Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 1 | //===-- MachODump.cpp - Object file dumping utility for llvm --------------===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This file implements the MachO-specific dumper for llvm-objdump. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #include "llvm-objdump.h" |
| 15 | #include "MCFunction.h" |
| 16 | #include "llvm/Support/MachO.h" |
| 17 | #include "llvm/Object/MachOObject.h" |
| 18 | #include "llvm/ADT/OwningPtr.h" |
| 19 | #include "llvm/ADT/Triple.h" |
| 20 | #include "llvm/ADT/STLExtras.h" |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 21 | #include "llvm/DebugInfo/DIContext.h" |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 22 | #include "llvm/MC/MCAsmInfo.h" |
| 23 | #include "llvm/MC/MCDisassembler.h" |
| 24 | #include "llvm/MC/MCInst.h" |
| 25 | #include "llvm/MC/MCInstPrinter.h" |
| 26 | #include "llvm/MC/MCInstrAnalysis.h" |
| 27 | #include "llvm/MC/MCInstrDesc.h" |
| 28 | #include "llvm/MC/MCInstrInfo.h" |
| 29 | #include "llvm/MC/MCSubtargetInfo.h" |
| 30 | #include "llvm/Support/CommandLine.h" |
| 31 | #include "llvm/Support/Debug.h" |
| 32 | #include "llvm/Support/Format.h" |
| 33 | #include "llvm/Support/GraphWriter.h" |
| 34 | #include "llvm/Support/MemoryBuffer.h" |
| 35 | #include "llvm/Support/TargetRegistry.h" |
| 36 | #include "llvm/Support/TargetSelect.h" |
| 37 | #include "llvm/Support/raw_ostream.h" |
| 38 | #include "llvm/Support/system_error.h" |
| 39 | #include <algorithm> |
| 40 | #include <cstring> |
| 41 | using namespace llvm; |
| 42 | using namespace object; |
| 43 | |
| 44 | static cl::opt<bool> |
| 45 | CFG("cfg", cl::desc("Create a CFG for every symbol in the object file and" |
| 46 | "write it to a graphviz file (MachO-only)")); |
| 47 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 48 | static cl::opt<bool> |
| 49 | UseDbg("g", cl::desc("Print line information from debug info if available")); |
| 50 | |
| 51 | static cl::opt<std::string> |
| 52 | DSYMFile("dsym", cl::desc("Use .dSYM file for debug info")); |
| 53 | |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 54 | static const Target *GetTarget(const MachOObject *MachOObj) { |
| 55 | // Figure out the target triple. |
| 56 | llvm::Triple TT("unknown-unknown-unknown"); |
| 57 | switch (MachOObj->getHeader().CPUType) { |
| 58 | case llvm::MachO::CPUTypeI386: |
| 59 | TT.setArch(Triple::ArchType(Triple::x86)); |
| 60 | break; |
| 61 | case llvm::MachO::CPUTypeX86_64: |
| 62 | TT.setArch(Triple::ArchType(Triple::x86_64)); |
| 63 | break; |
| 64 | case llvm::MachO::CPUTypeARM: |
| 65 | TT.setArch(Triple::ArchType(Triple::arm)); |
| 66 | break; |
| 67 | case llvm::MachO::CPUTypePowerPC: |
| 68 | TT.setArch(Triple::ArchType(Triple::ppc)); |
| 69 | break; |
| 70 | case llvm::MachO::CPUTypePowerPC64: |
| 71 | TT.setArch(Triple::ArchType(Triple::ppc64)); |
| 72 | break; |
| 73 | } |
| 74 | |
| 75 | TripleName = TT.str(); |
| 76 | |
| 77 | // Get the target specific parser. |
| 78 | std::string Error; |
| 79 | const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error); |
| 80 | if (TheTarget) |
| 81 | return TheTarget; |
| 82 | |
| 83 | errs() << "llvm-objdump: error: unable to get target for '" << TripleName |
| 84 | << "', see --version and --triple.\n"; |
| 85 | return 0; |
| 86 | } |
| 87 | |
| 88 | struct Section { |
| 89 | char Name[16]; |
| 90 | uint64_t Address; |
| 91 | uint64_t Size; |
| 92 | uint32_t Offset; |
| 93 | uint32_t NumRelocs; |
| 94 | uint64_t RelocTableOffset; |
| 95 | }; |
| 96 | |
| 97 | struct Symbol { |
| 98 | uint64_t Value; |
| 99 | uint32_t StringIndex; |
| 100 | uint8_t SectionIndex; |
| 101 | bool operator<(const Symbol &RHS) const { return Value < RHS.Value; } |
| 102 | }; |
| 103 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 104 | template <typename T> |
| 105 | static Section copySection(const T &Sect) { |
| 106 | Section S; |
| 107 | memcpy(S.Name, Sect->Name, 16); |
| 108 | S.Address = Sect->Address; |
| 109 | S.Size = Sect->Size; |
| 110 | S.Offset = Sect->Offset; |
| 111 | S.NumRelocs = Sect->NumRelocationTableEntries; |
| 112 | S.RelocTableOffset = Sect->RelocationTableOffset; |
| 113 | return S; |
| 114 | } |
| 115 | |
| 116 | template <typename T> |
| 117 | static Symbol copySymbol(const T &STE) { |
| 118 | Symbol S; |
| 119 | S.StringIndex = STE->StringIndex; |
| 120 | S.SectionIndex = STE->SectionIndex; |
| 121 | S.Value = STE->Value; |
| 122 | return S; |
| 123 | } |
| 124 | |
| 125 | // Print addtitional information about an address, if available. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 126 | static void DumpAddress(uint64_t Address, ArrayRef<Section> Sections, |
| 127 | MachOObject *MachOObj, raw_ostream &OS) { |
| 128 | for (unsigned i = 0; i != Sections.size(); ++i) { |
| 129 | uint64_t addr = Address-Sections[i].Address; |
| 130 | if (Sections[i].Address <= Address && |
| 131 | Sections[i].Address + Sections[i].Size > Address) { |
| 132 | StringRef bytes = MachOObj->getData(Sections[i].Offset, |
| 133 | Sections[i].Size); |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 134 | // Print constant strings. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 135 | if (!strcmp(Sections[i].Name, "__cstring")) |
| 136 | OS << '"' << bytes.substr(addr, bytes.find('\0', addr)) << '"'; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 137 | // Print constant CFStrings. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 138 | if (!strcmp(Sections[i].Name, "__cfstring")) |
| 139 | OS << "@\"" << bytes.substr(addr, bytes.find('\0', addr)) << '"'; |
| 140 | } |
| 141 | } |
| 142 | } |
| 143 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 144 | typedef std::map<uint64_t, MCFunction*> FunctionMapTy; |
| 145 | typedef SmallVector<MCFunction, 16> FunctionListTy; |
| 146 | static void createMCFunctionAndSaveCalls(StringRef Name, |
| 147 | const MCDisassembler *DisAsm, |
| 148 | MemoryObject &Object, uint64_t Start, |
| 149 | uint64_t End, |
| 150 | MCInstrAnalysis *InstrAnalysis, |
| 151 | uint64_t Address, |
| 152 | raw_ostream &DebugOut, |
| 153 | FunctionMapTy &FunctionMap, |
| 154 | FunctionListTy &Functions) { |
| 155 | SmallVector<uint64_t, 16> Calls; |
| 156 | MCFunction f = |
| 157 | MCFunction::createFunctionFromMC(Name, DisAsm, Object, Start, End, |
| 158 | InstrAnalysis, DebugOut, Calls); |
| 159 | Functions.push_back(f); |
| 160 | FunctionMap[Address] = &Functions.back(); |
| 161 | |
| 162 | // Add the gathered callees to the map. |
| 163 | for (unsigned i = 0, e = Calls.size(); i != e; ++i) |
| 164 | FunctionMap.insert(std::make_pair(Calls[i], (MCFunction*)0)); |
| 165 | } |
| 166 | |
| 167 | // Write a graphviz file for the CFG inside an MCFunction. |
| 168 | static void emitDOTFile(const char *FileName, const MCFunction &f, |
| 169 | MCInstPrinter *IP) { |
| 170 | // Start a new dot file. |
| 171 | std::string Error; |
| 172 | raw_fd_ostream Out(FileName, Error); |
| 173 | if (!Error.empty()) { |
| 174 | errs() << "llvm-objdump: warning: " << Error << '\n'; |
| 175 | return; |
| 176 | } |
| 177 | |
| 178 | Out << "digraph " << f.getName() << " {\n"; |
| 179 | Out << "graph [ rankdir = \"LR\" ];\n"; |
| 180 | for (MCFunction::iterator i = f.begin(), e = f.end(); i != e; ++i) { |
| 181 | bool hasPreds = false; |
| 182 | // Only print blocks that have predecessors. |
| 183 | // FIXME: Slow. |
| 184 | for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe; |
| 185 | ++pi) |
| 186 | if (pi->second.contains(i->first)) { |
| 187 | hasPreds = true; |
| 188 | break; |
| 189 | } |
| 190 | |
| 191 | if (!hasPreds && i != f.begin()) |
| 192 | continue; |
| 193 | |
| 194 | Out << '"' << i->first << "\" [ label=\"<a>"; |
| 195 | // Print instructions. |
| 196 | for (unsigned ii = 0, ie = i->second.getInsts().size(); ii != ie; |
| 197 | ++ii) { |
| 198 | // Escape special chars and print the instruction in mnemonic form. |
| 199 | std::string Str; |
| 200 | raw_string_ostream OS(Str); |
| 201 | IP->printInst(&i->second.getInsts()[ii].Inst, OS, ""); |
| 202 | Out << DOT::EscapeString(OS.str()) << '|'; |
| 203 | } |
| 204 | Out << "<o>\" shape=\"record\" ];\n"; |
| 205 | |
| 206 | // Add edges. |
| 207 | for (MCBasicBlock::succ_iterator si = i->second.succ_begin(), |
| 208 | se = i->second.succ_end(); si != se; ++si) |
| 209 | Out << i->first << ":o -> " << *si <<":a\n"; |
| 210 | } |
| 211 | Out << "}\n"; |
| 212 | } |
| 213 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 214 | static void getSectionsAndSymbols(const macho::Header &Header, |
| 215 | MachOObject *MachOObj, |
| 216 | InMemoryStruct<macho::SymtabLoadCommand> *SymtabLC, |
| 217 | std::vector<Section> &Sections, |
| 218 | std::vector<Symbol> &Symbols, |
| 219 | SmallVectorImpl<uint64_t> &FoundFns) { |
| 220 | // Make a list of all symbols in the object file. |
| 221 | for (unsigned i = 0; i != Header.NumLoadCommands; ++i) { |
| 222 | const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i); |
| 223 | if (LCI.Command.Type == macho::LCT_Segment) { |
| 224 | InMemoryStruct<macho::SegmentLoadCommand> SegmentLC; |
| 225 | MachOObj->ReadSegmentLoadCommand(LCI, SegmentLC); |
| 226 | |
| 227 | // Store the sections in this segment. |
| 228 | for (unsigned SectNum = 0; SectNum != SegmentLC->NumSections; ++SectNum) { |
| 229 | InMemoryStruct<macho::Section> Sect; |
| 230 | MachOObj->ReadSection(LCI, SectNum, Sect); |
| 231 | Sections.push_back(copySection(Sect)); |
| 232 | |
| 233 | // Store the symbols in this section. |
| 234 | if (SymtabLC) { |
| 235 | for (unsigned i = 0; i != (*SymtabLC)->NumSymbolTableEntries; ++i) { |
| 236 | InMemoryStruct<macho::SymbolTableEntry> STE; |
| 237 | MachOObj->ReadSymbolTableEntry((*SymtabLC)->SymbolTableOffset, i, |
| 238 | STE); |
| 239 | Symbols.push_back(copySymbol(STE)); |
| 240 | } |
| 241 | } |
| 242 | } |
| 243 | } else if (LCI.Command.Type == macho::LCT_Segment64) { |
| 244 | InMemoryStruct<macho::Segment64LoadCommand> Segment64LC; |
| 245 | MachOObj->ReadSegment64LoadCommand(LCI, Segment64LC); |
| 246 | |
| 247 | // Store the sections in this segment. |
| 248 | for (unsigned SectNum = 0; SectNum != Segment64LC->NumSections; |
| 249 | ++SectNum) { |
| 250 | InMemoryStruct<macho::Section64> Sect64; |
| 251 | MachOObj->ReadSection64(LCI, SectNum, Sect64); |
| 252 | Sections.push_back(copySection(Sect64)); |
| 253 | |
| 254 | // Store the symbols in this section. |
| 255 | if (SymtabLC) { |
| 256 | for (unsigned i = 0; i != (*SymtabLC)->NumSymbolTableEntries; ++i) { |
| 257 | InMemoryStruct<macho::Symbol64TableEntry> STE; |
| 258 | MachOObj->ReadSymbol64TableEntry((*SymtabLC)->SymbolTableOffset, i, |
| 259 | STE); |
| 260 | Symbols.push_back(copySymbol(STE)); |
| 261 | } |
| 262 | } |
| 263 | } |
| 264 | } else if (LCI.Command.Type == macho::LCT_FunctionStarts) { |
| 265 | // We found a function starts segment, parse the addresses for later |
| 266 | // consumption. |
| 267 | InMemoryStruct<macho::LinkeditDataLoadCommand> LLC; |
| 268 | MachOObj->ReadLinkeditDataLoadCommand(LCI, LLC); |
| 269 | |
| 270 | MachOObj->ReadULEB128s(LLC->DataOffset, FoundFns); |
| 271 | } |
| 272 | } |
| 273 | } |
| 274 | |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 275 | void llvm::DisassembleInputMachO(StringRef Filename) { |
| 276 | OwningPtr<MemoryBuffer> Buff; |
| 277 | |
| 278 | if (error_code ec = MemoryBuffer::getFileOrSTDIN(Filename, Buff)) { |
| 279 | errs() << "llvm-objdump: " << Filename << ": " << ec.message() << "\n"; |
| 280 | return; |
| 281 | } |
| 282 | |
| 283 | OwningPtr<MachOObject> MachOObj(MachOObject::LoadFromBuffer(Buff.take())); |
| 284 | |
| 285 | const Target *TheTarget = GetTarget(MachOObj.get()); |
| 286 | if (!TheTarget) { |
| 287 | // GetTarget prints out stuff. |
| 288 | return; |
| 289 | } |
| 290 | const MCInstrInfo *InstrInfo = TheTarget->createMCInstrInfo(); |
| 291 | OwningPtr<MCInstrAnalysis> |
| 292 | InstrAnalysis(TheTarget->createMCInstrAnalysis(InstrInfo)); |
| 293 | |
| 294 | // Set up disassembler. |
| 295 | OwningPtr<const MCAsmInfo> AsmInfo(TheTarget->createMCAsmInfo(TripleName)); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 296 | OwningPtr<const MCSubtargetInfo> |
| 297 | STI(TheTarget->createMCSubtargetInfo(TripleName, "", "")); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 298 | OwningPtr<const MCDisassembler> DisAsm(TheTarget->createMCDisassembler(*STI)); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 299 | int AsmPrinterVariant = AsmInfo->getAssemblerDialect(); |
| 300 | OwningPtr<MCInstPrinter> IP(TheTarget->createMCInstPrinter( |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 301 | AsmPrinterVariant, *AsmInfo, *STI)); |
| 302 | |
| 303 | if (!InstrAnalysis || !AsmInfo || !STI || !DisAsm || !IP) { |
| 304 | errs() << "error: couldn't initialize disassmbler for target " |
| 305 | << TripleName << '\n'; |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 306 | return; |
| 307 | } |
| 308 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 309 | outs() << '\n' << Filename << ":\n\n"; |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 310 | |
| 311 | const macho::Header &Header = MachOObj->getHeader(); |
| 312 | |
| 313 | const MachOObject::LoadCommandInfo *SymtabLCI = 0; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 314 | // First, find the symbol table segment. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 315 | for (unsigned i = 0; i != Header.NumLoadCommands; ++i) { |
| 316 | const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i); |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 317 | if (LCI.Command.Type == macho::LCT_Symtab) { |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 318 | SymtabLCI = &LCI; |
| 319 | break; |
| 320 | } |
| 321 | } |
| 322 | |
| 323 | // Read and register the symbol table data. |
| 324 | InMemoryStruct<macho::SymtabLoadCommand> SymtabLC; |
| 325 | MachOObj->ReadSymtabLoadCommand(*SymtabLCI, SymtabLC); |
| 326 | MachOObj->RegisterStringTable(*SymtabLC); |
| 327 | |
| 328 | std::vector<Section> Sections; |
| 329 | std::vector<Symbol> Symbols; |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 330 | SmallVector<uint64_t, 8> FoundFns; |
| 331 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 332 | getSectionsAndSymbols(Header, MachOObj.get(), &SymtabLC, Sections, Symbols, |
| 333 | FoundFns); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 334 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 335 | // Make a copy of the unsorted symbol list. FIXME: duplication |
| 336 | std::vector<Symbol> UnsortedSymbols(Symbols); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 337 | // Sort the symbols by address, just in case they didn't come in that way. |
| 338 | array_pod_sort(Symbols.begin(), Symbols.end()); |
| 339 | |
| 340 | #ifndef NDEBUG |
| 341 | raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls(); |
| 342 | #else |
| 343 | raw_ostream &DebugOut = nulls(); |
| 344 | #endif |
| 345 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 346 | StringRef DebugAbbrevSection, DebugInfoSection, DebugArangesSection, |
| 347 | DebugLineSection, DebugStrSection; |
| 348 | OwningPtr<DIContext> diContext; |
| 349 | // Try to find debug info and set up the DIContext for it. |
| 350 | if (UseDbg) { |
| 351 | ArrayRef<Section> DebugSections = Sections; |
| 352 | std::vector<Section> DSYMSections; |
| 353 | OwningPtr<MachOObject> DSYMObj; |
| 354 | |
| 355 | // A separate DSym file path was specified, parse it as a macho file, |
| 356 | // get the sections and supply it to the section name parsing machinery. |
| 357 | if (!DSYMFile.empty()) { |
| 358 | OwningPtr<MemoryBuffer> Buf; |
| 359 | if (error_code ec = MemoryBuffer::getFileOrSTDIN(DSYMFile.c_str(), Buf)) { |
| 360 | errs() << "llvm-objdump: " << Filename << ": " << ec.message() << '\n'; |
| 361 | return; |
| 362 | } |
| 363 | DSYMObj.reset(MachOObject::LoadFromBuffer(Buf.take())); |
| 364 | const macho::Header &Header = DSYMObj->getHeader(); |
| 365 | |
| 366 | std::vector<Symbol> Symbols; |
| 367 | SmallVector<uint64_t, 8> FoundFns; |
| 368 | getSectionsAndSymbols(Header, DSYMObj.get(), 0, DSYMSections, Symbols, |
| 369 | FoundFns); |
| 370 | DebugSections = DSYMSections; |
| 371 | } |
| 372 | |
| 373 | // Find the named debug info sections. |
| 374 | for (unsigned SectIdx = 0; SectIdx != DebugSections.size(); SectIdx++) { |
| 375 | if (!strcmp(DebugSections[SectIdx].Name, "__debug_abbrev")) |
| 376 | DebugAbbrevSection = DSYMObj->getData(DebugSections[SectIdx].Offset, |
| 377 | DebugSections[SectIdx].Size); |
| 378 | else if (!strcmp(DebugSections[SectIdx].Name, "__debug_info")) |
| 379 | DebugInfoSection = DSYMObj->getData(DebugSections[SectIdx].Offset, |
| 380 | DebugSections[SectIdx].Size); |
| 381 | else if (!strcmp(DebugSections[SectIdx].Name, "__debug_aranges")) |
| 382 | DebugArangesSection = DSYMObj->getData(DebugSections[SectIdx].Offset, |
| 383 | DebugSections[SectIdx].Size); |
| 384 | else if (!strcmp(DebugSections[SectIdx].Name, "__debug_line")) |
| 385 | DebugLineSection = DSYMObj->getData(DebugSections[SectIdx].Offset, |
| 386 | DebugSections[SectIdx].Size); |
| 387 | else if (!strcmp(DebugSections[SectIdx].Name, "__debug_str")) |
| 388 | DebugStrSection = DSYMObj->getData(DebugSections[SectIdx].Offset, |
| 389 | DebugSections[SectIdx].Size); |
| 390 | } |
| 391 | |
| 392 | // Setup the DIContext. |
| 393 | diContext.reset(DIContext::getDWARFContext(MachOObj->isLittleEndian(), |
| 394 | DebugInfoSection, |
| 395 | DebugAbbrevSection, |
| 396 | DebugArangesSection, |
| 397 | DebugLineSection, |
| 398 | DebugStrSection)); |
| 399 | } |
| 400 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 401 | FunctionMapTy FunctionMap; |
| 402 | FunctionListTy Functions; |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 403 | |
| 404 | for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) { |
| 405 | if (strcmp(Sections[SectIdx].Name, "__text")) |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 406 | continue; // Skip non-text sections |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 407 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 408 | // Insert the functions from the function starts segment into our map. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 409 | uint64_t VMAddr = Sections[SectIdx].Address - Sections[SectIdx].Offset; |
| 410 | for (unsigned i = 0, e = FoundFns.size(); i != e; ++i) |
Benjamin Kramer | 4d90638 | 2011-09-19 20:08:52 +0000 | [diff] [blame] | 411 | FunctionMap.insert(std::make_pair(FoundFns[i]+VMAddr, (MCFunction*)0)); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 412 | |
| 413 | StringRef Bytes = MachOObj->getData(Sections[SectIdx].Offset, |
| 414 | Sections[SectIdx].Size); |
| 415 | StringRefMemoryObject memoryObject(Bytes); |
| 416 | bool symbolTableWorked = false; |
| 417 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 418 | // Parse relocations. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 419 | std::vector<std::pair<uint64_t, uint32_t> > Relocs; |
| 420 | for (unsigned j = 0; j != Sections[SectIdx].NumRelocs; ++j) { |
| 421 | InMemoryStruct<macho::RelocationEntry> RE; |
| 422 | MachOObj->ReadRelocationEntry(Sections[SectIdx].RelocTableOffset, j, RE); |
| 423 | Relocs.push_back(std::make_pair(RE->Word0, RE->Word1 & 0xffffff)); |
| 424 | } |
| 425 | array_pod_sort(Relocs.begin(), Relocs.end()); |
| 426 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 427 | // Disassemble symbol by symbol. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 428 | for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) { |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 429 | // Make sure the symbol is defined in this section. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 430 | if ((unsigned)Symbols[SymIdx].SectionIndex - 1 != SectIdx) |
| 431 | continue; |
| 432 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 433 | // Start at the address of the symbol relative to the section's address. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 434 | uint64_t Start = Symbols[SymIdx].Value - Sections[SectIdx].Address; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 435 | // Stop disassembling either at the beginning of the next symbol or at |
| 436 | // the end of the section. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 437 | uint64_t End = (SymIdx+1 == Symbols.size() || |
| 438 | Symbols[SymIdx].SectionIndex != Symbols[SymIdx+1].SectionIndex) ? |
| 439 | Sections[SectIdx].Size : |
| 440 | Symbols[SymIdx+1].Value - Sections[SectIdx].Address; |
| 441 | uint64_t Size; |
| 442 | |
| 443 | if (Start >= End) |
| 444 | continue; |
| 445 | |
| 446 | symbolTableWorked = true; |
| 447 | |
| 448 | if (!CFG) { |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 449 | // Normal disassembly, print addresses, bytes and mnemonic form. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 450 | outs() << MachOObj->getStringAtIndex(Symbols[SymIdx].StringIndex) |
| 451 | << ":\n"; |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 452 | DILineInfo lastLine; |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 453 | for (uint64_t Index = Start; Index < End; Index += Size) { |
| 454 | MCInst Inst; |
| 455 | |
| 456 | if (DisAsm->getInstruction(Inst, Size, memoryObject, Index, |
| 457 | DebugOut, nulls())) { |
| 458 | outs() << format("%8llx:\t", Sections[SectIdx].Address + Index); |
| 459 | DumpBytes(StringRef(Bytes.data() + Index, Size)); |
| 460 | IP->printInst(&Inst, outs(), ""); |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 461 | |
| 462 | // Print debug info. |
| 463 | if (diContext) { |
| 464 | DILineInfo dli = |
| 465 | diContext->getLineInfoForAddress(Sections[SectIdx].Address + |
| 466 | Index); |
| 467 | // Print valid line info if it changed. |
| 468 | if (dli != lastLine && dli.getLine() != 0) |
| 469 | outs() << "\t## " << dli.getFileName() << ':' |
| 470 | << dli.getLine() << ':' << dli.getColumn(); |
| 471 | lastLine = dli; |
| 472 | } |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 473 | outs() << "\n"; |
| 474 | } else { |
| 475 | errs() << "llvm-objdump: warning: invalid instruction encoding\n"; |
| 476 | if (Size == 0) |
| 477 | Size = 1; // skip illegible bytes |
| 478 | } |
| 479 | } |
| 480 | } else { |
| 481 | // Create CFG and use it for disassembly. |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 482 | createMCFunctionAndSaveCalls( |
| 483 | MachOObj->getStringAtIndex(Symbols[SymIdx].StringIndex), |
| 484 | DisAsm.get(), memoryObject, Start, End, InstrAnalysis.get(), |
| 485 | Start, DebugOut, FunctionMap, Functions); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 486 | } |
| 487 | } |
| 488 | |
| 489 | if (CFG) { |
| 490 | if (!symbolTableWorked) { |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 491 | // Reading the symbol table didn't work, create a big __TEXT symbol. |
| 492 | createMCFunctionAndSaveCalls("__TEXT", DisAsm.get(), memoryObject, |
| 493 | 0, Sections[SectIdx].Size, |
| 494 | InstrAnalysis.get(), |
| 495 | Sections[SectIdx].Offset, DebugOut, |
| 496 | FunctionMap, Functions); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 497 | } |
| 498 | for (std::map<uint64_t, MCFunction*>::iterator mi = FunctionMap.begin(), |
| 499 | me = FunctionMap.end(); mi != me; ++mi) |
| 500 | if (mi->second == 0) { |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 501 | // Create functions for the remaining callees we have gathered, |
| 502 | // but we didn't find a name for them. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 503 | SmallVector<uint64_t, 16> Calls; |
| 504 | MCFunction f = |
| 505 | MCFunction::createFunctionFromMC("unknown", DisAsm.get(), |
| 506 | memoryObject, mi->first, |
| 507 | Sections[SectIdx].Size, |
| 508 | InstrAnalysis.get(), DebugOut, |
| 509 | Calls); |
| 510 | Functions.push_back(f); |
| 511 | mi->second = &Functions.back(); |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 512 | for (unsigned i = 0, e = Calls.size(); i != e; ++i) { |
| 513 | std::pair<uint64_t, MCFunction*> p(Calls[i], (MCFunction*)0); |
| 514 | if (FunctionMap.insert(p).second) |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 515 | mi = FunctionMap.begin(); |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 516 | } |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 517 | } |
| 518 | |
| 519 | DenseSet<uint64_t> PrintedBlocks; |
| 520 | for (unsigned ffi = 0, ffe = Functions.size(); ffi != ffe; ++ffi) { |
| 521 | MCFunction &f = Functions[ffi]; |
| 522 | for (MCFunction::iterator fi = f.begin(), fe = f.end(); fi != fe; ++fi){ |
| 523 | if (!PrintedBlocks.insert(fi->first).second) |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 524 | continue; // We already printed this block. |
| 525 | |
| 526 | // We assume a block has predecessors when it's the first block after |
| 527 | // a symbol. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 528 | bool hasPreds = FunctionMap.find(fi->first) != FunctionMap.end(); |
| 529 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 530 | // See if this block has predecessors. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 531 | // FIXME: Slow. |
| 532 | for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe; |
| 533 | ++pi) |
| 534 | if (pi->second.contains(fi->first)) { |
| 535 | hasPreds = true; |
| 536 | break; |
| 537 | } |
| 538 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 539 | // No predecessors, this is a data block. Print as .byte directives. |
| 540 | if (!hasPreds) { |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 541 | uint64_t End = llvm::next(fi) == fe ? Sections[SectIdx].Size : |
| 542 | llvm::next(fi)->first; |
| 543 | outs() << "# " << End-fi->first << " bytes of data:\n"; |
| 544 | for (unsigned pos = fi->first; pos != End; ++pos) { |
| 545 | outs() << format("%8x:\t", Sections[SectIdx].Address + pos); |
| 546 | DumpBytes(StringRef(Bytes.data() + pos, 1)); |
| 547 | outs() << format("\t.byte 0x%02x\n", (uint8_t)Bytes[pos]); |
| 548 | } |
| 549 | continue; |
| 550 | } |
| 551 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 552 | if (fi->second.contains(fi->first)) // Print a header for simple loops |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 553 | outs() << "# Loop begin:\n"; |
| 554 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 555 | DILineInfo lastLine; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 556 | // Walk over the instructions and print them. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 557 | for (unsigned ii = 0, ie = fi->second.getInsts().size(); ii != ie; |
| 558 | ++ii) { |
| 559 | const MCDecodedInst &Inst = fi->second.getInsts()[ii]; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 560 | |
| 561 | // If there's a symbol at this address, print its name. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 562 | if (FunctionMap.find(Sections[SectIdx].Address + Inst.Address) != |
| 563 | FunctionMap.end()) |
| 564 | outs() << FunctionMap[Sections[SectIdx].Address + Inst.Address]-> |
| 565 | getName() << ":\n"; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 566 | |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 567 | outs() << format("%8llx:\t", Sections[SectIdx].Address + |
| 568 | Inst.Address); |
| 569 | DumpBytes(StringRef(Bytes.data() + Inst.Address, Inst.Size)); |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 570 | |
| 571 | if (fi->second.contains(fi->first)) // Indent simple loops. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 572 | outs() << '\t'; |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 573 | |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 574 | IP->printInst(&Inst.Inst, outs(), ""); |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 575 | |
| 576 | // Look for relocations inside this instructions, if there is one |
| 577 | // print its target and additional information if availbable. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 578 | for (unsigned j = 0; j != Relocs.size(); ++j) |
| 579 | if (Relocs[j].first >= Sections[SectIdx].Address + Inst.Address && |
| 580 | Relocs[j].first < Sections[SectIdx].Address + Inst.Address + |
| 581 | Inst.Size) { |
| 582 | outs() << "\t# " |
| 583 | << MachOObj->getStringAtIndex( |
| 584 | UnsortedSymbols[Relocs[j].second].StringIndex) |
| 585 | << ' '; |
| 586 | DumpAddress(UnsortedSymbols[Relocs[j].second].Value, Sections, |
| 587 | MachOObj.get(), outs()); |
| 588 | } |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 589 | |
| 590 | // If this instructions contains an address, see if we can evaluate |
| 591 | // it and print additional information. |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 592 | uint64_t targ = InstrAnalysis->evaluateBranch(Inst.Inst, |
| 593 | Inst.Address, |
| 594 | Inst.Size); |
| 595 | if (targ != -1ULL) |
| 596 | DumpAddress(targ, Sections, MachOObj.get(), outs()); |
| 597 | |
Benjamin Kramer | 8c93097 | 2011-09-21 01:13:19 +0000 | [diff] [blame^] | 598 | // Print debug info. |
| 599 | if (diContext) { |
| 600 | DILineInfo dli = |
| 601 | diContext->getLineInfoForAddress(Sections[SectIdx].Address + |
| 602 | Inst.Address); |
| 603 | // Print valid line info if it changed. |
| 604 | if (dli != lastLine && dli.getLine() != 0) |
| 605 | outs() << "\t## " << dli.getFileName() << ':' |
| 606 | << dli.getLine() << ':' << dli.getColumn(); |
| 607 | lastLine = dli; |
| 608 | } |
| 609 | |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 610 | outs() << '\n'; |
| 611 | } |
| 612 | } |
| 613 | |
Benjamin Kramer | a894c8e | 2011-09-20 17:53:01 +0000 | [diff] [blame] | 614 | emitDOTFile((f.getName().str() + ".dot").c_str(), f, IP.get()); |
Benjamin Kramer | 0b8b771 | 2011-09-19 17:56:04 +0000 | [diff] [blame] | 615 | } |
| 616 | } |
| 617 | } |
| 618 | } |