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 
12c94d393aSSam Clegg #include "Config.h"
135fa274beSSam Clegg #include "InputChunks.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 
32c94d393aSSam Clegg   if (auto *F = dyn_cast<ObjFile>(File))
33c94d393aSSam Clegg     ObjectFiles.push_back(F);
34c94d393aSSam Clegg }
35c94d393aSSam Clegg 
36c94d393aSSam Clegg void SymbolTable::reportRemainingUndefines() {
377d67dd11SRui Ueyama   SetVector<Symbol *> Undefs;
3874fe0ba1SSam Clegg   for (Symbol *Sym : SymVector) {
39c94d393aSSam Clegg     if (Sym->isUndefined() && !Sym->isWeak() &&
40c94d393aSSam Clegg         Config->AllowUndefinedSymbols.count(Sym->getName()) == 0) {
41c94d393aSSam Clegg       Undefs.insert(Sym);
42c94d393aSSam Clegg     }
43c94d393aSSam Clegg   }
44c94d393aSSam Clegg 
45c94d393aSSam Clegg   if (Undefs.empty())
46c94d393aSSam Clegg     return;
47c94d393aSSam Clegg 
48c94d393aSSam Clegg   for (ObjFile *File : ObjectFiles)
49c94d393aSSam Clegg     for (Symbol *Sym : File->getSymbols())
50c94d393aSSam Clegg       if (Undefs.count(Sym))
51c94d393aSSam Clegg         error(toString(File) + ": undefined symbol: " + toString(*Sym));
52c94d393aSSam Clegg 
53c94d393aSSam Clegg   for (Symbol *Sym : Undefs)
54c94d393aSSam Clegg     if (!Sym->getFile())
55c94d393aSSam Clegg       error("undefined symbol: " + toString(*Sym));
56c94d393aSSam Clegg }
57c94d393aSSam Clegg 
58c94d393aSSam Clegg Symbol *SymbolTable::find(StringRef Name) {
59a80d94d5SSam Clegg   auto It = SymMap.find(CachedHashStringRef(Name));
60a80d94d5SSam Clegg   if (It == SymMap.end())
61c94d393aSSam Clegg     return nullptr;
62c94d393aSSam Clegg   return It->second;
63c94d393aSSam Clegg }
64c94d393aSSam Clegg 
65c94d393aSSam Clegg std::pair<Symbol *, bool> SymbolTable::insert(StringRef Name) {
66a80d94d5SSam Clegg   Symbol *&Sym = SymMap[CachedHashStringRef(Name)];
67c94d393aSSam Clegg   if (Sym)
68c94d393aSSam Clegg     return {Sym, false};
69dfb0b2c7SSam Clegg   Sym = reinterpret_cast<Symbol *>(make<SymbolUnion>());
7074fe0ba1SSam Clegg   SymVector.emplace_back(Sym);
71c94d393aSSam Clegg   return {Sym, true};
72c94d393aSSam Clegg }
73c94d393aSSam Clegg 
74c94d393aSSam Clegg void SymbolTable::reportDuplicate(Symbol *Existing, InputFile *NewFile) {
75c94d393aSSam Clegg   error("duplicate symbol: " + toString(*Existing) + "\n>>> defined in " +
76c94d393aSSam Clegg         toString(Existing->getFile()) + "\n>>> defined in " +
77e2ed092eSSam Clegg         toString(NewFile));
78c94d393aSSam Clegg }
79c94d393aSSam Clegg 
80b8621596SSam Clegg // Check the type of new symbol matches that of the symbol is replacing.
81b8621596SSam Clegg // For functions this can also involve verifying that the signatures match.
82b8621596SSam Clegg static void checkSymbolTypes(const Symbol &Existing, const InputFile &F,
83dfb0b2c7SSam Clegg                              bool NewIsFunction, const WasmSignature *NewSig) {
84b8621596SSam Clegg   if (Existing.isLazy())
85c94d393aSSam Clegg     return;
86c94d393aSSam Clegg 
87b8621596SSam Clegg   // First check the symbol types match (i.e. either both are function
88b8621596SSam Clegg   // symbols or both are data symbols).
89caca8d5dSSam Clegg   if (isa<FunctionSymbol>(Existing) != NewIsFunction) {
9020db381bSSam Clegg     error("symbol type mismatch: " + Existing.getName() + "\n>>> defined as " +
91caca8d5dSSam Clegg           (isa<FunctionSymbol>(Existing) ? "Function" : "Global") + " in " +
92b8621596SSam Clegg           toString(Existing.getFile()) + "\n>>> defined as " +
93b8621596SSam Clegg           (NewIsFunction ? "Function" : "Global") + " in " + F.getName());
94b8621596SSam Clegg     return;
95b8621596SSam Clegg   }
96b8621596SSam Clegg 
97b8621596SSam Clegg   // For function symbols, optionally check the function signature matches too.
98dfb0b2c7SSam Clegg   auto *ExistingFunc = dyn_cast<FunctionSymbol>(&Existing);
99dfb0b2c7SSam Clegg   if (!ExistingFunc || !Config->CheckSignatures)
100c94d393aSSam Clegg     return;
101dfb0b2c7SSam Clegg 
1023f8db988SSam Clegg   const WasmSignature *OldSig = ExistingFunc->getFunctionType();
1033f8db988SSam Clegg 
10431de2f0cSSam Clegg   // Skip the signature check if the existing function has no signature (e.g.
10531de2f0cSSam Clegg   // if it is an undefined symbol generated by --undefined command line flag).
1063f8db988SSam Clegg   if (OldSig == nullptr)
10731de2f0cSSam Clegg     return;
108c94d393aSSam Clegg 
109dfb0b2c7SSam Clegg   DEBUG(dbgs() << "checkSymbolTypes: " << ExistingFunc->getName() << "\n");
110b8621596SSam Clegg   assert(NewSig);
111b8621596SSam Clegg 
1123f8db988SSam Clegg   if (*NewSig == *OldSig)
113b8621596SSam Clegg     return;
114b8621596SSam Clegg 
115dfb0b2c7SSam Clegg   error("function signature mismatch: " + ExistingFunc->getName() +
1163f8db988SSam Clegg         "\n>>> defined as " + toString(*OldSig) + " in " +
117dfb0b2c7SSam Clegg         toString(ExistingFunc->getFile()) + "\n>>> defined as " +
118dfb0b2c7SSam Clegg         toString(*NewSig) + " in " + F.getName());
119c94d393aSSam Clegg }
120c94d393aSSam Clegg 
12124b3dcd4SSam Clegg static void checkSymbolTypes(const Symbol &Existing, const InputFile &F,
122dfb0b2c7SSam Clegg                              bool IsFunction, const InputChunk *Chunk) {
12324b3dcd4SSam Clegg   const WasmSignature *Sig = nullptr;
12424b3dcd4SSam Clegg   if (auto *F = dyn_cast_or_null<InputFunction>(Chunk))
12524b3dcd4SSam Clegg     Sig = &F->Signature;
126dfb0b2c7SSam Clegg   return checkSymbolTypes(Existing, F, IsFunction, Sig);
12724b3dcd4SSam Clegg }
12824b3dcd4SSam Clegg 
129dfb0b2c7SSam Clegg DefinedFunction *SymbolTable::addSyntheticFunction(StringRef Name,
13050686856SSam Clegg                                                    const WasmSignature *Type,
13150686856SSam Clegg                                                    uint32_t Flags) {
132a189230bSSam Clegg   DEBUG(dbgs() << "addSyntheticFunction: " << Name << "\n");
13350686856SSam Clegg   Symbol *S;
13450686856SSam Clegg   bool WasInserted;
13550686856SSam Clegg   std::tie(S, WasInserted) = insert(Name);
136a189230bSSam Clegg   assert(WasInserted);
137dfb0b2c7SSam Clegg   return replaceSymbol<DefinedFunction>(S, Name, Flags, Type);
13850686856SSam Clegg }
13950686856SSam Clegg 
140dfb0b2c7SSam Clegg DefinedGlobal *SymbolTable::addSyntheticGlobal(StringRef Name, uint32_t Flags) {
141a189230bSSam Clegg   DEBUG(dbgs() << "addSyntheticGlobal: " << Name << "\n");
142c94d393aSSam Clegg   Symbol *S;
143c94d393aSSam Clegg   bool WasInserted;
144c94d393aSSam Clegg   std::tie(S, WasInserted) = insert(Name);
145a189230bSSam Clegg   assert(WasInserted);
146dfb0b2c7SSam Clegg   return replaceSymbol<DefinedGlobal>(S, Name, Flags);
147c94d393aSSam Clegg }
148c94d393aSSam Clegg 
149dfb0b2c7SSam Clegg Symbol *SymbolTable::addDefined(bool IsFunction, StringRef Name, uint32_t Flags,
150dfb0b2c7SSam Clegg                                 InputFile *F, InputChunk *Chunk,
151fc50c623SSam Clegg                                 uint32_t Address) {
152dfb0b2c7SSam Clegg   if (IsFunction)
153dfb0b2c7SSam Clegg     DEBUG(dbgs() << "addDefined: func:" << Name << "\n");
154dfb0b2c7SSam Clegg   else
155dfb0b2c7SSam Clegg     DEBUG(dbgs() << "addDefined: global:" << Name << " addr:" << Address
156dfb0b2c7SSam Clegg                  << "\n");
157c94d393aSSam Clegg   Symbol *S;
158c94d393aSSam Clegg   bool WasInserted;
159dfb0b2c7SSam Clegg   bool Replace = false;
160dfb0b2c7SSam Clegg   bool CheckTypes = false;
161c94d393aSSam Clegg 
16220db381bSSam Clegg   std::tie(S, WasInserted) = insert(Name);
163c94d393aSSam Clegg   if (WasInserted) {
164dfb0b2c7SSam Clegg     Replace = true;
165b8621596SSam Clegg   } else if (S->isLazy()) {
166dfb0b2c7SSam Clegg     // Existing symbol is lazy. Replace it without checking types since
167b8621596SSam Clegg     // lazy symbols don't have any type information.
16820db381bSSam Clegg     DEBUG(dbgs() << "replacing existing lazy symbol: " << Name << "\n");
169dfb0b2c7SSam Clegg     Replace = true;
170c94d393aSSam Clegg   } else if (!S->isDefined()) {
171dfb0b2c7SSam Clegg     // Existing symbol is undefined: replace it, while check types.
17220db381bSSam Clegg     DEBUG(dbgs() << "resolving existing undefined symbol: " << Name << "\n");
173dfb0b2c7SSam Clegg     Replace = true;
174dfb0b2c7SSam Clegg     CheckTypes = true;
17520db381bSSam Clegg   } else if ((Flags & WASM_SYMBOL_BINDING_MASK) == WASM_SYMBOL_BINDING_WEAK) {
176c94d393aSSam Clegg     // the new symbol is weak we can ignore it
17749ed9262SSam Clegg     DEBUG(dbgs() << "existing symbol takes precedence\n");
178*c7e885b5SSam Clegg     CheckTypes = true;
179c94d393aSSam Clegg   } else if (S->isWeak()) {
180dfb0b2c7SSam Clegg     // the existing symbol is, so we replace it
181c94d393aSSam Clegg     DEBUG(dbgs() << "replacing existing weak symbol\n");
182dfb0b2c7SSam Clegg     Replace = true;
183dfb0b2c7SSam Clegg     CheckTypes = true;
184c94d393aSSam Clegg   } else {
18549ed9262SSam Clegg     // neither symbol is week. They conflict.
186c94d393aSSam Clegg     reportDuplicate(S, F);
187c94d393aSSam Clegg   }
188c94d393aSSam Clegg 
189dfb0b2c7SSam Clegg   if (Replace) {
190dfb0b2c7SSam Clegg     if (CheckTypes)
191dfb0b2c7SSam Clegg       checkSymbolTypes(*S, *F, IsFunction, Chunk);
192dfb0b2c7SSam Clegg     if (IsFunction)
1933f8db988SSam Clegg       replaceSymbol<DefinedFunction>(S, Name, Flags, F,
1943f8db988SSam Clegg                                      cast<InputFunction>(Chunk));
195dfb0b2c7SSam Clegg     else
1963f8db988SSam Clegg       replaceSymbol<DefinedGlobal>(S, Name, Flags, F, cast<InputSegment>(Chunk),
1973f8db988SSam Clegg                                    Address);
198b8621596SSam Clegg   }
199c94d393aSSam Clegg   return S;
200c94d393aSSam Clegg }
201c94d393aSSam Clegg 
20220db381bSSam Clegg Symbol *SymbolTable::addUndefined(StringRef Name, Symbol::Kind Kind,
20320db381bSSam Clegg                                   uint32_t Flags, InputFile *F,
2048d146bbcSSam Clegg                                   const WasmSignature *Type) {
20520db381bSSam Clegg   DEBUG(dbgs() << "addUndefined: " << Name << "\n");
206c94d393aSSam Clegg   Symbol *S;
207c94d393aSSam Clegg   bool WasInserted;
20820db381bSSam Clegg   std::tie(S, WasInserted) = insert(Name);
209dfb0b2c7SSam Clegg   bool IsFunction = Kind == Symbol::UndefinedFunctionKind;
210c94d393aSSam Clegg   if (WasInserted) {
211dfb0b2c7SSam Clegg     if (IsFunction)
212dfb0b2c7SSam Clegg       replaceSymbol<UndefinedFunction>(S, Name, Flags, F, Type);
213dfb0b2c7SSam Clegg     else
214dfb0b2c7SSam Clegg       replaceSymbol<UndefinedGlobal>(S, Name, Flags, F);
215dfb0b2c7SSam Clegg   } else if (auto *LazySym = dyn_cast<LazySymbol>(S)) {
216c94d393aSSam Clegg     DEBUG(dbgs() << "resolved by existing lazy\n");
217dfb0b2c7SSam Clegg     auto *AF = cast<ArchiveFile>(LazySym->getFile());
218dfb0b2c7SSam Clegg     AF->addMember(&LazySym->getArchiveSymbol());
219c94d393aSSam Clegg   } else if (S->isDefined()) {
220c94d393aSSam Clegg     DEBUG(dbgs() << "resolved by existing\n");
221dfb0b2c7SSam Clegg     checkSymbolTypes(*S, *F, IsFunction, Type);
222c94d393aSSam Clegg   }
223c94d393aSSam Clegg   return S;
224c94d393aSSam Clegg }
225c94d393aSSam Clegg 
226c94d393aSSam Clegg void SymbolTable::addLazy(ArchiveFile *F, const Archive::Symbol *Sym) {
227a681a116SSam Clegg   DEBUG(dbgs() << "addLazy: " << Sym->getName() << "\n");
228c94d393aSSam Clegg   StringRef Name = Sym->getName();
229c94d393aSSam Clegg   Symbol *S;
230c94d393aSSam Clegg   bool WasInserted;
231c94d393aSSam Clegg   std::tie(S, WasInserted) = insert(Name);
232c94d393aSSam Clegg   if (WasInserted) {
233dfb0b2c7SSam Clegg     replaceSymbol<LazySymbol>(S, Name, F, *Sym);
234c94d393aSSam Clegg   } else if (S->isUndefined()) {
235c94d393aSSam Clegg     // There is an existing undefined symbol.  The can load from the
236c94d393aSSam Clegg     // archive.
237c94d393aSSam Clegg     DEBUG(dbgs() << "replacing existing undefined\n");
238c94d393aSSam Clegg     F->addMember(Sym);
239c94d393aSSam Clegg   }
240c94d393aSSam Clegg }
241e0f6fcd0SSam Clegg 
242e0f6fcd0SSam Clegg bool SymbolTable::addComdat(StringRef Name, ObjFile *F) {
243e0f6fcd0SSam Clegg   DEBUG(dbgs() << "addComdat: " << Name << "\n");
244e0f6fcd0SSam Clegg   ObjFile *&File = ComdatMap[CachedHashStringRef(Name)];
245e0f6fcd0SSam Clegg   if (File) {
246e0f6fcd0SSam Clegg     DEBUG(dbgs() << "COMDAT already defined\n");
247e0f6fcd0SSam Clegg     return false;
248e0f6fcd0SSam Clegg   }
249e0f6fcd0SSam Clegg   File = F;
250e0f6fcd0SSam Clegg   return true;
251e0f6fcd0SSam Clegg }
252e0f6fcd0SSam Clegg 
253e0f6fcd0SSam Clegg ObjFile *SymbolTable::findComdat(StringRef Name) const {
254e0f6fcd0SSam Clegg   auto It = ComdatMap.find(CachedHashStringRef(Name));
255e0f6fcd0SSam Clegg   return It == ComdatMap.end() ? nullptr : It->second;
256e0f6fcd0SSam Clegg }
257