diff options
Diffstat (limited to 'COFF')
| -rw-r--r-- | COFF/ICF.cpp | 37 | ||||
| -rw-r--r-- | COFF/InputFiles.cpp | 21 | ||||
| -rw-r--r-- | COFF/InputFiles.h | 13 | ||||
| -rw-r--r-- | COFF/MarkLive.cpp | 17 | ||||
| -rw-r--r-- | COFF/PDB.cpp | 15 | ||||
| -rw-r--r-- | COFF/Symbols.cpp | 19 | ||||
| -rw-r--r-- | COFF/Symbols.h | 3 | ||||
| -rw-r--r-- | COFF/Writer.cpp | 55 |
8 files changed, 111 insertions, 69 deletions
diff --git a/COFF/ICF.cpp b/COFF/ICF.cpp index 3b7cc424f0a2..da8ca360542a 100644 --- a/COFF/ICF.cpp +++ b/COFF/ICF.cpp @@ -56,7 +56,6 @@ private: std::vector<SectionChunk *> Chunks; int Cnt = 0; - std::atomic<uint32_t> NextId = {1}; std::atomic<bool> Repeat = {false}; }; @@ -98,10 +97,10 @@ void ICF::segregate(size_t Begin, size_t End, bool Constant) { }); size_t Mid = Bound - Chunks.begin(); - // Split [Begin, End) into [Begin, Mid) and [Mid, End). - uint32_t Id = NextId++; + // Split [Begin, End) into [Begin, Mid) and [Mid, End). We use Mid as an + // equivalence class ID because every group ends with a unique index. for (size_t I = Begin; I < Mid; ++I) - Chunks[I]->Class[(Cnt + 1) % 2] = Id; + Chunks[I]->Class[(Cnt + 1) % 2] = Mid; // If we created a group, we need to iterate the main loop again. if (Mid != End) @@ -186,6 +185,7 @@ void ICF::forEachClass(std::function<void(size_t, size_t)> Fn) { // call Fn sequentially. if (Chunks.size() < 1024) { forEachClassRange(0, Chunks.size(), Fn); + ++Cnt; return; } @@ -193,9 +193,10 @@ void ICF::forEachClass(std::function<void(size_t, size_t)> Fn) { size_t NumShards = 256; size_t Step = Chunks.size() / NumShards; for_each_n(parallel::par, size_t(0), NumShards, [&](size_t I) { - forEachClassRange(I * Step, (I + 1) * Step, Fn); + size_t End = (I == NumShards - 1) ? Chunks.size() : (I + 1) * Step; + forEachClassRange(I * Step, End, Fn); }); - forEachClassRange(Step * NumShards, Chunks.size(), Fn); + ++Cnt; } // Merge identical COMDAT sections. @@ -203,22 +204,20 @@ void ICF::forEachClass(std::function<void(size_t, size_t)> Fn) { // contents and relocations are all the same. void ICF::run(const std::vector<Chunk *> &Vec) { // Collect only mergeable sections and group by hash value. + uint32_t NextId = 1; for (Chunk *C : Vec) { - auto *SC = dyn_cast<SectionChunk>(C); - if (!SC) - continue; - - if (isEligible(SC)) { - // Set MSB to 1 to avoid collisions with non-hash classs. - SC->Class[0] = getHash(SC) | (1 << 31); - Chunks.push_back(SC); - } else { - SC->Class[0] = NextId++; + if (auto *SC = dyn_cast<SectionChunk>(C)) { + if (isEligible(SC)) + Chunks.push_back(SC); + else + SC->Class[0] = NextId++; } } - if (Chunks.empty()) - return; + // Initially, we use hash values to partition sections. + for (SectionChunk *SC : Chunks) + // Set MSB to 1 to avoid collisions with non-hash classs. + SC->Class[0] = getHash(SC) | (1 << 31); // From now on, sections in Chunks are ordered so that sections in // the same group are consecutive in the vector. @@ -229,14 +228,12 @@ void ICF::run(const std::vector<Chunk *> &Vec) { // Compare static contents and assign unique IDs for each static content. forEachClass([&](size_t Begin, size_t End) { segregate(Begin, End, true); }); - ++Cnt; // Split groups by comparing relocations until convergence is obtained. do { Repeat = false; forEachClass( [&](size_t Begin, size_t End) { segregate(Begin, End, false); }); - ++Cnt; } while (Repeat); log("ICF needed " + Twine(Cnt) + " iterations"); diff --git a/COFF/InputFiles.cpp b/COFF/InputFiles.cpp index 6e6465cd5d62..258d9fd74b4d 100644 --- a/COFF/InputFiles.cpp +++ b/COFF/InputFiles.cpp @@ -48,13 +48,11 @@ namespace coff { /// alias to Target. static void checkAndSetWeakAlias(SymbolTable *Symtab, InputFile *F, SymbolBody *Source, SymbolBody *Target) { - auto *U = dyn_cast<Undefined>(Source); - if (!U) - return; - else if (!U->WeakAlias) + if (auto *U = dyn_cast<Undefined>(Source)) { + if (U->WeakAlias && U->WeakAlias != Target) + Symtab->reportDuplicate(Source->symbol(), F); U->WeakAlias = Target; - else if (U->WeakAlias != Target) - Symtab->reportDuplicate(Source->symbol(), F); + } } ArchiveFile::ArchiveFile(MemoryBufferRef M) : InputFile(ArchiveKind, M) {} @@ -153,8 +151,10 @@ void ObjectFile::initializeSymbols() { uint32_t NumSymbols = COFFObj->getNumberOfSymbols(); SymbolBodies.reserve(NumSymbols); SparseSymbolBodies.resize(NumSymbols); + SmallVector<std::pair<SymbolBody *, uint32_t>, 8> WeakAliases; int32_t LastSectionNumber = 0; + for (uint32_t I = 0; I < NumSymbols; ++I) { // Get a COFFSymbolRef object. ErrorOr<COFFSymbolRef> SymOrErr = COFFObj->getSymbol(I); @@ -185,9 +185,12 @@ void ObjectFile::initializeSymbols() { I += Sym.getNumberOfAuxSymbols(); LastSectionNumber = Sym.getSectionNumber(); } - for (auto WeakAlias : WeakAliases) - checkAndSetWeakAlias(Symtab, this, WeakAlias.first, - SparseSymbolBodies[WeakAlias.second]); + + for (auto &KV : WeakAliases) { + SymbolBody *Sym = KV.first; + uint32_t Idx = KV.second; + checkAndSetWeakAlias(Symtab, this, Sym, SparseSymbolBodies[Idx]); + } } SymbolBody *ObjectFile::createUndefined(COFFSymbolRef Sym) { diff --git a/COFF/InputFiles.h b/COFF/InputFiles.h index 9e32b3b9f9d6..9449f24ac241 100644 --- a/COFF/InputFiles.h +++ b/COFF/InputFiles.h @@ -10,6 +10,7 @@ #ifndef LLD_COFF_INPUT_FILES_H #define LLD_COFF_INPUT_FILES_H +#include "Config.h" #include "lld/Core/LLVM.h" #include "llvm/ADT/ArrayRef.h" #include "llvm/ADT/DenseSet.h" @@ -161,7 +162,9 @@ private: // for details about the format. class ImportFile : public InputFile { public: - explicit ImportFile(MemoryBufferRef M) : InputFile(ImportKind, M) {} + explicit ImportFile(MemoryBufferRef M) + : InputFile(ImportKind, M), Live(!Config->DoGC) {} + static bool classof(const InputFile *F) { return F->kind() == ImportKind; } DefinedImportData *ImpSym = nullptr; @@ -176,6 +179,14 @@ public: StringRef ExternalName; const coff_import_header *Hdr; Chunk *Location = nullptr; + + // We want to eliminate dllimported symbols if no one actually refers them. + // This "Live" bit is used to keep track of which import library members + // are actually in use. + // + // If the Live bit is turned off by MarkLive, Writer will ignore dllimported + // symbols provided by this import library member. + bool Live; }; // Used for LTO. diff --git a/COFF/MarkLive.cpp b/COFF/MarkLive.cpp index 0156d238b672..25e5cc350673 100644 --- a/COFF/MarkLive.cpp +++ b/COFF/MarkLive.cpp @@ -37,19 +37,26 @@ void markLive(const std::vector<Chunk *> &Chunks) { Worklist.push_back(C); }; + auto AddSym = [&](SymbolBody *B) { + if (auto *Sym = dyn_cast<DefinedRegular>(B)) + Enqueue(Sym->getChunk()); + else if (auto *Sym = dyn_cast<DefinedImportData>(B)) + Sym->File->Live = true; + else if (auto *Sym = dyn_cast<DefinedImportThunk>(B)) + Sym->WrappedSym->File->Live = true; + }; + // Add GC root chunks. for (SymbolBody *B : Config->GCRoot) - if (auto *D = dyn_cast<DefinedRegular>(B)) - Enqueue(D->getChunk()); + AddSym(B); while (!Worklist.empty()) { SectionChunk *SC = Worklist.pop_back_val(); assert(SC->isLive() && "We mark as live when pushing onto the worklist!"); // Mark all symbols listed in the relocation table for this section. - for (SymbolBody *S : SC->symbols()) - if (auto *D = dyn_cast<DefinedRegular>(S)) - Enqueue(D->getChunk()); + for (SymbolBody *B : SC->symbols()) + AddSym(B); // Mark associative sections if any. for (SectionChunk *C : SC->children()) diff --git a/COFF/PDB.cpp b/COFF/PDB.cpp index 0266148cc6c9..a3b3ab7bbab0 100644 --- a/COFF/PDB.cpp +++ b/COFF/PDB.cpp @@ -99,6 +99,12 @@ static void addTypeInfo(pdb::TpiStreamBuilder &TpiBuilder, static void mergeDebugT(SymbolTable *Symtab, pdb::PDBFileBuilder &Builder, codeview::TypeTableBuilder &TypeTable, codeview::TypeTableBuilder &IDTable) { + // Follow type servers. If the same type server is encountered more than + // once for this instance of `PDBTypeServerHandler` (for example if many + // object files reference the same TypeServer), the types from the + // TypeServer will only be visited once. + pdb::PDBTypeServerHandler Handler; + // Visit all .debug$T sections to add them to Builder. for (ObjectFile *File : Symtab->ObjectFiles) { ArrayRef<uint8_t> Data = getDebugSection(File, ".debug$T"); @@ -109,16 +115,11 @@ static void mergeDebugT(SymbolTable *Symtab, pdb::PDBFileBuilder &Builder, codeview::CVTypeArray Types; BinaryStreamReader Reader(Stream); SmallVector<TypeIndex, 128> SourceToDest; - // Follow type servers. If the same type server is encountered more than - // once for this instance of `PDBTypeServerHandler` (for example if many - // object files reference the same TypeServer), the types from the - // TypeServer will only be visited once. - pdb::PDBTypeServerHandler Handler; Handler.addSearchPath(llvm::sys::path::parent_path(File->getName())); if (auto EC = Reader.readArray(Types, Reader.getLength())) fatal(EC, "Reader::readArray failed"); - if (auto Err = codeview::mergeTypeStreams(IDTable, TypeTable, SourceToDest, - &Handler, Types)) + if (auto Err = codeview::mergeTypeAndIdRecords( + IDTable, TypeTable, SourceToDest, &Handler, Types)) fatal(Err, "codeview::mergeTypeStreams failed"); } diff --git a/COFF/Symbols.cpp b/COFF/Symbols.cpp index 993e920ce7f7..5c185a511dd7 100644 --- a/COFF/Symbols.cpp +++ b/COFF/Symbols.cpp @@ -61,16 +61,19 @@ COFFSymbolRef DefinedCOFF::getCOFFSymbol() { return COFFSymbolRef(reinterpret_cast<const coff_symbol32 *>(Sym)); } +static Chunk *makeImportThunk(DefinedImportData *S, uint16_t Machine) { + if (Machine == AMD64) + return make<ImportThunkChunkX64>(S); + if (Machine == I386) + return make<ImportThunkChunkX86>(S); + assert(Machine == ARMNT); + return make<ImportThunkChunkARM>(S); +} + DefinedImportThunk::DefinedImportThunk(StringRef Name, DefinedImportData *S, uint16_t Machine) - : Defined(DefinedImportThunkKind, Name) { - switch (Machine) { - case AMD64: Data = make<ImportThunkChunkX64>(S); return; - case I386: Data = make<ImportThunkChunkX86>(S); return; - case ARMNT: Data = make<ImportThunkChunkARM>(S); return; - default: llvm_unreachable("unknown machine type"); - } -} + : Defined(DefinedImportThunkKind, Name), WrappedSym(S), + Data(makeImportThunk(S, Machine)) {} Defined *Undefined::getWeakAlias() { // A weak alias may be a weak alias to another symbol, so check recursively. diff --git a/COFF/Symbols.h b/COFF/Symbols.h index 1b83f73ff20c..801fc87f91d9 100644 --- a/COFF/Symbols.h +++ b/COFF/Symbols.h @@ -300,7 +300,6 @@ public: void setLocation(Chunk *AddressTable) { File->Location = AddressTable; } uint16_t getOrdinal() { return File->Hdr->OrdinalHint; } -private: ImportFile *File; }; @@ -320,6 +319,8 @@ public: uint64_t getRVA() { return Data->getRVA(); } Chunk *getChunk() { return Data; } + DefinedImportData *WrappedSym; + private: Chunk *Data; }; diff --git a/COFF/Writer.cpp b/COFF/Writer.cpp index cf3ad7ef045c..fb1f3cae5bb2 100644 --- a/COFF/Writer.cpp +++ b/COFF/Writer.cpp @@ -365,6 +365,9 @@ void Writer::createImportTables() { // the same order as in the command line. (That affects DLL // initialization order, and this ordering is MSVC-compatible.) for (ImportFile *File : Symtab->ImportFiles) { + if (!File->Live) + continue; + std::string DLL = StringRef(File->DLLName).lower(); if (Config->DLLOrder.count(DLL) == 0) Config->DLLOrder[DLL] = Config->DLLOrder.size(); @@ -372,19 +375,28 @@ void Writer::createImportTables() { OutputSection *Text = createSection(".text"); for (ImportFile *File : Symtab->ImportFiles) { + if (!File->Live) + continue; + if (DefinedImportThunk *Thunk = File->ThunkSym) Text->addChunk(Thunk->getChunk()); + if (Config->DelayLoads.count(StringRef(File->DLLName).lower())) { + if (!File->ThunkSym) + fatal("cannot delay-load " + toString(File) + + " due to import of data: " + toString(*File->ImpSym)); DelayIdata.add(File->ImpSym); } else { Idata.add(File->ImpSym); } } + if (!Idata.empty()) { OutputSection *Sec = createSection(".idata"); for (Chunk *C : Idata.getChunks()) Sec->addChunk(C); } + if (!DelayIdata.empty()) { Defined *Helper = cast<Defined>(Config->DelayLoadHelper); DelayIdata.create(Helper); @@ -437,6 +449,14 @@ Optional<coff_symbol16> Writer::createSymbol(Defined *Def) { if (!D->getChunk()->isLive()) return None; + if (auto *Sym = dyn_cast<DefinedImportData>(Def)) + if (!Sym->File->Live) + return None; + + if (auto *Sym = dyn_cast<DefinedImportThunk>(Def)) + if (!Sym->WrappedSym->File->Live) + return None; + coff_symbol16 Sym; StringRef Name = Def->getName(); if (Name.size() > COFF::NameSize) { @@ -491,14 +511,17 @@ void Writer::createSymbolAndStringTable() { Sec->setStringTableOff(addEntryToStringTable(Name)); } - for (lld::coff::ObjectFile *File : Symtab->ObjectFiles) - for (SymbolBody *B : File->getSymbols()) - if (auto *D = dyn_cast<Defined>(B)) - if (!D->WrittenToSymtab) { - D->WrittenToSymtab = true; - if (Optional<coff_symbol16> Sym = createSymbol(D)) - OutputSymtab.push_back(*Sym); - } + for (lld::coff::ObjectFile *File : Symtab->ObjectFiles) { + for (SymbolBody *B : File->getSymbols()) { + auto *D = dyn_cast<Defined>(B); + if (!D || D->WrittenToSymtab) + continue; + D->WrittenToSymtab = true; + + if (Optional<coff_symbol16> Sym = createSymbol(D)) + OutputSymtab.push_back(*Sym); + } + } OutputSection *LastSection = OutputSections.back(); // We position the symbol table to be adjacent to the end of the last section. @@ -782,19 +805,15 @@ void Writer::writeBuildId() { if (BuildId == nullptr) return; - MD5 Hash; - MD5::MD5Result Res; - - Hash.update(ArrayRef<uint8_t>{Buffer->getBufferStart(), - Buffer->getBufferEnd()}); - Hash.final(Res); - assert(BuildId->DI->Signature.CVSignature == OMF::Signature::PDB70 && "only PDB 7.0 is supported"); - assert(sizeof(Res) == sizeof(BuildId->DI->PDB70.Signature) && + assert(sizeof(BuildId->DI->PDB70.Signature) == 16 && "signature size mismatch"); - memcpy(BuildId->DI->PDB70.Signature, Res.Bytes.data(), - sizeof(codeview::PDB70DebugInfo::Signature)); + + // Compute an MD5 hash. + ArrayRef<uint8_t> Buf(Buffer->getBufferStart(), Buffer->getBufferEnd()); + memcpy(BuildId->DI->PDB70.Signature, MD5::hash(Buf).data(), 16); + // TODO(compnerd) track the Age BuildId->DI->PDB70.Age = 1; } |
