| 551 | Result.Def = PIC; |
| 552 | } else { |
| 553 | auto *PAssume = dyn_cast<PredicateAssume>(Val: ValInfo); |
| 554 | assert(PAssume && |
| 555 | "Should not have gotten here without it being an assume" ); |
| 556 | // Insert the predicate directly after the assume. While it also holds |
| 557 | // directly before it, assume(i1 true) is not a useful fact. |
| 558 | BitCastInst *PIC = CreateSSACopy(PAssume->AssumeInst->getNextNode(), Op); |
| 559 | PI.PredicateMap.insert(KV: {PIC, ValInfo}); |
| 560 | Result.Def = PIC; |
| 561 | } |
| 562 | } |
| 563 | return RenameStack.back().Def; |
| 564 | } |
| 565 | |
| 566 | // Instead of the standard SSA renaming algorithm, which is O(Number of |
| 567 | // instructions), and walks the entire dominator tree, we walk only the defs + |
| 568 | // uses. The standard SSA renaming algorithm does not really rely on the |
| 569 | // dominator tree except to order the stack push/pops of the renaming stacks, so |
| 570 | // that defs end up getting pushed before hitting the correct uses. This does |
| 571 | // not require the dominator tree, only the *order* of the dominator tree. The |
| 572 | // complete and correct ordering of the defs and uses, in dominator tree is |
| 573 | // contained in the DFS numbering of the dominator tree. So we sort the defs and |
| 574 | // uses into the DFS ordering, and then just use the renaming stack as per |
| 575 | // normal, pushing when we hit a def (which is a predicateinfo instruction), |
| 576 | // popping when we are out of the dfs scope for that def, and replacing any uses |
| 577 | // with top of stack if it exists. In order to handle liveness without |
| 578 | // propagating liveness info, we don't actually insert the predicateinfo |
| 579 | // instruction def until we see a use that it would dominate. Once we see such |
| 580 | // a use, we materialize the predicateinfo instruction in the right place and |
| 581 | // use it. |
| 582 | // |
| 583 | // TODO: Use this algorithm to perform fast single-variable renaming in |
| 584 | // promotememtoreg and memoryssa. |
| 585 | void PredicateInfoBuilder::renameUses(SmallVectorImpl<Value *> &OpsToRename) { |
| 586 | ValueDFS_Compare Compare(DT); |
| 587 | // Compute liveness, and rename in O(uses) per Op. |
| 588 | for (auto *Op : OpsToRename) { |
| 589 | LLVM_DEBUG(dbgs() << "Visiting " << *Op << "\n" ); |
| 590 | unsigned Counter = 0; |
| 591 | SmallVector<ValueDFS, 16> OrderedUses; |
| 592 | const auto &ValueInfo = getValueInfo(Op); |
| 593 | // Insert the possible copies into the def/use list. |
| 594 | // They will become real copies if we find a real use for them, and never |
| 595 | // created otherwise. |
| 596 | for (const auto &PossibleCopy : ValueInfo.Infos) { |
| 597 | ValueDFS VD; |
| 598 | // Determine where we are going to place the copy by the copy type. |
| 599 | // The predicate info for branches always come first, they will get |
| 600 | // materialized in the split block at the top of the block. |
| 601 | // The predicate info for assumes will be somewhere in the middle, |
| 602 | // it will get materialized right after the assume. |
| 603 | if (const auto *PAssume = dyn_cast<PredicateAssume>(Val: PossibleCopy)) { |
| 604 | VD.LocalNum = LN_Middle; |
| 605 | DomTreeNode *DomNode = DT.getNode(BB: PAssume->AssumeInst->getParent()); |
| 606 | if (!DomNode) |
| 607 | continue; |
| 608 | VD.DFSIn = DomNode->getDFSNumIn(); |
| 609 | VD.DFSOut = DomNode->getDFSNumOut(); |
| 610 | VD.PInfo = PossibleCopy; |
| 611 | OrderedUses.push_back(Elt: VD); |
| 612 | } else if (isa<PredicateWithEdge>(Val: PossibleCopy)) { |
| 613 | // If we can only do phi uses, we treat it like it's in the branch |
| 614 | // block, and handle it specially. We know that it goes last, and only |
| 615 | // dominate phi uses. |
| 616 | auto BlockEdge = getBlockEdge(PB: PossibleCopy); |
| 617 | if (!BlockEdge.second->getSinglePredecessor()) { |
| 618 | VD.LocalNum = LN_Last; |
| 619 | auto *DomNode = DT.getNode(BB: BlockEdge.first); |
| 620 | if (DomNode) { |
| 621 | VD.DFSIn = DomNode->getDFSNumIn(); |
| 622 | VD.DFSOut = DomNode->getDFSNumOut(); |
| 623 | VD.PInfo = PossibleCopy; |
| 624 | OrderedUses.push_back(Elt: VD); |
| 625 | } |
| 626 | } else { |
| 627 | // Otherwise, we are in the split block (even though we perform |
| 628 | // insertion in the branch block). |
| 629 | // Insert a possible copy at the split block and before the branch. |
| 630 | VD.LocalNum = LN_First; |
| 631 | auto *DomNode = DT.getNode(BB: BlockEdge.second); |
| 632 | if (DomNode) { |
| 633 | VD.DFSIn = DomNode->getDFSNumIn(); |
| 634 | VD.DFSOut = DomNode->getDFSNumOut(); |
| 635 | VD.PInfo = PossibleCopy; |
| 636 | OrderedUses.push_back(Elt: VD); |
| 637 | } |
| 638 | } |
| 639 | } |
| 640 | } |
| 641 | |
| 642 | convertUsesToDFSOrdered(Op, DFSOrderedSet&: OrderedUses); |
| 643 | // Here we require a stable sort because we do not bother to try to |
| 644 | // assign an order to the operands the uses represent. Thus, two |
| 645 | // uses in the same instruction do not have a strict sort order |
| 646 | // currently and will be considered equal. We could get rid of the |
| 647 | // stable sort by creating one if we wanted. |
| 648 | llvm::stable_sort(Range&: OrderedUses, C: Compare); |
| 649 | SmallVector<StackEntry, 8> RenameStack; |
| 650 | // For each use, sorted into dfs order, push values and replaces uses with |
| 651 | // top of stack, which will represent the reaching def. |
| 652 | for (const ValueDFS &VD : OrderedUses) { |
| 653 | // We currently do not materialize copy over copy, but we should decide if |
| 654 | // we want to. |
| 655 | if (RenameStack.empty()) { |
| 656 | LLVM_DEBUG(dbgs() << "Rename Stack is empty\n" ); |
| 657 | } else { |
| 658 | LLVM_DEBUG(dbgs() << "Rename Stack Top DFS numbers are (" |
| 659 | << RenameStack.back().V->DFSIn << "," |
| 660 | << RenameStack.back().V->DFSOut << ")\n" ); |
| 661 | } |
| 662 | |
| 663 | LLVM_DEBUG(dbgs() << "Current DFS numbers are (" << VD.DFSIn << "," |
| 664 | << VD.DFSOut << ")\n" ); |
| 665 | |
| 666 | // Sync to our current scope. |
| 667 | popStackUntilDFSScope(Stack&: RenameStack, VD); |
| 668 | |
| 669 | if (VD.PInfo) { |
| 670 | RenameStack.push_back(Elt: &VD); |
| 671 | continue; |
| 672 | } |
| 673 | |
| 674 | // If we get to this point, and the stack is empty we must have a use |
| 675 | // with no renaming needed, just skip it. |
| 676 | if (RenameStack.empty()) |
| 677 | continue; |
| 678 | if (!DebugCounter::shouldExecute(Counter&: RenameCounter)) { |
| 679 | LLVM_DEBUG(dbgs() << "Skipping execution due to debug counter\n" ); |
| 680 | continue; |
| 681 | } |
| 682 | StackEntry &Result = RenameStack.back(); |
| 683 | |
| 684 | // If the possible copy dominates something, materialize our stack up to |
| 685 | // this point. This ensures every comparison that affects our operation |
| 686 | // ends up with predicateinfo. |
| 687 | if (!Result.Def) |
| 688 | Result.Def = materializeStack(Counter, RenameStack, OrigOp: Op); |
| 689 | |
| 690 | LLVM_DEBUG(dbgs() << "Found replacement " << *Result.Def << " for " |
| 691 | << *VD.U->get() << " in " << *(VD.U->getUser()) |
| 692 | << "\n" ); |
| 693 | assert(DT.dominates(cast<Instruction>(Result.Def), *VD.U) && |
| 694 | "Predicateinfo def should have dominated this use" ); |
| 695 | VD.U->set(Result.Def); |
| 696 | } |
| 697 | } |
| 698 | } |
| 699 | |
| 700 | PredicateInfoBuilder::ValueInfo & |
| 701 | PredicateInfoBuilder::getOrCreateValueInfo(Value *Operand) { |
| 702 | auto Res = ValueInfoNums.try_emplace(Key: Operand, Args: ValueInfos.size()); |
| 703 | if (Res.second) { |
| 704 | // Allocate space for new ValueInfo. |
| 705 | ValueInfos.resize(N: ValueInfos.size() + 1); |
| 706 | } |
| 707 | return ValueInfos[Res.first->second]; |
| 708 | } |
| 709 | |
| 710 | const PredicateInfoBuilder::ValueInfo & |
| 711 | PredicateInfoBuilder::getValueInfo(Value *Operand) const { |
| 712 | auto OINI = ValueInfoNums.lookup(Val: Operand); |
| 713 | assert(OINI != 0 && "Operand was not really in the Value Info Numbers" ); |
| 714 | assert(OINI < ValueInfos.size() && |
| 715 | "Value Info Number greater than size of Value Info Table" ); |
| 716 | return ValueInfos[OINI]; |
| 717 | } |
| 718 | |
| 719 | PredicateInfo::PredicateInfo(Function &F, DominatorTree &DT, |
| 720 | AssumptionCache &AC, BumpPtrAllocator &Allocator) |
| 721 | : F(F) { |
| 722 | PredicateInfoBuilder Builder(*this, F, DT, AC, Allocator); |
| 723 | Builder.buildPredicateInfo(); |
| 724 | } |
| 725 | |
| 726 | std::optional<PredicateConstraint> PredicateBase::getConstraint() const { |
| 727 | switch (Type) { |
| 728 | case PT_BundleAssume: { |
| 729 | assert(cast<PredicateBundleAssume>(this)->AttrKind == Attribute::NonNull && |
| 730 | "Cannot handle anything other than NonNull" ); |
| 731 | return {{.Predicate: CmpInst::ICMP_NE, .OtherOp: ConstantPointerNull::get( |
| 732 | T: cast<PointerType>(Val: OriginalOp->getType()))}}; |
| 733 | } |
| 734 | |
| 735 | case PT_ConditionAssume: |
| 736 | case PT_Branch: { |
| 737 | bool TrueEdge = true; |
| 738 | if (auto *PBranch = dyn_cast<PredicateBranch>(Val: this)) |
| 739 | TrueEdge = PBranch->TrueEdge; |
| 740 | |
| 741 | if (Condition == RenamedOp) { |
| 742 | return {{.Predicate: CmpInst::ICMP_EQ, |
| 743 | .OtherOp: TrueEdge ? ConstantInt::getTrue(Ty: Condition->getType()) |
| 744 | : ConstantInt::getFalse(Ty: Condition->getType())}}; |
| 745 | } |
| 746 | |
| 747 | if (match(V: Condition, P: m_NUWTrunc(Op: m_Specific(V: RenamedOp)))) { |
| 748 | return {{.Predicate: TrueEdge ? CmpInst::ICMP_NE : CmpInst::ICMP_EQ, |
| 749 | .OtherOp: ConstantInt::getNullValue(Ty: RenamedOp->getType())}}; |
| 750 | } |
| 751 | |
| 752 | CmpInst *Cmp = dyn_cast<CmpInst>(Val: Condition); |
| 753 | if (!Cmp) { |
| 754 | // TODO: Make this an assertion once RenamedOp is fully accurate. |
| 755 | return std::nullopt; |
| 756 | } |
| 757 | |
| 758 | CmpInst::Predicate Pred; |
| 759 | Value *OtherOp; |
| 760 | if (Cmp->getOperand(i_nocapture: 0) == RenamedOp) { |
| 761 | Pred = Cmp->getPredicate(); |
| 762 | OtherOp = Cmp->getOperand(i_nocapture: 1); |
| 763 | } else if (Cmp->getOperand(i_nocapture: 1) == RenamedOp) { |
| 764 | Pred = Cmp->getSwappedPredicate(); |
| 765 | OtherOp = Cmp->getOperand(i_nocapture: 0); |
| 766 | } else { |
| 767 | // TODO: Make this an assertion once RenamedOp is fully accurate. |
| 768 | return std::nullopt; |
| 769 | } |
| 770 | |
| 771 | // Invert predicate along false edge. |
| 772 | if (!TrueEdge) |
| 773 | Pred = CmpInst::getInversePredicate(pred: Pred); |
| 774 | |
| 775 | return {{.Predicate: Pred, .OtherOp: OtherOp}}; |
| 776 | } |
| 777 | case PT_Switch: |
| 778 | if (Condition != RenamedOp) { |
| 779 | // TODO: Make this an assertion once RenamedOp is fully accurate. |
| 780 | return std::nullopt; |
| 781 | } |
| 782 | |
| 783 | return {{.Predicate: CmpInst::ICMP_EQ, .OtherOp: cast<PredicateSwitch>(Val: this)->CaseValue}}; |
| 784 | } |
| 785 | llvm_unreachable("Unknown predicate type" ); |
| 786 | } |
| 787 | |
| 788 | void PredicateInfo::verifyPredicateInfo() const {} |
| 789 | |
| 790 | // Replace bitcasts created by PredicateInfo with their operand. |
| 791 | static void replaceCreatedSSACopys(PredicateInfo &PredInfo, Function &F) { |
| 792 | for (Instruction &Inst : llvm::make_early_inc_range(Range: instructions(F))) { |
| 793 | const auto *PI = PredInfo.getPredicateInfoFor(V: &Inst); |
| 794 | if (!PI) |
| 795 | continue; |
| 796 | |
| 797 | assert(isa<BitCastInst>(Inst) && |
| 798 | Inst.getType() == Inst.getOperand(0)->getType()); |
| 799 | Inst.replaceAllUsesWith(V: Inst.getOperand(i: 0)); |
| 800 | Inst.eraseFromParent(); |
| 801 | } |
| 802 | } |
| 803 | |
| 804 | PreservedAnalyses PredicateInfoPrinterPass::run(Function &F, |
| 805 | FunctionAnalysisManager &AM) { |
| 806 | auto &DT = AM.getResult<DominatorTreeAnalysis>(IR&: F); |
| 807 | auto &AC = AM.getResult<AssumptionAnalysis>(IR&: F); |
| 808 | OS << "PredicateInfo for function: " << F.getName() << "\n" ; |
| 809 | BumpPtrAllocator Allocator; |
| 810 | auto PredInfo = std::make_unique<PredicateInfo>(args&: F, args&: DT, args&: AC, args&: Allocator); |
| 811 | PredInfo->print(OS); |
| 812 | |
| 813 | replaceCreatedSSACopys(PredInfo&: *PredInfo, F); |
| 814 | return PreservedAnalyses::all(); |
| 815 | } |
| 816 | |
| 817 | /// An assembly annotator class to print PredicateInfo information in |
| 818 | /// comments. |
| 819 | class PredicateInfoAnnotatedWriter : public AssemblyAnnotationWriter { |
| 820 | friend class PredicateInfo; |
| 821 | const PredicateInfo *PredInfo; |
| 822 | |
| 823 | public: |
| 824 | PredicateInfoAnnotatedWriter(const PredicateInfo *M) : PredInfo(M) {} |
| 825 | |
| 826 | void emitBasicBlockStartAnnot(const BasicBlock *BB, |
| 827 | formatted_raw_ostream &OS) override {} |
| 828 | |
| 829 | void emitInstructionAnnot(const Instruction *I, |
| 830 | formatted_raw_ostream &OS) override { |
| 831 | if (const auto *PI = PredInfo->getPredicateInfoFor(V: I)) { |
| 832 | if (const auto *PB = dyn_cast<PredicateBranch>(Val: PI)) { |
| 833 | OS << "; branch predicate info { TrueEdge: " << PB->TrueEdge |
| 834 | << " Comparison:" << *PB->Condition << " Edge: [" ; |
| 835 | PB->From->printAsOperand(O&: OS); |
| 836 | OS << "," ; |
| 837 | PB->To->printAsOperand(O&: OS); |
| 838 | OS << "]" ; |
| 839 | } else if (const auto *PS = dyn_cast<PredicateSwitch>(Val: PI)) { |
| 840 | OS << "; switch predicate info { CaseValue: " << *PS->CaseValue |
| 841 | << " Edge: [" ; |
| 842 | PS->From->printAsOperand(O&: OS); |
| 843 | OS << "," ; |
| 844 | PS->To->printAsOperand(O&: OS); |
| 845 | OS << "]" ; |
| 846 | } else if (const auto *PA = dyn_cast<PredicateAssume>(Val: PI)) { |
| 847 | OS << "; assume predicate info {" ; |
| 848 | if (auto *PBA = dyn_cast<PredicateBundleAssume>(Val: PA)) { |
| 849 | OS << " Attribute: " << Attribute::getNameFromAttrKind(AttrKind: PBA->AttrKind); |
| 850 | } else { |
| 851 | assert(isa<PredicateConditionAssume>(PA)); |
| 852 | OS << " Comparison:" << *PA->Condition; |
| 853 | } |
| 854 | } |
| 855 | OS << ", RenamedOp: " ; |
| 856 | PI->RenamedOp->printAsOperand(O&: OS, PrintType: false); |
| 857 | OS << " }\n" ; |
| 858 | } |
| 859 | } |
| 860 | }; |
| 861 | |
| 862 | void PredicateInfo::print(raw_ostream &OS) const { |
| 863 | PredicateInfoAnnotatedWriter Writer(this); |
| 864 | F.print(OS, AAW: &Writer); |
| 865 | } |
| 866 | |
| 867 | void PredicateInfo::dump() const { |
| 868 | PredicateInfoAnnotatedWriter Writer(this); |
| 869 | F.print(OS&: dbgs(), AAW: &Writer); |
| 870 | } |
| 871 | |
| 872 | PreservedAnalyses PredicateInfoVerifierPass::run(Function &F, |
| 873 | FunctionAnalysisManager &AM) { |
| 874 | auto &DT = AM.getResult<DominatorTreeAnalysis>(IR&: F); |
| 875 | auto &AC = AM.getResult<AssumptionAnalysis>(IR&: F); |
| 876 | BumpPtrAllocator Allocator; |
| 877 | std::make_unique<PredicateInfo>(args&: F, args&: DT, args&: AC, args&: Allocator)->verifyPredicateInfo(); |
| 878 | |
| 879 | return PreservedAnalyses::all(); |
| 880 | } |
| 881 | } |
| 882 | |