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" 13b8621596SSam Clegg #include "WriterUtils.h" 14c94d393aSSam Clegg #include "lld/Common/ErrorHandler.h" 152017d52bSRui Ueyama #include "lld/Common/Memory.h" 167d67dd11SRui Ueyama #include "llvm/ADT/SetVector.h" 17c94d393aSSam Clegg 18c94d393aSSam Clegg #define DEBUG_TYPE "lld" 19c94d393aSSam Clegg 20c94d393aSSam Clegg using namespace llvm; 2120db381bSSam Clegg using namespace llvm::wasm; 22c94d393aSSam Clegg using namespace lld; 23c94d393aSSam Clegg using namespace lld::wasm; 24c94d393aSSam Clegg 25c94d393aSSam Clegg SymbolTable *lld::wasm::Symtab; 26c94d393aSSam Clegg 27c94d393aSSam Clegg void SymbolTable::addFile(InputFile *File) { 28c94d393aSSam Clegg log("Processing: " + toString(File)); 29c94d393aSSam Clegg File->parse(); 30c94d393aSSam Clegg 31c94d393aSSam Clegg if (auto *F = dyn_cast<ObjFile>(File)) 32c94d393aSSam Clegg ObjectFiles.push_back(F); 33c94d393aSSam Clegg } 34c94d393aSSam Clegg 35c94d393aSSam Clegg void SymbolTable::reportRemainingUndefines() { 367d67dd11SRui Ueyama SetVector<Symbol *> Undefs; 3774fe0ba1SSam Clegg for (Symbol *Sym : SymVector) { 38c94d393aSSam Clegg if (Sym->isUndefined() && !Sym->isWeak() && 39c94d393aSSam Clegg Config->AllowUndefinedSymbols.count(Sym->getName()) == 0) { 40c94d393aSSam Clegg Undefs.insert(Sym); 41c94d393aSSam Clegg } 42c94d393aSSam Clegg } 43c94d393aSSam Clegg 44c94d393aSSam Clegg if (Undefs.empty()) 45c94d393aSSam Clegg return; 46c94d393aSSam Clegg 47c94d393aSSam Clegg for (ObjFile *File : ObjectFiles) 48c94d393aSSam Clegg for (Symbol *Sym : File->getSymbols()) 49c94d393aSSam Clegg if (Undefs.count(Sym)) 50c94d393aSSam Clegg error(toString(File) + ": undefined symbol: " + toString(*Sym)); 51c94d393aSSam Clegg 52c94d393aSSam Clegg for (Symbol *Sym : Undefs) 53c94d393aSSam Clegg if (!Sym->getFile()) 54c94d393aSSam Clegg error("undefined symbol: " + toString(*Sym)); 55c94d393aSSam Clegg } 56c94d393aSSam Clegg 57c94d393aSSam Clegg Symbol *SymbolTable::find(StringRef Name) { 58a80d94d5SSam Clegg auto It = SymMap.find(CachedHashStringRef(Name)); 59a80d94d5SSam Clegg if (It == SymMap.end()) 60c94d393aSSam Clegg return nullptr; 61c94d393aSSam Clegg return It->second; 62c94d393aSSam Clegg } 63c94d393aSSam Clegg 64c94d393aSSam Clegg std::pair<Symbol *, bool> SymbolTable::insert(StringRef Name) { 65a80d94d5SSam Clegg Symbol *&Sym = SymMap[CachedHashStringRef(Name)]; 66c94d393aSSam Clegg if (Sym) 67c94d393aSSam Clegg return {Sym, false}; 68dfb0b2c7SSam Clegg Sym = reinterpret_cast<Symbol *>(make<SymbolUnion>()); 6974fe0ba1SSam Clegg SymVector.emplace_back(Sym); 70c94d393aSSam Clegg return {Sym, true}; 71c94d393aSSam Clegg } 72c94d393aSSam Clegg 73b8621596SSam Clegg // Check the type of new symbol matches that of the symbol is replacing. 74b8621596SSam Clegg // For functions this can also involve verifying that the signatures match. 75b8621596SSam Clegg static void checkSymbolTypes(const Symbol &Existing, const InputFile &F, 76dfb0b2c7SSam Clegg bool NewIsFunction, const WasmSignature *NewSig) { 77b8621596SSam Clegg if (Existing.isLazy()) 78c94d393aSSam Clegg return; 79c94d393aSSam Clegg 80b8621596SSam Clegg // First check the symbol types match (i.e. either both are function 81b8621596SSam Clegg // symbols or both are data symbols). 82caca8d5dSSam Clegg if (isa<FunctionSymbol>(Existing) != NewIsFunction) { 8320db381bSSam Clegg error("symbol type mismatch: " + Existing.getName() + "\n>>> defined as " + 84*00245539SSam Clegg (isa<FunctionSymbol>(Existing) ? "Function" : "Data") + " in " + 85b8621596SSam Clegg toString(Existing.getFile()) + "\n>>> defined as " + 86*00245539SSam Clegg (NewIsFunction ? "Function" : "Data") + " in " + F.getName()); 87b8621596SSam Clegg return; 88b8621596SSam Clegg } 89b8621596SSam Clegg 90b8621596SSam Clegg // For function symbols, optionally check the function signature matches too. 91dfb0b2c7SSam Clegg auto *ExistingFunc = dyn_cast<FunctionSymbol>(&Existing); 92dfb0b2c7SSam Clegg if (!ExistingFunc || !Config->CheckSignatures) 93c94d393aSSam Clegg return; 94dfb0b2c7SSam Clegg 953f8db988SSam Clegg const WasmSignature *OldSig = ExistingFunc->getFunctionType(); 963f8db988SSam Clegg 9731de2f0cSSam Clegg // Skip the signature check if the existing function has no signature (e.g. 9831de2f0cSSam Clegg // if it is an undefined symbol generated by --undefined command line flag). 993f8db988SSam Clegg if (OldSig == nullptr) 10031de2f0cSSam Clegg return; 101c94d393aSSam Clegg 102dfb0b2c7SSam Clegg DEBUG(dbgs() << "checkSymbolTypes: " << ExistingFunc->getName() << "\n"); 103b8621596SSam Clegg assert(NewSig); 104b8621596SSam Clegg 1053f8db988SSam Clegg if (*NewSig == *OldSig) 106b8621596SSam Clegg return; 107b8621596SSam Clegg 108dfb0b2c7SSam Clegg error("function signature mismatch: " + ExistingFunc->getName() + 1093f8db988SSam Clegg "\n>>> defined as " + toString(*OldSig) + " in " + 110dfb0b2c7SSam Clegg toString(ExistingFunc->getFile()) + "\n>>> defined as " + 111dfb0b2c7SSam Clegg toString(*NewSig) + " in " + F.getName()); 112c94d393aSSam Clegg } 113c94d393aSSam Clegg 11424b3dcd4SSam Clegg static void checkSymbolTypes(const Symbol &Existing, const InputFile &F, 115dfb0b2c7SSam Clegg bool IsFunction, const InputChunk *Chunk) { 11624b3dcd4SSam Clegg const WasmSignature *Sig = nullptr; 11724b3dcd4SSam Clegg if (auto *F = dyn_cast_or_null<InputFunction>(Chunk)) 11824b3dcd4SSam Clegg Sig = &F->Signature; 119dfb0b2c7SSam Clegg return checkSymbolTypes(Existing, F, IsFunction, Sig); 12024b3dcd4SSam Clegg } 12124b3dcd4SSam Clegg 122dfb0b2c7SSam Clegg DefinedFunction *SymbolTable::addSyntheticFunction(StringRef Name, 12350686856SSam Clegg const WasmSignature *Type, 12450686856SSam Clegg uint32_t Flags) { 125a189230bSSam Clegg DEBUG(dbgs() << "addSyntheticFunction: " << Name << "\n"); 12650686856SSam Clegg Symbol *S; 12750686856SSam Clegg bool WasInserted; 12850686856SSam Clegg std::tie(S, WasInserted) = insert(Name); 129a189230bSSam Clegg assert(WasInserted); 130dfb0b2c7SSam Clegg return replaceSymbol<DefinedFunction>(S, Name, Flags, Type); 13150686856SSam Clegg } 13250686856SSam Clegg 133*00245539SSam Clegg DefinedData *SymbolTable::addSyntheticDataSymbol(StringRef Name, 134*00245539SSam Clegg uint32_t Flags) { 135*00245539SSam Clegg DEBUG(dbgs() << "addSyntheticDataSymbol: " << Name << "\n"); 136c94d393aSSam Clegg Symbol *S; 137c94d393aSSam Clegg bool WasInserted; 138c94d393aSSam Clegg std::tie(S, WasInserted) = insert(Name); 139a189230bSSam Clegg assert(WasInserted); 140*00245539SSam Clegg return replaceSymbol<DefinedData>(S, Name, Flags); 141c94d393aSSam Clegg } 142c94d393aSSam Clegg 143f60b0d5cSSam Clegg static bool shouldReplace(const Symbol &Existing, InputFile *NewFile, 144f60b0d5cSSam Clegg uint32_t NewFlags, InputChunk *NewChunk, 145f60b0d5cSSam Clegg bool NewIsFunction) { 146c03c9040SRui Ueyama // If existing symbol is lazy, replace it without checking types since 147b8621596SSam Clegg // lazy symbols don't have any type information. 148c03c9040SRui Ueyama if (Existing.isLazy()) { 14993e559b1SSam Clegg DEBUG(dbgs() << "replacing existing lazy symbol: " << Existing.getName() 15093e559b1SSam Clegg << "\n"); 151c03c9040SRui Ueyama return true; 152c94d393aSSam Clegg } 153c94d393aSSam Clegg 154c03c9040SRui Ueyama // Now we have two wasm symbols, and all wasm symbols that have the same 155c03c9040SRui Ueyama // symbol name must have the same type, even if they are undefined. This 156c03c9040SRui Ueyama // is different from ELF because symbol types are not that significant 157c03c9040SRui Ueyama // in ELF, and undefined symbols in ELF don't have type in the first place. 158f60b0d5cSSam Clegg checkSymbolTypes(Existing, *NewFile, NewIsFunction, NewChunk); 15993e559b1SSam Clegg 160c03c9040SRui Ueyama // If existing symbol is undefined, replace it. 161c03c9040SRui Ueyama if (!Existing.isDefined()) { 162c03c9040SRui Ueyama DEBUG(dbgs() << "resolving existing undefined symbol: " 163c03c9040SRui Ueyama << Existing.getName() << "\n"); 164c03c9040SRui Ueyama return true; 165c03c9040SRui Ueyama } 166c03c9040SRui Ueyama 167c03c9040SRui Ueyama // Now we have two defined symbols. If the new one is weak, we can ignore it. 168c03c9040SRui Ueyama if ((NewFlags & WASM_SYMBOL_BINDING_MASK) == WASM_SYMBOL_BINDING_WEAK) { 169c03c9040SRui Ueyama DEBUG(dbgs() << "existing symbol takes precedence\n"); 170c03c9040SRui Ueyama return false; 171c03c9040SRui Ueyama } 172c03c9040SRui Ueyama 173c03c9040SRui Ueyama // If the existing symbol is weak, we should replace it. 174c03c9040SRui Ueyama if (Existing.isWeak()) { 175c03c9040SRui Ueyama DEBUG(dbgs() << "replacing existing weak symbol\n"); 176c03c9040SRui Ueyama return true; 177c03c9040SRui Ueyama } 178c03c9040SRui Ueyama 179c03c9040SRui Ueyama // Neither symbol is week. They conflict. 180c03c9040SRui Ueyama error("duplicate symbol: " + toString(Existing) + "\n>>> defined in " + 181c03c9040SRui Ueyama toString(Existing.getFile()) + "\n>>> defined in " + toString(NewFile)); 182c03c9040SRui Ueyama return true; 183b8621596SSam Clegg } 18493e559b1SSam Clegg 18593e559b1SSam Clegg Symbol *SymbolTable::addDefinedFunction(StringRef Name, uint32_t Flags, 18693e559b1SSam Clegg InputFile *F, InputFunction *Function) { 18793e559b1SSam Clegg DEBUG(dbgs() << "addDefinedFunction: " << Name << "\n"); 18893e559b1SSam Clegg Symbol *S; 18993e559b1SSam Clegg bool WasInserted; 19093e559b1SSam Clegg std::tie(S, WasInserted) = insert(Name); 191f60b0d5cSSam Clegg if (WasInserted || shouldReplace(*S, F, Flags, Function, true)) 19293e559b1SSam Clegg replaceSymbol<DefinedFunction>(S, Name, Flags, F, Function); 19393e559b1SSam Clegg return S; 19493e559b1SSam Clegg } 19593e559b1SSam Clegg 196*00245539SSam Clegg Symbol *SymbolTable::addDefinedData(StringRef Name, uint32_t Flags, 19793e559b1SSam Clegg InputFile *F, InputSegment *Segment, 19893e559b1SSam Clegg uint32_t Address) { 199*00245539SSam Clegg DEBUG(dbgs() << "addDefinedData:" << Name << " addr:" << Address << "\n"); 20093e559b1SSam Clegg Symbol *S; 20193e559b1SSam Clegg bool WasInserted; 20293e559b1SSam Clegg std::tie(S, WasInserted) = insert(Name); 203f60b0d5cSSam Clegg if (WasInserted || shouldReplace(*S, F, Flags, Segment, false)) 204*00245539SSam Clegg replaceSymbol<DefinedData>(S, Name, Flags, F, Segment, Address); 205c94d393aSSam Clegg return S; 206c94d393aSSam Clegg } 207c94d393aSSam Clegg 20820db381bSSam Clegg Symbol *SymbolTable::addUndefined(StringRef Name, Symbol::Kind Kind, 20920db381bSSam Clegg uint32_t Flags, InputFile *F, 2108d146bbcSSam Clegg const WasmSignature *Type) { 21120db381bSSam Clegg DEBUG(dbgs() << "addUndefined: " << Name << "\n"); 212c03c9040SRui Ueyama 213c94d393aSSam Clegg Symbol *S; 214c94d393aSSam Clegg bool WasInserted; 21520db381bSSam Clegg std::tie(S, WasInserted) = insert(Name); 216c03c9040SRui Ueyama 217dfb0b2c7SSam Clegg bool IsFunction = Kind == Symbol::UndefinedFunctionKind; 218c94d393aSSam Clegg if (WasInserted) { 219dfb0b2c7SSam Clegg if (IsFunction) 220dfb0b2c7SSam Clegg replaceSymbol<UndefinedFunction>(S, Name, Flags, F, Type); 221dfb0b2c7SSam Clegg else 222*00245539SSam Clegg replaceSymbol<UndefinedData>(S, Name, Flags, F); 223c03c9040SRui Ueyama return S; 224c03c9040SRui Ueyama } 225c03c9040SRui Ueyama 226c03c9040SRui Ueyama if (auto *Lazy = dyn_cast<LazySymbol>(S)) { 227c94d393aSSam Clegg DEBUG(dbgs() << "resolved by existing lazy\n"); 228c03c9040SRui Ueyama cast<ArchiveFile>(Lazy->getFile())->addMember(&Lazy->getArchiveSymbol()); 229c03c9040SRui Ueyama return S; 230c03c9040SRui Ueyama } 231c03c9040SRui Ueyama 232c03c9040SRui Ueyama if (S->isDefined()) { 233c94d393aSSam Clegg DEBUG(dbgs() << "resolved by existing\n"); 234dfb0b2c7SSam Clegg checkSymbolTypes(*S, *F, IsFunction, Type); 235c94d393aSSam Clegg } 236c94d393aSSam Clegg return S; 237c94d393aSSam Clegg } 238c94d393aSSam Clegg 239c94d393aSSam Clegg void SymbolTable::addLazy(ArchiveFile *F, const Archive::Symbol *Sym) { 240a681a116SSam Clegg DEBUG(dbgs() << "addLazy: " << Sym->getName() << "\n"); 241c94d393aSSam Clegg StringRef Name = Sym->getName(); 242c03c9040SRui Ueyama 243c94d393aSSam Clegg Symbol *S; 244c94d393aSSam Clegg bool WasInserted; 245c94d393aSSam Clegg std::tie(S, WasInserted) = insert(Name); 246c03c9040SRui Ueyama 247c94d393aSSam Clegg if (WasInserted) { 248dfb0b2c7SSam Clegg replaceSymbol<LazySymbol>(S, Name, F, *Sym); 249c03c9040SRui Ueyama return; 250c03c9040SRui Ueyama } 251c03c9040SRui Ueyama 252c03c9040SRui Ueyama // If there is an existing undefined symbol, load a new one from the archive. 253c03c9040SRui Ueyama if (S->isUndefined()) { 254c94d393aSSam Clegg DEBUG(dbgs() << "replacing existing undefined\n"); 255c94d393aSSam Clegg F->addMember(Sym); 256c94d393aSSam Clegg } 257c94d393aSSam Clegg } 258e0f6fcd0SSam Clegg 259e0f6fcd0SSam Clegg bool SymbolTable::addComdat(StringRef Name, ObjFile *F) { 260e0f6fcd0SSam Clegg DEBUG(dbgs() << "addComdat: " << Name << "\n"); 261e0f6fcd0SSam Clegg ObjFile *&File = ComdatMap[CachedHashStringRef(Name)]; 262e0f6fcd0SSam Clegg if (File) { 263e0f6fcd0SSam Clegg DEBUG(dbgs() << "COMDAT already defined\n"); 264e0f6fcd0SSam Clegg return false; 265e0f6fcd0SSam Clegg } 266e0f6fcd0SSam Clegg File = F; 267e0f6fcd0SSam Clegg return true; 268e0f6fcd0SSam Clegg } 269e0f6fcd0SSam Clegg 270e0f6fcd0SSam Clegg ObjFile *SymbolTable::findComdat(StringRef Name) const { 271e0f6fcd0SSam Clegg auto It = ComdatMap.find(CachedHashStringRef(Name)); 272e0f6fcd0SSam Clegg return It == ComdatMap.end() ? nullptr : It->second; 273e0f6fcd0SSam Clegg } 274