1 //===- SymbolTable.cpp ----------------------------------------------------===//
2 //
3 //                             The LLVM Linker
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 
10 #include "SymbolTable.h"
11 #include "Config.h"
12 #include "Driver.h"
13 #include "LTO.h"
14 #include "PDB.h"
15 #include "Symbols.h"
16 #include "lld/Common/ErrorHandler.h"
17 #include "lld/Common/Memory.h"
18 #include "lld/Common/Timer.h"
19 #include "llvm/IR/LLVMContext.h"
20 #include "llvm/Support/Debug.h"
21 #include "llvm/Support/raw_ostream.h"
22 #include <utility>
23 
24 using namespace llvm;
25 
26 namespace lld {
27 namespace coff {
28 
29 static Timer LTOTimer("LTO", Timer::root());
30 
31 SymbolTable *Symtab;
32 
33 void SymbolTable::addFile(InputFile *File) {
34   log("Reading " + toString(File));
35   File->parse();
36 
37   MachineTypes MT = File->getMachineType();
38   if (Config->Machine == IMAGE_FILE_MACHINE_UNKNOWN) {
39     Config->Machine = MT;
40   } else if (MT != IMAGE_FILE_MACHINE_UNKNOWN && Config->Machine != MT) {
41     error(toString(File) + ": machine type " + machineToStr(MT) +
42           " conflicts with " + machineToStr(Config->Machine));
43     return;
44   }
45 
46   if (auto *F = dyn_cast<ObjFile>(File)) {
47     ObjFile::Instances.push_back(F);
48   } else if (auto *F = dyn_cast<BitcodeFile>(File)) {
49     BitcodeFile::Instances.push_back(F);
50   } else if (auto *F = dyn_cast<ImportFile>(File)) {
51     ImportFile::Instances.push_back(F);
52   }
53 
54   StringRef S = File->getDirectives();
55   if (S.empty())
56     return;
57 
58   log("Directives: " + toString(File) + ": " + S);
59   Driver->parseDirectives(S);
60 }
61 
62 static void errorOrWarn(const Twine &S) {
63   if (Config->Force)
64     warn(S);
65   else
66     error(S);
67 }
68 
69 // Returns the name of the symbol in SC whose value is <= Addr that is closest
70 // to Addr. This is generally the name of the global variable or function whose
71 // definition contains Addr.
72 static StringRef getSymbolName(SectionChunk *SC, uint32_t Addr) {
73   DefinedRegular *Candidate = nullptr;
74 
75   for (Symbol *S : SC->File->getSymbols()) {
76     auto *D = dyn_cast_or_null<DefinedRegular>(S);
77     if (!D || D->getChunk() != SC || D->getValue() > Addr ||
78         (Candidate && D->getValue() < Candidate->getValue()))
79       continue;
80 
81     Candidate = D;
82   }
83 
84   if (!Candidate)
85     return "";
86   return Candidate->getName();
87 }
88 
89 static std::string getSymbolLocations(ObjFile *File, uint32_t SymIndex) {
90   struct Location {
91     StringRef SymName;
92     std::pair<StringRef, uint32_t> FileLine;
93   };
94   std::vector<Location> Locations;
95 
96   for (Chunk *C : File->getChunks()) {
97     auto *SC = dyn_cast<SectionChunk>(C);
98     if (!SC)
99       continue;
100     for (const coff_relocation &R : SC->Relocs) {
101       if (R.SymbolTableIndex != SymIndex)
102         continue;
103       std::pair<StringRef, uint32_t> FileLine =
104           getFileLine(SC, R.VirtualAddress);
105       StringRef SymName = getSymbolName(SC, R.VirtualAddress);
106       if (!FileLine.first.empty() || !SymName.empty())
107         Locations.push_back({SymName, FileLine});
108     }
109   }
110 
111   if (Locations.empty())
112     return "\n>>> referenced by " + toString(File);
113 
114   std::string Out;
115   llvm::raw_string_ostream OS(Out);
116   for (Location Loc : Locations) {
117     OS << "\n>>> referenced by ";
118     if (!Loc.FileLine.first.empty())
119       OS << Loc.FileLine.first << ":" << Loc.FileLine.second
120          << "\n>>>               ";
121     OS << toString(File);
122     if (!Loc.SymName.empty())
123       OS << ":(" << Loc.SymName << ')';
124   }
125   return OS.str();
126 }
127 
128 void SymbolTable::reportRemainingUndefines() {
129   SmallPtrSet<Symbol *, 8> Undefs;
130   DenseMap<Symbol *, Symbol *> LocalImports;
131 
132   for (auto &I : SymMap) {
133     Symbol *Sym = I.second;
134     auto *Undef = dyn_cast<Undefined>(Sym);
135     if (!Undef)
136       continue;
137     if (!Sym->IsUsedInRegularObj)
138       continue;
139 
140     StringRef Name = Undef->getName();
141 
142     // A weak alias may have been resolved, so check for that.
143     if (Defined *D = Undef->getWeakAlias()) {
144       // We want to replace Sym with D. However, we can't just blindly
145       // copy sizeof(SymbolUnion) bytes from D to Sym because D may be an
146       // internal symbol, and internal symbols are stored as "unparented"
147       // Symbols. For that reason we need to check which type of symbol we
148       // are dealing with and copy the correct number of bytes.
149       if (isa<DefinedRegular>(D))
150         memcpy(Sym, D, sizeof(DefinedRegular));
151       else if (isa<DefinedAbsolute>(D))
152         memcpy(Sym, D, sizeof(DefinedAbsolute));
153       else
154         memcpy(Sym, D, sizeof(SymbolUnion));
155       continue;
156     }
157 
158     // If we can resolve a symbol by removing __imp_ prefix, do that.
159     // This odd rule is for compatibility with MSVC linker.
160     if (Name.startswith("__imp_")) {
161       Symbol *Imp = find(Name.substr(strlen("__imp_")));
162       if (Imp && isa<Defined>(Imp)) {
163         auto *D = cast<Defined>(Imp);
164         replaceSymbol<DefinedLocalImport>(Sym, Name, D);
165         LocalImportChunks.push_back(cast<DefinedLocalImport>(Sym)->getChunk());
166         LocalImports[Sym] = D;
167         continue;
168       }
169     }
170 
171     // Remaining undefined symbols are not fatal if /force is specified.
172     // They are replaced with dummy defined symbols.
173     if (Config->Force)
174       replaceSymbol<DefinedAbsolute>(Sym, Name, 0);
175     Undefs.insert(Sym);
176   }
177 
178   if (Undefs.empty() && LocalImports.empty())
179     return;
180 
181   for (Symbol *B : Config->GCRoot) {
182     if (Undefs.count(B))
183       errorOrWarn("<root>: undefined symbol: " + B->getName());
184     if (Config->WarnLocallyDefinedImported)
185       if (Symbol *Imp = LocalImports.lookup(B))
186         warn("<root>: locally defined symbol imported: " + Imp->getName() +
187              " (defined in " + toString(Imp->getFile()) + ") [LNK4217]");
188   }
189 
190   for (ObjFile *File : ObjFile::Instances) {
191     size_t SymIndex = (size_t)-1;
192     for (Symbol *Sym : File->getSymbols()) {
193       ++SymIndex;
194       if (!Sym)
195         continue;
196       if (Undefs.count(Sym))
197         errorOrWarn("undefined symbol: " + Sym->getName() +
198                     getSymbolLocations(File, SymIndex));
199       if (Config->WarnLocallyDefinedImported)
200         if (Symbol *Imp = LocalImports.lookup(Sym))
201           warn(toString(File) + ": locally defined symbol imported: " +
202                Imp->getName() + " (defined in " + toString(Imp->getFile()) +
203                ") [LNK4217]");
204     }
205   }
206 }
207 
208 std::pair<Symbol *, bool> SymbolTable::insert(StringRef Name, InputFile *File) {
209   bool Inserted = false;
210   Symbol *&Sym = SymMap[CachedHashStringRef(Name)];
211   if (!Sym) {
212     Sym = reinterpret_cast<Symbol *>(make<SymbolUnion>());
213     Sym->IsUsedInRegularObj = false;
214     Sym->PendingArchiveLoad = false;
215     Inserted = true;
216   }
217   if (!File || !isa<BitcodeFile>(File))
218     Sym->IsUsedInRegularObj = true;
219   return {Sym, Inserted};
220 }
221 
222 Symbol *SymbolTable::addUndefined(StringRef Name, InputFile *F,
223                                   bool IsWeakAlias) {
224   Symbol *S;
225   bool WasInserted;
226   std::tie(S, WasInserted) = insert(Name, F);
227   if (WasInserted || (isa<Lazy>(S) && IsWeakAlias)) {
228     replaceSymbol<Undefined>(S, Name);
229     return S;
230   }
231   if (auto *L = dyn_cast<Lazy>(S)) {
232     if (!S->PendingArchiveLoad) {
233       S->PendingArchiveLoad = true;
234       L->File->addMember(&L->Sym);
235     }
236   }
237   return S;
238 }
239 
240 void SymbolTable::addLazy(ArchiveFile *F, const Archive::Symbol Sym) {
241   StringRef Name = Sym.getName();
242   Symbol *S;
243   bool WasInserted;
244   std::tie(S, WasInserted) = insert(Name, nullptr);
245   if (WasInserted) {
246     replaceSymbol<Lazy>(S, F, Sym);
247     return;
248   }
249   auto *U = dyn_cast<Undefined>(S);
250   if (!U || U->WeakAlias || S->PendingArchiveLoad)
251     return;
252   S->PendingArchiveLoad = true;
253   F->addMember(&Sym);
254 }
255 
256 void SymbolTable::reportDuplicate(Symbol *Existing, InputFile *NewFile) {
257   error("duplicate symbol: " + toString(*Existing) + " in " +
258         toString(Existing->getFile()) + " and in " + toString(NewFile));
259 }
260 
261 Symbol *SymbolTable::addAbsolute(StringRef N, COFFSymbolRef Sym) {
262   Symbol *S;
263   bool WasInserted;
264   std::tie(S, WasInserted) = insert(N, nullptr);
265   S->IsUsedInRegularObj = true;
266   if (WasInserted || isa<Undefined>(S) || isa<Lazy>(S))
267     replaceSymbol<DefinedAbsolute>(S, N, Sym);
268   else if (!isa<DefinedCOFF>(S))
269     reportDuplicate(S, nullptr);
270   return S;
271 }
272 
273 Symbol *SymbolTable::addAbsolute(StringRef N, uint64_t VA) {
274   Symbol *S;
275   bool WasInserted;
276   std::tie(S, WasInserted) = insert(N, nullptr);
277   S->IsUsedInRegularObj = true;
278   if (WasInserted || isa<Undefined>(S) || isa<Lazy>(S))
279     replaceSymbol<DefinedAbsolute>(S, N, VA);
280   else if (!isa<DefinedCOFF>(S))
281     reportDuplicate(S, nullptr);
282   return S;
283 }
284 
285 Symbol *SymbolTable::addSynthetic(StringRef N, Chunk *C) {
286   Symbol *S;
287   bool WasInserted;
288   std::tie(S, WasInserted) = insert(N, nullptr);
289   S->IsUsedInRegularObj = true;
290   if (WasInserted || isa<Undefined>(S) || isa<Lazy>(S))
291     replaceSymbol<DefinedSynthetic>(S, N, C);
292   else if (!isa<DefinedCOFF>(S))
293     reportDuplicate(S, nullptr);
294   return S;
295 }
296 
297 Symbol *SymbolTable::addRegular(InputFile *F, StringRef N,
298                                 const coff_symbol_generic *Sym,
299                                 SectionChunk *C) {
300   Symbol *S;
301   bool WasInserted;
302   std::tie(S, WasInserted) = insert(N, F);
303   if (WasInserted || !isa<DefinedRegular>(S))
304     replaceSymbol<DefinedRegular>(S, F, N, /*IsCOMDAT*/ false,
305                                   /*IsExternal*/ true, Sym, C);
306   else
307     reportDuplicate(S, F);
308   return S;
309 }
310 
311 std::pair<Symbol *, bool>
312 SymbolTable::addComdat(InputFile *F, StringRef N,
313                        const coff_symbol_generic *Sym) {
314   Symbol *S;
315   bool WasInserted;
316   std::tie(S, WasInserted) = insert(N, F);
317   if (WasInserted || !isa<DefinedRegular>(S)) {
318     replaceSymbol<DefinedRegular>(S, F, N, /*IsCOMDAT*/ true,
319                                   /*IsExternal*/ true, Sym, nullptr);
320     return {S, true};
321   }
322   if (!cast<DefinedRegular>(S)->isCOMDAT())
323     reportDuplicate(S, F);
324   return {S, false};
325 }
326 
327 Symbol *SymbolTable::addCommon(InputFile *F, StringRef N, uint64_t Size,
328                                const coff_symbol_generic *Sym, CommonChunk *C) {
329   Symbol *S;
330   bool WasInserted;
331   std::tie(S, WasInserted) = insert(N, F);
332   if (WasInserted || !isa<DefinedCOFF>(S))
333     replaceSymbol<DefinedCommon>(S, F, N, Size, Sym, C);
334   else if (auto *DC = dyn_cast<DefinedCommon>(S))
335     if (Size > DC->getSize())
336       replaceSymbol<DefinedCommon>(S, F, N, Size, Sym, C);
337   return S;
338 }
339 
340 Symbol *SymbolTable::addImportData(StringRef N, ImportFile *F) {
341   Symbol *S;
342   bool WasInserted;
343   std::tie(S, WasInserted) = insert(N, nullptr);
344   S->IsUsedInRegularObj = true;
345   if (WasInserted || isa<Undefined>(S) || isa<Lazy>(S)) {
346     replaceSymbol<DefinedImportData>(S, N, F);
347     return S;
348   }
349 
350   reportDuplicate(S, F);
351   return nullptr;
352 }
353 
354 Symbol *SymbolTable::addImportThunk(StringRef Name, DefinedImportData *ID,
355                                     uint16_t Machine) {
356   Symbol *S;
357   bool WasInserted;
358   std::tie(S, WasInserted) = insert(Name, nullptr);
359   S->IsUsedInRegularObj = true;
360   if (WasInserted || isa<Undefined>(S) || isa<Lazy>(S)) {
361     replaceSymbol<DefinedImportThunk>(S, Name, ID, Machine);
362     return S;
363   }
364 
365   reportDuplicate(S, ID->File);
366   return nullptr;
367 }
368 
369 std::vector<Chunk *> SymbolTable::getChunks() {
370   std::vector<Chunk *> Res;
371   for (ObjFile *File : ObjFile::Instances) {
372     ArrayRef<Chunk *> V = File->getChunks();
373     Res.insert(Res.end(), V.begin(), V.end());
374   }
375   return Res;
376 }
377 
378 Symbol *SymbolTable::find(StringRef Name) {
379   return SymMap.lookup(CachedHashStringRef(Name));
380 }
381 
382 Symbol *SymbolTable::findUnderscore(StringRef Name) {
383   if (Config->Machine == I386)
384     return find(("_" + Name).str());
385   return find(Name);
386 }
387 
388 StringRef SymbolTable::findByPrefix(StringRef Prefix) {
389   for (auto Pair : SymMap) {
390     StringRef Name = Pair.first.val();
391     if (Name.startswith(Prefix))
392       return Name;
393   }
394   return "";
395 }
396 
397 StringRef SymbolTable::findMangle(StringRef Name) {
398   if (Symbol *Sym = find(Name))
399     if (!isa<Undefined>(Sym))
400       return Name;
401   if (Config->Machine != I386)
402     return findByPrefix(("?" + Name + "@@Y").str());
403   if (!Name.startswith("_"))
404     return "";
405   // Search for x86 stdcall function.
406   StringRef S = findByPrefix((Name + "@").str());
407   if (!S.empty())
408     return S;
409   // Search for x86 fastcall function.
410   S = findByPrefix(("@" + Name.substr(1) + "@").str());
411   if (!S.empty())
412     return S;
413   // Search for x86 vectorcall function.
414   S = findByPrefix((Name.substr(1) + "@@").str());
415   if (!S.empty())
416     return S;
417   // Search for x86 C++ non-member function.
418   return findByPrefix(("?" + Name.substr(1) + "@@Y").str());
419 }
420 
421 void SymbolTable::mangleMaybe(Symbol *B) {
422   auto *U = dyn_cast<Undefined>(B);
423   if (!U || U->WeakAlias)
424     return;
425   StringRef Alias = findMangle(U->getName());
426   if (!Alias.empty()) {
427     log(U->getName() + " aliased to " + Alias);
428     U->WeakAlias = addUndefined(Alias);
429   }
430 }
431 
432 Symbol *SymbolTable::addUndefined(StringRef Name) {
433   return addUndefined(Name, nullptr, false);
434 }
435 
436 std::vector<StringRef> SymbolTable::compileBitcodeFiles() {
437   LTO.reset(new BitcodeCompiler);
438   for (BitcodeFile *F : BitcodeFile::Instances)
439     LTO->add(*F);
440   return LTO->compile();
441 }
442 
443 void SymbolTable::addCombinedLTOObjects() {
444   if (BitcodeFile::Instances.empty())
445     return;
446 
447   ScopedTimer T(LTOTimer);
448   for (StringRef Object : compileBitcodeFiles()) {
449     auto *Obj = make<ObjFile>(MemoryBufferRef(Object, "lto.tmp"));
450     Obj->parse();
451     ObjFile::Instances.push_back(Obj);
452   }
453 }
454 
455 } // namespace coff
456 } // namespace lld
457