blob: 6ea03232edd47e21b020ddf4ab1e0abf59053fd5 [file] [log] [blame]
Benjamin Kramer0b8b7712011-09-19 17:56:04 +00001//===-- 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"
Owen Anderson481837a2011-10-17 21:37:35 +000017#include "llvm/Object/MachO.h"
Benjamin Kramer0b8b7712011-09-19 17:56:04 +000018#include "llvm/ADT/OwningPtr.h"
19#include "llvm/ADT/Triple.h"
20#include "llvm/ADT/STLExtras.h"
Benjamin Kramer8c930972011-09-21 01:13:19 +000021#include "llvm/DebugInfo/DIContext.h"
Benjamin Kramer0b8b7712011-09-19 17:56:04 +000022#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>
41using namespace llvm;
42using namespace object;
43
44static 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 Kramer8c930972011-09-21 01:13:19 +000048static cl::opt<bool>
49 UseDbg("g", cl::desc("Print line information from debug info if available"));
50
51static cl::opt<std::string>
52 DSYMFile("dsym", cl::desc("Use .dSYM file for debug info"));
53
Benjamin Kramer0b8b7712011-09-19 17:56:04 +000054static 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
Owen Anderson481837a2011-10-17 21:37:35 +000088struct SymbolSorter {
89 bool operator()(const SymbolRef &A, const SymbolRef &B) {
90 SymbolRef::Type AType, BType;
91 A.getType(AType);
92 B.getType(BType);
93
94 uint64_t AAddr, BAddr;
95 if (AType != SymbolRef::ST_Function)
96 AAddr = 0;
97 else
98 A.getAddress(AAddr);
99 if (BType != SymbolRef::ST_Function)
100 BAddr = 0;
101 else
102 B.getAddress(BAddr);
103 return AAddr < BAddr;
104 }
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000105};
106
Michael J. Spencer3773fb42011-10-07 19:25:47 +0000107// Print additional information about an address, if available.
Owen Anderson481837a2011-10-17 21:37:35 +0000108static void DumpAddress(uint64_t Address, ArrayRef<SectionRef> Sections,
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000109 MachOObject *MachOObj, raw_ostream &OS) {
110 for (unsigned i = 0; i != Sections.size(); ++i) {
Owen Anderson481837a2011-10-17 21:37:35 +0000111 uint64_t SectAddr = 0, SectSize = 0;
112 Sections[i].getAddress(SectAddr);
113 Sections[i].getSize(SectSize);
114 uint64_t addr = SectAddr;
115 if (SectAddr <= Address &&
116 SectAddr + SectSize > Address) {
117 StringRef bytes, name;
118 Sections[i].getContents(bytes);
119 Sections[i].getName(name);
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000120 // Print constant strings.
Owen Anderson481837a2011-10-17 21:37:35 +0000121 if (!name.compare("__cstring"))
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000122 OS << '"' << bytes.substr(addr, bytes.find('\0', addr)) << '"';
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000123 // Print constant CFStrings.
Owen Anderson481837a2011-10-17 21:37:35 +0000124 if (!name.compare("__cfstring"))
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000125 OS << "@\"" << bytes.substr(addr, bytes.find('\0', addr)) << '"';
126 }
127 }
128}
129
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000130typedef std::map<uint64_t, MCFunction*> FunctionMapTy;
131typedef SmallVector<MCFunction, 16> FunctionListTy;
132static void createMCFunctionAndSaveCalls(StringRef Name,
133 const MCDisassembler *DisAsm,
134 MemoryObject &Object, uint64_t Start,
135 uint64_t End,
136 MCInstrAnalysis *InstrAnalysis,
137 uint64_t Address,
138 raw_ostream &DebugOut,
139 FunctionMapTy &FunctionMap,
140 FunctionListTy &Functions) {
141 SmallVector<uint64_t, 16> Calls;
142 MCFunction f =
143 MCFunction::createFunctionFromMC(Name, DisAsm, Object, Start, End,
144 InstrAnalysis, DebugOut, Calls);
145 Functions.push_back(f);
146 FunctionMap[Address] = &Functions.back();
147
148 // Add the gathered callees to the map.
149 for (unsigned i = 0, e = Calls.size(); i != e; ++i)
150 FunctionMap.insert(std::make_pair(Calls[i], (MCFunction*)0));
151}
152
153// Write a graphviz file for the CFG inside an MCFunction.
154static void emitDOTFile(const char *FileName, const MCFunction &f,
155 MCInstPrinter *IP) {
156 // Start a new dot file.
157 std::string Error;
158 raw_fd_ostream Out(FileName, Error);
159 if (!Error.empty()) {
160 errs() << "llvm-objdump: warning: " << Error << '\n';
161 return;
162 }
163
164 Out << "digraph " << f.getName() << " {\n";
165 Out << "graph [ rankdir = \"LR\" ];\n";
166 for (MCFunction::iterator i = f.begin(), e = f.end(); i != e; ++i) {
167 bool hasPreds = false;
168 // Only print blocks that have predecessors.
169 // FIXME: Slow.
170 for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe;
171 ++pi)
172 if (pi->second.contains(i->first)) {
173 hasPreds = true;
174 break;
175 }
176
177 if (!hasPreds && i != f.begin())
178 continue;
179
180 Out << '"' << i->first << "\" [ label=\"<a>";
181 // Print instructions.
182 for (unsigned ii = 0, ie = i->second.getInsts().size(); ii != ie;
183 ++ii) {
184 // Escape special chars and print the instruction in mnemonic form.
185 std::string Str;
186 raw_string_ostream OS(Str);
187 IP->printInst(&i->second.getInsts()[ii].Inst, OS, "");
188 Out << DOT::EscapeString(OS.str()) << '|';
189 }
190 Out << "<o>\" shape=\"record\" ];\n";
191
192 // Add edges.
193 for (MCBasicBlock::succ_iterator si = i->second.succ_begin(),
194 se = i->second.succ_end(); si != se; ++si)
195 Out << i->first << ":o -> " << *si <<":a\n";
196 }
197 Out << "}\n";
198}
199
Benjamin Kramer8c930972011-09-21 01:13:19 +0000200static void getSectionsAndSymbols(const macho::Header &Header,
Owen Anderson481837a2011-10-17 21:37:35 +0000201 MachOObjectFile *MachOObj,
Benjamin Kramer8c930972011-09-21 01:13:19 +0000202 InMemoryStruct<macho::SymtabLoadCommand> *SymtabLC,
Owen Anderson481837a2011-10-17 21:37:35 +0000203 std::vector<SectionRef> &Sections,
204 std::vector<SymbolRef> &Symbols,
Benjamin Kramer8c930972011-09-21 01:13:19 +0000205 SmallVectorImpl<uint64_t> &FoundFns) {
Owen Anderson481837a2011-10-17 21:37:35 +0000206 error_code ec;
207 for (symbol_iterator SI = MachOObj->begin_symbols(),
208 SE = MachOObj->end_symbols(); SI != SE; SI.increment(ec))
209 Symbols.push_back(*SI);
210
211 for (section_iterator SI = MachOObj->begin_sections(),
212 SE = MachOObj->end_sections(); SI != SE; SI.increment(ec)) {
213 SectionRef SR = *SI;
214 StringRef SectName;
215 SR.getName(SectName);
216 Sections.push_back(*SI);
217 }
218
Benjamin Kramer8c930972011-09-21 01:13:19 +0000219 for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
Owen Anderson481837a2011-10-17 21:37:35 +0000220 const MachOObject::LoadCommandInfo &LCI =
221 MachOObj->getObject()->getLoadCommandInfo(i);
222 if (LCI.Command.Type == macho::LCT_FunctionStarts) {
Benjamin Kramer8c930972011-09-21 01:13:19 +0000223 // We found a function starts segment, parse the addresses for later
224 // consumption.
225 InMemoryStruct<macho::LinkeditDataLoadCommand> LLC;
Owen Anderson481837a2011-10-17 21:37:35 +0000226 MachOObj->getObject()->ReadLinkeditDataLoadCommand(LCI, LLC);
Benjamin Kramer8c930972011-09-21 01:13:19 +0000227
Owen Anderson481837a2011-10-17 21:37:35 +0000228 MachOObj->getObject()->ReadULEB128s(LLC->DataOffset, FoundFns);
Benjamin Kramerafbaf482011-09-21 22:16:43 +0000229 }
230 }
Benjamin Kramer8c930972011-09-21 01:13:19 +0000231}
232
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000233void llvm::DisassembleInputMachO(StringRef Filename) {
234 OwningPtr<MemoryBuffer> Buff;
235
236 if (error_code ec = MemoryBuffer::getFileOrSTDIN(Filename, Buff)) {
237 errs() << "llvm-objdump: " << Filename << ": " << ec.message() << "\n";
238 return;
239 }
240
Owen Anderson481837a2011-10-17 21:37:35 +0000241 OwningPtr<MachOObjectFile> MachOOF(static_cast<MachOObjectFile*>(
242 ObjectFile::createMachOObjectFile(Buff.take())));
243 MachOObject *MachOObj = MachOOF->getObject();
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000244
Owen Anderson481837a2011-10-17 21:37:35 +0000245 const Target *TheTarget = GetTarget(MachOObj);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000246 if (!TheTarget) {
247 // GetTarget prints out stuff.
248 return;
249 }
Benjamin Kramerd226ed712011-10-10 13:10:09 +0000250 OwningPtr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000251 OwningPtr<MCInstrAnalysis>
Benjamin Kramerd226ed712011-10-10 13:10:09 +0000252 InstrAnalysis(TheTarget->createMCInstrAnalysis(InstrInfo.get()));
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000253
254 // Set up disassembler.
255 OwningPtr<const MCAsmInfo> AsmInfo(TheTarget->createMCAsmInfo(TripleName));
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000256 OwningPtr<const MCSubtargetInfo>
257 STI(TheTarget->createMCSubtargetInfo(TripleName, "", ""));
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000258 OwningPtr<const MCDisassembler> DisAsm(TheTarget->createMCDisassembler(*STI));
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000259 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
260 OwningPtr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000261 AsmPrinterVariant, *AsmInfo, *STI));
262
263 if (!InstrAnalysis || !AsmInfo || !STI || !DisAsm || !IP) {
Michael J. Spencer3773fb42011-10-07 19:25:47 +0000264 errs() << "error: couldn't initialize disassembler for target "
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000265 << TripleName << '\n';
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000266 return;
267 }
268
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000269 outs() << '\n' << Filename << ":\n\n";
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000270
271 const macho::Header &Header = MachOObj->getHeader();
272
273 const MachOObject::LoadCommandInfo *SymtabLCI = 0;
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000274 // First, find the symbol table segment.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000275 for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
276 const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i);
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000277 if (LCI.Command.Type == macho::LCT_Symtab) {
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000278 SymtabLCI = &LCI;
279 break;
280 }
281 }
282
283 // Read and register the symbol table data.
284 InMemoryStruct<macho::SymtabLoadCommand> SymtabLC;
285 MachOObj->ReadSymtabLoadCommand(*SymtabLCI, SymtabLC);
286 MachOObj->RegisterStringTable(*SymtabLC);
287
Owen Anderson481837a2011-10-17 21:37:35 +0000288 std::vector<SectionRef> Sections;
289 std::vector<SymbolRef> Symbols;
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000290 SmallVector<uint64_t, 8> FoundFns;
291
Owen Anderson481837a2011-10-17 21:37:35 +0000292 getSectionsAndSymbols(Header, MachOOF.get(), &SymtabLC, Sections, Symbols,
Benjamin Kramer8c930972011-09-21 01:13:19 +0000293 FoundFns);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000294
Benjamin Kramer8c930972011-09-21 01:13:19 +0000295 // Make a copy of the unsorted symbol list. FIXME: duplication
Owen Anderson481837a2011-10-17 21:37:35 +0000296 std::vector<SymbolRef> UnsortedSymbols(Symbols);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000297 // Sort the symbols by address, just in case they didn't come in that way.
Owen Anderson481837a2011-10-17 21:37:35 +0000298 std::sort(Symbols.begin(), Symbols.end(), SymbolSorter());
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000299
300#ifndef NDEBUG
301 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
302#else
303 raw_ostream &DebugOut = nulls();
304#endif
305
Benjamin Kramer8c930972011-09-21 01:13:19 +0000306 StringRef DebugAbbrevSection, DebugInfoSection, DebugArangesSection,
307 DebugLineSection, DebugStrSection;
308 OwningPtr<DIContext> diContext;
Owen Anderson481837a2011-10-17 21:37:35 +0000309 OwningPtr<MachOObjectFile> DSYMObj;
310 MachOObject *DbgInfoObj = MachOObj;
Benjamin Kramer8c930972011-09-21 01:13:19 +0000311 // Try to find debug info and set up the DIContext for it.
312 if (UseDbg) {
Owen Anderson481837a2011-10-17 21:37:35 +0000313 ArrayRef<SectionRef> DebugSections = Sections;
314 std::vector<SectionRef> DSYMSections;
Benjamin Kramer8c930972011-09-21 01:13:19 +0000315
316 // A separate DSym file path was specified, parse it as a macho file,
317 // get the sections and supply it to the section name parsing machinery.
318 if (!DSYMFile.empty()) {
319 OwningPtr<MemoryBuffer> Buf;
320 if (error_code ec = MemoryBuffer::getFileOrSTDIN(DSYMFile.c_str(), Buf)) {
321 errs() << "llvm-objdump: " << Filename << ": " << ec.message() << '\n';
322 return;
323 }
Owen Anderson481837a2011-10-17 21:37:35 +0000324 DSYMObj.reset(static_cast<MachOObjectFile*>(
325 ObjectFile::createMachOObjectFile(Buf.take())));
326 const macho::Header &Header = DSYMObj->getObject()->getHeader();
Benjamin Kramer8c930972011-09-21 01:13:19 +0000327
Owen Anderson481837a2011-10-17 21:37:35 +0000328 std::vector<SymbolRef> Symbols;
Benjamin Kramer8c930972011-09-21 01:13:19 +0000329 SmallVector<uint64_t, 8> FoundFns;
330 getSectionsAndSymbols(Header, DSYMObj.get(), 0, DSYMSections, Symbols,
331 FoundFns);
332 DebugSections = DSYMSections;
Owen Anderson481837a2011-10-17 21:37:35 +0000333 DbgInfoObj = DSYMObj.get()->getObject();
Benjamin Kramer8c930972011-09-21 01:13:19 +0000334 }
335
336 // Find the named debug info sections.
337 for (unsigned SectIdx = 0; SectIdx != DebugSections.size(); SectIdx++) {
Owen Anderson481837a2011-10-17 21:37:35 +0000338 StringRef SectName;
339 if (!DebugSections[SectIdx].getName(SectName)) {
340 if (SectName.equals("__DWARF,__debug_abbrev"))
341 DebugSections[SectIdx].getContents(DebugAbbrevSection);
342 else if (SectName.equals("__DWARF,__debug_info"))
343 DebugSections[SectIdx].getContents(DebugInfoSection);
344 else if (SectName.equals("__DWARF,__debug_aranges"))
345 DebugSections[SectIdx].getContents(DebugArangesSection);
346 else if (SectName.equals("__DWARF,__debug_line"))
347 DebugSections[SectIdx].getContents(DebugLineSection);
348 else if (SectName.equals("__DWARF,__debug_str"))
349 DebugSections[SectIdx].getContents(DebugStrSection);
350 }
Benjamin Kramer8c930972011-09-21 01:13:19 +0000351 }
352
353 // Setup the DIContext.
Benjamin Kramer91c603b2011-09-21 18:18:53 +0000354 diContext.reset(DIContext::getDWARFContext(DbgInfoObj->isLittleEndian(),
Benjamin Kramer8c930972011-09-21 01:13:19 +0000355 DebugInfoSection,
356 DebugAbbrevSection,
357 DebugArangesSection,
358 DebugLineSection,
359 DebugStrSection));
360 }
361
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000362 FunctionMapTy FunctionMap;
363 FunctionListTy Functions;
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000364
365 for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
Owen Anderson481837a2011-10-17 21:37:35 +0000366 StringRef SectName;
367 if (Sections[SectIdx].getName(SectName) ||
368 SectName.compare("__TEXT,__text"))
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000369 continue; // Skip non-text sections
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000370
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000371 // Insert the functions from the function starts segment into our map.
Owen Anderson481837a2011-10-17 21:37:35 +0000372 uint64_t VMAddr;
373 Sections[SectIdx].getAddress(VMAddr);
374 for (unsigned i = 0, e = FoundFns.size(); i != e; ++i) {
375 StringRef SectBegin;
376 Sections[SectIdx].getContents(SectBegin);
377 uint64_t Offset = (uint64_t)SectBegin.data();
378 FunctionMap.insert(std::make_pair(VMAddr + FoundFns[i]-Offset,
379 (MCFunction*)0));
380 }
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000381
Owen Anderson481837a2011-10-17 21:37:35 +0000382 StringRef Bytes;
383 Sections[SectIdx].getContents(Bytes);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000384 StringRefMemoryObject memoryObject(Bytes);
385 bool symbolTableWorked = false;
386
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000387 // Parse relocations.
Owen Anderson7d3f8b82011-11-07 17:21:36 +0000388 std::vector<std::pair<uint64_t, SymbolRef> > Relocs;
Owen Anderson481837a2011-10-17 21:37:35 +0000389 error_code ec;
390 for (relocation_iterator RI = Sections[SectIdx].begin_relocations(),
391 RE = Sections[SectIdx].end_relocations(); RI != RE; RI.increment(ec)) {
392 uint64_t RelocOffset, SectionAddress;
393 RI->getAddress(RelocOffset);
394 Sections[SectIdx].getAddress(SectionAddress);
395 RelocOffset -= SectionAddress;
396
Owen Anderson7d3f8b82011-11-07 17:21:36 +0000397 SymbolRef RelocSym;
398 RI->getSymbol(RelocSym);
Owen Anderson481837a2011-10-17 21:37:35 +0000399
Owen Anderson7d3f8b82011-11-07 17:21:36 +0000400 Relocs.push_back(std::make_pair(RelocOffset, RelocSym));
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000401 }
402 array_pod_sort(Relocs.begin(), Relocs.end());
403
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000404 // Disassemble symbol by symbol.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000405 for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
Owen Anderson481837a2011-10-17 21:37:35 +0000406 StringRef SymName;
407 Symbols[SymIdx].getName(SymName);
408
409 SymbolRef::Type ST;
410 Symbols[SymIdx].getType(ST);
411 if (ST != SymbolRef::ST_Function)
412 continue;
413
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000414 // Make sure the symbol is defined in this section.
Owen Anderson481837a2011-10-17 21:37:35 +0000415 bool containsSym = false;
416 Sections[SectIdx].containsSymbol(Symbols[SymIdx], containsSym);
417 if (!containsSym)
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000418 continue;
419
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000420 // Start at the address of the symbol relative to the section's address.
Cameron Zwarichec8eac62012-02-03 05:42:17 +0000421 uint64_t SectionAddress = 0;
Owen Anderson481837a2011-10-17 21:37:35 +0000422 uint64_t Start = 0;
Cameron Zwarichec8eac62012-02-03 05:42:17 +0000423 Sections[SectIdx].getAddress(SectionAddress);
Danil Malyshevb0436a72011-11-29 17:40:10 +0000424 Symbols[SymIdx].getAddress(Start);
Cameron Zwarichec8eac62012-02-03 05:42:17 +0000425 Start -= SectionAddress;
Owen Anderson481837a2011-10-17 21:37:35 +0000426
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000427 // Stop disassembling either at the beginning of the next symbol or at
428 // the end of the section.
Owen Anderson481837a2011-10-17 21:37:35 +0000429 bool containsNextSym = true;
430 uint64_t NextSym = 0;
431 uint64_t NextSymIdx = SymIdx+1;
432 while (Symbols.size() > NextSymIdx) {
433 SymbolRef::Type NextSymType;
434 Symbols[NextSymIdx].getType(NextSymType);
435 if (NextSymType == SymbolRef::ST_Function) {
436 Sections[SectIdx].containsSymbol(Symbols[NextSymIdx],
437 containsNextSym);
Danil Malyshevb0436a72011-11-29 17:40:10 +0000438 Symbols[NextSymIdx].getAddress(NextSym);
Cameron Zwarichec8eac62012-02-03 05:42:17 +0000439 NextSym -= SectionAddress;
Owen Anderson481837a2011-10-17 21:37:35 +0000440 break;
441 }
442 ++NextSymIdx;
443 }
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000444
Owen Anderson481837a2011-10-17 21:37:35 +0000445 uint64_t SectSize;
446 Sections[SectIdx].getSize(SectSize);
447 uint64_t End = containsNextSym ? NextSym : SectSize;
448 uint64_t Size;
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000449
450 symbolTableWorked = true;
451
452 if (!CFG) {
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000453 // Normal disassembly, print addresses, bytes and mnemonic form.
Owen Anderson481837a2011-10-17 21:37:35 +0000454 StringRef SymName;
455 Symbols[SymIdx].getName(SymName);
456
457 outs() << SymName << ":\n";
Benjamin Kramer8c930972011-09-21 01:13:19 +0000458 DILineInfo lastLine;
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000459 for (uint64_t Index = Start; Index < End; Index += Size) {
460 MCInst Inst;
461
462 if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
463 DebugOut, nulls())) {
Owen Anderson481837a2011-10-17 21:37:35 +0000464 uint64_t SectAddress = 0;
465 Sections[SectIdx].getAddress(SectAddress);
Benjamin Kramer41a96492011-11-05 08:57:40 +0000466 outs() << format("%8" PRIx64 ":\t", SectAddress + Index);
Owen Anderson481837a2011-10-17 21:37:35 +0000467
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000468 DumpBytes(StringRef(Bytes.data() + Index, Size));
469 IP->printInst(&Inst, outs(), "");
Benjamin Kramer8c930972011-09-21 01:13:19 +0000470
471 // Print debug info.
472 if (diContext) {
473 DILineInfo dli =
Owen Anderson481837a2011-10-17 21:37:35 +0000474 diContext->getLineInfoForAddress(SectAddress + Index);
Benjamin Kramer8c930972011-09-21 01:13:19 +0000475 // Print valid line info if it changed.
476 if (dli != lastLine && dli.getLine() != 0)
477 outs() << "\t## " << dli.getFileName() << ':'
478 << dli.getLine() << ':' << dli.getColumn();
479 lastLine = dli;
480 }
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000481 outs() << "\n";
482 } else {
483 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
484 if (Size == 0)
485 Size = 1; // skip illegible bytes
486 }
487 }
488 } else {
489 // Create CFG and use it for disassembly.
Owen Anderson481837a2011-10-17 21:37:35 +0000490 StringRef SymName;
491 Symbols[SymIdx].getName(SymName);
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000492 createMCFunctionAndSaveCalls(
Owen Anderson481837a2011-10-17 21:37:35 +0000493 SymName, DisAsm.get(), memoryObject, Start, End,
494 InstrAnalysis.get(), Start, DebugOut, FunctionMap, Functions);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000495 }
496 }
497
498 if (CFG) {
499 if (!symbolTableWorked) {
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000500 // Reading the symbol table didn't work, create a big __TEXT symbol.
Owen Anderson481837a2011-10-17 21:37:35 +0000501 uint64_t SectSize = 0, SectAddress = 0;
502 Sections[SectIdx].getSize(SectSize);
503 Sections[SectIdx].getAddress(SectAddress);
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000504 createMCFunctionAndSaveCalls("__TEXT", DisAsm.get(), memoryObject,
Owen Anderson481837a2011-10-17 21:37:35 +0000505 0, SectSize,
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000506 InstrAnalysis.get(),
Owen Anderson481837a2011-10-17 21:37:35 +0000507 SectAddress, DebugOut,
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000508 FunctionMap, Functions);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000509 }
510 for (std::map<uint64_t, MCFunction*>::iterator mi = FunctionMap.begin(),
511 me = FunctionMap.end(); mi != me; ++mi)
512 if (mi->second == 0) {
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000513 // Create functions for the remaining callees we have gathered,
514 // but we didn't find a name for them.
Owen Anderson481837a2011-10-17 21:37:35 +0000515 uint64_t SectSize = 0;
516 Sections[SectIdx].getSize(SectSize);
517
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000518 SmallVector<uint64_t, 16> Calls;
519 MCFunction f =
520 MCFunction::createFunctionFromMC("unknown", DisAsm.get(),
521 memoryObject, mi->first,
Owen Anderson481837a2011-10-17 21:37:35 +0000522 SectSize,
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000523 InstrAnalysis.get(), DebugOut,
524 Calls);
525 Functions.push_back(f);
526 mi->second = &Functions.back();
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000527 for (unsigned i = 0, e = Calls.size(); i != e; ++i) {
528 std::pair<uint64_t, MCFunction*> p(Calls[i], (MCFunction*)0);
529 if (FunctionMap.insert(p).second)
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000530 mi = FunctionMap.begin();
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000531 }
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000532 }
533
534 DenseSet<uint64_t> PrintedBlocks;
535 for (unsigned ffi = 0, ffe = Functions.size(); ffi != ffe; ++ffi) {
536 MCFunction &f = Functions[ffi];
537 for (MCFunction::iterator fi = f.begin(), fe = f.end(); fi != fe; ++fi){
538 if (!PrintedBlocks.insert(fi->first).second)
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000539 continue; // We already printed this block.
540
541 // We assume a block has predecessors when it's the first block after
542 // a symbol.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000543 bool hasPreds = FunctionMap.find(fi->first) != FunctionMap.end();
544
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000545 // See if this block has predecessors.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000546 // FIXME: Slow.
547 for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe;
548 ++pi)
549 if (pi->second.contains(fi->first)) {
550 hasPreds = true;
551 break;
552 }
553
Owen Anderson481837a2011-10-17 21:37:35 +0000554 uint64_t SectSize = 0, SectAddress;
555 Sections[SectIdx].getSize(SectSize);
556 Sections[SectIdx].getAddress(SectAddress);
557
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000558 // No predecessors, this is a data block. Print as .byte directives.
559 if (!hasPreds) {
Owen Anderson481837a2011-10-17 21:37:35 +0000560 uint64_t End = llvm::next(fi) == fe ? SectSize :
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000561 llvm::next(fi)->first;
562 outs() << "# " << End-fi->first << " bytes of data:\n";
563 for (unsigned pos = fi->first; pos != End; ++pos) {
Owen Anderson481837a2011-10-17 21:37:35 +0000564 outs() << format("%8x:\t", SectAddress + pos);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000565 DumpBytes(StringRef(Bytes.data() + pos, 1));
566 outs() << format("\t.byte 0x%02x\n", (uint8_t)Bytes[pos]);
567 }
568 continue;
569 }
570
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000571 if (fi->second.contains(fi->first)) // Print a header for simple loops
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000572 outs() << "# Loop begin:\n";
573
Benjamin Kramer8c930972011-09-21 01:13:19 +0000574 DILineInfo lastLine;
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000575 // Walk over the instructions and print them.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000576 for (unsigned ii = 0, ie = fi->second.getInsts().size(); ii != ie;
577 ++ii) {
578 const MCDecodedInst &Inst = fi->second.getInsts()[ii];
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000579
580 // If there's a symbol at this address, print its name.
Owen Anderson481837a2011-10-17 21:37:35 +0000581 if (FunctionMap.find(SectAddress + Inst.Address) !=
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000582 FunctionMap.end())
Owen Anderson481837a2011-10-17 21:37:35 +0000583 outs() << FunctionMap[SectAddress + Inst.Address]-> getName()
584 << ":\n";
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000585
Benjamin Kramer41a96492011-11-05 08:57:40 +0000586 outs() << format("%8" PRIx64 ":\t", SectAddress + Inst.Address);
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000587 DumpBytes(StringRef(Bytes.data() + Inst.Address, Inst.Size));
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000588
589 if (fi->second.contains(fi->first)) // Indent simple loops.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000590 outs() << '\t';
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000591
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000592 IP->printInst(&Inst.Inst, outs(), "");
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000593
594 // Look for relocations inside this instructions, if there is one
Michael J. Spencer3773fb42011-10-07 19:25:47 +0000595 // print its target and additional information if available.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000596 for (unsigned j = 0; j != Relocs.size(); ++j)
Owen Anderson481837a2011-10-17 21:37:35 +0000597 if (Relocs[j].first >= SectAddress + Inst.Address &&
598 Relocs[j].first < SectAddress + Inst.Address + Inst.Size) {
599 StringRef SymName;
600 uint64_t Addr;
Owen Anderson7d3f8b82011-11-07 17:21:36 +0000601 Relocs[j].second.getAddress(Addr);
602 Relocs[j].second.getName(SymName);
Owen Anderson481837a2011-10-17 21:37:35 +0000603
604 outs() << "\t# " << SymName << ' ';
605 DumpAddress(Addr, Sections, MachOObj, outs());
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000606 }
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000607
608 // If this instructions contains an address, see if we can evaluate
609 // it and print additional information.
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000610 uint64_t targ = InstrAnalysis->evaluateBranch(Inst.Inst,
611 Inst.Address,
612 Inst.Size);
613 if (targ != -1ULL)
Owen Anderson481837a2011-10-17 21:37:35 +0000614 DumpAddress(targ, Sections, MachOObj, outs());
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000615
Benjamin Kramer8c930972011-09-21 01:13:19 +0000616 // Print debug info.
617 if (diContext) {
618 DILineInfo dli =
Owen Anderson481837a2011-10-17 21:37:35 +0000619 diContext->getLineInfoForAddress(SectAddress + Inst.Address);
Benjamin Kramer8c930972011-09-21 01:13:19 +0000620 // Print valid line info if it changed.
621 if (dli != lastLine && dli.getLine() != 0)
622 outs() << "\t## " << dli.getFileName() << ':'
623 << dli.getLine() << ':' << dli.getColumn();
624 lastLine = dli;
625 }
626
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000627 outs() << '\n';
628 }
629 }
630
Benjamin Kramera894c8e2011-09-20 17:53:01 +0000631 emitDOTFile((f.getName().str() + ".dot").c_str(), f, IP.get());
Benjamin Kramer0b8b7712011-09-19 17:56:04 +0000632 }
633 }
634 }
635}