1c94d393aSSam Clegg //===- SymbolTable.cpp ----------------------------------------------------===// 2c94d393aSSam Clegg // 32946cd70SChandler Carruth // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 42946cd70SChandler Carruth // See https://llvm.org/LICENSE.txt for license information. 52946cd70SChandler Carruth // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6c94d393aSSam Clegg // 7c94d393aSSam Clegg //===----------------------------------------------------------------------===// 8c94d393aSSam Clegg 9c94d393aSSam Clegg #include "SymbolTable.h" 10c94d393aSSam Clegg #include "Config.h" 115fa274beSSam Clegg #include "InputChunks.h" 12e915a71fSHeejin Ahn #include "InputEvent.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; 2345218f4aSSam Clegg using namespace llvm::object; 24c94d393aSSam Clegg 2533c59abfSFangrui Song namespace lld { 2633c59abfSFangrui Song namespace wasm { 2733c59abfSFangrui Song SymbolTable *symtab; 28c94d393aSSam Clegg 29136d27abSRui Ueyama void SymbolTable::addFile(InputFile *file) { 30136d27abSRui Ueyama log("Processing: " + toString(file)); 31a282a61bSSam Clegg 32a282a61bSSam Clegg // .a file 33136d27abSRui Ueyama if (auto *f = dyn_cast<ArchiveFile>(file)) { 34136d27abSRui Ueyama f->parse(); 35a282a61bSSam Clegg return; 36a282a61bSSam Clegg } 37a282a61bSSam Clegg 38a282a61bSSam Clegg // .so file 39136d27abSRui Ueyama if (auto *f = dyn_cast<SharedFile>(file)) { 40136d27abSRui Ueyama sharedFiles.push_back(f); 41a282a61bSSam Clegg return; 42a282a61bSSam Clegg } 43a282a61bSSam Clegg 44136d27abSRui Ueyama if (config->trace) 45136d27abSRui Ueyama message(toString(file)); 46c94d393aSSam Clegg 47c729c1b4SSam Clegg // LLVM bitcode file 48136d27abSRui Ueyama if (auto *f = dyn_cast<BitcodeFile>(file)) { 49136d27abSRui Ueyama f->parse(); 50136d27abSRui Ueyama bitcodeFiles.push_back(f); 51a282a61bSSam Clegg return; 52a282a61bSSam Clegg } 53a282a61bSSam Clegg 54a282a61bSSam Clegg // Regular object file 55136d27abSRui Ueyama auto *f = cast<ObjFile>(file); 56136d27abSRui Ueyama f->parse(false); 57136d27abSRui Ueyama objectFiles.push_back(f); 58c94d393aSSam Clegg } 59c94d393aSSam Clegg 60c729c1b4SSam Clegg // This function is where all the optimizations of link-time 61c729c1b4SSam Clegg // optimization happens. When LTO is in use, some input files are 62c729c1b4SSam Clegg // not in native object file format but in the LLVM bitcode format. 63c729c1b4SSam Clegg // This function compiles bitcode files into a few big native files 64c729c1b4SSam Clegg // using LLVM functions and replaces bitcode symbols with the results. 65c729c1b4SSam Clegg // Because all bitcode files that the program consists of are passed 66c729c1b4SSam Clegg // to the compiler at once, it can do whole-program optimization. 67c729c1b4SSam Clegg void SymbolTable::addCombinedLTOObject() { 68b062fe18SSam Clegg // Prevent further LTO objects being included 69b062fe18SSam Clegg BitcodeFile::doneLTO = true; 70b062fe18SSam Clegg 71136d27abSRui Ueyama if (bitcodeFiles.empty()) 72c729c1b4SSam Clegg return; 73c729c1b4SSam Clegg 74c729c1b4SSam Clegg // Compile bitcode files and replace bitcode symbols. 75136d27abSRui Ueyama lto.reset(new BitcodeCompiler); 76136d27abSRui Ueyama for (BitcodeFile *f : bitcodeFiles) 77136d27abSRui Ueyama lto->add(*f); 78c729c1b4SSam Clegg 79136d27abSRui Ueyama for (StringRef filename : lto->compile()) { 80136d27abSRui Ueyama auto *obj = make<ObjFile>(MemoryBufferRef(filename, "lto.tmp"), ""); 81136d27abSRui Ueyama obj->parse(true); 82136d27abSRui Ueyama objectFiles.push_back(obj); 83c729c1b4SSam Clegg } 84c729c1b4SSam Clegg } 85c729c1b4SSam Clegg 86136d27abSRui Ueyama Symbol *SymbolTable::find(StringRef name) { 87136d27abSRui Ueyama auto it = symMap.find(CachedHashStringRef(name)); 88136d27abSRui Ueyama if (it == symMap.end() || it->second == -1) 891f3f774fSSam Clegg return nullptr; 90136d27abSRui Ueyama return symVector[it->second]; 911f3f774fSSam Clegg } 921f3f774fSSam Clegg 93136d27abSRui Ueyama void SymbolTable::replace(StringRef name, Symbol* sym) { 94136d27abSRui Ueyama auto it = symMap.find(CachedHashStringRef(name)); 95136d27abSRui Ueyama symVector[it->second] = sym; 966540e570SSam Clegg } 976540e570SSam Clegg 98136d27abSRui Ueyama std::pair<Symbol *, bool> SymbolTable::insertName(StringRef name) { 99136d27abSRui Ueyama bool trace = false; 100136d27abSRui Ueyama auto p = symMap.insert({CachedHashStringRef(name), (int)symVector.size()}); 101136d27abSRui Ueyama int &symIndex = p.first->second; 102136d27abSRui Ueyama bool isNew = p.second; 103136d27abSRui Ueyama if (symIndex == -1) { 104136d27abSRui Ueyama symIndex = symVector.size(); 105136d27abSRui Ueyama trace = true; 106136d27abSRui Ueyama isNew = true; 1071f3f774fSSam Clegg } 1081f3f774fSSam Clegg 109136d27abSRui Ueyama if (!isNew) 110136d27abSRui Ueyama return {symVector[symIndex], false}; 1111f3f774fSSam Clegg 112136d27abSRui Ueyama Symbol *sym = reinterpret_cast<Symbol *>(make<SymbolUnion>()); 113136d27abSRui Ueyama sym->isUsedInRegularObj = false; 114136d27abSRui Ueyama sym->canInline = true; 115136d27abSRui Ueyama sym->traced = trace; 116064e9907SSam Clegg sym->forceExport = false; 117136d27abSRui Ueyama symVector.emplace_back(sym); 118136d27abSRui Ueyama return {sym, true}; 119c94d393aSSam Clegg } 120c94d393aSSam Clegg 121136d27abSRui Ueyama std::pair<Symbol *, bool> SymbolTable::insert(StringRef name, 122136d27abSRui Ueyama const InputFile *file) { 123136d27abSRui Ueyama Symbol *s; 124136d27abSRui Ueyama bool wasInserted; 125136d27abSRui Ueyama std::tie(s, wasInserted) = insertName(name); 1261f3f774fSSam Clegg 127136d27abSRui Ueyama if (!file || file->kind() == InputFile::ObjectKind) 128136d27abSRui Ueyama s->isUsedInRegularObj = true; 1291f3f774fSSam Clegg 130136d27abSRui Ueyama return {s, wasInserted}; 131c94d393aSSam Clegg } 132c94d393aSSam Clegg 133136d27abSRui Ueyama static void reportTypeError(const Symbol *existing, const InputFile *file, 134136d27abSRui Ueyama llvm::wasm::WasmSymbolType type) { 135136d27abSRui Ueyama error("symbol type mismatch: " + toString(*existing) + "\n>>> defined as " + 136136d27abSRui Ueyama toString(existing->getWasmType()) + " in " + 137136d27abSRui Ueyama toString(existing->getFile()) + "\n>>> defined as " + toString(type) + 138136d27abSRui Ueyama " in " + toString(file)); 139e3498ec5SRui Ueyama } 140e3498ec5SRui Ueyama 1416f4286fbSHeejin Ahn // Check the type of new symbol matches that of the symbol is replacing. 1427c5fcb35SKazuaki Ishizaki // Returns true if the function types match, false is there is a signature 1436540e570SSam Clegg // mismatch. 144136d27abSRui Ueyama static bool signatureMatches(FunctionSymbol *existing, 145136d27abSRui Ueyama const WasmSignature *newSig) { 146136d27abSRui Ueyama const WasmSignature *oldSig = existing->signature; 147d506b0a4SSam Clegg 148d506b0a4SSam Clegg // If either function is missing a signature (this happend for bitcode 149d506b0a4SSam Clegg // symbols) then assume they match. Any mismatch will be reported later 150d506b0a4SSam Clegg // when the LTO objects are added. 151136d27abSRui Ueyama if (!newSig || !oldSig) 1526540e570SSam Clegg return true; 153cefbf9acSSam Clegg 154136d27abSRui Ueyama return *newSig == *oldSig; 155e3498ec5SRui Ueyama } 156e3498ec5SRui Ueyama 157136d27abSRui Ueyama static void checkGlobalType(const Symbol *existing, const InputFile *file, 158136d27abSRui Ueyama const WasmGlobalType *newType) { 159136d27abSRui Ueyama if (!isa<GlobalSymbol>(existing)) { 160136d27abSRui Ueyama reportTypeError(existing, file, WASM_SYMBOL_TYPE_GLOBAL); 161b8621596SSam Clegg return; 162b8621596SSam Clegg } 163b8621596SSam Clegg 164136d27abSRui Ueyama const WasmGlobalType *oldType = cast<GlobalSymbol>(existing)->getGlobalType(); 165136d27abSRui Ueyama if (*newType != *oldType) { 166136d27abSRui Ueyama error("Global type mismatch: " + existing->getName() + "\n>>> defined as " + 167136d27abSRui Ueyama toString(*oldType) + " in " + toString(existing->getFile()) + 168136d27abSRui Ueyama "\n>>> defined as " + toString(*newType) + " in " + toString(file)); 169c94d393aSSam Clegg } 17024b3dcd4SSam Clegg } 17124b3dcd4SSam Clegg 172136d27abSRui Ueyama static void checkEventType(const Symbol *existing, const InputFile *file, 173136d27abSRui Ueyama const WasmEventType *newType, 174136d27abSRui Ueyama const WasmSignature *newSig) { 175136d27abSRui Ueyama auto existingEvent = dyn_cast<EventSymbol>(existing); 176136d27abSRui Ueyama if (!isa<EventSymbol>(existing)) { 177136d27abSRui Ueyama reportTypeError(existing, file, WASM_SYMBOL_TYPE_EVENT); 178e915a71fSHeejin Ahn return; 179e915a71fSHeejin Ahn } 180e915a71fSHeejin Ahn 181136d27abSRui Ueyama const WasmEventType *oldType = cast<EventSymbol>(existing)->getEventType(); 182136d27abSRui Ueyama const WasmSignature *oldSig = existingEvent->signature; 183136d27abSRui Ueyama if (newType->Attribute != oldType->Attribute) 184136d27abSRui Ueyama error("Event type mismatch: " + existing->getName() + "\n>>> defined as " + 185136d27abSRui Ueyama toString(*oldType) + " in " + toString(existing->getFile()) + 186136d27abSRui Ueyama "\n>>> defined as " + toString(*newType) + " in " + toString(file)); 187136d27abSRui Ueyama if (*newSig != *oldSig) 188136d27abSRui Ueyama warn("Event signature mismatch: " + existing->getName() + 189136d27abSRui Ueyama "\n>>> defined as " + toString(*oldSig) + " in " + 190136d27abSRui Ueyama toString(existing->getFile()) + "\n>>> defined as " + 191136d27abSRui Ueyama toString(*newSig) + " in " + toString(file)); 192e915a71fSHeejin Ahn } 193e915a71fSHeejin Ahn 194136d27abSRui Ueyama static void checkDataType(const Symbol *existing, const InputFile *file) { 195136d27abSRui Ueyama if (!isa<DataSymbol>(existing)) 196136d27abSRui Ueyama reportTypeError(existing, file, WASM_SYMBOL_TYPE_DATA); 197e3498ec5SRui Ueyama } 198e3498ec5SRui Ueyama 199136d27abSRui Ueyama DefinedFunction *SymbolTable::addSyntheticFunction(StringRef name, 200136d27abSRui Ueyama uint32_t flags, 201136d27abSRui Ueyama InputFunction *function) { 202136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addSyntheticFunction: " << name << "\n"); 203136d27abSRui Ueyama assert(!find(name)); 204136d27abSRui Ueyama syntheticFunctions.emplace_back(function); 205136d27abSRui Ueyama return replaceSymbol<DefinedFunction>(insertName(name).first, name, 206136d27abSRui Ueyama flags, nullptr, function); 20750686856SSam Clegg } 20850686856SSam Clegg 2098e8ddaa3SSam Clegg // Adds an optional, linker generated, data symbols. The symbol will only be 2107ae3d335SKazuaki Ishizaki // added if there is an undefine reference to it, or if it is explicitly 2117ae3d335SKazuaki Ishizaki // exported via the --export flag. Otherwise we don't add the symbol and return 2127ae3d335SKazuaki Ishizaki // nullptr. 213caa0db13SSam Clegg DefinedData *SymbolTable::addOptionalDataSymbol(StringRef name, 214caa0db13SSam Clegg uint32_t value) { 215136d27abSRui Ueyama Symbol *s = find(name); 216136d27abSRui Ueyama if (!s && (config->exportAll || config->exportedSymbols.count(name) != 0)) 217136d27abSRui Ueyama s = insertName(name).first; 218136d27abSRui Ueyama else if (!s || s->isDefined()) 2194bce63a0SSam Clegg return nullptr; 220136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addOptionalDataSymbol: " << name << "\n"); 221caa0db13SSam Clegg auto *rtn = replaceSymbol<DefinedData>(s, name, WASM_SYMBOL_VISIBILITY_HIDDEN); 222136d27abSRui Ueyama rtn->setVirtualAddress(value); 223136d27abSRui Ueyama rtn->referenced = true; 2244bce63a0SSam Clegg return rtn; 2254bce63a0SSam Clegg } 2264bce63a0SSam Clegg 227136d27abSRui Ueyama DefinedData *SymbolTable::addSyntheticDataSymbol(StringRef name, 228136d27abSRui Ueyama uint32_t flags) { 229136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addSyntheticDataSymbol: " << name << "\n"); 230136d27abSRui Ueyama assert(!find(name)); 231136d27abSRui Ueyama return replaceSymbol<DefinedData>(insertName(name).first, name, flags); 232c94d393aSSam Clegg } 233c94d393aSSam Clegg 234136d27abSRui Ueyama DefinedGlobal *SymbolTable::addSyntheticGlobal(StringRef name, uint32_t flags, 235136d27abSRui Ueyama InputGlobal *global) { 236136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addSyntheticGlobal: " << name << " -> " << global 237e7245b42SNicola Zaghen << "\n"); 238136d27abSRui Ueyama assert(!find(name)); 239136d27abSRui Ueyama syntheticGlobals.emplace_back(global); 240136d27abSRui Ueyama return replaceSymbol<DefinedGlobal>(insertName(name).first, name, flags, 241136d27abSRui Ueyama nullptr, global); 24293102974SSam Clegg } 24393102974SSam Clegg 244136d27abSRui Ueyama static bool shouldReplace(const Symbol *existing, InputFile *newFile, 245136d27abSRui Ueyama uint32_t newFlags) { 246c03c9040SRui Ueyama // If existing symbol is undefined, replace it. 247136d27abSRui Ueyama if (!existing->isDefined()) { 248e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "resolving existing undefined symbol: " 249136d27abSRui Ueyama << existing->getName() << "\n"); 250c03c9040SRui Ueyama return true; 251c03c9040SRui Ueyama } 252c03c9040SRui Ueyama 253c03c9040SRui Ueyama // Now we have two defined symbols. If the new one is weak, we can ignore it. 254136d27abSRui Ueyama if ((newFlags & WASM_SYMBOL_BINDING_MASK) == WASM_SYMBOL_BINDING_WEAK) { 255e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "existing symbol takes precedence\n"); 256c03c9040SRui Ueyama return false; 257c03c9040SRui Ueyama } 258c03c9040SRui Ueyama 259c03c9040SRui Ueyama // If the existing symbol is weak, we should replace it. 260136d27abSRui Ueyama if (existing->isWeak()) { 261e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "replacing existing weak symbol\n"); 262c03c9040SRui Ueyama return true; 263c03c9040SRui Ueyama } 264c03c9040SRui Ueyama 265c03c9040SRui Ueyama // Neither symbol is week. They conflict. 266136d27abSRui Ueyama error("duplicate symbol: " + toString(*existing) + "\n>>> defined in " + 267136d27abSRui Ueyama toString(existing->getFile()) + "\n>>> defined in " + 268136d27abSRui Ueyama toString(newFile)); 269c03c9040SRui Ueyama return true; 270b8621596SSam Clegg } 27193e559b1SSam Clegg 272136d27abSRui Ueyama Symbol *SymbolTable::addDefinedFunction(StringRef name, uint32_t flags, 273136d27abSRui Ueyama InputFile *file, 274136d27abSRui Ueyama InputFunction *function) { 275136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addDefinedFunction: " << name << " [" 276136d27abSRui Ueyama << (function ? toString(function->signature) : "none") 2778b0b48f3SSam Clegg << "]\n"); 278136d27abSRui Ueyama Symbol *s; 279136d27abSRui Ueyama bool wasInserted; 280136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 281c729c1b4SSam Clegg 282136d27abSRui Ueyama auto replaceSym = [&](Symbol *sym) { 2837c5fcb35SKazuaki Ishizaki // If the new defined function doesn't have signature (i.e. bitcode 2846540e570SSam Clegg // functions) but the old symbol does, then preserve the old signature 285136d27abSRui Ueyama const WasmSignature *oldSig = s->getSignature(); 286136d27abSRui Ueyama auto* newSym = replaceSymbol<DefinedFunction>(sym, name, flags, file, function); 287136d27abSRui Ueyama if (!newSym->signature) 288136d27abSRui Ueyama newSym->signature = oldSig; 2896540e570SSam Clegg }; 2906540e570SSam Clegg 291136d27abSRui Ueyama if (wasInserted || s->isLazy()) { 292136d27abSRui Ueyama replaceSym(s); 293136d27abSRui Ueyama return s; 294e3498ec5SRui Ueyama } 295e3498ec5SRui Ueyama 296136d27abSRui Ueyama auto existingFunction = dyn_cast<FunctionSymbol>(s); 297136d27abSRui Ueyama if (!existingFunction) { 298136d27abSRui Ueyama reportTypeError(s, file, WASM_SYMBOL_TYPE_FUNCTION); 299136d27abSRui Ueyama return s; 3008b0b48f3SSam Clegg } 3016540e570SSam Clegg 302136d27abSRui Ueyama bool checkSig = true; 303136d27abSRui Ueyama if (auto ud = dyn_cast<UndefinedFunction>(existingFunction)) 304136d27abSRui Ueyama checkSig = ud->isCalledDirectly; 30559f959ffSSam Clegg 306136d27abSRui Ueyama if (checkSig && function && !signatureMatches(existingFunction, &function->signature)) { 307136d27abSRui Ueyama Symbol* variant; 308136d27abSRui Ueyama if (getFunctionVariant(s, &function->signature, file, &variant)) 3096540e570SSam Clegg // New variant, always replace 310136d27abSRui Ueyama replaceSym(variant); 311136d27abSRui Ueyama else if (shouldReplace(s, file, flags)) 3126540e570SSam Clegg // Variant already exists, replace it after checking shouldReplace 313136d27abSRui Ueyama replaceSym(variant); 3146540e570SSam Clegg 3156540e570SSam Clegg // This variant we found take the place in the symbol table as the primary 3166540e570SSam Clegg // variant. 317136d27abSRui Ueyama replace(name, variant); 318136d27abSRui Ueyama return variant; 3196540e570SSam Clegg } 3206540e570SSam Clegg 3216540e570SSam Clegg // Existing function with matching signature. 322136d27abSRui Ueyama if (shouldReplace(s, file, flags)) 323136d27abSRui Ueyama replaceSym(s); 3246540e570SSam Clegg 325136d27abSRui Ueyama return s; 32693e559b1SSam Clegg } 32793e559b1SSam Clegg 328136d27abSRui Ueyama Symbol *SymbolTable::addDefinedData(StringRef name, uint32_t flags, 329136d27abSRui Ueyama InputFile *file, InputSegment *segment, 330*3b29376eSWouter van Oortmerssen uint64_t address, uint64_t size) { 331136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addDefinedData:" << name << " addr:" << address 332e7245b42SNicola Zaghen << "\n"); 333136d27abSRui Ueyama Symbol *s; 334136d27abSRui Ueyama bool wasInserted; 335136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 336c729c1b4SSam Clegg 337136d27abSRui Ueyama auto replaceSym = [&]() { 338136d27abSRui Ueyama replaceSymbol<DefinedData>(s, name, flags, file, segment, address, size); 3396540e570SSam Clegg }; 3406540e570SSam Clegg 341136d27abSRui Ueyama if (wasInserted || s->isLazy()) { 342136d27abSRui Ueyama replaceSym(); 343136d27abSRui Ueyama return s; 344e3498ec5SRui Ueyama } 345e3498ec5SRui Ueyama 346136d27abSRui Ueyama checkDataType(s, file); 347e3498ec5SRui Ueyama 348136d27abSRui Ueyama if (shouldReplace(s, file, flags)) 349136d27abSRui Ueyama replaceSym(); 350136d27abSRui Ueyama return s; 351c94d393aSSam Clegg } 352c94d393aSSam Clegg 353136d27abSRui Ueyama Symbol *SymbolTable::addDefinedGlobal(StringRef name, uint32_t flags, 354136d27abSRui Ueyama InputFile *file, InputGlobal *global) { 355136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addDefinedGlobal:" << name << "\n"); 3564c2cbfe6SSam Clegg 357136d27abSRui Ueyama Symbol *s; 358136d27abSRui Ueyama bool wasInserted; 359136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 360c729c1b4SSam Clegg 361136d27abSRui Ueyama auto replaceSym = [&]() { 362136d27abSRui Ueyama replaceSymbol<DefinedGlobal>(s, name, flags, file, global); 3636540e570SSam Clegg }; 3646540e570SSam Clegg 365136d27abSRui Ueyama if (wasInserted || s->isLazy()) { 366136d27abSRui Ueyama replaceSym(); 367136d27abSRui Ueyama return s; 36893102974SSam Clegg } 36993102974SSam Clegg 370136d27abSRui Ueyama checkGlobalType(s, file, &global->getType()); 371e3498ec5SRui Ueyama 372136d27abSRui Ueyama if (shouldReplace(s, file, flags)) 373136d27abSRui Ueyama replaceSym(); 374136d27abSRui Ueyama return s; 375e3498ec5SRui Ueyama } 376e3498ec5SRui Ueyama 377136d27abSRui Ueyama Symbol *SymbolTable::addDefinedEvent(StringRef name, uint32_t flags, 378136d27abSRui Ueyama InputFile *file, InputEvent *event) { 379136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addDefinedEvent:" << name << "\n"); 380e915a71fSHeejin Ahn 381136d27abSRui Ueyama Symbol *s; 382136d27abSRui Ueyama bool wasInserted; 383136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 384e915a71fSHeejin Ahn 385136d27abSRui Ueyama auto replaceSym = [&]() { 386136d27abSRui Ueyama replaceSymbol<DefinedEvent>(s, name, flags, file, event); 3876540e570SSam Clegg }; 3886540e570SSam Clegg 389136d27abSRui Ueyama if (wasInserted || s->isLazy()) { 390136d27abSRui Ueyama replaceSym(); 391136d27abSRui Ueyama return s; 392e915a71fSHeejin Ahn } 393e915a71fSHeejin Ahn 394136d27abSRui Ueyama checkEventType(s, file, &event->getType(), &event->signature); 395e915a71fSHeejin Ahn 396136d27abSRui Ueyama if (shouldReplace(s, file, flags)) 397136d27abSRui Ueyama replaceSym(); 398136d27abSRui Ueyama return s; 399e915a71fSHeejin Ahn } 400e915a71fSHeejin Ahn 40167b05584SSam Clegg // This function get called when an undefined symbol is added, and there is 40267b05584SSam Clegg // already an existing one in the symbols table. In this case we check that 40367b05584SSam Clegg // custom 'import-module' and 'import-field' symbol attributes agree. 4047ae3d335SKazuaki Ishizaki // With LTO these attributes are not available when the bitcode is read and only 40567b05584SSam Clegg // become available when the LTO object is read. In this case we silently 40667b05584SSam Clegg // replace the empty attributes with the valid ones. 40767b05584SSam Clegg template <typename T> 408bd481277SSam Clegg static void setImportAttributes(T *existing, Optional<StringRef> importName, 409bd481277SSam Clegg Optional<StringRef> importModule, 410a57f1a54SSam Clegg uint32_t flags, InputFile *file) { 411bd481277SSam Clegg if (importName) { 412bd481277SSam Clegg if (!existing->importName) 41367b05584SSam Clegg existing->importName = importName; 41467b05584SSam Clegg if (existing->importName != importName) 41567b05584SSam Clegg error("import name mismatch for symbol: " + toString(*existing) + 416bd481277SSam Clegg "\n>>> defined as " + *existing->importName + " in " + 417bd481277SSam Clegg toString(existing->getFile()) + "\n>>> defined as " + *importName + 41867b05584SSam Clegg " in " + toString(file)); 41967b05584SSam Clegg } 42067b05584SSam Clegg 421bd481277SSam Clegg if (importModule) { 422bd481277SSam Clegg if (!existing->importModule) 42367b05584SSam Clegg existing->importModule = importModule; 42467b05584SSam Clegg if (existing->importModule != importModule) 42567b05584SSam Clegg error("import module mismatch for symbol: " + toString(*existing) + 426bd481277SSam Clegg "\n>>> defined as " + *existing->importModule + " in " + 427bd481277SSam Clegg toString(existing->getFile()) + "\n>>> defined as " + 428bd481277SSam Clegg *importModule + " in " + toString(file)); 42967b05584SSam Clegg } 430a57f1a54SSam Clegg 431a57f1a54SSam Clegg // Update symbol binding, if the existing symbol is weak 432a57f1a54SSam Clegg uint32_t binding = flags & WASM_SYMBOL_BINDING_MASK; 433a57f1a54SSam Clegg if (existing->isWeak() && binding != WASM_SYMBOL_BINDING_WEAK) { 434a57f1a54SSam Clegg existing->flags = (existing->flags & ~WASM_SYMBOL_BINDING_MASK) | binding; 435a57f1a54SSam Clegg } 43667b05584SSam Clegg } 43767b05584SSam Clegg 438bd481277SSam Clegg Symbol *SymbolTable::addUndefinedFunction(StringRef name, 439bd481277SSam Clegg Optional<StringRef> importName, 440bd481277SSam Clegg Optional<StringRef> importModule, 441136d27abSRui Ueyama uint32_t flags, InputFile *file, 442136d27abSRui Ueyama const WasmSignature *sig, 443136d27abSRui Ueyama bool isCalledDirectly) { 444136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addUndefinedFunction: " << name << " [" 445136d27abSRui Ueyama << (sig ? toString(*sig) : "none") 446a57f1a54SSam Clegg << "] IsCalledDirectly:" << isCalledDirectly << " flags=0x" 447a57f1a54SSam Clegg << utohexstr(flags) << "\n"); 448f6f4b98fSSam Clegg assert(flags & WASM_SYMBOL_UNDEFINED); 449c03c9040SRui Ueyama 450136d27abSRui Ueyama Symbol *s; 451136d27abSRui Ueyama bool wasInserted; 452136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 453136d27abSRui Ueyama if (s->traced) 454136d27abSRui Ueyama printTraceSymbolUndefined(name, file); 455c729c1b4SSam Clegg 456136d27abSRui Ueyama auto replaceSym = [&]() { 457136d27abSRui Ueyama replaceSymbol<UndefinedFunction>(s, name, importName, importModule, flags, 458136d27abSRui Ueyama file, sig, isCalledDirectly); 4596540e570SSam Clegg }; 4606540e570SSam Clegg 461136d27abSRui Ueyama if (wasInserted) 462136d27abSRui Ueyama replaceSym(); 463136d27abSRui Ueyama else if (auto *lazy = dyn_cast<LazySymbol>(s)) 464136d27abSRui Ueyama lazy->fetch(); 4656540e570SSam Clegg else { 466136d27abSRui Ueyama auto existingFunction = dyn_cast<FunctionSymbol>(s); 467136d27abSRui Ueyama if (!existingFunction) { 468136d27abSRui Ueyama reportTypeError(s, file, WASM_SYMBOL_TYPE_FUNCTION); 469136d27abSRui Ueyama return s; 4706540e570SSam Clegg } 471136d27abSRui Ueyama if (!existingFunction->signature && sig) 472136d27abSRui Ueyama existingFunction->signature = sig; 473a57f1a54SSam Clegg auto *existingUndefined = dyn_cast<UndefinedFunction>(existingFunction); 474fecfc592SSam Clegg if (isCalledDirectly && !signatureMatches(existingFunction, sig)) { 475a57f1a54SSam Clegg // If the existing undefined functions is not called directly then let 476fecfc592SSam Clegg // this one take precedence. Otherwise the existing function is either 477a57f1a54SSam Clegg // directly called or defined, in which case we need a function variant. 478fecfc592SSam Clegg if (existingUndefined && !existingUndefined->isCalledDirectly) 479136d27abSRui Ueyama replaceSym(); 480fecfc592SSam Clegg else if (getFunctionVariant(s, sig, file, &s)) 481fecfc592SSam Clegg replaceSym(); 482fecfc592SSam Clegg } 48367b05584SSam Clegg if (existingUndefined) 484a57f1a54SSam Clegg setImportAttributes(existingUndefined, importName, importModule, flags, 485a57f1a54SSam Clegg file); 4866540e570SSam Clegg } 487cefbf9acSSam Clegg 488136d27abSRui Ueyama return s; 489c03c9040SRui Ueyama } 490c03c9040SRui Ueyama 491136d27abSRui Ueyama Symbol *SymbolTable::addUndefinedData(StringRef name, uint32_t flags, 492136d27abSRui Ueyama InputFile *file) { 493136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addUndefinedData: " << name << "\n"); 494f6f4b98fSSam Clegg assert(flags & WASM_SYMBOL_UNDEFINED); 49593102974SSam Clegg 496136d27abSRui Ueyama Symbol *s; 497136d27abSRui Ueyama bool wasInserted; 498136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 499136d27abSRui Ueyama if (s->traced) 500136d27abSRui Ueyama printTraceSymbolUndefined(name, file); 501f989a92dSSam Clegg 502136d27abSRui Ueyama if (wasInserted) 503136d27abSRui Ueyama replaceSymbol<UndefinedData>(s, name, flags, file); 504136d27abSRui Ueyama else if (auto *lazy = dyn_cast<LazySymbol>(s)) 505136d27abSRui Ueyama lazy->fetch(); 506136d27abSRui Ueyama else if (s->isDefined()) 507136d27abSRui Ueyama checkDataType(s, file); 508136d27abSRui Ueyama return s; 509c94d393aSSam Clegg } 510c94d393aSSam Clegg 511bd481277SSam Clegg Symbol *SymbolTable::addUndefinedGlobal(StringRef name, 512bd481277SSam Clegg Optional<StringRef> importName, 513bd481277SSam Clegg Optional<StringRef> importModule, 514bd481277SSam Clegg uint32_t flags, InputFile *file, 515136d27abSRui Ueyama const WasmGlobalType *type) { 516136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addUndefinedGlobal: " << name << "\n"); 517f6f4b98fSSam Clegg assert(flags & WASM_SYMBOL_UNDEFINED); 518e3498ec5SRui Ueyama 519136d27abSRui Ueyama Symbol *s; 520136d27abSRui Ueyama bool wasInserted; 521136d27abSRui Ueyama std::tie(s, wasInserted) = insert(name, file); 522136d27abSRui Ueyama if (s->traced) 523136d27abSRui Ueyama printTraceSymbolUndefined(name, file); 524c729c1b4SSam Clegg 525136d27abSRui Ueyama if (wasInserted) 526136d27abSRui Ueyama replaceSymbol<UndefinedGlobal>(s, name, importName, importModule, flags, 527136d27abSRui Ueyama file, type); 528136d27abSRui Ueyama else if (auto *lazy = dyn_cast<LazySymbol>(s)) 529136d27abSRui Ueyama lazy->fetch(); 530136d27abSRui Ueyama else if (s->isDefined()) 531136d27abSRui Ueyama checkGlobalType(s, file, type); 532136d27abSRui Ueyama return s; 533e3498ec5SRui Ueyama } 534e3498ec5SRui Ueyama 535136d27abSRui Ueyama void SymbolTable::addLazy(ArchiveFile *file, const Archive::Symbol *sym) { 536136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "addLazy: " << sym->getName() << "\n"); 537136d27abSRui Ueyama StringRef name = sym->getName(); 538c03c9040SRui Ueyama 539136d27abSRui Ueyama Symbol *s; 540136d27abSRui Ueyama bool wasInserted; 541136d27abSRui Ueyama std::tie(s, wasInserted) = insertName(name); 542c03c9040SRui Ueyama 543136d27abSRui Ueyama if (wasInserted) { 544136d27abSRui Ueyama replaceSymbol<LazySymbol>(s, name, 0, file, *sym); 545c03c9040SRui Ueyama return; 546c03c9040SRui Ueyama } 547c03c9040SRui Ueyama 548136d27abSRui Ueyama if (!s->isUndefined()) 54937b4ee52SSam Clegg return; 55037b4ee52SSam Clegg 55137b4ee52SSam Clegg // The existing symbol is undefined, load a new one from the archive, 5527ae3d335SKazuaki Ishizaki // unless the existing symbol is weak in which case replace the undefined 55337b4ee52SSam Clegg // symbols with a LazySymbol. 554136d27abSRui Ueyama if (s->isWeak()) { 555136d27abSRui Ueyama const WasmSignature *oldSig = nullptr; 55637b4ee52SSam Clegg // In the case of an UndefinedFunction we need to preserve the expected 55737b4ee52SSam Clegg // signature. 558136d27abSRui Ueyama if (auto *f = dyn_cast<UndefinedFunction>(s)) 559136d27abSRui Ueyama oldSig = f->signature; 56037b4ee52SSam Clegg LLVM_DEBUG(dbgs() << "replacing existing weak undefined symbol\n"); 561136d27abSRui Ueyama auto newSym = replaceSymbol<LazySymbol>(s, name, WASM_SYMBOL_BINDING_WEAK, 562136d27abSRui Ueyama file, *sym); 563136d27abSRui Ueyama newSym->signature = oldSig; 56437b4ee52SSam Clegg return; 56537b4ee52SSam Clegg } 56637b4ee52SSam Clegg 567e7245b42SNicola Zaghen LLVM_DEBUG(dbgs() << "replacing existing undefined\n"); 568136d27abSRui Ueyama file->addMember(sym); 569c94d393aSSam Clegg } 570e0f6fcd0SSam Clegg 571136d27abSRui Ueyama bool SymbolTable::addComdat(StringRef name) { 572136d27abSRui Ueyama return comdatGroups.insert(CachedHashStringRef(name)).second; 573e0f6fcd0SSam Clegg } 5741f3f774fSSam Clegg 5756540e570SSam Clegg // The new signature doesn't match. Create a variant to the symbol with the 5766540e570SSam Clegg // signature encoded in the name and return that instead. These symbols are 5776540e570SSam Clegg // then unified later in handleSymbolVariants. 578136d27abSRui Ueyama bool SymbolTable::getFunctionVariant(Symbol* sym, const WasmSignature *sig, 579136d27abSRui Ueyama const InputFile *file, Symbol **out) { 580136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "getFunctionVariant: " << sym->getName() << " -> " 581136d27abSRui Ueyama << " " << toString(*sig) << "\n"); 582136d27abSRui Ueyama Symbol *variant = nullptr; 5836540e570SSam Clegg 5846540e570SSam Clegg // Linear search through symbol variants. Should never be more than two 5856540e570SSam Clegg // or three entries here. 586136d27abSRui Ueyama auto &variants = symVariants[CachedHashStringRef(sym->getName())]; 587136d27abSRui Ueyama if (variants.empty()) 588136d27abSRui Ueyama variants.push_back(sym); 5896540e570SSam Clegg 590136d27abSRui Ueyama for (Symbol* v : variants) { 591136d27abSRui Ueyama if (*v->getSignature() == *sig) { 592136d27abSRui Ueyama variant = v; 5936540e570SSam Clegg break; 5946540e570SSam Clegg } 5956540e570SSam Clegg } 5966540e570SSam Clegg 597136d27abSRui Ueyama bool wasAdded = !variant; 598136d27abSRui Ueyama if (wasAdded) { 5996540e570SSam Clegg // Create a new variant; 6006540e570SSam Clegg LLVM_DEBUG(dbgs() << "added new variant\n"); 601136d27abSRui Ueyama variant = reinterpret_cast<Symbol *>(make<SymbolUnion>()); 602064e9907SSam Clegg variant->isUsedInRegularObj = 603064e9907SSam Clegg !file || file->kind() == InputFile::ObjectKind; 604064e9907SSam Clegg variant->canInline = true; 605064e9907SSam Clegg variant->traced = false; 606064e9907SSam Clegg variant->forceExport = false; 607136d27abSRui Ueyama variants.push_back(variant); 6086540e570SSam Clegg } else { 609136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "variant already exists: " << toString(*variant) << "\n"); 610136d27abSRui Ueyama assert(*variant->getSignature() == *sig); 6116540e570SSam Clegg } 6126540e570SSam Clegg 613136d27abSRui Ueyama *out = variant; 614136d27abSRui Ueyama return wasAdded; 6156540e570SSam Clegg } 6166540e570SSam Clegg 6171f3f774fSSam Clegg // Set a flag for --trace-symbol so that we can print out a log message 6181f3f774fSSam Clegg // if a new symbol with the same name is inserted into the symbol table. 619136d27abSRui Ueyama void SymbolTable::trace(StringRef name) { 620136d27abSRui Ueyama symMap.insert({CachedHashStringRef(name), -1}); 6211f3f774fSSam Clegg } 622230dc11dSSam Clegg 623136d27abSRui Ueyama void SymbolTable::wrap(Symbol *sym, Symbol *real, Symbol *wrap) { 624a5ca34e6SSam Clegg // Swap symbols as instructed by -wrap. 625136d27abSRui Ueyama int &origIdx = symMap[CachedHashStringRef(sym->getName())]; 626136d27abSRui Ueyama int &realIdx= symMap[CachedHashStringRef(real->getName())]; 627136d27abSRui Ueyama int &wrapIdx = symMap[CachedHashStringRef(wrap->getName())]; 628136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "wrap: " << sym->getName() << "\n"); 629a5ca34e6SSam Clegg 630a5ca34e6SSam Clegg // Anyone looking up __real symbols should get the original 631136d27abSRui Ueyama realIdx = origIdx; 632a5ca34e6SSam Clegg // Anyone looking up the original should get the __wrap symbol 633136d27abSRui Ueyama origIdx = wrapIdx; 634a5ca34e6SSam Clegg } 635a5ca34e6SSam Clegg 636136d27abSRui Ueyama static const uint8_t unreachableFn[] = { 637230dc11dSSam Clegg 0x03 /* ULEB length */, 0x00 /* ULEB num locals */, 638230dc11dSSam Clegg 0x00 /* opcode unreachable */, 0x0b /* opcode end */ 639230dc11dSSam Clegg }; 640230dc11dSSam Clegg 641230dc11dSSam Clegg // Replace the given symbol body with an unreachable function. 642230dc11dSSam Clegg // This is used by handleWeakUndefines in order to generate a callable 6436540e570SSam Clegg // equivalent of an undefined function and also handleSymbolVariants for 6446540e570SSam Clegg // undefined functions that don't match the signature of the definition. 645136d27abSRui Ueyama InputFunction *SymbolTable::replaceWithUnreachable(Symbol *sym, 646136d27abSRui Ueyama const WasmSignature &sig, 647136d27abSRui Ueyama StringRef debugName) { 648136d27abSRui Ueyama auto *func = make<SyntheticFunction>(sig, sym->getName(), debugName); 649136d27abSRui Ueyama func->setBody(unreachableFn); 650136d27abSRui Ueyama syntheticFunctions.emplace_back(func); 6514b8e2d8eSSam Clegg // Mark new symbols as local. For relocatable output we don't want them 6524b8e2d8eSSam Clegg // to be exported outside the object file. 6534b8e2d8eSSam Clegg replaceSymbol<DefinedFunction>(sym, debugName, WASM_SYMBOL_BINDING_LOCAL, 6544b8e2d8eSSam Clegg nullptr, func); 655136d27abSRui Ueyama return func; 656230dc11dSSam Clegg } 657230dc11dSSam Clegg 658230dc11dSSam Clegg // For weak undefined functions, there may be "call" instructions that reference 659230dc11dSSam Clegg // the symbol. In this case, we need to synthesise a dummy/stub function that 660230dc11dSSam Clegg // will abort at runtime, so that relocations can still provided an operand to 661230dc11dSSam Clegg // the call instruction that passes Wasm validation. 662230dc11dSSam Clegg void SymbolTable::handleWeakUndefines() { 663136d27abSRui Ueyama for (Symbol *sym : getSymbols()) { 664136d27abSRui Ueyama if (!sym->isUndefWeak()) 665230dc11dSSam Clegg continue; 666230dc11dSSam Clegg 667136d27abSRui Ueyama const WasmSignature *sig = sym->getSignature(); 668136d27abSRui Ueyama if (!sig) { 669230dc11dSSam Clegg // It is possible for undefined functions not to have a signature (eg. if 670230dc11dSSam Clegg // added via "--undefined"), but weak undefined ones do have a signature. 6716540e570SSam Clegg // Lazy symbols may not be functions and therefore Sig can still be null 6727ae3d335SKazuaki Ishizaki // in some circumstance. 673136d27abSRui Ueyama assert(!isa<FunctionSymbol>(sym)); 674230dc11dSSam Clegg continue; 6756540e570SSam Clegg } 676230dc11dSSam Clegg 677230dc11dSSam Clegg // Add a synthetic dummy for weak undefined functions. These dummies will 678230dc11dSSam Clegg // be GC'd if not used as the target of any "call" instructions. 679136d27abSRui Ueyama StringRef debugName = saver.save("undefined:" + toString(*sym)); 680136d27abSRui Ueyama InputFunction* func = replaceWithUnreachable(sym, *sig, debugName); 681230dc11dSSam Clegg // Ensure it compares equal to the null pointer, and so that table relocs 682230dc11dSSam Clegg // don't pull in the stub body (only call-operand relocs should do that). 683136d27abSRui Ueyama func->setTableIndex(0); 684230dc11dSSam Clegg // Hide our dummy to prevent export. 685136d27abSRui Ueyama sym->setHidden(true); 686230dc11dSSam Clegg } 687230dc11dSSam Clegg } 6886540e570SSam Clegg 689136d27abSRui Ueyama static void reportFunctionSignatureMismatch(StringRef symName, 690136d27abSRui Ueyama FunctionSymbol *a, 691136d27abSRui Ueyama FunctionSymbol *b, bool isError) { 692136d27abSRui Ueyama std::string msg = ("function signature mismatch: " + symName + 693136d27abSRui Ueyama "\n>>> defined as " + toString(*a->signature) + " in " + 694136d27abSRui Ueyama toString(a->getFile()) + "\n>>> defined as " + 695136d27abSRui Ueyama toString(*b->signature) + " in " + toString(b->getFile())) 6966540e570SSam Clegg .str(); 697136d27abSRui Ueyama if (isError) 6986540e570SSam Clegg error(msg); 6996540e570SSam Clegg else 7006540e570SSam Clegg warn(msg); 7016540e570SSam Clegg } 7026540e570SSam Clegg 7036540e570SSam Clegg // Remove any variant symbols that were created due to function signature 7046540e570SSam Clegg // mismatches. 7056540e570SSam Clegg void SymbolTable::handleSymbolVariants() { 706136d27abSRui Ueyama for (auto pair : symVariants) { 7076540e570SSam Clegg // Push the initial symbol onto the list of variants. 708136d27abSRui Ueyama StringRef symName = pair.first.val(); 709136d27abSRui Ueyama std::vector<Symbol *> &variants = pair.second; 7106540e570SSam Clegg 7116540e570SSam Clegg #ifndef NDEBUG 712136d27abSRui Ueyama LLVM_DEBUG(dbgs() << "symbol with (" << variants.size() 713136d27abSRui Ueyama << ") variants: " << symName << "\n"); 714136d27abSRui Ueyama for (auto *s: variants) { 715136d27abSRui Ueyama auto *f = cast<FunctionSymbol>(s); 716136d27abSRui Ueyama LLVM_DEBUG(dbgs() << " variant: " + f->getName() << " " 717136d27abSRui Ueyama << toString(*f->signature) << "\n"); 7186540e570SSam Clegg } 7196540e570SSam Clegg #endif 7206540e570SSam Clegg 7216540e570SSam Clegg // Find the one definition. 722136d27abSRui Ueyama DefinedFunction *defined = nullptr; 723136d27abSRui Ueyama for (auto *symbol : variants) { 724136d27abSRui Ueyama if (auto f = dyn_cast<DefinedFunction>(symbol)) { 725136d27abSRui Ueyama defined = f; 7266540e570SSam Clegg break; 7276540e570SSam Clegg } 7286540e570SSam Clegg } 7296540e570SSam Clegg 7306540e570SSam Clegg // If there are no definitions, and the undefined symbols disagree on 7316540e570SSam Clegg // the signature, there is not we can do since we don't know which one 7326540e570SSam Clegg // to use as the signature on the import. 733136d27abSRui Ueyama if (!defined) { 734136d27abSRui Ueyama reportFunctionSignatureMismatch(symName, 735136d27abSRui Ueyama cast<FunctionSymbol>(variants[0]), 736136d27abSRui Ueyama cast<FunctionSymbol>(variants[1]), true); 7376540e570SSam Clegg return; 7386540e570SSam Clegg } 7396540e570SSam Clegg 740136d27abSRui Ueyama for (auto *symbol : variants) { 741136d27abSRui Ueyama if (symbol != defined) { 742136d27abSRui Ueyama auto *f = cast<FunctionSymbol>(symbol); 743136d27abSRui Ueyama reportFunctionSignatureMismatch(symName, f, defined, false); 74451b521c0SSam Clegg StringRef debugName = saver.save("signature_mismatch:" + toString(*f)); 745136d27abSRui Ueyama replaceWithUnreachable(f, *f->signature, debugName); 7466540e570SSam Clegg } 7476540e570SSam Clegg } 7486540e570SSam Clegg } 7496540e570SSam Clegg } 75033c59abfSFangrui Song 75133c59abfSFangrui Song } // namespace wasm 75233c59abfSFangrui Song } // namespace lld 753