1 //===- InputFiles.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 "Chunks.h"
11 #include "Error.h"
12 #include "InputFiles.h"
13 #include "Writer.h"
14 #include "llvm/ADT/STLExtras.h"
15 #include "llvm/LTO/LTOModule.h"
16 #include "llvm/Object/COFF.h"
17 #include "llvm/Support/COFF.h"
18 #include "llvm/Support/Debug.h"
19 #include "llvm/Support/Endian.h"
20 #include "llvm/Support/raw_ostream.h"
21 
22 using namespace llvm::COFF;
23 using namespace llvm::object;
24 using namespace llvm::support::endian;
25 using llvm::RoundUpToAlignment;
26 using llvm::Triple;
27 using llvm::support::ulittle32_t;
28 using llvm::sys::fs::file_magic;
29 using llvm::sys::fs::identify_magic;
30 
31 namespace lld {
32 namespace coff {
33 
34 int InputFile::NextIndex = 0;
35 
36 // Returns the last element of a path, which is supposed to be a filename.
37 static StringRef getBasename(StringRef Path) {
38   size_t Pos = Path.find_last_of("\\/");
39   if (Pos == StringRef::npos)
40     return Path;
41   return Path.substr(Pos + 1);
42 }
43 
44 // Returns a string in the format of "foo.obj" or "foo.obj(bar.lib)".
45 std::string InputFile::getShortName() {
46   if (ParentName == "")
47     return getName().lower();
48   std::string Res = (getBasename(ParentName) + "(" +
49                      getBasename(getName()) + ")").str();
50   return StringRef(Res).lower();
51 }
52 
53 std::error_code ArchiveFile::parse() {
54   // Parse a MemoryBufferRef as an archive file.
55   auto ArchiveOrErr = Archive::create(MB);
56   if (auto EC = ArchiveOrErr.getError())
57     return EC;
58   File = std::move(ArchiveOrErr.get());
59 
60   // Allocate a buffer for Lazy objects.
61   size_t NumSyms = File->getNumberOfSymbols();
62   size_t BufSize = NumSyms * sizeof(Lazy);
63   Lazy *Buf = (Lazy *)Alloc.Allocate(BufSize, llvm::alignOf<Lazy>());
64   LazySymbols.reserve(NumSyms);
65 
66   // Read the symbol table to construct Lazy objects.
67   uint32_t I = 0;
68   for (const Archive::Symbol &Sym : File->symbols()) {
69     auto *B = new (&Buf[I++]) Lazy(this, Sym);
70     // Skip special symbol exists in import library files.
71     if (B->getName() != "__NULL_IMPORT_DESCRIPTOR")
72       LazySymbols.push_back(B);
73   }
74 
75   // Seen is a map from member files to boolean values. Initially
76   // all members are mapped to false, which indicates all these files
77   // are not read yet.
78   for (const Archive::Child &Child : File->children())
79     Seen[Child.getChildOffset()].clear();
80   return std::error_code();
81 }
82 
83 // Returns a buffer pointing to a member file containing a given symbol.
84 // This function is thread-safe.
85 ErrorOr<MemoryBufferRef> ArchiveFile::getMember(const Archive::Symbol *Sym) {
86   auto ItOrErr = Sym->getMember();
87   if (auto EC = ItOrErr.getError())
88     return EC;
89   Archive::child_iterator It = ItOrErr.get();
90 
91   // Return an empty buffer if we have already returned the same buffer.
92   if (Seen[It->getChildOffset()].test_and_set())
93     return MemoryBufferRef();
94   return It->getMemoryBufferRef();
95 }
96 
97 std::error_code ObjectFile::parse() {
98   // Parse a memory buffer as a COFF file.
99   auto BinOrErr = createBinary(MB);
100   if (auto EC = BinOrErr.getError())
101     return EC;
102   std::unique_ptr<Binary> Bin = std::move(BinOrErr.get());
103 
104   if (auto *Obj = dyn_cast<COFFObjectFile>(Bin.get())) {
105     Bin.release();
106     COFFObj.reset(Obj);
107   } else {
108     llvm::errs() << getName() << " is not a COFF file.\n";
109     return make_error_code(LLDError::InvalidFile);
110   }
111 
112   // Read section and symbol tables.
113   if (auto EC = initializeChunks())
114     return EC;
115   if (auto EC = initializeSymbols())
116     return EC;
117   return initializeSEH();
118 }
119 
120 std::error_code ObjectFile::initializeChunks() {
121   uint32_t NumSections = COFFObj->getNumberOfSections();
122   Chunks.reserve(NumSections);
123   SparseChunks.resize(NumSections + 1);
124   for (uint32_t I = 1; I < NumSections + 1; ++I) {
125     const coff_section *Sec;
126     StringRef Name;
127     if (auto EC = COFFObj->getSection(I, Sec)) {
128       llvm::errs() << "getSection failed: " << Name << ": "
129                    << EC.message() << "\n";
130       return make_error_code(LLDError::BrokenFile);
131     }
132     if (auto EC = COFFObj->getSectionName(Sec, Name)) {
133       llvm::errs() << "getSectionName failed: " << Name << ": "
134                    << EC.message() << "\n";
135       return make_error_code(LLDError::BrokenFile);
136     }
137     if (Name == ".sxdata") {
138       SXData = Sec;
139       continue;
140     }
141     if (Name == ".drectve") {
142       ArrayRef<uint8_t> Data;
143       COFFObj->getSectionContents(Sec, Data);
144       Directives = std::string((const char *)Data.data(), Data.size());
145       continue;
146     }
147     // Skip non-DWARF debug info. MSVC linker converts the sections into
148     // a PDB file, but we don't support that.
149     if (Name == ".debug" || Name.startswith(".debug$"))
150       continue;
151     // We want to preserve DWARF debug sections only when /debug is on.
152     if (!Config->Debug && Name.startswith(".debug"))
153       continue;
154     if (Sec->Characteristics & llvm::COFF::IMAGE_SCN_LNK_REMOVE)
155       continue;
156     auto *C = new (Alloc) SectionChunk(this, Sec);
157     Chunks.push_back(C);
158     SparseChunks[I] = C;
159   }
160   return std::error_code();
161 }
162 
163 std::error_code ObjectFile::initializeSymbols() {
164   uint32_t NumSymbols = COFFObj->getNumberOfSymbols();
165   SymbolBodies.reserve(NumSymbols);
166   SparseSymbolBodies.resize(NumSymbols);
167   int32_t LastSectionNumber = 0;
168   for (uint32_t I = 0; I < NumSymbols; ++I) {
169     // Get a COFFSymbolRef object.
170     auto SymOrErr = COFFObj->getSymbol(I);
171     if (auto EC = SymOrErr.getError()) {
172       llvm::errs() << "broken object file: " << getName() << ": "
173                    << EC.message() << "\n";
174       return make_error_code(LLDError::BrokenFile);
175     }
176     COFFSymbolRef Sym = SymOrErr.get();
177 
178     const void *AuxP = nullptr;
179     if (Sym.getNumberOfAuxSymbols())
180       AuxP = COFFObj->getSymbol(I + 1)->getRawPtr();
181     bool IsFirst = (LastSectionNumber != Sym.getSectionNumber());
182 
183     SymbolBody *Body = nullptr;
184     if (Sym.isUndefined()) {
185       Body = createUndefined(Sym);
186     } else if (Sym.isWeakExternal()) {
187       Body = createWeakExternal(Sym, AuxP);
188     } else {
189       Body = createDefined(Sym, AuxP, IsFirst);
190     }
191     if (Body) {
192       SymbolBodies.push_back(Body);
193       SparseSymbolBodies[I] = Body;
194     }
195     I += Sym.getNumberOfAuxSymbols();
196     LastSectionNumber = Sym.getSectionNumber();
197   }
198   return std::error_code();
199 }
200 
201 Undefined *ObjectFile::createUndefined(COFFSymbolRef Sym) {
202   StringRef Name;
203   COFFObj->getSymbolName(Sym, Name);
204   return new (Alloc) Undefined(Name);
205 }
206 
207 Undefined *ObjectFile::createWeakExternal(COFFSymbolRef Sym, const void *AuxP) {
208   StringRef Name;
209   COFFObj->getSymbolName(Sym, Name);
210   auto *U = new (Alloc) Undefined(Name);
211   auto *Aux = (const coff_aux_weak_external *)AuxP;
212   U->WeakAlias = SparseSymbolBodies[Aux->TagIndex];
213   return U;
214 }
215 
216 Defined *ObjectFile::createDefined(COFFSymbolRef Sym, const void *AuxP,
217                                    bool IsFirst) {
218   StringRef Name;
219   if (Sym.isCommon()) {
220     auto *C = new (Alloc) CommonChunk(Sym);
221     Chunks.push_back(C);
222     return new (Alloc) DefinedCommon(this, Sym, C);
223   }
224   if (Sym.isAbsolute()) {
225     COFFObj->getSymbolName(Sym, Name);
226     // Skip special symbols.
227     if (Name == "@comp.id")
228       return nullptr;
229     // COFF spec 5.10.1. The .sxdata section.
230     if (Name == "@feat.00") {
231       if (Sym.getValue() & 1)
232         SEHCompat = true;
233       return nullptr;
234     }
235     return new (Alloc) DefinedAbsolute(Name, Sym);
236   }
237   if (Sym.getSectionNumber() == llvm::COFF::IMAGE_SYM_DEBUG)
238     return nullptr;
239 
240   // Nothing else to do without a section chunk.
241   auto *SC = cast_or_null<SectionChunk>(SparseChunks[Sym.getSectionNumber()]);
242   if (!SC)
243     return nullptr;
244 
245   // Handle associative sections
246   if (IsFirst && AuxP) {
247     auto *Aux = reinterpret_cast<const coff_aux_section_definition *>(AuxP);
248     if (Aux->Selection == IMAGE_COMDAT_SELECT_ASSOCIATIVE)
249       if (auto *ParentSC = cast_or_null<SectionChunk>(
250               SparseChunks[Aux->getNumber(Sym.isBigObj())]))
251         ParentSC->addAssociative(SC);
252   }
253 
254   auto *B = new (Alloc) DefinedRegular(this, Sym, SC);
255   if (SC->isCOMDAT() && Sym.getValue() == 0 && !AuxP)
256     SC->setSymbol(B);
257 
258   return B;
259 }
260 
261 std::error_code ObjectFile::initializeSEH() {
262   if (!SEHCompat || !SXData)
263     return std::error_code();
264   ArrayRef<uint8_t> A;
265   COFFObj->getSectionContents(SXData, A);
266   if (A.size() % 4 != 0) {
267     llvm::errs() << ".sxdata must be an array of symbol table indices\n";
268     return make_error_code(LLDError::BrokenFile);
269   }
270   auto *I = reinterpret_cast<const ulittle32_t *>(A.data());
271   auto *E = reinterpret_cast<const ulittle32_t *>(A.data() + A.size());
272   for (; I != E; ++I)
273     SEHandlers.insert(SparseSymbolBodies[*I]);
274   return std::error_code();
275 }
276 
277 MachineTypes ObjectFile::getMachineType() {
278   if (COFFObj)
279     return static_cast<MachineTypes>(COFFObj->getMachine());
280   return IMAGE_FILE_MACHINE_UNKNOWN;
281 }
282 
283 StringRef ltrim1(StringRef S, const char *Chars) {
284   if (!S.empty() && strchr(Chars, S[0]))
285     return S.substr(1);
286   return S;
287 }
288 
289 std::error_code ImportFile::parse() {
290   const char *Buf = MB.getBufferStart();
291   const char *End = MB.getBufferEnd();
292   const auto *Hdr = reinterpret_cast<const coff_import_header *>(Buf);
293 
294   // Check if the total size is valid.
295   if ((size_t)(End - Buf) != (sizeof(*Hdr) + Hdr->SizeOfData)) {
296     llvm::errs() << "broken import library\n";
297     return make_error_code(LLDError::BrokenFile);
298   }
299 
300   // Read names and create an __imp_ symbol.
301   StringRef Name = StringAlloc.save(StringRef(Buf + sizeof(*Hdr)));
302   StringRef ImpName = StringAlloc.save(Twine("__imp_") + Name);
303   StringRef DLLName(Buf + sizeof(coff_import_header) + Name.size() + 1);
304   StringRef ExtName;
305   switch (Hdr->getNameType()) {
306   case IMPORT_ORDINAL:
307     ExtName = "";
308     break;
309   case IMPORT_NAME:
310     ExtName = Name;
311     break;
312   case IMPORT_NAME_NOPREFIX:
313     ExtName = ltrim1(Name, "?@_");
314     break;
315   case IMPORT_NAME_UNDECORATE:
316     ExtName = ltrim1(Name, "?@_");
317     ExtName = ExtName.substr(0, ExtName.find('@'));
318     break;
319   }
320   auto *ImpSym = new (Alloc) DefinedImportData(DLLName, ImpName, ExtName, Hdr);
321   SymbolBodies.push_back(ImpSym);
322 
323   // If type is function, we need to create a thunk which jump to an
324   // address pointed by the __imp_ symbol. (This allows you to call
325   // DLL functions just like regular non-DLL functions.)
326   if (Hdr->getType() == llvm::COFF::IMPORT_CODE) {
327     auto *B = new (Alloc) DefinedImportThunk(Name, ImpSym, Hdr->Machine);
328     SymbolBodies.push_back(B);
329   }
330   return std::error_code();
331 }
332 
333 std::error_code BitcodeFile::parse() {
334   std::string Err;
335   M.reset(LTOModule::createFromBuffer(MB.getBufferStart(),
336                                       MB.getBufferSize(),
337                                       llvm::TargetOptions(), Err));
338   if (!Err.empty()) {
339     llvm::errs() << Err << '\n';
340     return make_error_code(LLDError::BrokenFile);
341   }
342 
343   llvm::BumpPtrStringSaver Saver(Alloc);
344   for (unsigned I = 0, E = M->getSymbolCount(); I != E; ++I) {
345     lto_symbol_attributes Attrs = M->getSymbolAttributes(I);
346     if ((Attrs & LTO_SYMBOL_SCOPE_MASK) == LTO_SYMBOL_SCOPE_INTERNAL)
347       continue;
348 
349     StringRef SymName = Saver.save(M->getSymbolName(I));
350     int SymbolDef = Attrs & LTO_SYMBOL_DEFINITION_MASK;
351     if (SymbolDef == LTO_SYMBOL_DEFINITION_UNDEFINED) {
352       SymbolBodies.push_back(new (Alloc) Undefined(SymName));
353     } else {
354       bool Replaceable =
355           (SymbolDef == LTO_SYMBOL_DEFINITION_TENTATIVE || // common
356            (Attrs & LTO_SYMBOL_COMDAT) ||                  // comdat
357            (SymbolDef == LTO_SYMBOL_DEFINITION_WEAK &&     // weak external
358             (Attrs & LTO_SYMBOL_ALIAS)));
359       SymbolBodies.push_back(new (Alloc) DefinedBitcode(this, SymName,
360                                                         Replaceable));
361     }
362   }
363 
364   Directives = M->getLinkerOpts();
365   return std::error_code();
366 }
367 
368 MachineTypes BitcodeFile::getMachineType() {
369   if (!M)
370     return IMAGE_FILE_MACHINE_UNKNOWN;
371   switch (Triple(M->getTargetTriple()).getArch()) {
372   case Triple::x86_64:
373     return AMD64;
374   case Triple::x86:
375     return I386;
376   case Triple::arm:
377     return ARMNT;
378   default:
379     return IMAGE_FILE_MACHINE_UNKNOWN;
380   }
381 }
382 
383 } // namespace coff
384 } // namespace lld
385