1 //===-- llvm-objdump.cpp - Object file dumping utility for llvm -----------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This program is a utility that works like binutils "objdump", that is, it
11 // dumps out a plethora of information about an object file depending on the
14 //===----------------------------------------------------------------------===//
16 #include "llvm-objdump.h"
17 #include "MCFunction.h"
18 #include "llvm/Object/Archive.h"
19 #include "llvm/Object/COFF.h"
20 #include "llvm/Object/ObjectFile.h"
21 #include "llvm/ADT/OwningPtr.h"
22 #include "llvm/ADT/StringExtras.h"
23 #include "llvm/ADT/Triple.h"
24 #include "llvm/ADT/STLExtras.h"
25 #include "llvm/MC/MCAsmInfo.h"
26 #include "llvm/MC/MCContext.h"
27 #include "llvm/MC/MCDisassembler.h"
28 #include "llvm/MC/MCInst.h"
29 #include "llvm/MC/MCInstPrinter.h"
30 #include "llvm/MC/MCRegisterInfo.h"
31 #include "llvm/MC/MCSubtargetInfo.h"
32 #include "llvm/Support/Casting.h"
33 #include "llvm/Support/CommandLine.h"
34 #include "llvm/Support/Debug.h"
35 #include "llvm/Support/FileSystem.h"
36 #include "llvm/Support/Format.h"
37 #include "llvm/Support/GraphWriter.h"
38 #include "llvm/Support/Host.h"
39 #include "llvm/Support/ManagedStatic.h"
40 #include "llvm/Support/MemoryBuffer.h"
41 #include "llvm/Support/MemoryObject.h"
42 #include "llvm/Support/PrettyStackTrace.h"
43 #include "llvm/Support/Signals.h"
44 #include "llvm/Support/SourceMgr.h"
45 #include "llvm/Support/TargetRegistry.h"
46 #include "llvm/Support/TargetSelect.h"
47 #include "llvm/Support/raw_ostream.h"
48 #include "llvm/Support/system_error.h"
52 using namespace object;
54 static cl::list<std::string>
55 InputFilenames(cl::Positional, cl::desc("<input object files>"),cl::ZeroOrMore);
58 Disassemble("disassemble",
59 cl::desc("Display assembler mnemonics for the machine instructions"));
61 Disassembled("d", cl::desc("Alias for --disassemble"),
62 cl::aliasopt(Disassemble));
65 Relocations("r", cl::desc("Display the relocation entries in the file"));
68 SectionContents("s", cl::desc("Display the content of each section"));
71 SymbolTable("t", cl::desc("Display the symbol table"));
74 MachO("macho", cl::desc("Use MachO specific object file parser"));
76 MachOm("m", cl::desc("Alias for --macho"), cl::aliasopt(MachO));
79 llvm::TripleName("triple", cl::desc("Target triple to disassemble for, "
80 "see -version for available targets"));
83 llvm::ArchName("arch", cl::desc("Target arch to disassemble for, "
84 "see -version for available targets"));
87 SectionHeaders("section-headers", cl::desc("Display summaries of the headers "
88 "for each section."));
90 SectionHeadersShort("headers", cl::desc("Alias for --section-headers"),
91 cl::aliasopt(SectionHeaders));
93 SectionHeadersShorter("h", cl::desc("Alias for --section-headers"),
94 cl::aliasopt(SectionHeaders));
96 static StringRef ToolName;
98 static bool error(error_code ec) {
99 if (!ec) return false;
101 outs() << ToolName << ": error reading file: " << ec.message() << ".\n";
106 static const Target *GetTarget(const ObjectFile *Obj = NULL) {
107 // Figure out the target triple.
108 llvm::Triple TT("unknown-unknown-unknown");
109 if (TripleName.empty()) {
111 TT.setArch(Triple::ArchType(Obj->getArch()));
113 TT.setTriple(Triple::normalize(TripleName));
115 if (!ArchName.empty())
116 TT.setArchName(ArchName);
118 TripleName = TT.str();
120 // Get the target specific parser.
122 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
126 errs() << ToolName << ": error: unable to get target for '" << TripleName
127 << "', see --version and --triple.\n";
131 void llvm::DumpBytes(StringRef bytes) {
132 static const char hex_rep[] = "0123456789abcdef";
133 // FIXME: The real way to do this is to figure out the longest instruction
134 // and align to that size before printing. I'll fix this when I get
135 // around to outputting relocations.
136 // 15 is the longest x86 instruction
137 // 3 is for the hex rep of a byte + a space.
138 // 1 is for the null terminator.
139 enum { OutputSize = (15 * 3) + 1 };
140 char output[OutputSize];
142 assert(bytes.size() <= 15
143 && "DumpBytes only supports instructions of up to 15 bytes");
144 memset(output, ' ', sizeof(output));
146 for (StringRef::iterator i = bytes.begin(),
147 e = bytes.end(); i != e; ++i) {
148 output[index] = hex_rep[(*i & 0xF0) >> 4];
149 output[index + 1] = hex_rep[*i & 0xF];
153 output[sizeof(output) - 1] = 0;
157 static bool RelocAddressLess(RelocationRef a, RelocationRef b) {
158 uint64_t a_addr, b_addr;
159 if (error(a.getAddress(a_addr))) return false;
160 if (error(b.getAddress(b_addr))) return false;
161 return a_addr < b_addr;
164 // OperandInfoCallback - Callback from the MCDisassembler to perform
165 // relocation resolution for operands while disassembling.
166 int OperandInfoCallback(void *ObjectFile, uint64_t PC, uint64_t Offset,
167 uint64_t Size, int TagType, void *TagBuf) {
168 LLVMOpInfo1 *OpInfo = static_cast<LLVMOpInfo1*>(TagBuf);
169 OpInfo->AddSymbol.Present = 0;
170 OpInfo->SubtractSymbol.Present = 0;
171 OpInfo->VariantKind = LLVMDisassembler_VariantKind_None;
176 // SymbolLookupCallback - Callback from the MCDisassembler to convert
177 // offsets to symbolic references while disassembling.
178 const char* SymbolLookupCallback(void *ObjectFile, uint64_t ReferenceValue,
179 uint64_t *ReferenceType, uint64_t PC,
180 const char **ReferenceName) {
181 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
185 static void DisassembleObject(const ObjectFile *Obj, bool InlineRelocs) {
186 const Target *TheTarget = GetTarget(Obj);
189 // GetTarget prints out stuff.
194 for (section_iterator i = Obj->begin_sections(),
195 e = Obj->end_sections();
196 i != e; i.increment(ec)) {
197 if (error(ec)) break;
199 if (error(i->isText(text))) break;
202 uint64_t SectionAddr;
203 if (error(i->getAddress(SectionAddr))) break;
205 // Make a list of all the symbols in this section.
206 std::vector<std::pair<uint64_t, StringRef> > Symbols;
207 for (symbol_iterator si = Obj->begin_symbols(),
208 se = Obj->end_symbols();
209 si != se; si.increment(ec)) {
211 if (!error(i->containsSymbol(*si, contains)) && contains) {
213 if (error(si->getOffset(Address))) break;
215 if (error(si->getName(Name))) break;
216 Symbols.push_back(std::make_pair(Address, Name));
220 // Sort the symbols by address, just in case they didn't come in that way.
221 array_pod_sort(Symbols.begin(), Symbols.end());
223 // Make a list of all the relocations for this section.
224 std::vector<RelocationRef> Rels;
226 for (relocation_iterator ri = i->begin_relocations(),
227 re = i->end_relocations();
228 ri != re; ri.increment(ec)) {
229 if (error(ec)) break;
234 // Sort relocations by address.
235 std::sort(Rels.begin(), Rels.end(), RelocAddressLess);
238 if (error(i->getName(name))) break;
239 outs() << "Disassembly of section " << name << ':';
241 // If the section has no symbols just insert a dummy one and disassemble
242 // the whole section.
244 Symbols.push_back(std::make_pair(0, name));
246 // Set up disassembler.
247 OwningPtr<const MCAsmInfo> AsmInfo(TheTarget->createMCAsmInfo(TripleName));
250 errs() << "error: no assembly info for target " << TripleName << "\n";
254 OwningPtr<const MCRegisterInfo>
255 RegInfo(TheTarget->createMCRegInfo(TripleName));
258 errs() << "error: no register info for target " << TripleName << "\n";
262 OwningPtr<const MCSubtargetInfo> STI(
263 TheTarget->createMCSubtargetInfo(TripleName, "", ""));
266 errs() << "error: no subtarget info for target " << TripleName << "\n";
270 MCContext Context(*AsmInfo, *RegInfo, 0);
271 OwningPtr<MCDisassembler> DisAsm(
272 TheTarget->createMCDisassembler(*STI));
274 errs() << "error: no disassembler for target " << TripleName << "\n";
278 // Setup the disassembler for symbolic decoding.
279 DisAsm->setupForSymbolicDisassembly(OperandInfoCallback,
280 SymbolLookupCallback,
281 const_cast<ObjectFile*>(Obj),
284 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
285 OwningPtr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
286 AsmPrinterVariant, *AsmInfo, *STI));
288 errs() << "error: no instruction printer for target " << TripleName
294 if (error(i->getContents(Bytes))) break;
295 StringRefMemoryObject memoryObject(Bytes);
299 if (error(i->getSize(SectSize))) break;
301 std::vector<RelocationRef>::const_iterator rel_cur = Rels.begin();
302 std::vector<RelocationRef>::const_iterator rel_end = Rels.end();
303 // Disassemble symbol by symbol.
304 for (unsigned si = 0, se = Symbols.size(); si != se; ++si) {
305 uint64_t Start = Symbols[si].first;
307 // The end is either the size of the section or the beginning of the next
311 // Make sure this symbol takes up space.
312 else if (Symbols[si + 1].first != Start)
313 End = Symbols[si + 1].first - 1;
315 // This symbol has the same address as the next symbol. Skip it.
318 outs() << '\n' << Symbols[si].second << ":\n";
321 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
323 raw_ostream &DebugOut = nulls();
326 for (Index = Start; Index < End; Index += Size) {
329 if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
330 DebugOut, nulls())) {
331 outs() << format("%8x:\t", SectionAddr + Index);
332 DumpBytes(StringRef(Bytes.data() + Index, Size));
333 IP->printInst(&Inst, outs(), "");
336 errs() << ToolName << ": warning: invalid instruction encoding\n";
338 Size = 1; // skip illegible bytes
341 // Print relocation for instruction.
342 while (rel_cur != rel_end) {
345 SmallString<16> name;
348 // If this relocation is hidden, skip it.
349 if (error(rel_cur->getHidden(hidden))) goto skip_print_rel;
350 if (hidden) goto skip_print_rel;
352 if (error(rel_cur->getAddress(addr))) goto skip_print_rel;
353 // Stop when rel_cur's address is past the current instruction.
354 if (addr >= Index + Size) break;
355 if (error(rel_cur->getTypeName(name))) goto skip_print_rel;
356 if (error(rel_cur->getValueString(val))) goto skip_print_rel;
358 outs() << format("\t\t\t%8x: ", SectionAddr + addr) << name << "\t"
369 static void PrintRelocations(const ObjectFile *o) {
371 for (section_iterator si = o->begin_sections(), se = o->end_sections();
372 si != se; si.increment(ec)){
373 if (error(ec)) return;
374 if (si->begin_relocations() == si->end_relocations())
377 if (error(si->getName(secname))) continue;
378 outs() << "RELOCATION RECORDS FOR [" << secname << "]:\n";
379 for (relocation_iterator ri = si->begin_relocations(),
380 re = si->end_relocations();
381 ri != re; ri.increment(ec)) {
382 if (error(ec)) return;
386 SmallString<32> relocname;
387 SmallString<32> valuestr;
388 if (error(ri->getHidden(hidden))) continue;
389 if (hidden) continue;
390 if (error(ri->getTypeName(relocname))) continue;
391 if (error(ri->getAddress(address))) continue;
392 if (error(ri->getValueString(valuestr))) continue;
393 outs() << address << " " << relocname << " " << valuestr << "\n";
399 static void PrintSectionHeaders(const ObjectFile *o) {
400 outs() << "Sections:\n"
401 "Idx Name Size Address Type\n";
404 for (section_iterator si = o->begin_sections(), se = o->end_sections();
405 si != se; si.increment(ec)) {
406 if (error(ec)) return;
408 if (error(si->getName(Name))) return;
410 if (error(si->getAddress(Address))) return;
412 if (error(si->getSize(Size))) return;
413 bool Text, Data, BSS;
414 if (error(si->isText(Text))) return;
415 if (error(si->isData(Data))) return;
416 if (error(si->isBSS(BSS))) return;
417 std::string Type = (std::string(Text ? "TEXT " : "") +
418 (Data ? "DATA " : "") + (BSS ? "BSS" : ""));
419 outs() << format("%3d %-13s %09"PRIx64" %017"PRIx64" %s\n", i, Name.str().c_str(), Size,
420 Address, Type.c_str());
425 static void PrintSectionContents(const ObjectFile *o) {
427 for (section_iterator si = o->begin_sections(),
428 se = o->end_sections();
429 si != se; si.increment(ec)) {
430 if (error(ec)) return;
434 if (error(si->getName(Name))) continue;
435 if (error(si->getContents(Contents))) continue;
436 if (error(si->getAddress(BaseAddr))) continue;
438 outs() << "Contents of section " << Name << ":\n";
440 // Dump out the content as hex and printable ascii characters.
441 for (std::size_t addr = 0, end = Contents.size(); addr < end; addr += 16) {
442 outs() << format(" %04x ", BaseAddr + addr);
444 for (std::size_t i = 0; i < 16; ++i) {
445 if (i != 0 && i % 4 == 0)
448 outs() << hexdigit((Contents[addr + i] >> 4) & 0xF, true)
449 << hexdigit(Contents[addr + i] & 0xF, true);
455 for (std::size_t i = 0; i < 16 && addr + i < end; ++i) {
456 if (std::isprint(Contents[addr + i] & 0xFF))
457 outs() << Contents[addr + i];
466 static void PrintCOFFSymbolTable(const COFFObjectFile *coff) {
467 const coff_file_header *header;
468 if (error(coff->getHeader(header))) return;
470 const coff_symbol *symbol = 0;
471 for (int i = 0, e = header->NumberOfSymbols; i != e; ++i) {
473 // Figure out which type of aux this is.
474 if (symbol->StorageClass == COFF::IMAGE_SYM_CLASS_STATIC
475 && symbol->Value == 0) { // Section definition.
476 const coff_aux_section_definition *asd;
477 if (error(coff->getAuxSymbol<coff_aux_section_definition>(i, asd)))
480 << format("scnlen 0x%x nreloc %d nlnno %d checksum 0x%x "
481 , unsigned(asd->Length)
482 , unsigned(asd->NumberOfRelocations)
483 , unsigned(asd->NumberOfLinenumbers)
484 , unsigned(asd->CheckSum))
485 << format("assoc %d comdat %d\n"
486 , unsigned(asd->Number)
487 , unsigned(asd->Selection));
489 outs() << "AUX Unknown\n";
493 if (error(coff->getSymbol(i, symbol))) return;
494 if (error(coff->getSymbolName(symbol, name))) return;
495 outs() << "[" << format("%2d", i) << "]"
496 << "(sec " << format("%2d", int(symbol->SectionNumber)) << ")"
497 << "(fl 0x00)" // Flag bits, which COFF doesn't have.
498 << "(ty " << format("%3x", unsigned(symbol->Type)) << ")"
499 << "(scl " << format("%3x", unsigned(symbol->StorageClass)) << ") "
500 << "(nx " << unsigned(symbol->NumberOfAuxSymbols) << ") "
501 << "0x" << format("%08x", unsigned(symbol->Value)) << " "
503 aux_count = symbol->NumberOfAuxSymbols;
508 static void PrintSymbolTable(const ObjectFile *o) {
509 outs() << "SYMBOL TABLE:\n";
511 if (const COFFObjectFile *coff = dyn_cast<const COFFObjectFile>(o))
512 PrintCOFFSymbolTable(coff);
515 for (symbol_iterator si = o->begin_symbols(),
516 se = o->end_symbols(); si != se; si.increment(ec)) {
517 if (error(ec)) return;
521 SymbolRef::Type Type;
525 section_iterator Section = o->end_sections();
526 if (error(si->getName(Name))) continue;
527 if (error(si->getOffset(Offset))) continue;
528 if (error(si->isGlobal(Global))) continue;
529 if (error(si->getType(Type))) continue;
530 if (error(si->isWeak(Weak))) continue;
531 if (error(si->isAbsolute(Absolute))) continue;
532 if (error(si->getSize(Size))) continue;
533 if (error(si->getSection(Section))) continue;
535 if (Offset == UnknownAddressOrSize)
538 if (Type != SymbolRef::ST_External)
539 GlobLoc = Global ? 'g' : 'l';
540 char Debug = (Type == SymbolRef::ST_Debug || Type == SymbolRef::ST_File)
543 if (Type == SymbolRef::ST_File)
545 else if (Type == SymbolRef::ST_Function)
548 outs() << format("%08x", Offset) << " "
549 << GlobLoc // Local -> 'l', Global -> 'g', Neither -> ' '
550 << (Weak ? 'w' : ' ') // Weak?
551 << ' ' // Constructor. Not supported yet.
552 << ' ' // Warning. Not supported yet.
553 << ' ' // Indirect reference to another symbol.
554 << Debug // Debugging (d) or dynamic (D) symbol.
555 << FileFunc // Name of function (F), file (f) or object (O).
559 else if (Section == o->end_sections())
562 StringRef SectionName;
563 if (error(Section->getName(SectionName)))
565 outs() << SectionName;
568 << format("%08x ", Size)
575 static void DumpObject(const ObjectFile *o) {
577 outs() << o->getFileName()
578 << ":\tfile format " << o->getFileFormatName() << "\n\n";
581 DisassembleObject(o, Relocations);
582 if (Relocations && !Disassemble)
585 PrintSectionHeaders(o);
587 PrintSectionContents(o);
592 /// @brief Dump each object file in \a a;
593 static void DumpArchive(const Archive *a) {
594 for (Archive::child_iterator i = a->begin_children(),
595 e = a->end_children(); i != e; ++i) {
596 OwningPtr<Binary> child;
597 if (error_code ec = i->getAsBinary(child)) {
598 errs() << ToolName << ": '" << a->getFileName() << "': " << ec.message()
602 if (ObjectFile *o = dyn_cast<ObjectFile>(child.get()))
605 errs() << ToolName << ": '" << a->getFileName() << "': "
606 << "Unrecognized file type.\n";
610 /// @brief Open file and figure out how to dump it.
611 static void DumpInput(StringRef file) {
612 // If file isn't stdin, check that it exists.
613 if (file != "-" && !sys::fs::exists(file)) {
614 errs() << ToolName << ": '" << file << "': " << "No such file\n";
618 if (MachO && Disassemble) {
619 DisassembleInputMachO(file);
623 // Attempt to open the binary.
624 OwningPtr<Binary> binary;
625 if (error_code ec = createBinary(file, binary)) {
626 errs() << ToolName << ": '" << file << "': " << ec.message() << ".\n";
630 if (Archive *a = dyn_cast<Archive>(binary.get())) {
632 } else if (ObjectFile *o = dyn_cast<ObjectFile>(binary.get())) {
635 errs() << ToolName << ": '" << file << "': " << "Unrecognized file type.\n";
639 int main(int argc, char **argv) {
640 // Print a stack trace if we signal out.
641 sys::PrintStackTraceOnErrorSignal();
642 PrettyStackTraceProgram X(argc, argv);
643 llvm_shutdown_obj Y; // Call llvm_shutdown() on exit.
645 // Initialize targets and assembly printers/parsers.
646 llvm::InitializeAllTargetInfos();
647 llvm::InitializeAllTargetMCs();
648 llvm::InitializeAllAsmParsers();
649 llvm::InitializeAllDisassemblers();
651 cl::ParseCommandLineOptions(argc, argv, "llvm object file dumper\n");
652 TripleName = Triple::normalize(TripleName);
656 // Defaults to a.out if no filenames specified.
657 if (InputFilenames.size() == 0)
658 InputFilenames.push_back("a.out");
665 cl::PrintHelpMessage();
669 std::for_each(InputFilenames.begin(), InputFilenames.end(),