1c94d393aSSam Clegg //===- SymbolTable.cpp ----------------------------------------------------===// 2c94d393aSSam Clegg // 3c94d393aSSam Clegg // The LLVM Linker 4c94d393aSSam Clegg // 5c94d393aSSam Clegg // This file is distributed under the University of Illinois Open Source 6c94d393aSSam Clegg // License. See LICENSE.TXT for details. 7c94d393aSSam Clegg // 8c94d393aSSam Clegg //===----------------------------------------------------------------------===// 9c94d393aSSam Clegg 10c94d393aSSam Clegg #include "SymbolTable.h" 11c94d393aSSam Clegg #include "Config.h" 125fa274beSSam Clegg #include "InputChunks.h" 1393102974SSam Clegg #include "InputGlobal.h" 14b8621596SSam Clegg #include "WriterUtils.h" 15c94d393aSSam Clegg #include "lld/Common/ErrorHandler.h" 162017d52bSRui Ueyama #include "lld/Common/Memory.h" 177d67dd11SRui Ueyama #include "llvm/ADT/SetVector.h" 18c94d393aSSam Clegg 19c94d393aSSam Clegg #define DEBUG_TYPE "lld" 20c94d393aSSam Clegg 21c94d393aSSam Clegg using namespace llvm; 2220db381bSSam Clegg using namespace llvm::wasm; 23c94d393aSSam Clegg using namespace lld; 24c94d393aSSam Clegg using namespace lld::wasm; 25c94d393aSSam Clegg 26c94d393aSSam Clegg SymbolTable *lld::wasm::Symtab; 27c94d393aSSam Clegg 28c94d393aSSam Clegg void SymbolTable::addFile(InputFile *File) { 29c94d393aSSam Clegg log("Processing: " + toString(File)); 30c94d393aSSam Clegg File->parse(); 31c94d393aSSam Clegg 32c729c1b4SSam Clegg // LLVM bitcode file 33c729c1b4SSam Clegg if (auto *F = dyn_cast<BitcodeFile>(File)) 34c729c1b4SSam Clegg BitcodeFiles.push_back(F); 35c729c1b4SSam Clegg else if (auto *F = dyn_cast<ObjFile>(File)) 36c94d393aSSam Clegg ObjectFiles.push_back(F); 37c94d393aSSam Clegg } 38c94d393aSSam Clegg 39c729c1b4SSam Clegg // This function is where all the optimizations of link-time 40c729c1b4SSam Clegg // optimization happens. When LTO is in use, some input files are 41c729c1b4SSam Clegg // not in native object file format but in the LLVM bitcode format. 42c729c1b4SSam Clegg // This function compiles bitcode files into a few big native files 43c729c1b4SSam Clegg // using LLVM functions and replaces bitcode symbols with the results. 44c729c1b4SSam Clegg // Because all bitcode files that the program consists of are passed 45c729c1b4SSam Clegg // to the compiler at once, it can do whole-program optimization. 46c729c1b4SSam Clegg void SymbolTable::addCombinedLTOObject() { 47c729c1b4SSam Clegg if (BitcodeFiles.empty()) 48c729c1b4SSam Clegg return; 49c729c1b4SSam Clegg 50c729c1b4SSam Clegg // Compile bitcode files and replace bitcode symbols. 51c729c1b4SSam Clegg LTO.reset(new BitcodeCompiler); 52c729c1b4SSam Clegg for (BitcodeFile *F : BitcodeFiles) 53c729c1b4SSam Clegg LTO->add(*F); 54c729c1b4SSam Clegg 55c729c1b4SSam Clegg for (StringRef Filename : LTO->compile()) { 56c729c1b4SSam Clegg auto *Obj = make<ObjFile>(MemoryBufferRef(Filename, "lto.tmp")); 57c729c1b4SSam Clegg Obj->parse(); 58c729c1b4SSam Clegg ObjectFiles.push_back(Obj); 59c729c1b4SSam Clegg } 60c729c1b4SSam Clegg } 61c729c1b4SSam Clegg 62c94d393aSSam Clegg void SymbolTable::reportRemainingUndefines() { 637d67dd11SRui Ueyama SetVector<Symbol *> Undefs; 6474fe0ba1SSam Clegg for (Symbol *Sym : SymVector) { 65c729c1b4SSam Clegg if (!Sym->isUndefined() || Sym->isWeak()) 66c729c1b4SSam Clegg continue; 67c729c1b4SSam Clegg if (Config->AllowUndefinedSymbols.count(Sym->getName()) != 0) 68c729c1b4SSam Clegg continue; 69c729c1b4SSam Clegg if (!Sym->IsUsedInRegularObj) 70c729c1b4SSam Clegg continue; 71c94d393aSSam Clegg Undefs.insert(Sym); 72c94d393aSSam Clegg } 73c94d393aSSam Clegg 74c94d393aSSam Clegg if (Undefs.empty()) 75c94d393aSSam Clegg return; 76c94d393aSSam Clegg 77c94d393aSSam Clegg for (ObjFile *File : ObjectFiles) 78c94d393aSSam Clegg for (Symbol *Sym : File->getSymbols()) 79c94d393aSSam Clegg if (Undefs.count(Sym)) 80c94d393aSSam Clegg error(toString(File) + ": undefined symbol: " + toString(*Sym)); 81c94d393aSSam Clegg 82c94d393aSSam Clegg for (Symbol *Sym : Undefs) 83c94d393aSSam Clegg if (!Sym->getFile()) 84c94d393aSSam Clegg error("undefined symbol: " + toString(*Sym)); 85c94d393aSSam Clegg } 86c94d393aSSam Clegg 87c94d393aSSam Clegg Symbol *SymbolTable::find(StringRef Name) { 8867d6908fSRui Ueyama return SymMap.lookup(CachedHashStringRef(Name)); 89c94d393aSSam Clegg } 90c94d393aSSam Clegg 91*4c2cbfe6SSam Clegg std::pair<Symbol *, bool> SymbolTable::insert(StringRef Name, InputFile *File) { 92*4c2cbfe6SSam Clegg bool Inserted = false; 93a80d94d5SSam Clegg Symbol *&Sym = SymMap[CachedHashStringRef(Name)]; 94*4c2cbfe6SSam Clegg if (!Sym) { 95dfb0b2c7SSam Clegg Sym = reinterpret_cast<Symbol *>(make<SymbolUnion>()); 96c729c1b4SSam Clegg Sym->IsUsedInRegularObj = false; 9774fe0ba1SSam Clegg SymVector.emplace_back(Sym); 98*4c2cbfe6SSam Clegg Inserted = true; 99*4c2cbfe6SSam Clegg } 100*4c2cbfe6SSam Clegg if (!File || File->kind() == InputFile::ObjectKind) 101*4c2cbfe6SSam Clegg Sym->IsUsedInRegularObj = true; 102*4c2cbfe6SSam Clegg return {Sym, Inserted}; 103c94d393aSSam Clegg } 104c94d393aSSam Clegg 105e3498ec5SRui Ueyama static void reportTypeError(const Symbol *Existing, const InputFile *File, 1063876d89aSSam Clegg llvm::wasm::WasmSymbolType Type) { 107e3498ec5SRui Ueyama error("symbol type mismatch: " + toString(*Existing) + "\n>>> defined as " + 108e3498ec5SRui Ueyama toString(Existing->getWasmType()) + " in " + 1093876d89aSSam Clegg toString(Existing->getFile()) + "\n>>> defined as " + toString(Type) + 1103876d89aSSam Clegg " in " + toString(File)); 111e3498ec5SRui Ueyama } 112e3498ec5SRui Ueyama 113cefbf9acSSam Clegg static void checkFunctionType(Symbol *Existing, const InputFile *File, 114e3498ec5SRui Ueyama const WasmSignature *NewSig) { 11565d6380cSSam Clegg auto ExistingFunction = dyn_cast<FunctionSymbol>(Existing); 11665d6380cSSam Clegg if (!ExistingFunction) { 1173876d89aSSam Clegg reportTypeError(Existing, File, WASM_SYMBOL_TYPE_FUNCTION); 118e3498ec5SRui Ueyama return; 119e3498ec5SRui Ueyama } 120e3498ec5SRui Ueyama 121cefbf9acSSam Clegg if (!NewSig) 122cefbf9acSSam Clegg return; 123cefbf9acSSam Clegg 124cefbf9acSSam Clegg const WasmSignature *OldSig = ExistingFunction->FunctionType; 125cefbf9acSSam Clegg if (!OldSig) { 126cefbf9acSSam Clegg ExistingFunction->FunctionType = NewSig; 127cefbf9acSSam Clegg return; 128cefbf9acSSam Clegg } 129cefbf9acSSam Clegg 130cefbf9acSSam Clegg if (*NewSig != *OldSig) 131ffd0aaf8SSam Clegg warn("function signature mismatch: " + Existing->getName() + 132e3498ec5SRui Ueyama "\n>>> defined as " + toString(*OldSig) + " in " + 133e3498ec5SRui Ueyama toString(Existing->getFile()) + "\n>>> defined as " + 134ffd0aaf8SSam Clegg toString(*NewSig) + " in " + toString(File)); 135e3498ec5SRui Ueyama } 136e3498ec5SRui Ueyama 137b8621596SSam Clegg // Check the type of new symbol matches that of the symbol is replacing. 138b8621596SSam Clegg // For functions this can also involve verifying that the signatures match. 139e3498ec5SRui Ueyama static void checkGlobalType(const Symbol *Existing, const InputFile *File, 140e3498ec5SRui Ueyama const WasmGlobalType *NewType) { 141e3498ec5SRui Ueyama if (!isa<GlobalSymbol>(Existing)) { 1423876d89aSSam Clegg reportTypeError(Existing, File, WASM_SYMBOL_TYPE_GLOBAL); 143b8621596SSam Clegg return; 144b8621596SSam Clegg } 145b8621596SSam Clegg 146e3498ec5SRui Ueyama const WasmGlobalType *OldType = cast<GlobalSymbol>(Existing)->getGlobalType(); 147e3498ec5SRui Ueyama if (*NewType != *OldType) { 148e3498ec5SRui Ueyama error("Global type mismatch: " + Existing->getName() + "\n>>> defined as " + 149e3498ec5SRui Ueyama toString(*OldType) + " in " + toString(Existing->getFile()) + 150e3498ec5SRui Ueyama "\n>>> defined as " + toString(*NewType) + " in " + toString(File)); 151c94d393aSSam Clegg } 15224b3dcd4SSam Clegg } 15324b3dcd4SSam Clegg 154e3498ec5SRui Ueyama static void checkDataType(const Symbol *Existing, const InputFile *File) { 155e3498ec5SRui Ueyama if (!isa<DataSymbol>(Existing)) 1563876d89aSSam Clegg reportTypeError(Existing, File, WASM_SYMBOL_TYPE_DATA); 157e3498ec5SRui Ueyama } 158e3498ec5SRui Ueyama 159dfb0b2c7SSam Clegg DefinedFunction *SymbolTable::addSyntheticFunction(StringRef Name, 160ebda41f8SNicholas Wilson uint32_t Flags, 161ebda41f8SNicholas Wilson InputFunction *Function) { 162e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addSyntheticFunction: " << Name << "\n"); 163b961abc8SRui Ueyama assert(!find(Name)); 164ebda41f8SNicholas Wilson SyntheticFunctions.emplace_back(Function); 165*4c2cbfe6SSam Clegg return replaceSymbol<DefinedFunction>(insert(Name, nullptr).first, Name, 166*4c2cbfe6SSam Clegg Flags, nullptr, Function); 16750686856SSam Clegg } 16850686856SSam Clegg 16900245539SSam Clegg DefinedData *SymbolTable::addSyntheticDataSymbol(StringRef Name, 17000245539SSam Clegg uint32_t Flags) { 171e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addSyntheticDataSymbol: " << Name << "\n"); 172b961abc8SRui Ueyama assert(!find(Name)); 173*4c2cbfe6SSam Clegg return replaceSymbol<DefinedData>(insert(Name, nullptr).first, Name, Flags); 174c94d393aSSam Clegg } 175c94d393aSSam Clegg 17693102974SSam Clegg DefinedGlobal *SymbolTable::addSyntheticGlobal(StringRef Name, uint32_t Flags, 17793102974SSam Clegg InputGlobal *Global) { 178e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addSyntheticGlobal: " << Name << " -> " << Global 179e7245b42SNicola Zaghen << "\n"); 180b961abc8SRui Ueyama assert(!find(Name)); 181ebda41f8SNicholas Wilson SyntheticGlobals.emplace_back(Global); 182*4c2cbfe6SSam Clegg return replaceSymbol<DefinedGlobal>(insert(Name, nullptr).first, Name, Flags, 183*4c2cbfe6SSam Clegg nullptr, Global); 18493102974SSam Clegg } 18593102974SSam Clegg 186e3498ec5SRui Ueyama static bool shouldReplace(const Symbol *Existing, InputFile *NewFile, 187e3498ec5SRui Ueyama uint32_t NewFlags) { 188c03c9040SRui Ueyama // If existing symbol is undefined, replace it. 189e3498ec5SRui Ueyama if (!Existing->isDefined()) { 190e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "resolving existing undefined symbol: " 191e3498ec5SRui Ueyama << Existing->getName() << "\n"); 192c03c9040SRui Ueyama return true; 193c03c9040SRui Ueyama } 194c03c9040SRui Ueyama 195c03c9040SRui Ueyama // Now we have two defined symbols. If the new one is weak, we can ignore it. 196c03c9040SRui Ueyama if ((NewFlags & WASM_SYMBOL_BINDING_MASK) == WASM_SYMBOL_BINDING_WEAK) { 197e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "existing symbol takes precedence\n"); 198c03c9040SRui Ueyama return false; 199c03c9040SRui Ueyama } 200c03c9040SRui Ueyama 201c03c9040SRui Ueyama // If the existing symbol is weak, we should replace it. 202e3498ec5SRui Ueyama if (Existing->isWeak()) { 203e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "replacing existing weak symbol\n"); 204c03c9040SRui Ueyama return true; 205c03c9040SRui Ueyama } 206c03c9040SRui Ueyama 207c03c9040SRui Ueyama // Neither symbol is week. They conflict. 208e3498ec5SRui Ueyama error("duplicate symbol: " + toString(*Existing) + "\n>>> defined in " + 209e3498ec5SRui Ueyama toString(Existing->getFile()) + "\n>>> defined in " + 210e3498ec5SRui Ueyama toString(NewFile)); 211c03c9040SRui Ueyama return true; 212b8621596SSam Clegg } 21393e559b1SSam Clegg 21493e559b1SSam Clegg Symbol *SymbolTable::addDefinedFunction(StringRef Name, uint32_t Flags, 215e3498ec5SRui Ueyama InputFile *File, 216e3498ec5SRui Ueyama InputFunction *Function) { 217e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addDefinedFunction: " << Name << "\n"); 21893e559b1SSam Clegg Symbol *S; 21993e559b1SSam Clegg bool WasInserted; 220*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 221c729c1b4SSam Clegg 222e3498ec5SRui Ueyama if (WasInserted || S->isLazy()) { 223e3498ec5SRui Ueyama replaceSymbol<DefinedFunction>(S, Name, Flags, File, Function); 224e3498ec5SRui Ueyama return S; 225e3498ec5SRui Ueyama } 226e3498ec5SRui Ueyama 227c729c1b4SSam Clegg if (Function) 228e3498ec5SRui Ueyama checkFunctionType(S, File, &Function->Signature); 229e3498ec5SRui Ueyama 230e3498ec5SRui Ueyama if (shouldReplace(S, File, Flags)) 231e3498ec5SRui Ueyama replaceSymbol<DefinedFunction>(S, Name, Flags, File, Function); 23293e559b1SSam Clegg return S; 23393e559b1SSam Clegg } 23493e559b1SSam Clegg 23500245539SSam Clegg Symbol *SymbolTable::addDefinedData(StringRef Name, uint32_t Flags, 236e3498ec5SRui Ueyama InputFile *File, InputSegment *Segment, 23793102974SSam Clegg uint32_t Address, uint32_t Size) { 238e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addDefinedData:" << Name << " addr:" << Address 239e7245b42SNicola Zaghen << "\n"); 24093e559b1SSam Clegg Symbol *S; 24193e559b1SSam Clegg bool WasInserted; 242*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 243c729c1b4SSam Clegg 244e3498ec5SRui Ueyama if (WasInserted || S->isLazy()) { 245e3498ec5SRui Ueyama replaceSymbol<DefinedData>(S, Name, Flags, File, Segment, Address, Size); 246e3498ec5SRui Ueyama return S; 247e3498ec5SRui Ueyama } 248e3498ec5SRui Ueyama 249e3498ec5SRui Ueyama checkDataType(S, File); 250e3498ec5SRui Ueyama 251e3498ec5SRui Ueyama if (shouldReplace(S, File, Flags)) 252e3498ec5SRui Ueyama replaceSymbol<DefinedData>(S, Name, Flags, File, Segment, Address, Size); 253c94d393aSSam Clegg return S; 254c94d393aSSam Clegg } 255c94d393aSSam Clegg 25693102974SSam Clegg Symbol *SymbolTable::addDefinedGlobal(StringRef Name, uint32_t Flags, 257e3498ec5SRui Ueyama InputFile *File, InputGlobal *Global) { 258e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addDefinedGlobal:" << Name << "\n"); 259*4c2cbfe6SSam Clegg 26093102974SSam Clegg Symbol *S; 26193102974SSam Clegg bool WasInserted; 262*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 263c729c1b4SSam Clegg 264e3498ec5SRui Ueyama if (WasInserted || S->isLazy()) { 265e3498ec5SRui Ueyama replaceSymbol<DefinedGlobal>(S, Name, Flags, File, Global); 26693102974SSam Clegg return S; 26793102974SSam Clegg } 26893102974SSam Clegg 269e3498ec5SRui Ueyama checkGlobalType(S, File, &Global->getType()); 270e3498ec5SRui Ueyama 271e3498ec5SRui Ueyama if (shouldReplace(S, File, Flags)) 272e3498ec5SRui Ueyama replaceSymbol<DefinedGlobal>(S, Name, Flags, File, Global); 273e3498ec5SRui Ueyama return S; 274e3498ec5SRui Ueyama } 275e3498ec5SRui Ueyama 276e3498ec5SRui Ueyama Symbol *SymbolTable::addUndefinedFunction(StringRef Name, uint32_t Flags, 277e3498ec5SRui Ueyama InputFile *File, 278e3498ec5SRui Ueyama const WasmSignature *Sig) { 279e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addUndefinedFunction: " << Name << "\n"); 280c03c9040SRui Ueyama 281c94d393aSSam Clegg Symbol *S; 282c94d393aSSam Clegg bool WasInserted; 283*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 284c729c1b4SSam Clegg 285e3498ec5SRui Ueyama if (WasInserted) 286e3498ec5SRui Ueyama replaceSymbol<UndefinedFunction>(S, Name, Flags, File, Sig); 287e3498ec5SRui Ueyama else if (auto *Lazy = dyn_cast<LazySymbol>(S)) 288b961abc8SRui Ueyama Lazy->fetch(); 289cefbf9acSSam Clegg else 290e3498ec5SRui Ueyama checkFunctionType(S, File, Sig); 291cefbf9acSSam Clegg 292c03c9040SRui Ueyama return S; 293c03c9040SRui Ueyama } 294c03c9040SRui Ueyama 295e3498ec5SRui Ueyama Symbol *SymbolTable::addUndefinedData(StringRef Name, uint32_t Flags, 296e3498ec5SRui Ueyama InputFile *File) { 297e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addUndefinedData: " << Name << "\n"); 29893102974SSam Clegg 299e3498ec5SRui Ueyama Symbol *S; 300e3498ec5SRui Ueyama bool WasInserted; 301*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 302f989a92dSSam Clegg 303e3498ec5SRui Ueyama if (WasInserted) 304e3498ec5SRui Ueyama replaceSymbol<UndefinedData>(S, Name, Flags, File); 305e3498ec5SRui Ueyama else if (auto *Lazy = dyn_cast<LazySymbol>(S)) 306b961abc8SRui Ueyama Lazy->fetch(); 307e3498ec5SRui Ueyama else if (S->isDefined()) 308e3498ec5SRui Ueyama checkDataType(S, File); 309c94d393aSSam Clegg return S; 310c94d393aSSam Clegg } 311c94d393aSSam Clegg 312e3498ec5SRui Ueyama Symbol *SymbolTable::addUndefinedGlobal(StringRef Name, uint32_t Flags, 313e3498ec5SRui Ueyama InputFile *File, 314e3498ec5SRui Ueyama const WasmGlobalType *Type) { 315e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addUndefinedGlobal: " << Name << "\n"); 316e3498ec5SRui Ueyama 317e3498ec5SRui Ueyama Symbol *S; 318e3498ec5SRui Ueyama bool WasInserted; 319*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 320c729c1b4SSam Clegg 321e3498ec5SRui Ueyama if (WasInserted) 322e3498ec5SRui Ueyama replaceSymbol<UndefinedGlobal>(S, Name, Flags, File, Type); 323e3498ec5SRui Ueyama else if (auto *Lazy = dyn_cast<LazySymbol>(S)) 324b961abc8SRui Ueyama Lazy->fetch(); 325e3498ec5SRui Ueyama else if (S->isDefined()) 326e3498ec5SRui Ueyama checkGlobalType(S, File, Type); 327e3498ec5SRui Ueyama return S; 328e3498ec5SRui Ueyama } 329e3498ec5SRui Ueyama 330e3498ec5SRui Ueyama void SymbolTable::addLazy(ArchiveFile *File, const Archive::Symbol *Sym) { 331e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addLazy: " << Sym->getName() << "\n"); 332c94d393aSSam Clegg StringRef Name = Sym->getName(); 333c03c9040SRui Ueyama 334c94d393aSSam Clegg Symbol *S; 335c94d393aSSam Clegg bool WasInserted; 336*4c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, nullptr); 337c03c9040SRui Ueyama 338c94d393aSSam Clegg if (WasInserted) { 339e3498ec5SRui Ueyama replaceSymbol<LazySymbol>(S, Name, File, *Sym); 340c03c9040SRui Ueyama return; 341c03c9040SRui Ueyama } 342c03c9040SRui Ueyama 343c03c9040SRui Ueyama // If there is an existing undefined symbol, load a new one from the archive. 344c03c9040SRui Ueyama if (S->isUndefined()) { 345e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "replacing existing undefined\n"); 346e3498ec5SRui Ueyama File->addMember(Sym); 347c94d393aSSam Clegg } 348c94d393aSSam Clegg } 349e0f6fcd0SSam Clegg 350c4d9aa1bSNicholas Wilson bool SymbolTable::addComdat(StringRef Name) { 351c4d9aa1bSNicholas Wilson return Comdats.insert(CachedHashStringRef(Name)).second; 352e0f6fcd0SSam Clegg } 353