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() { 6374fe0ba1SSam Clegg for (Symbol *Sym : SymVector) { 64c729c1b4SSam Clegg if (!Sym->isUndefined() || Sym->isWeak()) 65c729c1b4SSam Clegg continue; 66c729c1b4SSam Clegg if (Config->AllowUndefinedSymbols.count(Sym->getName()) != 0) 67c729c1b4SSam Clegg continue; 68c729c1b4SSam Clegg if (!Sym->IsUsedInRegularObj) 69c729c1b4SSam Clegg continue; 7047e2b6b2SSam Clegg error(toString(Sym->getFile()) + ": undefined symbol: " + toString(*Sym)); 71c94d393aSSam Clegg } 72c94d393aSSam Clegg } 73c94d393aSSam Clegg 74c94d393aSSam Clegg Symbol *SymbolTable::find(StringRef Name) { 7567d6908fSRui Ueyama return SymMap.lookup(CachedHashStringRef(Name)); 76c94d393aSSam Clegg } 77c94d393aSSam Clegg 784c2cbfe6SSam Clegg std::pair<Symbol *, bool> SymbolTable::insert(StringRef Name, InputFile *File) { 794c2cbfe6SSam Clegg bool Inserted = false; 80a80d94d5SSam Clegg Symbol *&Sym = SymMap[CachedHashStringRef(Name)]; 814c2cbfe6SSam Clegg if (!Sym) { 82dfb0b2c7SSam Clegg Sym = reinterpret_cast<Symbol *>(make<SymbolUnion>()); 83c729c1b4SSam Clegg Sym->IsUsedInRegularObj = false; 8474fe0ba1SSam Clegg SymVector.emplace_back(Sym); 854c2cbfe6SSam Clegg Inserted = true; 864c2cbfe6SSam Clegg } 874c2cbfe6SSam Clegg if (!File || File->kind() == InputFile::ObjectKind) 884c2cbfe6SSam Clegg Sym->IsUsedInRegularObj = true; 894c2cbfe6SSam Clegg return {Sym, Inserted}; 90c94d393aSSam Clegg } 91c94d393aSSam Clegg 92e3498ec5SRui Ueyama static void reportTypeError(const Symbol *Existing, const InputFile *File, 933876d89aSSam Clegg llvm::wasm::WasmSymbolType Type) { 94e3498ec5SRui Ueyama error("symbol type mismatch: " + toString(*Existing) + "\n>>> defined as " + 95e3498ec5SRui Ueyama toString(Existing->getWasmType()) + " in " + 963876d89aSSam Clegg toString(Existing->getFile()) + "\n>>> defined as " + toString(Type) + 973876d89aSSam Clegg " in " + toString(File)); 98e3498ec5SRui Ueyama } 99e3498ec5SRui Ueyama 100cefbf9acSSam Clegg static void checkFunctionType(Symbol *Existing, const InputFile *File, 101e3498ec5SRui Ueyama const WasmSignature *NewSig) { 10265d6380cSSam Clegg auto ExistingFunction = dyn_cast<FunctionSymbol>(Existing); 10365d6380cSSam Clegg if (!ExistingFunction) { 1043876d89aSSam Clegg reportTypeError(Existing, File, WASM_SYMBOL_TYPE_FUNCTION); 105e3498ec5SRui Ueyama return; 106e3498ec5SRui Ueyama } 107e3498ec5SRui Ueyama 108cefbf9acSSam Clegg if (!NewSig) 109cefbf9acSSam Clegg return; 110cefbf9acSSam Clegg 111cefbf9acSSam Clegg const WasmSignature *OldSig = ExistingFunction->FunctionType; 112cefbf9acSSam Clegg if (!OldSig) { 113cefbf9acSSam Clegg ExistingFunction->FunctionType = NewSig; 114cefbf9acSSam Clegg return; 115cefbf9acSSam Clegg } 116cefbf9acSSam Clegg 117cefbf9acSSam Clegg if (*NewSig != *OldSig) 118ffd0aaf8SSam Clegg warn("function signature mismatch: " + Existing->getName() + 119e3498ec5SRui Ueyama "\n>>> defined as " + toString(*OldSig) + " in " + 120e3498ec5SRui Ueyama toString(Existing->getFile()) + "\n>>> defined as " + 121ffd0aaf8SSam Clegg toString(*NewSig) + " in " + toString(File)); 122e3498ec5SRui Ueyama } 123e3498ec5SRui Ueyama 124b8621596SSam Clegg // Check the type of new symbol matches that of the symbol is replacing. 125b8621596SSam Clegg // For functions this can also involve verifying that the signatures match. 126e3498ec5SRui Ueyama static void checkGlobalType(const Symbol *Existing, const InputFile *File, 127e3498ec5SRui Ueyama const WasmGlobalType *NewType) { 128e3498ec5SRui Ueyama if (!isa<GlobalSymbol>(Existing)) { 1293876d89aSSam Clegg reportTypeError(Existing, File, WASM_SYMBOL_TYPE_GLOBAL); 130b8621596SSam Clegg return; 131b8621596SSam Clegg } 132b8621596SSam Clegg 133e3498ec5SRui Ueyama const WasmGlobalType *OldType = cast<GlobalSymbol>(Existing)->getGlobalType(); 134e3498ec5SRui Ueyama if (*NewType != *OldType) { 135e3498ec5SRui Ueyama error("Global type mismatch: " + Existing->getName() + "\n>>> defined as " + 136e3498ec5SRui Ueyama toString(*OldType) + " in " + toString(Existing->getFile()) + 137e3498ec5SRui Ueyama "\n>>> defined as " + toString(*NewType) + " in " + toString(File)); 138c94d393aSSam Clegg } 13924b3dcd4SSam Clegg } 14024b3dcd4SSam Clegg 141e3498ec5SRui Ueyama static void checkDataType(const Symbol *Existing, const InputFile *File) { 142e3498ec5SRui Ueyama if (!isa<DataSymbol>(Existing)) 1433876d89aSSam Clegg reportTypeError(Existing, File, WASM_SYMBOL_TYPE_DATA); 144e3498ec5SRui Ueyama } 145e3498ec5SRui Ueyama 146dfb0b2c7SSam Clegg DefinedFunction *SymbolTable::addSyntheticFunction(StringRef Name, 147ebda41f8SNicholas Wilson uint32_t Flags, 148ebda41f8SNicholas Wilson InputFunction *Function) { 149e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addSyntheticFunction: " << Name << "\n"); 150b961abc8SRui Ueyama assert(!find(Name)); 151ebda41f8SNicholas Wilson SyntheticFunctions.emplace_back(Function); 1524c2cbfe6SSam Clegg return replaceSymbol<DefinedFunction>(insert(Name, nullptr).first, Name, 1534c2cbfe6SSam Clegg Flags, nullptr, Function); 15450686856SSam Clegg } 15550686856SSam Clegg 15600245539SSam Clegg DefinedData *SymbolTable::addSyntheticDataSymbol(StringRef Name, 15700245539SSam Clegg uint32_t Flags) { 158e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addSyntheticDataSymbol: " << Name << "\n"); 159b961abc8SRui Ueyama assert(!find(Name)); 1604c2cbfe6SSam Clegg return replaceSymbol<DefinedData>(insert(Name, nullptr).first, Name, Flags); 161c94d393aSSam Clegg } 162c94d393aSSam Clegg 16393102974SSam Clegg DefinedGlobal *SymbolTable::addSyntheticGlobal(StringRef Name, uint32_t Flags, 16493102974SSam Clegg InputGlobal *Global) { 165e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addSyntheticGlobal: " << Name << " -> " << Global 166e7245b42SNicola Zaghen << "\n"); 167b961abc8SRui Ueyama assert(!find(Name)); 168ebda41f8SNicholas Wilson SyntheticGlobals.emplace_back(Global); 1694c2cbfe6SSam Clegg return replaceSymbol<DefinedGlobal>(insert(Name, nullptr).first, Name, Flags, 1704c2cbfe6SSam Clegg nullptr, Global); 17193102974SSam Clegg } 17293102974SSam Clegg 173e3498ec5SRui Ueyama static bool shouldReplace(const Symbol *Existing, InputFile *NewFile, 174e3498ec5SRui Ueyama uint32_t NewFlags) { 175c03c9040SRui Ueyama // If existing symbol is undefined, replace it. 176e3498ec5SRui Ueyama if (!Existing->isDefined()) { 177e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "resolving existing undefined symbol: " 178e3498ec5SRui Ueyama << Existing->getName() << "\n"); 179c03c9040SRui Ueyama return true; 180c03c9040SRui Ueyama } 181c03c9040SRui Ueyama 182c03c9040SRui Ueyama // Now we have two defined symbols. If the new one is weak, we can ignore it. 183c03c9040SRui Ueyama if ((NewFlags & WASM_SYMBOL_BINDING_MASK) == WASM_SYMBOL_BINDING_WEAK) { 184e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "existing symbol takes precedence\n"); 185c03c9040SRui Ueyama return false; 186c03c9040SRui Ueyama } 187c03c9040SRui Ueyama 188c03c9040SRui Ueyama // If the existing symbol is weak, we should replace it. 189e3498ec5SRui Ueyama if (Existing->isWeak()) { 190e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "replacing existing weak symbol\n"); 191c03c9040SRui Ueyama return true; 192c03c9040SRui Ueyama } 193c03c9040SRui Ueyama 194c03c9040SRui Ueyama // Neither symbol is week. They conflict. 195e3498ec5SRui Ueyama error("duplicate symbol: " + toString(*Existing) + "\n>>> defined in " + 196e3498ec5SRui Ueyama toString(Existing->getFile()) + "\n>>> defined in " + 197e3498ec5SRui Ueyama toString(NewFile)); 198c03c9040SRui Ueyama return true; 199b8621596SSam Clegg } 20093e559b1SSam Clegg 20193e559b1SSam Clegg Symbol *SymbolTable::addDefinedFunction(StringRef Name, uint32_t Flags, 202e3498ec5SRui Ueyama InputFile *File, 203e3498ec5SRui Ueyama InputFunction *Function) { 204*8b0b48f3SSam Clegg LLVM_DEBUG(dbgs() << "addDefinedFunction: " << Name << " [" 205*8b0b48f3SSam Clegg << (Function ? toString(Function->Signature) : "none") 206*8b0b48f3SSam Clegg << "]\n"); 20793e559b1SSam Clegg Symbol *S; 20893e559b1SSam Clegg bool WasInserted; 2094c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 210c729c1b4SSam Clegg 211e3498ec5SRui Ueyama if (WasInserted || S->isLazy()) { 212e3498ec5SRui Ueyama replaceSymbol<DefinedFunction>(S, Name, Flags, File, Function); 213e3498ec5SRui Ueyama return S; 214e3498ec5SRui Ueyama } 215e3498ec5SRui Ueyama 216c729c1b4SSam Clegg if (Function) 217e3498ec5SRui Ueyama checkFunctionType(S, File, &Function->Signature); 218e3498ec5SRui Ueyama 219*8b0b48f3SSam Clegg if (shouldReplace(S, File, Flags)) { 220*8b0b48f3SSam Clegg // If the new defined function doesn't have signture (i.e. bitcode 221*8b0b48f3SSam Clegg // functions) but the old symbols does then preserve the old signature 222*8b0b48f3SSam Clegg const WasmSignature *OldSig = nullptr; 223*8b0b48f3SSam Clegg if (auto* F = dyn_cast<FunctionSymbol>(S)) 224*8b0b48f3SSam Clegg OldSig = F->FunctionType; 225*8b0b48f3SSam Clegg auto NewSym = replaceSymbol<DefinedFunction>(S, Name, Flags, File, Function); 226*8b0b48f3SSam Clegg if (!NewSym->FunctionType) 227*8b0b48f3SSam Clegg NewSym->FunctionType = OldSig; 228*8b0b48f3SSam Clegg } 22993e559b1SSam Clegg return S; 23093e559b1SSam Clegg } 23193e559b1SSam Clegg 23200245539SSam Clegg Symbol *SymbolTable::addDefinedData(StringRef Name, uint32_t Flags, 233e3498ec5SRui Ueyama InputFile *File, InputSegment *Segment, 23493102974SSam Clegg uint32_t Address, uint32_t Size) { 235e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addDefinedData:" << Name << " addr:" << Address 236e7245b42SNicola Zaghen << "\n"); 23793e559b1SSam Clegg Symbol *S; 23893e559b1SSam Clegg bool WasInserted; 2394c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 240c729c1b4SSam Clegg 241e3498ec5SRui Ueyama if (WasInserted || S->isLazy()) { 242e3498ec5SRui Ueyama replaceSymbol<DefinedData>(S, Name, Flags, File, Segment, Address, Size); 243e3498ec5SRui Ueyama return S; 244e3498ec5SRui Ueyama } 245e3498ec5SRui Ueyama 246e3498ec5SRui Ueyama checkDataType(S, File); 247e3498ec5SRui Ueyama 248e3498ec5SRui Ueyama if (shouldReplace(S, File, Flags)) 249e3498ec5SRui Ueyama replaceSymbol<DefinedData>(S, Name, Flags, File, Segment, Address, Size); 250c94d393aSSam Clegg return S; 251c94d393aSSam Clegg } 252c94d393aSSam Clegg 25393102974SSam Clegg Symbol *SymbolTable::addDefinedGlobal(StringRef Name, uint32_t Flags, 254e3498ec5SRui Ueyama InputFile *File, InputGlobal *Global) { 255e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addDefinedGlobal:" << Name << "\n"); 2564c2cbfe6SSam Clegg 25793102974SSam Clegg Symbol *S; 25893102974SSam Clegg bool WasInserted; 2594c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 260c729c1b4SSam Clegg 261e3498ec5SRui Ueyama if (WasInserted || S->isLazy()) { 262e3498ec5SRui Ueyama replaceSymbol<DefinedGlobal>(S, Name, Flags, File, Global); 26393102974SSam Clegg return S; 26493102974SSam Clegg } 26593102974SSam Clegg 266e3498ec5SRui Ueyama checkGlobalType(S, File, &Global->getType()); 267e3498ec5SRui Ueyama 268e3498ec5SRui Ueyama if (shouldReplace(S, File, Flags)) 269e3498ec5SRui Ueyama replaceSymbol<DefinedGlobal>(S, Name, Flags, File, Global); 270e3498ec5SRui Ueyama return S; 271e3498ec5SRui Ueyama } 272e3498ec5SRui Ueyama 273e3498ec5SRui Ueyama Symbol *SymbolTable::addUndefinedFunction(StringRef Name, uint32_t Flags, 274e3498ec5SRui Ueyama InputFile *File, 275e3498ec5SRui Ueyama const WasmSignature *Sig) { 276*8b0b48f3SSam Clegg LLVM_DEBUG(dbgs() << "addUndefinedFunction: " << Name << 277*8b0b48f3SSam Clegg " [" << (Sig ? toString(*Sig) : "none") << "]\n"); 278c03c9040SRui Ueyama 279c94d393aSSam Clegg Symbol *S; 280c94d393aSSam Clegg bool WasInserted; 2814c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 282c729c1b4SSam Clegg 283e3498ec5SRui Ueyama if (WasInserted) 284e3498ec5SRui Ueyama replaceSymbol<UndefinedFunction>(S, Name, Flags, File, Sig); 285e3498ec5SRui Ueyama else if (auto *Lazy = dyn_cast<LazySymbol>(S)) 286b961abc8SRui Ueyama Lazy->fetch(); 287cefbf9acSSam Clegg else 288e3498ec5SRui Ueyama checkFunctionType(S, File, Sig); 289cefbf9acSSam Clegg 290c03c9040SRui Ueyama return S; 291c03c9040SRui Ueyama } 292c03c9040SRui Ueyama 293e3498ec5SRui Ueyama Symbol *SymbolTable::addUndefinedData(StringRef Name, uint32_t Flags, 294e3498ec5SRui Ueyama InputFile *File) { 295e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addUndefinedData: " << Name << "\n"); 29693102974SSam Clegg 297e3498ec5SRui Ueyama Symbol *S; 298e3498ec5SRui Ueyama bool WasInserted; 2994c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 300f989a92dSSam Clegg 301e3498ec5SRui Ueyama if (WasInserted) 302e3498ec5SRui Ueyama replaceSymbol<UndefinedData>(S, Name, Flags, File); 303e3498ec5SRui Ueyama else if (auto *Lazy = dyn_cast<LazySymbol>(S)) 304b961abc8SRui Ueyama Lazy->fetch(); 305e3498ec5SRui Ueyama else if (S->isDefined()) 306e3498ec5SRui Ueyama checkDataType(S, File); 307c94d393aSSam Clegg return S; 308c94d393aSSam Clegg } 309c94d393aSSam Clegg 310e3498ec5SRui Ueyama Symbol *SymbolTable::addUndefinedGlobal(StringRef Name, uint32_t Flags, 311e3498ec5SRui Ueyama InputFile *File, 312e3498ec5SRui Ueyama const WasmGlobalType *Type) { 313e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addUndefinedGlobal: " << Name << "\n"); 314e3498ec5SRui Ueyama 315e3498ec5SRui Ueyama Symbol *S; 316e3498ec5SRui Ueyama bool WasInserted; 3174c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, File); 318c729c1b4SSam Clegg 319e3498ec5SRui Ueyama if (WasInserted) 320e3498ec5SRui Ueyama replaceSymbol<UndefinedGlobal>(S, Name, Flags, File, Type); 321e3498ec5SRui Ueyama else if (auto *Lazy = dyn_cast<LazySymbol>(S)) 322b961abc8SRui Ueyama Lazy->fetch(); 323e3498ec5SRui Ueyama else if (S->isDefined()) 324e3498ec5SRui Ueyama checkGlobalType(S, File, Type); 325e3498ec5SRui Ueyama return S; 326e3498ec5SRui Ueyama } 327e3498ec5SRui Ueyama 328e3498ec5SRui Ueyama void SymbolTable::addLazy(ArchiveFile *File, const Archive::Symbol *Sym) { 329e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "addLazy: " << Sym->getName() << "\n"); 330c94d393aSSam Clegg StringRef Name = Sym->getName(); 331c03c9040SRui Ueyama 332c94d393aSSam Clegg Symbol *S; 333c94d393aSSam Clegg bool WasInserted; 3344c2cbfe6SSam Clegg std::tie(S, WasInserted) = insert(Name, nullptr); 335c03c9040SRui Ueyama 336c94d393aSSam Clegg if (WasInserted) { 337e3498ec5SRui Ueyama replaceSymbol<LazySymbol>(S, Name, File, *Sym); 338c03c9040SRui Ueyama return; 339c03c9040SRui Ueyama } 340c03c9040SRui Ueyama 341c03c9040SRui Ueyama // If there is an existing undefined symbol, load a new one from the archive. 342c03c9040SRui Ueyama if (S->isUndefined()) { 343e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "replacing existing undefined\n"); 344e3498ec5SRui Ueyama File->addMember(Sym); 345c94d393aSSam Clegg } 346c94d393aSSam Clegg } 347e0f6fcd0SSam Clegg 348c4d9aa1bSNicholas Wilson bool SymbolTable::addComdat(StringRef Name) { 349c4d9aa1bSNicholas Wilson return Comdats.insert(CachedHashStringRef(Name)).second; 350e0f6fcd0SSam Clegg } 351