1 //===-- MachODump.cpp - Object file dumping utility for llvm --------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file implements the MachO-specific dumper for llvm-objdump.
10 //
11 //===----------------------------------------------------------------------===//
12
13 #include "MachODump.h"
14
15 #include "ObjdumpOptID.h"
16 #include "llvm-objdump.h"
17 #include "llvm-c/Disassembler.h"
18 #include "llvm/ADT/STLExtras.h"
19 #include "llvm/ADT/StringExtras.h"
20 #include "llvm/ADT/Triple.h"
21 #include "llvm/BinaryFormat/MachO.h"
22 #include "llvm/Config/config.h"
23 #include "llvm/DebugInfo/DIContext.h"
24 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
25 #include "llvm/Demangle/Demangle.h"
26 #include "llvm/MC/MCAsmInfo.h"
27 #include "llvm/MC/MCContext.h"
28 #include "llvm/MC/MCDisassembler/MCDisassembler.h"
29 #include "llvm/MC/MCInst.h"
30 #include "llvm/MC/MCInstPrinter.h"
31 #include "llvm/MC/MCInstrDesc.h"
32 #include "llvm/MC/MCInstrInfo.h"
33 #include "llvm/MC/MCRegisterInfo.h"
34 #include "llvm/MC/MCSubtargetInfo.h"
35 #include "llvm/MC/MCTargetOptions.h"
36 #include "llvm/MC/TargetRegistry.h"
37 #include "llvm/Object/MachO.h"
38 #include "llvm/Object/MachOUniversal.h"
39 #include "llvm/Option/ArgList.h"
40 #include "llvm/Support/Casting.h"
41 #include "llvm/Support/Debug.h"
42 #include "llvm/Support/Endian.h"
43 #include "llvm/Support/Format.h"
44 #include "llvm/Support/FormattedStream.h"
45 #include "llvm/Support/GraphWriter.h"
46 #include "llvm/Support/LEB128.h"
47 #include "llvm/Support/MemoryBuffer.h"
48 #include "llvm/Support/TargetSelect.h"
49 #include "llvm/Support/ToolOutputFile.h"
50 #include "llvm/Support/WithColor.h"
51 #include "llvm/Support/raw_ostream.h"
52 #include <algorithm>
53 #include <cstring>
54 #include <system_error>
55
56 #ifdef LLVM_HAVE_LIBXAR
57 extern "C" {
58 #include <xar/xar.h>
59 }
60 #endif
61
62 using namespace llvm;
63 using namespace llvm::object;
64 using namespace llvm::objdump;
65
66 bool objdump::FirstPrivateHeader;
67 bool objdump::ExportsTrie;
68 bool objdump::Rebase;
69 bool objdump::Rpaths;
70 bool objdump::Bind;
71 bool objdump::LazyBind;
72 bool objdump::WeakBind;
73 static bool UseDbg;
74 static std::string DSYMFile;
75 bool objdump::FullLeadingAddr;
76 bool objdump::LeadingHeaders;
77 bool objdump::UniversalHeaders;
78 static bool ArchiveMemberOffsets;
79 bool objdump::IndirectSymbols;
80 bool objdump::DataInCode;
81 bool objdump::FunctionStarts;
82 bool objdump::LinkOptHints;
83 bool objdump::InfoPlist;
84 bool objdump::DyldInfo;
85 bool objdump::DylibsUsed;
86 bool objdump::DylibId;
87 bool objdump::Verbose;
88 bool objdump::ObjcMetaData;
89 std::string objdump::DisSymName;
90 bool objdump::SymbolicOperands;
91 static std::vector<std::string> ArchFlags;
92
93 static bool ArchAll = false;
94 static std::string ThumbTripleName;
95
parseMachOOptions(const llvm::opt::InputArgList & InputArgs)96 void objdump::parseMachOOptions(const llvm::opt::InputArgList &InputArgs) {
97 FirstPrivateHeader = InputArgs.hasArg(OBJDUMP_private_header);
98 ExportsTrie = InputArgs.hasArg(OBJDUMP_exports_trie);
99 Rebase = InputArgs.hasArg(OBJDUMP_rebase);
100 Rpaths = InputArgs.hasArg(OBJDUMP_rpaths);
101 Bind = InputArgs.hasArg(OBJDUMP_bind);
102 LazyBind = InputArgs.hasArg(OBJDUMP_lazy_bind);
103 WeakBind = InputArgs.hasArg(OBJDUMP_weak_bind);
104 UseDbg = InputArgs.hasArg(OBJDUMP_g);
105 DSYMFile = InputArgs.getLastArgValue(OBJDUMP_dsym_EQ).str();
106 FullLeadingAddr = InputArgs.hasArg(OBJDUMP_full_leading_addr);
107 LeadingHeaders = !InputArgs.hasArg(OBJDUMP_no_leading_headers);
108 UniversalHeaders = InputArgs.hasArg(OBJDUMP_universal_headers);
109 ArchiveMemberOffsets = InputArgs.hasArg(OBJDUMP_archive_member_offsets);
110 IndirectSymbols = InputArgs.hasArg(OBJDUMP_indirect_symbols);
111 DataInCode = InputArgs.hasArg(OBJDUMP_data_in_code);
112 FunctionStarts = InputArgs.hasArg(OBJDUMP_function_starts);
113 LinkOptHints = InputArgs.hasArg(OBJDUMP_link_opt_hints);
114 InfoPlist = InputArgs.hasArg(OBJDUMP_info_plist);
115 DyldInfo = InputArgs.hasArg(OBJDUMP_dyld_info);
116 DylibsUsed = InputArgs.hasArg(OBJDUMP_dylibs_used);
117 DylibId = InputArgs.hasArg(OBJDUMP_dylib_id);
118 Verbose = !InputArgs.hasArg(OBJDUMP_non_verbose);
119 ObjcMetaData = InputArgs.hasArg(OBJDUMP_objc_meta_data);
120 DisSymName = InputArgs.getLastArgValue(OBJDUMP_dis_symname).str();
121 SymbolicOperands = !InputArgs.hasArg(OBJDUMP_no_symbolic_operands);
122 ArchFlags = InputArgs.getAllArgValues(OBJDUMP_arch_EQ);
123 }
124
GetTarget(const MachOObjectFile * MachOObj,const char ** McpuDefault,const Target ** ThumbTarget)125 static const Target *GetTarget(const MachOObjectFile *MachOObj,
126 const char **McpuDefault,
127 const Target **ThumbTarget) {
128 // Figure out the target triple.
129 Triple TT(TripleName);
130 if (TripleName.empty()) {
131 TT = MachOObj->getArchTriple(McpuDefault);
132 TripleName = TT.str();
133 }
134
135 if (TT.getArch() == Triple::arm) {
136 // We've inferred a 32-bit ARM target from the object file. All MachO CPUs
137 // that support ARM are also capable of Thumb mode.
138 Triple ThumbTriple = TT;
139 std::string ThumbName = (Twine("thumb") + TT.getArchName().substr(3)).str();
140 ThumbTriple.setArchName(ThumbName);
141 ThumbTripleName = ThumbTriple.str();
142 }
143
144 // Get the target specific parser.
145 std::string Error;
146 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
147 if (TheTarget && ThumbTripleName.empty())
148 return TheTarget;
149
150 *ThumbTarget = TargetRegistry::lookupTarget(ThumbTripleName, Error);
151 if (*ThumbTarget)
152 return TheTarget;
153
154 WithColor::error(errs(), "llvm-objdump") << "unable to get target for '";
155 if (!TheTarget)
156 errs() << TripleName;
157 else
158 errs() << ThumbTripleName;
159 errs() << "', see --version and --triple.\n";
160 return nullptr;
161 }
162
163 namespace {
164 struct SymbolSorter {
operator ()__anon4a9557200111::SymbolSorter165 bool operator()(const SymbolRef &A, const SymbolRef &B) {
166 Expected<SymbolRef::Type> ATypeOrErr = A.getType();
167 if (!ATypeOrErr)
168 reportError(ATypeOrErr.takeError(), A.getObject()->getFileName());
169 SymbolRef::Type AType = *ATypeOrErr;
170 Expected<SymbolRef::Type> BTypeOrErr = B.getType();
171 if (!BTypeOrErr)
172 reportError(BTypeOrErr.takeError(), B.getObject()->getFileName());
173 SymbolRef::Type BType = *BTypeOrErr;
174 uint64_t AAddr =
175 (AType != SymbolRef::ST_Function) ? 0 : cantFail(A.getValue());
176 uint64_t BAddr =
177 (BType != SymbolRef::ST_Function) ? 0 : cantFail(B.getValue());
178 return AAddr < BAddr;
179 }
180 };
181 } // namespace
182
183 // Types for the storted data in code table that is built before disassembly
184 // and the predicate function to sort them.
185 typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
186 typedef std::vector<DiceTableEntry> DiceTable;
187 typedef DiceTable::iterator dice_table_iterator;
188
189 #ifdef LLVM_HAVE_LIBXAR
190 namespace {
191 struct ScopedXarFile {
192 xar_t xar;
ScopedXarFile__anon4a9557200211::ScopedXarFile193 ScopedXarFile(const char *filename, int32_t flags) {
194 #pragma clang diagnostic push
195 #pragma clang diagnostic ignored "-Wdeprecated-declarations"
196 xar = xar_open(filename, flags);
197 #pragma clang diagnostic pop
198 }
~ScopedXarFile__anon4a9557200211::ScopedXarFile199 ~ScopedXarFile() {
200 if (xar)
201 xar_close(xar);
202 }
203 ScopedXarFile(const ScopedXarFile &) = delete;
204 ScopedXarFile &operator=(const ScopedXarFile &) = delete;
operator xar_t__anon4a9557200211::ScopedXarFile205 operator xar_t() { return xar; }
206 };
207
208 struct ScopedXarIter {
209 xar_iter_t iter;
ScopedXarIter__anon4a9557200211::ScopedXarIter210 ScopedXarIter() : iter(xar_iter_new()) {}
~ScopedXarIter__anon4a9557200211::ScopedXarIter211 ~ScopedXarIter() {
212 if (iter)
213 xar_iter_free(iter);
214 }
215 ScopedXarIter(const ScopedXarIter &) = delete;
216 ScopedXarIter &operator=(const ScopedXarIter &) = delete;
operator xar_iter_t__anon4a9557200211::ScopedXarIter217 operator xar_iter_t() { return iter; }
218 };
219 } // namespace
220 #endif // defined(LLVM_HAVE_LIBXAR)
221
222 // This is used to search for a data in code table entry for the PC being
223 // disassembled. The j parameter has the PC in j.first. A single data in code
224 // table entry can cover many bytes for each of its Kind's. So if the offset,
225 // aka the i.first value, of the data in code table entry plus its Length
226 // covers the PC being searched for this will return true. If not it will
227 // return false.
compareDiceTableEntries(const DiceTableEntry & i,const DiceTableEntry & j)228 static bool compareDiceTableEntries(const DiceTableEntry &i,
229 const DiceTableEntry &j) {
230 uint16_t Length;
231 i.second.getLength(Length);
232
233 return j.first >= i.first && j.first < i.first + Length;
234 }
235
DumpDataInCode(const uint8_t * bytes,uint64_t Length,unsigned short Kind)236 static uint64_t DumpDataInCode(const uint8_t *bytes, uint64_t Length,
237 unsigned short Kind) {
238 uint32_t Value, Size = 1;
239
240 switch (Kind) {
241 default:
242 case MachO::DICE_KIND_DATA:
243 if (Length >= 4) {
244 if (ShowRawInsn)
245 dumpBytes(makeArrayRef(bytes, 4), outs());
246 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
247 outs() << "\t.long " << Value;
248 Size = 4;
249 } else if (Length >= 2) {
250 if (ShowRawInsn)
251 dumpBytes(makeArrayRef(bytes, 2), outs());
252 Value = bytes[1] << 8 | bytes[0];
253 outs() << "\t.short " << Value;
254 Size = 2;
255 } else {
256 if (ShowRawInsn)
257 dumpBytes(makeArrayRef(bytes, 2), outs());
258 Value = bytes[0];
259 outs() << "\t.byte " << Value;
260 Size = 1;
261 }
262 if (Kind == MachO::DICE_KIND_DATA)
263 outs() << "\t@ KIND_DATA\n";
264 else
265 outs() << "\t@ data in code kind = " << Kind << "\n";
266 break;
267 case MachO::DICE_KIND_JUMP_TABLE8:
268 if (ShowRawInsn)
269 dumpBytes(makeArrayRef(bytes, 1), outs());
270 Value = bytes[0];
271 outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n";
272 Size = 1;
273 break;
274 case MachO::DICE_KIND_JUMP_TABLE16:
275 if (ShowRawInsn)
276 dumpBytes(makeArrayRef(bytes, 2), outs());
277 Value = bytes[1] << 8 | bytes[0];
278 outs() << "\t.short " << format("%5u", Value & 0xffff)
279 << "\t@ KIND_JUMP_TABLE16\n";
280 Size = 2;
281 break;
282 case MachO::DICE_KIND_JUMP_TABLE32:
283 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
284 if (ShowRawInsn)
285 dumpBytes(makeArrayRef(bytes, 4), outs());
286 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
287 outs() << "\t.long " << Value;
288 if (Kind == MachO::DICE_KIND_JUMP_TABLE32)
289 outs() << "\t@ KIND_JUMP_TABLE32\n";
290 else
291 outs() << "\t@ KIND_ABS_JUMP_TABLE32\n";
292 Size = 4;
293 break;
294 }
295 return Size;
296 }
297
getSectionsAndSymbols(MachOObjectFile * MachOObj,std::vector<SectionRef> & Sections,std::vector<SymbolRef> & Symbols,SmallVectorImpl<uint64_t> & FoundFns,uint64_t & BaseSegmentAddress)298 static void getSectionsAndSymbols(MachOObjectFile *MachOObj,
299 std::vector<SectionRef> &Sections,
300 std::vector<SymbolRef> &Symbols,
301 SmallVectorImpl<uint64_t> &FoundFns,
302 uint64_t &BaseSegmentAddress) {
303 const StringRef FileName = MachOObj->getFileName();
304 for (const SymbolRef &Symbol : MachOObj->symbols()) {
305 StringRef SymName = unwrapOrError(Symbol.getName(), FileName);
306 if (!SymName.startswith("ltmp"))
307 Symbols.push_back(Symbol);
308 }
309
310 append_range(Sections, MachOObj->sections());
311
312 bool BaseSegmentAddressSet = false;
313 for (const auto &Command : MachOObj->load_commands()) {
314 if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) {
315 // We found a function starts segment, parse the addresses for later
316 // consumption.
317 MachO::linkedit_data_command LLC =
318 MachOObj->getLinkeditDataLoadCommand(Command);
319
320 MachOObj->ReadULEB128s(LLC.dataoff, FoundFns);
321 } else if (Command.C.cmd == MachO::LC_SEGMENT) {
322 MachO::segment_command SLC = MachOObj->getSegmentLoadCommand(Command);
323 StringRef SegName = SLC.segname;
324 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
325 BaseSegmentAddressSet = true;
326 BaseSegmentAddress = SLC.vmaddr;
327 }
328 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
329 MachO::segment_command_64 SLC = MachOObj->getSegment64LoadCommand(Command);
330 StringRef SegName = SLC.segname;
331 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
332 BaseSegmentAddressSet = true;
333 BaseSegmentAddress = SLC.vmaddr;
334 }
335 }
336 }
337 }
338
DumpAndSkipDataInCode(uint64_t PC,const uint8_t * bytes,DiceTable & Dices,uint64_t & InstSize)339 static bool DumpAndSkipDataInCode(uint64_t PC, const uint8_t *bytes,
340 DiceTable &Dices, uint64_t &InstSize) {
341 // Check the data in code table here to see if this is data not an
342 // instruction to be disassembled.
343 DiceTable Dice;
344 Dice.push_back(std::make_pair(PC, DiceRef()));
345 dice_table_iterator DTI =
346 std::search(Dices.begin(), Dices.end(), Dice.begin(), Dice.end(),
347 compareDiceTableEntries);
348 if (DTI != Dices.end()) {
349 uint16_t Length;
350 DTI->second.getLength(Length);
351 uint16_t Kind;
352 DTI->second.getKind(Kind);
353 InstSize = DumpDataInCode(bytes, Length, Kind);
354 if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) &&
355 (PC == (DTI->first + Length - 1)) && (Length & 1))
356 InstSize++;
357 return true;
358 }
359 return false;
360 }
361
printRelocationTargetName(const MachOObjectFile * O,const MachO::any_relocation_info & RE,raw_string_ostream & Fmt)362 static void printRelocationTargetName(const MachOObjectFile *O,
363 const MachO::any_relocation_info &RE,
364 raw_string_ostream &Fmt) {
365 // Target of a scattered relocation is an address. In the interest of
366 // generating pretty output, scan through the symbol table looking for a
367 // symbol that aligns with that address. If we find one, print it.
368 // Otherwise, we just print the hex address of the target.
369 const StringRef FileName = O->getFileName();
370 if (O->isRelocationScattered(RE)) {
371 uint32_t Val = O->getPlainRelocationSymbolNum(RE);
372
373 for (const SymbolRef &Symbol : O->symbols()) {
374 uint64_t Addr = unwrapOrError(Symbol.getAddress(), FileName);
375 if (Addr != Val)
376 continue;
377 Fmt << unwrapOrError(Symbol.getName(), FileName);
378 return;
379 }
380
381 // If we couldn't find a symbol that this relocation refers to, try
382 // to find a section beginning instead.
383 for (const SectionRef &Section : ToolSectionFilter(*O)) {
384 uint64_t Addr = Section.getAddress();
385 if (Addr != Val)
386 continue;
387 StringRef NameOrErr = unwrapOrError(Section.getName(), O->getFileName());
388 Fmt << NameOrErr;
389 return;
390 }
391
392 Fmt << format("0x%x", Val);
393 return;
394 }
395
396 StringRef S;
397 bool isExtern = O->getPlainRelocationExternal(RE);
398 uint64_t Val = O->getPlainRelocationSymbolNum(RE);
399
400 if (O->getAnyRelocationType(RE) == MachO::ARM64_RELOC_ADDEND &&
401 (O->getArch() == Triple::aarch64 || O->getArch() == Triple::aarch64_be)) {
402 Fmt << format("0x%0" PRIx64, Val);
403 return;
404 }
405
406 if (isExtern) {
407 symbol_iterator SI = O->symbol_begin();
408 std::advance(SI, Val);
409 S = unwrapOrError(SI->getName(), FileName);
410 } else {
411 section_iterator SI = O->section_begin();
412 // Adjust for the fact that sections are 1-indexed.
413 if (Val == 0) {
414 Fmt << "0 (?,?)";
415 return;
416 }
417 uint32_t I = Val - 1;
418 while (I != 0 && SI != O->section_end()) {
419 --I;
420 std::advance(SI, 1);
421 }
422 if (SI == O->section_end()) {
423 Fmt << Val << " (?,?)";
424 } else {
425 if (Expected<StringRef> NameOrErr = SI->getName())
426 S = *NameOrErr;
427 else
428 consumeError(NameOrErr.takeError());
429 }
430 }
431
432 Fmt << S;
433 }
434
getMachORelocationValueString(const MachOObjectFile * Obj,const RelocationRef & RelRef,SmallVectorImpl<char> & Result)435 Error objdump::getMachORelocationValueString(const MachOObjectFile *Obj,
436 const RelocationRef &RelRef,
437 SmallVectorImpl<char> &Result) {
438 DataRefImpl Rel = RelRef.getRawDataRefImpl();
439 MachO::any_relocation_info RE = Obj->getRelocation(Rel);
440
441 unsigned Arch = Obj->getArch();
442
443 std::string FmtBuf;
444 raw_string_ostream Fmt(FmtBuf);
445 unsigned Type = Obj->getAnyRelocationType(RE);
446 bool IsPCRel = Obj->getAnyRelocationPCRel(RE);
447
448 // Determine any addends that should be displayed with the relocation.
449 // These require decoding the relocation type, which is triple-specific.
450
451 // X86_64 has entirely custom relocation types.
452 if (Arch == Triple::x86_64) {
453 switch (Type) {
454 case MachO::X86_64_RELOC_GOT_LOAD:
455 case MachO::X86_64_RELOC_GOT: {
456 printRelocationTargetName(Obj, RE, Fmt);
457 Fmt << "@GOT";
458 if (IsPCRel)
459 Fmt << "PCREL";
460 break;
461 }
462 case MachO::X86_64_RELOC_SUBTRACTOR: {
463 DataRefImpl RelNext = Rel;
464 Obj->moveRelocationNext(RelNext);
465 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
466
467 // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type
468 // X86_64_RELOC_UNSIGNED.
469 // NOTE: Scattered relocations don't exist on x86_64.
470 unsigned RType = Obj->getAnyRelocationType(RENext);
471 if (RType != MachO::X86_64_RELOC_UNSIGNED)
472 reportError(Obj->getFileName(), "Expected X86_64_RELOC_UNSIGNED after "
473 "X86_64_RELOC_SUBTRACTOR.");
474
475 // The X86_64_RELOC_UNSIGNED contains the minuend symbol;
476 // X86_64_RELOC_SUBTRACTOR contains the subtrahend.
477 printRelocationTargetName(Obj, RENext, Fmt);
478 Fmt << "-";
479 printRelocationTargetName(Obj, RE, Fmt);
480 break;
481 }
482 case MachO::X86_64_RELOC_TLV:
483 printRelocationTargetName(Obj, RE, Fmt);
484 Fmt << "@TLV";
485 if (IsPCRel)
486 Fmt << "P";
487 break;
488 case MachO::X86_64_RELOC_SIGNED_1:
489 printRelocationTargetName(Obj, RE, Fmt);
490 Fmt << "-1";
491 break;
492 case MachO::X86_64_RELOC_SIGNED_2:
493 printRelocationTargetName(Obj, RE, Fmt);
494 Fmt << "-2";
495 break;
496 case MachO::X86_64_RELOC_SIGNED_4:
497 printRelocationTargetName(Obj, RE, Fmt);
498 Fmt << "-4";
499 break;
500 default:
501 printRelocationTargetName(Obj, RE, Fmt);
502 break;
503 }
504 // X86 and ARM share some relocation types in common.
505 } else if (Arch == Triple::x86 || Arch == Triple::arm ||
506 Arch == Triple::ppc) {
507 // Generic relocation types...
508 switch (Type) {
509 case MachO::GENERIC_RELOC_PAIR: // prints no info
510 return Error::success();
511 case MachO::GENERIC_RELOC_SECTDIFF: {
512 DataRefImpl RelNext = Rel;
513 Obj->moveRelocationNext(RelNext);
514 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
515
516 // X86 sect diff's must be followed by a relocation of type
517 // GENERIC_RELOC_PAIR.
518 unsigned RType = Obj->getAnyRelocationType(RENext);
519
520 if (RType != MachO::GENERIC_RELOC_PAIR)
521 reportError(Obj->getFileName(), "Expected GENERIC_RELOC_PAIR after "
522 "GENERIC_RELOC_SECTDIFF.");
523
524 printRelocationTargetName(Obj, RE, Fmt);
525 Fmt << "-";
526 printRelocationTargetName(Obj, RENext, Fmt);
527 break;
528 }
529 }
530
531 if (Arch == Triple::x86 || Arch == Triple::ppc) {
532 switch (Type) {
533 case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: {
534 DataRefImpl RelNext = Rel;
535 Obj->moveRelocationNext(RelNext);
536 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
537
538 // X86 sect diff's must be followed by a relocation of type
539 // GENERIC_RELOC_PAIR.
540 unsigned RType = Obj->getAnyRelocationType(RENext);
541 if (RType != MachO::GENERIC_RELOC_PAIR)
542 reportError(Obj->getFileName(), "Expected GENERIC_RELOC_PAIR after "
543 "GENERIC_RELOC_LOCAL_SECTDIFF.");
544
545 printRelocationTargetName(Obj, RE, Fmt);
546 Fmt << "-";
547 printRelocationTargetName(Obj, RENext, Fmt);
548 break;
549 }
550 case MachO::GENERIC_RELOC_TLV: {
551 printRelocationTargetName(Obj, RE, Fmt);
552 Fmt << "@TLV";
553 if (IsPCRel)
554 Fmt << "P";
555 break;
556 }
557 default:
558 printRelocationTargetName(Obj, RE, Fmt);
559 }
560 } else { // ARM-specific relocations
561 switch (Type) {
562 case MachO::ARM_RELOC_HALF:
563 case MachO::ARM_RELOC_HALF_SECTDIFF: {
564 // Half relocations steal a bit from the length field to encode
565 // whether this is an upper16 or a lower16 relocation.
566 bool isUpper = (Obj->getAnyRelocationLength(RE) & 0x1) == 1;
567
568 if (isUpper)
569 Fmt << ":upper16:(";
570 else
571 Fmt << ":lower16:(";
572 printRelocationTargetName(Obj, RE, Fmt);
573
574 DataRefImpl RelNext = Rel;
575 Obj->moveRelocationNext(RelNext);
576 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
577
578 // ARM half relocs must be followed by a relocation of type
579 // ARM_RELOC_PAIR.
580 unsigned RType = Obj->getAnyRelocationType(RENext);
581 if (RType != MachO::ARM_RELOC_PAIR)
582 reportError(Obj->getFileName(), "Expected ARM_RELOC_PAIR after "
583 "ARM_RELOC_HALF");
584
585 // NOTE: The half of the target virtual address is stashed in the
586 // address field of the secondary relocation, but we can't reverse
587 // engineer the constant offset from it without decoding the movw/movt
588 // instruction to find the other half in its immediate field.
589
590 // ARM_RELOC_HALF_SECTDIFF encodes the second section in the
591 // symbol/section pointer of the follow-on relocation.
592 if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) {
593 Fmt << "-";
594 printRelocationTargetName(Obj, RENext, Fmt);
595 }
596
597 Fmt << ")";
598 break;
599 }
600 default: {
601 printRelocationTargetName(Obj, RE, Fmt);
602 }
603 }
604 }
605 } else
606 printRelocationTargetName(Obj, RE, Fmt);
607
608 Fmt.flush();
609 Result.append(FmtBuf.begin(), FmtBuf.end());
610 return Error::success();
611 }
612
PrintIndirectSymbolTable(MachOObjectFile * O,bool verbose,uint32_t n,uint32_t count,uint32_t stride,uint64_t addr)613 static void PrintIndirectSymbolTable(MachOObjectFile *O, bool verbose,
614 uint32_t n, uint32_t count,
615 uint32_t stride, uint64_t addr) {
616 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
617 uint32_t nindirectsyms = Dysymtab.nindirectsyms;
618 if (n > nindirectsyms)
619 outs() << " (entries start past the end of the indirect symbol "
620 "table) (reserved1 field greater than the table size)";
621 else if (n + count > nindirectsyms)
622 outs() << " (entries extends past the end of the indirect symbol "
623 "table)";
624 outs() << "\n";
625 uint32_t cputype = O->getHeader().cputype;
626 if (cputype & MachO::CPU_ARCH_ABI64)
627 outs() << "address index";
628 else
629 outs() << "address index";
630 if (verbose)
631 outs() << " name\n";
632 else
633 outs() << "\n";
634 for (uint32_t j = 0; j < count && n + j < nindirectsyms; j++) {
635 if (cputype & MachO::CPU_ARCH_ABI64)
636 outs() << format("0x%016" PRIx64, addr + j * stride) << " ";
637 else
638 outs() << format("0x%08" PRIx32, (uint32_t)addr + j * stride) << " ";
639 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
640 uint32_t indirect_symbol = O->getIndirectSymbolTableEntry(Dysymtab, n + j);
641 if (indirect_symbol == MachO::INDIRECT_SYMBOL_LOCAL) {
642 outs() << "LOCAL\n";
643 continue;
644 }
645 if (indirect_symbol ==
646 (MachO::INDIRECT_SYMBOL_LOCAL | MachO::INDIRECT_SYMBOL_ABS)) {
647 outs() << "LOCAL ABSOLUTE\n";
648 continue;
649 }
650 if (indirect_symbol == MachO::INDIRECT_SYMBOL_ABS) {
651 outs() << "ABSOLUTE\n";
652 continue;
653 }
654 outs() << format("%5u ", indirect_symbol);
655 if (verbose) {
656 MachO::symtab_command Symtab = O->getSymtabLoadCommand();
657 if (indirect_symbol < Symtab.nsyms) {
658 symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol);
659 SymbolRef Symbol = *Sym;
660 outs() << unwrapOrError(Symbol.getName(), O->getFileName());
661 } else {
662 outs() << "?";
663 }
664 }
665 outs() << "\n";
666 }
667 }
668
PrintIndirectSymbols(MachOObjectFile * O,bool verbose)669 static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) {
670 for (const auto &Load : O->load_commands()) {
671 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
672 MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load);
673 for (unsigned J = 0; J < Seg.nsects; ++J) {
674 MachO::section_64 Sec = O->getSection64(Load, J);
675 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
676 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
677 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
678 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
679 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
680 section_type == MachO::S_SYMBOL_STUBS) {
681 uint32_t stride;
682 if (section_type == MachO::S_SYMBOL_STUBS)
683 stride = Sec.reserved2;
684 else
685 stride = 8;
686 if (stride == 0) {
687 outs() << "Can't print indirect symbols for (" << Sec.segname << ","
688 << Sec.sectname << ") "
689 << "(size of stubs in reserved2 field is zero)\n";
690 continue;
691 }
692 uint32_t count = Sec.size / stride;
693 outs() << "Indirect symbols for (" << Sec.segname << ","
694 << Sec.sectname << ") " << count << " entries";
695 uint32_t n = Sec.reserved1;
696 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
697 }
698 }
699 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
700 MachO::segment_command Seg = O->getSegmentLoadCommand(Load);
701 for (unsigned J = 0; J < Seg.nsects; ++J) {
702 MachO::section Sec = O->getSection(Load, J);
703 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
704 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
705 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
706 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
707 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
708 section_type == MachO::S_SYMBOL_STUBS) {
709 uint32_t stride;
710 if (section_type == MachO::S_SYMBOL_STUBS)
711 stride = Sec.reserved2;
712 else
713 stride = 4;
714 if (stride == 0) {
715 outs() << "Can't print indirect symbols for (" << Sec.segname << ","
716 << Sec.sectname << ") "
717 << "(size of stubs in reserved2 field is zero)\n";
718 continue;
719 }
720 uint32_t count = Sec.size / stride;
721 outs() << "Indirect symbols for (" << Sec.segname << ","
722 << Sec.sectname << ") " << count << " entries";
723 uint32_t n = Sec.reserved1;
724 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
725 }
726 }
727 }
728 }
729 }
730
PrintRType(const uint64_t cputype,const unsigned r_type)731 static void PrintRType(const uint64_t cputype, const unsigned r_type) {
732 static char const *generic_r_types[] = {
733 "VANILLA ", "PAIR ", "SECTDIF ", "PBLAPTR ", "LOCSDIF ", "TLV ",
734 " 6 (?) ", " 7 (?) ", " 8 (?) ", " 9 (?) ", " 10 (?) ", " 11 (?) ",
735 " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
736 };
737 static char const *x86_64_r_types[] = {
738 "UNSIGND ", "SIGNED ", "BRANCH ", "GOT_LD ", "GOT ", "SUB ",
739 "SIGNED1 ", "SIGNED2 ", "SIGNED4 ", "TLV ", " 10 (?) ", " 11 (?) ",
740 " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
741 };
742 static char const *arm_r_types[] = {
743 "VANILLA ", "PAIR ", "SECTDIFF", "LOCSDIF ", "PBLAPTR ",
744 "BR24 ", "T_BR22 ", "T_BR32 ", "HALF ", "HALFDIF ",
745 " 10 (?) ", " 11 (?) ", " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
746 };
747 static char const *arm64_r_types[] = {
748 "UNSIGND ", "SUB ", "BR26 ", "PAGE21 ", "PAGOF12 ",
749 "GOTLDP ", "GOTLDPOF", "PTRTGOT ", "TLVLDP ", "TLVLDPOF",
750 "ADDEND ", " 11 (?) ", " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
751 };
752
753 if (r_type > 0xf){
754 outs() << format("%-7u", r_type) << " ";
755 return;
756 }
757 switch (cputype) {
758 case MachO::CPU_TYPE_I386:
759 outs() << generic_r_types[r_type];
760 break;
761 case MachO::CPU_TYPE_X86_64:
762 outs() << x86_64_r_types[r_type];
763 break;
764 case MachO::CPU_TYPE_ARM:
765 outs() << arm_r_types[r_type];
766 break;
767 case MachO::CPU_TYPE_ARM64:
768 case MachO::CPU_TYPE_ARM64_32:
769 outs() << arm64_r_types[r_type];
770 break;
771 default:
772 outs() << format("%-7u ", r_type);
773 }
774 }
775
PrintRLength(const uint64_t cputype,const unsigned r_type,const unsigned r_length,const bool previous_arm_half)776 static void PrintRLength(const uint64_t cputype, const unsigned r_type,
777 const unsigned r_length, const bool previous_arm_half){
778 if (cputype == MachO::CPU_TYPE_ARM &&
779 (r_type == MachO::ARM_RELOC_HALF ||
780 r_type == MachO::ARM_RELOC_HALF_SECTDIFF || previous_arm_half == true)) {
781 if ((r_length & 0x1) == 0)
782 outs() << "lo/";
783 else
784 outs() << "hi/";
785 if ((r_length & 0x1) == 0)
786 outs() << "arm ";
787 else
788 outs() << "thm ";
789 } else {
790 switch (r_length) {
791 case 0:
792 outs() << "byte ";
793 break;
794 case 1:
795 outs() << "word ";
796 break;
797 case 2:
798 outs() << "long ";
799 break;
800 case 3:
801 if (cputype == MachO::CPU_TYPE_X86_64)
802 outs() << "quad ";
803 else
804 outs() << format("?(%2d) ", r_length);
805 break;
806 default:
807 outs() << format("?(%2d) ", r_length);
808 }
809 }
810 }
811
PrintRelocationEntries(const MachOObjectFile * O,const relocation_iterator Begin,const relocation_iterator End,const uint64_t cputype,const bool verbose)812 static void PrintRelocationEntries(const MachOObjectFile *O,
813 const relocation_iterator Begin,
814 const relocation_iterator End,
815 const uint64_t cputype,
816 const bool verbose) {
817 const MachO::symtab_command Symtab = O->getSymtabLoadCommand();
818 bool previous_arm_half = false;
819 bool previous_sectdiff = false;
820 uint32_t sectdiff_r_type = 0;
821
822 for (relocation_iterator Reloc = Begin; Reloc != End; ++Reloc) {
823 const DataRefImpl Rel = Reloc->getRawDataRefImpl();
824 const MachO::any_relocation_info RE = O->getRelocation(Rel);
825 const unsigned r_type = O->getAnyRelocationType(RE);
826 const bool r_scattered = O->isRelocationScattered(RE);
827 const unsigned r_pcrel = O->getAnyRelocationPCRel(RE);
828 const unsigned r_length = O->getAnyRelocationLength(RE);
829 const unsigned r_address = O->getAnyRelocationAddress(RE);
830 const bool r_extern = (r_scattered ? false :
831 O->getPlainRelocationExternal(RE));
832 const uint32_t r_value = (r_scattered ?
833 O->getScatteredRelocationValue(RE) : 0);
834 const unsigned r_symbolnum = (r_scattered ? 0 :
835 O->getPlainRelocationSymbolNum(RE));
836
837 if (r_scattered && cputype != MachO::CPU_TYPE_X86_64) {
838 if (verbose) {
839 // scattered: address
840 if ((cputype == MachO::CPU_TYPE_I386 &&
841 r_type == MachO::GENERIC_RELOC_PAIR) ||
842 (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR))
843 outs() << " ";
844 else
845 outs() << format("%08x ", (unsigned int)r_address);
846
847 // scattered: pcrel
848 if (r_pcrel)
849 outs() << "True ";
850 else
851 outs() << "False ";
852
853 // scattered: length
854 PrintRLength(cputype, r_type, r_length, previous_arm_half);
855
856 // scattered: extern & type
857 outs() << "n/a ";
858 PrintRType(cputype, r_type);
859
860 // scattered: scattered & value
861 outs() << format("True 0x%08x", (unsigned int)r_value);
862 if (previous_sectdiff == false) {
863 if ((cputype == MachO::CPU_TYPE_ARM &&
864 r_type == MachO::ARM_RELOC_PAIR))
865 outs() << format(" half = 0x%04x ", (unsigned int)r_address);
866 } else if (cputype == MachO::CPU_TYPE_ARM &&
867 sectdiff_r_type == MachO::ARM_RELOC_HALF_SECTDIFF)
868 outs() << format(" other_half = 0x%04x ", (unsigned int)r_address);
869 if ((cputype == MachO::CPU_TYPE_I386 &&
870 (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
871 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) ||
872 (cputype == MachO::CPU_TYPE_ARM &&
873 (sectdiff_r_type == MachO::ARM_RELOC_SECTDIFF ||
874 sectdiff_r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
875 sectdiff_r_type == MachO::ARM_RELOC_HALF_SECTDIFF))) {
876 previous_sectdiff = true;
877 sectdiff_r_type = r_type;
878 } else {
879 previous_sectdiff = false;
880 sectdiff_r_type = 0;
881 }
882 if (cputype == MachO::CPU_TYPE_ARM &&
883 (r_type == MachO::ARM_RELOC_HALF ||
884 r_type == MachO::ARM_RELOC_HALF_SECTDIFF))
885 previous_arm_half = true;
886 else
887 previous_arm_half = false;
888 outs() << "\n";
889 }
890 else {
891 // scattered: address pcrel length extern type scattered value
892 outs() << format("%08x %1d %-2d n/a %-7d 1 0x%08x\n",
893 (unsigned int)r_address, r_pcrel, r_length, r_type,
894 (unsigned int)r_value);
895 }
896 }
897 else {
898 if (verbose) {
899 // plain: address
900 if (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR)
901 outs() << " ";
902 else
903 outs() << format("%08x ", (unsigned int)r_address);
904
905 // plain: pcrel
906 if (r_pcrel)
907 outs() << "True ";
908 else
909 outs() << "False ";
910
911 // plain: length
912 PrintRLength(cputype, r_type, r_length, previous_arm_half);
913
914 if (r_extern) {
915 // plain: extern & type & scattered
916 outs() << "True ";
917 PrintRType(cputype, r_type);
918 outs() << "False ";
919
920 // plain: symbolnum/value
921 if (r_symbolnum > Symtab.nsyms)
922 outs() << format("?(%d)\n", r_symbolnum);
923 else {
924 SymbolRef Symbol = *O->getSymbolByIndex(r_symbolnum);
925 Expected<StringRef> SymNameNext = Symbol.getName();
926 const char *name = nullptr;
927 if (SymNameNext)
928 name = SymNameNext->data();
929 if (name == nullptr)
930 outs() << format("?(%d)\n", r_symbolnum);
931 else
932 outs() << name << "\n";
933 }
934 }
935 else {
936 // plain: extern & type & scattered
937 outs() << "False ";
938 PrintRType(cputype, r_type);
939 outs() << "False ";
940
941 // plain: symbolnum/value
942 if (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR)
943 outs() << format("other_half = 0x%04x\n", (unsigned int)r_address);
944 else if ((cputype == MachO::CPU_TYPE_ARM64 ||
945 cputype == MachO::CPU_TYPE_ARM64_32) &&
946 r_type == MachO::ARM64_RELOC_ADDEND)
947 outs() << format("addend = 0x%06x\n", (unsigned int)r_symbolnum);
948 else {
949 outs() << format("%d ", r_symbolnum);
950 if (r_symbolnum == MachO::R_ABS)
951 outs() << "R_ABS\n";
952 else {
953 // in this case, r_symbolnum is actually a 1-based section number
954 uint32_t nsects = O->section_end()->getRawDataRefImpl().d.a;
955 if (r_symbolnum > 0 && r_symbolnum <= nsects) {
956 object::DataRefImpl DRI;
957 DRI.d.a = r_symbolnum-1;
958 StringRef SegName = O->getSectionFinalSegmentName(DRI);
959 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI))
960 outs() << "(" << SegName << "," << *NameOrErr << ")\n";
961 else
962 outs() << "(?,?)\n";
963 }
964 else {
965 outs() << "(?,?)\n";
966 }
967 }
968 }
969 }
970 if (cputype == MachO::CPU_TYPE_ARM &&
971 (r_type == MachO::ARM_RELOC_HALF ||
972 r_type == MachO::ARM_RELOC_HALF_SECTDIFF))
973 previous_arm_half = true;
974 else
975 previous_arm_half = false;
976 }
977 else {
978 // plain: address pcrel length extern type scattered symbolnum/section
979 outs() << format("%08x %1d %-2d %1d %-7d 0 %d\n",
980 (unsigned int)r_address, r_pcrel, r_length, r_extern,
981 r_type, r_symbolnum);
982 }
983 }
984 }
985 }
986
PrintRelocations(const MachOObjectFile * O,const bool verbose)987 static void PrintRelocations(const MachOObjectFile *O, const bool verbose) {
988 const uint64_t cputype = O->getHeader().cputype;
989 const MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
990 if (Dysymtab.nextrel != 0) {
991 outs() << "External relocation information " << Dysymtab.nextrel
992 << " entries";
993 outs() << "\naddress pcrel length extern type scattered "
994 "symbolnum/value\n";
995 PrintRelocationEntries(O, O->extrel_begin(), O->extrel_end(), cputype,
996 verbose);
997 }
998 if (Dysymtab.nlocrel != 0) {
999 outs() << format("Local relocation information %u entries",
1000 Dysymtab.nlocrel);
1001 outs() << "\naddress pcrel length extern type scattered "
1002 "symbolnum/value\n";
1003 PrintRelocationEntries(O, O->locrel_begin(), O->locrel_end(), cputype,
1004 verbose);
1005 }
1006 for (const auto &Load : O->load_commands()) {
1007 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
1008 const MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load);
1009 for (unsigned J = 0; J < Seg.nsects; ++J) {
1010 const MachO::section_64 Sec = O->getSection64(Load, J);
1011 if (Sec.nreloc != 0) {
1012 DataRefImpl DRI;
1013 DRI.d.a = J;
1014 const StringRef SegName = O->getSectionFinalSegmentName(DRI);
1015 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI))
1016 outs() << "Relocation information (" << SegName << "," << *NameOrErr
1017 << format(") %u entries", Sec.nreloc);
1018 else
1019 outs() << "Relocation information (" << SegName << ",?) "
1020 << format("%u entries", Sec.nreloc);
1021 outs() << "\naddress pcrel length extern type scattered "
1022 "symbolnum/value\n";
1023 PrintRelocationEntries(O, O->section_rel_begin(DRI),
1024 O->section_rel_end(DRI), cputype, verbose);
1025 }
1026 }
1027 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
1028 const MachO::segment_command Seg = O->getSegmentLoadCommand(Load);
1029 for (unsigned J = 0; J < Seg.nsects; ++J) {
1030 const MachO::section Sec = O->getSection(Load, J);
1031 if (Sec.nreloc != 0) {
1032 DataRefImpl DRI;
1033 DRI.d.a = J;
1034 const StringRef SegName = O->getSectionFinalSegmentName(DRI);
1035 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI))
1036 outs() << "Relocation information (" << SegName << "," << *NameOrErr
1037 << format(") %u entries", Sec.nreloc);
1038 else
1039 outs() << "Relocation information (" << SegName << ",?) "
1040 << format("%u entries", Sec.nreloc);
1041 outs() << "\naddress pcrel length extern type scattered "
1042 "symbolnum/value\n";
1043 PrintRelocationEntries(O, O->section_rel_begin(DRI),
1044 O->section_rel_end(DRI), cputype, verbose);
1045 }
1046 }
1047 }
1048 }
1049 }
1050
PrintFunctionStarts(MachOObjectFile * O)1051 static void PrintFunctionStarts(MachOObjectFile *O) {
1052 uint64_t BaseSegmentAddress = 0;
1053 for (const MachOObjectFile::LoadCommandInfo &Command : O->load_commands()) {
1054 if (Command.C.cmd == MachO::LC_SEGMENT) {
1055 MachO::segment_command SLC = O->getSegmentLoadCommand(Command);
1056 if (StringRef(SLC.segname) == "__TEXT") {
1057 BaseSegmentAddress = SLC.vmaddr;
1058 break;
1059 }
1060 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
1061 MachO::segment_command_64 SLC = O->getSegment64LoadCommand(Command);
1062 if (StringRef(SLC.segname) == "__TEXT") {
1063 BaseSegmentAddress = SLC.vmaddr;
1064 break;
1065 }
1066 }
1067 }
1068
1069 SmallVector<uint64_t, 8> FunctionStarts;
1070 for (const MachOObjectFile::LoadCommandInfo &LC : O->load_commands()) {
1071 if (LC.C.cmd == MachO::LC_FUNCTION_STARTS) {
1072 MachO::linkedit_data_command FunctionStartsLC =
1073 O->getLinkeditDataLoadCommand(LC);
1074 O->ReadULEB128s(FunctionStartsLC.dataoff, FunctionStarts);
1075 break;
1076 }
1077 }
1078
1079 for (uint64_t S : FunctionStarts) {
1080 uint64_t Addr = BaseSegmentAddress + S;
1081 if (O->is64Bit())
1082 outs() << format("%016" PRIx64, Addr) << "\n";
1083 else
1084 outs() << format("%08" PRIx32, static_cast<uint32_t>(Addr)) << "\n";
1085 }
1086 }
1087
PrintDataInCodeTable(MachOObjectFile * O,bool verbose)1088 static void PrintDataInCodeTable(MachOObjectFile *O, bool verbose) {
1089 MachO::linkedit_data_command DIC = O->getDataInCodeLoadCommand();
1090 uint32_t nentries = DIC.datasize / sizeof(struct MachO::data_in_code_entry);
1091 outs() << "Data in code table (" << nentries << " entries)\n";
1092 outs() << "offset length kind\n";
1093 for (dice_iterator DI = O->begin_dices(), DE = O->end_dices(); DI != DE;
1094 ++DI) {
1095 uint32_t Offset;
1096 DI->getOffset(Offset);
1097 outs() << format("0x%08" PRIx32, Offset) << " ";
1098 uint16_t Length;
1099 DI->getLength(Length);
1100 outs() << format("%6u", Length) << " ";
1101 uint16_t Kind;
1102 DI->getKind(Kind);
1103 if (verbose) {
1104 switch (Kind) {
1105 case MachO::DICE_KIND_DATA:
1106 outs() << "DATA";
1107 break;
1108 case MachO::DICE_KIND_JUMP_TABLE8:
1109 outs() << "JUMP_TABLE8";
1110 break;
1111 case MachO::DICE_KIND_JUMP_TABLE16:
1112 outs() << "JUMP_TABLE16";
1113 break;
1114 case MachO::DICE_KIND_JUMP_TABLE32:
1115 outs() << "JUMP_TABLE32";
1116 break;
1117 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
1118 outs() << "ABS_JUMP_TABLE32";
1119 break;
1120 default:
1121 outs() << format("0x%04" PRIx32, Kind);
1122 break;
1123 }
1124 } else
1125 outs() << format("0x%04" PRIx32, Kind);
1126 outs() << "\n";
1127 }
1128 }
1129
PrintLinkOptHints(MachOObjectFile * O)1130 static void PrintLinkOptHints(MachOObjectFile *O) {
1131 MachO::linkedit_data_command LohLC = O->getLinkOptHintsLoadCommand();
1132 const char *loh = O->getData().substr(LohLC.dataoff, 1).data();
1133 uint32_t nloh = LohLC.datasize;
1134 outs() << "Linker optimiztion hints (" << nloh << " total bytes)\n";
1135 for (uint32_t i = 0; i < nloh;) {
1136 unsigned n;
1137 uint64_t identifier = decodeULEB128((const uint8_t *)(loh + i), &n);
1138 i += n;
1139 outs() << " identifier " << identifier << " ";
1140 if (i >= nloh)
1141 return;
1142 switch (identifier) {
1143 case 1:
1144 outs() << "AdrpAdrp\n";
1145 break;
1146 case 2:
1147 outs() << "AdrpLdr\n";
1148 break;
1149 case 3:
1150 outs() << "AdrpAddLdr\n";
1151 break;
1152 case 4:
1153 outs() << "AdrpLdrGotLdr\n";
1154 break;
1155 case 5:
1156 outs() << "AdrpAddStr\n";
1157 break;
1158 case 6:
1159 outs() << "AdrpLdrGotStr\n";
1160 break;
1161 case 7:
1162 outs() << "AdrpAdd\n";
1163 break;
1164 case 8:
1165 outs() << "AdrpLdrGot\n";
1166 break;
1167 default:
1168 outs() << "Unknown identifier value\n";
1169 break;
1170 }
1171 uint64_t narguments = decodeULEB128((const uint8_t *)(loh + i), &n);
1172 i += n;
1173 outs() << " narguments " << narguments << "\n";
1174 if (i >= nloh)
1175 return;
1176
1177 for (uint32_t j = 0; j < narguments; j++) {
1178 uint64_t value = decodeULEB128((const uint8_t *)(loh + i), &n);
1179 i += n;
1180 outs() << "\tvalue " << format("0x%" PRIx64, value) << "\n";
1181 if (i >= nloh)
1182 return;
1183 }
1184 }
1185 }
1186
printMachOChainedFixups(object::MachOObjectFile * Obj)1187 static void printMachOChainedFixups(object::MachOObjectFile *Obj) {
1188 Error Err = Error::success();
1189 for (const object::MachOChainedFixupEntry &Entry : Obj->fixupTable(Err)) {
1190 (void)Entry;
1191 }
1192 if (Err)
1193 reportError(std::move(Err), Obj->getFileName());
1194 }
1195
PrintDyldInfo(MachOObjectFile * O)1196 static void PrintDyldInfo(MachOObjectFile *O) {
1197 outs() << "dyld information:" << '\n';
1198 printMachOChainedFixups(O);
1199 }
1200
PrintDylibs(MachOObjectFile * O,bool JustId)1201 static void PrintDylibs(MachOObjectFile *O, bool JustId) {
1202 unsigned Index = 0;
1203 for (const auto &Load : O->load_commands()) {
1204 if ((JustId && Load.C.cmd == MachO::LC_ID_DYLIB) ||
1205 (!JustId && (Load.C.cmd == MachO::LC_ID_DYLIB ||
1206 Load.C.cmd == MachO::LC_LOAD_DYLIB ||
1207 Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
1208 Load.C.cmd == MachO::LC_REEXPORT_DYLIB ||
1209 Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
1210 Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB))) {
1211 MachO::dylib_command dl = O->getDylibIDLoadCommand(Load);
1212 if (dl.dylib.name < dl.cmdsize) {
1213 const char *p = (const char *)(Load.Ptr) + dl.dylib.name;
1214 if (JustId)
1215 outs() << p << "\n";
1216 else {
1217 outs() << "\t" << p;
1218 outs() << " (compatibility version "
1219 << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
1220 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
1221 << (dl.dylib.compatibility_version & 0xff) << ",";
1222 outs() << " current version "
1223 << ((dl.dylib.current_version >> 16) & 0xffff) << "."
1224 << ((dl.dylib.current_version >> 8) & 0xff) << "."
1225 << (dl.dylib.current_version & 0xff);
1226 if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB)
1227 outs() << ", weak";
1228 if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB)
1229 outs() << ", reexport";
1230 if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
1231 outs() << ", upward";
1232 if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB)
1233 outs() << ", lazy";
1234 outs() << ")\n";
1235 }
1236 } else {
1237 outs() << "\tBad offset (" << dl.dylib.name << ") for name of ";
1238 if (Load.C.cmd == MachO::LC_ID_DYLIB)
1239 outs() << "LC_ID_DYLIB ";
1240 else if (Load.C.cmd == MachO::LC_LOAD_DYLIB)
1241 outs() << "LC_LOAD_DYLIB ";
1242 else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB)
1243 outs() << "LC_LOAD_WEAK_DYLIB ";
1244 else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB)
1245 outs() << "LC_LAZY_LOAD_DYLIB ";
1246 else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB)
1247 outs() << "LC_REEXPORT_DYLIB ";
1248 else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
1249 outs() << "LC_LOAD_UPWARD_DYLIB ";
1250 else
1251 outs() << "LC_??? ";
1252 outs() << "command " << Index++ << "\n";
1253 }
1254 }
1255 }
1256 }
1257
printRpaths(MachOObjectFile * O)1258 static void printRpaths(MachOObjectFile *O) {
1259 for (const auto &Command : O->load_commands()) {
1260 if (Command.C.cmd == MachO::LC_RPATH) {
1261 auto Rpath = O->getRpathCommand(Command);
1262 const char *P = (const char *)(Command.Ptr) + Rpath.path;
1263 outs() << P << "\n";
1264 }
1265 }
1266 }
1267
1268 typedef DenseMap<uint64_t, StringRef> SymbolAddressMap;
1269
CreateSymbolAddressMap(MachOObjectFile * O,SymbolAddressMap * AddrMap)1270 static void CreateSymbolAddressMap(MachOObjectFile *O,
1271 SymbolAddressMap *AddrMap) {
1272 // Create a map of symbol addresses to symbol names.
1273 const StringRef FileName = O->getFileName();
1274 for (const SymbolRef &Symbol : O->symbols()) {
1275 SymbolRef::Type ST = unwrapOrError(Symbol.getType(), FileName);
1276 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
1277 ST == SymbolRef::ST_Other) {
1278 uint64_t Address = cantFail(Symbol.getValue());
1279 StringRef SymName = unwrapOrError(Symbol.getName(), FileName);
1280 if (!SymName.startswith(".objc"))
1281 (*AddrMap)[Address] = SymName;
1282 }
1283 }
1284 }
1285
1286 // GuessSymbolName is passed the address of what might be a symbol and a
1287 // pointer to the SymbolAddressMap. It returns the name of a symbol
1288 // with that address or nullptr if no symbol is found with that address.
GuessSymbolName(uint64_t value,SymbolAddressMap * AddrMap)1289 static const char *GuessSymbolName(uint64_t value, SymbolAddressMap *AddrMap) {
1290 const char *SymbolName = nullptr;
1291 // A DenseMap can't lookup up some values.
1292 if (value != 0xffffffffffffffffULL && value != 0xfffffffffffffffeULL) {
1293 StringRef name = AddrMap->lookup(value);
1294 if (!name.empty())
1295 SymbolName = name.data();
1296 }
1297 return SymbolName;
1298 }
1299
DumpCstringChar(const char c)1300 static void DumpCstringChar(const char c) {
1301 char p[2];
1302 p[0] = c;
1303 p[1] = '\0';
1304 outs().write_escaped(p);
1305 }
1306
DumpCstringSection(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1307 static void DumpCstringSection(MachOObjectFile *O, const char *sect,
1308 uint32_t sect_size, uint64_t sect_addr,
1309 bool print_addresses) {
1310 for (uint32_t i = 0; i < sect_size; i++) {
1311 if (print_addresses) {
1312 if (O->is64Bit())
1313 outs() << format("%016" PRIx64, sect_addr + i) << " ";
1314 else
1315 outs() << format("%08" PRIx64, sect_addr + i) << " ";
1316 }
1317 for (; i < sect_size && sect[i] != '\0'; i++)
1318 DumpCstringChar(sect[i]);
1319 if (i < sect_size && sect[i] == '\0')
1320 outs() << "\n";
1321 }
1322 }
1323
DumpLiteral4(uint32_t l,float f)1324 static void DumpLiteral4(uint32_t l, float f) {
1325 outs() << format("0x%08" PRIx32, l);
1326 if ((l & 0x7f800000) != 0x7f800000)
1327 outs() << format(" (%.16e)\n", f);
1328 else {
1329 if (l == 0x7f800000)
1330 outs() << " (+Infinity)\n";
1331 else if (l == 0xff800000)
1332 outs() << " (-Infinity)\n";
1333 else if ((l & 0x00400000) == 0x00400000)
1334 outs() << " (non-signaling Not-a-Number)\n";
1335 else
1336 outs() << " (signaling Not-a-Number)\n";
1337 }
1338 }
1339
DumpLiteral4Section(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1340 static void DumpLiteral4Section(MachOObjectFile *O, const char *sect,
1341 uint32_t sect_size, uint64_t sect_addr,
1342 bool print_addresses) {
1343 for (uint32_t i = 0; i < sect_size; i += sizeof(float)) {
1344 if (print_addresses) {
1345 if (O->is64Bit())
1346 outs() << format("%016" PRIx64, sect_addr + i) << " ";
1347 else
1348 outs() << format("%08" PRIx64, sect_addr + i) << " ";
1349 }
1350 float f;
1351 memcpy(&f, sect + i, sizeof(float));
1352 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1353 sys::swapByteOrder(f);
1354 uint32_t l;
1355 memcpy(&l, sect + i, sizeof(uint32_t));
1356 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1357 sys::swapByteOrder(l);
1358 DumpLiteral4(l, f);
1359 }
1360 }
1361
DumpLiteral8(MachOObjectFile * O,uint32_t l0,uint32_t l1,double d)1362 static void DumpLiteral8(MachOObjectFile *O, uint32_t l0, uint32_t l1,
1363 double d) {
1364 outs() << format("0x%08" PRIx32, l0) << " " << format("0x%08" PRIx32, l1);
1365 uint32_t Hi, Lo;
1366 Hi = (O->isLittleEndian()) ? l1 : l0;
1367 Lo = (O->isLittleEndian()) ? l0 : l1;
1368
1369 // Hi is the high word, so this is equivalent to if(isfinite(d))
1370 if ((Hi & 0x7ff00000) != 0x7ff00000)
1371 outs() << format(" (%.16e)\n", d);
1372 else {
1373 if (Hi == 0x7ff00000 && Lo == 0)
1374 outs() << " (+Infinity)\n";
1375 else if (Hi == 0xfff00000 && Lo == 0)
1376 outs() << " (-Infinity)\n";
1377 else if ((Hi & 0x00080000) == 0x00080000)
1378 outs() << " (non-signaling Not-a-Number)\n";
1379 else
1380 outs() << " (signaling Not-a-Number)\n";
1381 }
1382 }
1383
DumpLiteral8Section(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1384 static void DumpLiteral8Section(MachOObjectFile *O, const char *sect,
1385 uint32_t sect_size, uint64_t sect_addr,
1386 bool print_addresses) {
1387 for (uint32_t i = 0; i < sect_size; i += sizeof(double)) {
1388 if (print_addresses) {
1389 if (O->is64Bit())
1390 outs() << format("%016" PRIx64, sect_addr + i) << " ";
1391 else
1392 outs() << format("%08" PRIx64, sect_addr + i) << " ";
1393 }
1394 double d;
1395 memcpy(&d, sect + i, sizeof(double));
1396 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1397 sys::swapByteOrder(d);
1398 uint32_t l0, l1;
1399 memcpy(&l0, sect + i, sizeof(uint32_t));
1400 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
1401 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1402 sys::swapByteOrder(l0);
1403 sys::swapByteOrder(l1);
1404 }
1405 DumpLiteral8(O, l0, l1, d);
1406 }
1407 }
1408
DumpLiteral16(uint32_t l0,uint32_t l1,uint32_t l2,uint32_t l3)1409 static void DumpLiteral16(uint32_t l0, uint32_t l1, uint32_t l2, uint32_t l3) {
1410 outs() << format("0x%08" PRIx32, l0) << " ";
1411 outs() << format("0x%08" PRIx32, l1) << " ";
1412 outs() << format("0x%08" PRIx32, l2) << " ";
1413 outs() << format("0x%08" PRIx32, l3) << "\n";
1414 }
1415
DumpLiteral16Section(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1416 static void DumpLiteral16Section(MachOObjectFile *O, const char *sect,
1417 uint32_t sect_size, uint64_t sect_addr,
1418 bool print_addresses) {
1419 for (uint32_t i = 0; i < sect_size; i += 16) {
1420 if (print_addresses) {
1421 if (O->is64Bit())
1422 outs() << format("%016" PRIx64, sect_addr + i) << " ";
1423 else
1424 outs() << format("%08" PRIx64, sect_addr + i) << " ";
1425 }
1426 uint32_t l0, l1, l2, l3;
1427 memcpy(&l0, sect + i, sizeof(uint32_t));
1428 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
1429 memcpy(&l2, sect + i + 2 * sizeof(uint32_t), sizeof(uint32_t));
1430 memcpy(&l3, sect + i + 3 * sizeof(uint32_t), sizeof(uint32_t));
1431 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1432 sys::swapByteOrder(l0);
1433 sys::swapByteOrder(l1);
1434 sys::swapByteOrder(l2);
1435 sys::swapByteOrder(l3);
1436 }
1437 DumpLiteral16(l0, l1, l2, l3);
1438 }
1439 }
1440
DumpLiteralPointerSection(MachOObjectFile * O,const SectionRef & Section,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1441 static void DumpLiteralPointerSection(MachOObjectFile *O,
1442 const SectionRef &Section,
1443 const char *sect, uint32_t sect_size,
1444 uint64_t sect_addr,
1445 bool print_addresses) {
1446 // Collect the literal sections in this Mach-O file.
1447 std::vector<SectionRef> LiteralSections;
1448 for (const SectionRef &Section : O->sections()) {
1449 DataRefImpl Ref = Section.getRawDataRefImpl();
1450 uint32_t section_type;
1451 if (O->is64Bit()) {
1452 const MachO::section_64 Sec = O->getSection64(Ref);
1453 section_type = Sec.flags & MachO::SECTION_TYPE;
1454 } else {
1455 const MachO::section Sec = O->getSection(Ref);
1456 section_type = Sec.flags & MachO::SECTION_TYPE;
1457 }
1458 if (section_type == MachO::S_CSTRING_LITERALS ||
1459 section_type == MachO::S_4BYTE_LITERALS ||
1460 section_type == MachO::S_8BYTE_LITERALS ||
1461 section_type == MachO::S_16BYTE_LITERALS)
1462 LiteralSections.push_back(Section);
1463 }
1464
1465 // Set the size of the literal pointer.
1466 uint32_t lp_size = O->is64Bit() ? 8 : 4;
1467
1468 // Collect the external relocation symbols for the literal pointers.
1469 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
1470 for (const RelocationRef &Reloc : Section.relocations()) {
1471 DataRefImpl Rel;
1472 MachO::any_relocation_info RE;
1473 bool isExtern = false;
1474 Rel = Reloc.getRawDataRefImpl();
1475 RE = O->getRelocation(Rel);
1476 isExtern = O->getPlainRelocationExternal(RE);
1477 if (isExtern) {
1478 uint64_t RelocOffset = Reloc.getOffset();
1479 symbol_iterator RelocSym = Reloc.getSymbol();
1480 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
1481 }
1482 }
1483 array_pod_sort(Relocs.begin(), Relocs.end());
1484
1485 // Dump each literal pointer.
1486 for (uint32_t i = 0; i < sect_size; i += lp_size) {
1487 if (print_addresses) {
1488 if (O->is64Bit())
1489 outs() << format("%016" PRIx64, sect_addr + i) << " ";
1490 else
1491 outs() << format("%08" PRIx64, sect_addr + i) << " ";
1492 }
1493 uint64_t lp;
1494 if (O->is64Bit()) {
1495 memcpy(&lp, sect + i, sizeof(uint64_t));
1496 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1497 sys::swapByteOrder(lp);
1498 } else {
1499 uint32_t li;
1500 memcpy(&li, sect + i, sizeof(uint32_t));
1501 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1502 sys::swapByteOrder(li);
1503 lp = li;
1504 }
1505
1506 // First look for an external relocation entry for this literal pointer.
1507 auto Reloc = find_if(Relocs, [&](const std::pair<uint64_t, SymbolRef> &P) {
1508 return P.first == i;
1509 });
1510 if (Reloc != Relocs.end()) {
1511 symbol_iterator RelocSym = Reloc->second;
1512 StringRef SymName = unwrapOrError(RelocSym->getName(), O->getFileName());
1513 outs() << "external relocation entry for symbol:" << SymName << "\n";
1514 continue;
1515 }
1516
1517 // For local references see what the section the literal pointer points to.
1518 auto Sect = find_if(LiteralSections, [&](const SectionRef &R) {
1519 return lp >= R.getAddress() && lp < R.getAddress() + R.getSize();
1520 });
1521 if (Sect == LiteralSections.end()) {
1522 outs() << format("0x%" PRIx64, lp) << " (not in a literal section)\n";
1523 continue;
1524 }
1525
1526 uint64_t SectAddress = Sect->getAddress();
1527 uint64_t SectSize = Sect->getSize();
1528
1529 StringRef SectName;
1530 Expected<StringRef> SectNameOrErr = Sect->getName();
1531 if (SectNameOrErr)
1532 SectName = *SectNameOrErr;
1533 else
1534 consumeError(SectNameOrErr.takeError());
1535
1536 DataRefImpl Ref = Sect->getRawDataRefImpl();
1537 StringRef SegmentName = O->getSectionFinalSegmentName(Ref);
1538 outs() << SegmentName << ":" << SectName << ":";
1539
1540 uint32_t section_type;
1541 if (O->is64Bit()) {
1542 const MachO::section_64 Sec = O->getSection64(Ref);
1543 section_type = Sec.flags & MachO::SECTION_TYPE;
1544 } else {
1545 const MachO::section Sec = O->getSection(Ref);
1546 section_type = Sec.flags & MachO::SECTION_TYPE;
1547 }
1548
1549 StringRef BytesStr = unwrapOrError(Sect->getContents(), O->getFileName());
1550
1551 const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
1552
1553 switch (section_type) {
1554 case MachO::S_CSTRING_LITERALS:
1555 for (uint64_t i = lp - SectAddress; i < SectSize && Contents[i] != '\0';
1556 i++) {
1557 DumpCstringChar(Contents[i]);
1558 }
1559 outs() << "\n";
1560 break;
1561 case MachO::S_4BYTE_LITERALS:
1562 float f;
1563 memcpy(&f, Contents + (lp - SectAddress), sizeof(float));
1564 uint32_t l;
1565 memcpy(&l, Contents + (lp - SectAddress), sizeof(uint32_t));
1566 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1567 sys::swapByteOrder(f);
1568 sys::swapByteOrder(l);
1569 }
1570 DumpLiteral4(l, f);
1571 break;
1572 case MachO::S_8BYTE_LITERALS: {
1573 double d;
1574 memcpy(&d, Contents + (lp - SectAddress), sizeof(double));
1575 uint32_t l0, l1;
1576 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
1577 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
1578 sizeof(uint32_t));
1579 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1580 sys::swapByteOrder(f);
1581 sys::swapByteOrder(l0);
1582 sys::swapByteOrder(l1);
1583 }
1584 DumpLiteral8(O, l0, l1, d);
1585 break;
1586 }
1587 case MachO::S_16BYTE_LITERALS: {
1588 uint32_t l0, l1, l2, l3;
1589 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
1590 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
1591 sizeof(uint32_t));
1592 memcpy(&l2, Contents + (lp - SectAddress) + 2 * sizeof(uint32_t),
1593 sizeof(uint32_t));
1594 memcpy(&l3, Contents + (lp - SectAddress) + 3 * sizeof(uint32_t),
1595 sizeof(uint32_t));
1596 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1597 sys::swapByteOrder(l0);
1598 sys::swapByteOrder(l1);
1599 sys::swapByteOrder(l2);
1600 sys::swapByteOrder(l3);
1601 }
1602 DumpLiteral16(l0, l1, l2, l3);
1603 break;
1604 }
1605 }
1606 }
1607 }
1608
DumpInitTermPointerSection(MachOObjectFile * O,const SectionRef & Section,const char * sect,uint32_t sect_size,uint64_t sect_addr,SymbolAddressMap * AddrMap,bool verbose)1609 static void DumpInitTermPointerSection(MachOObjectFile *O,
1610 const SectionRef &Section,
1611 const char *sect,
1612 uint32_t sect_size, uint64_t sect_addr,
1613 SymbolAddressMap *AddrMap,
1614 bool verbose) {
1615 uint32_t stride;
1616 stride = (O->is64Bit()) ? sizeof(uint64_t) : sizeof(uint32_t);
1617
1618 // Collect the external relocation symbols for the pointers.
1619 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
1620 for (const RelocationRef &Reloc : Section.relocations()) {
1621 DataRefImpl Rel;
1622 MachO::any_relocation_info RE;
1623 bool isExtern = false;
1624 Rel = Reloc.getRawDataRefImpl();
1625 RE = O->getRelocation(Rel);
1626 isExtern = O->getPlainRelocationExternal(RE);
1627 if (isExtern) {
1628 uint64_t RelocOffset = Reloc.getOffset();
1629 symbol_iterator RelocSym = Reloc.getSymbol();
1630 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
1631 }
1632 }
1633 array_pod_sort(Relocs.begin(), Relocs.end());
1634
1635 for (uint32_t i = 0; i < sect_size; i += stride) {
1636 const char *SymbolName = nullptr;
1637 uint64_t p;
1638 if (O->is64Bit()) {
1639 outs() << format("0x%016" PRIx64, sect_addr + i * stride) << " ";
1640 uint64_t pointer_value;
1641 memcpy(&pointer_value, sect + i, stride);
1642 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1643 sys::swapByteOrder(pointer_value);
1644 outs() << format("0x%016" PRIx64, pointer_value);
1645 p = pointer_value;
1646 } else {
1647 outs() << format("0x%08" PRIx64, sect_addr + i * stride) << " ";
1648 uint32_t pointer_value;
1649 memcpy(&pointer_value, sect + i, stride);
1650 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1651 sys::swapByteOrder(pointer_value);
1652 outs() << format("0x%08" PRIx32, pointer_value);
1653 p = pointer_value;
1654 }
1655 if (verbose) {
1656 // First look for an external relocation entry for this pointer.
1657 auto Reloc = find_if(Relocs, [&](const std::pair<uint64_t, SymbolRef> &P) {
1658 return P.first == i;
1659 });
1660 if (Reloc != Relocs.end()) {
1661 symbol_iterator RelocSym = Reloc->second;
1662 outs() << " " << unwrapOrError(RelocSym->getName(), O->getFileName());
1663 } else {
1664 SymbolName = GuessSymbolName(p, AddrMap);
1665 if (SymbolName)
1666 outs() << " " << SymbolName;
1667 }
1668 }
1669 outs() << "\n";
1670 }
1671 }
1672
DumpRawSectionContents(MachOObjectFile * O,const char * sect,uint32_t size,uint64_t addr)1673 static void DumpRawSectionContents(MachOObjectFile *O, const char *sect,
1674 uint32_t size, uint64_t addr) {
1675 uint32_t cputype = O->getHeader().cputype;
1676 if (cputype == MachO::CPU_TYPE_I386 || cputype == MachO::CPU_TYPE_X86_64) {
1677 uint32_t j;
1678 for (uint32_t i = 0; i < size; i += j, addr += j) {
1679 if (O->is64Bit())
1680 outs() << format("%016" PRIx64, addr) << "\t";
1681 else
1682 outs() << format("%08" PRIx64, addr) << "\t";
1683 for (j = 0; j < 16 && i + j < size; j++) {
1684 uint8_t byte_word = *(sect + i + j);
1685 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
1686 }
1687 outs() << "\n";
1688 }
1689 } else {
1690 uint32_t j;
1691 for (uint32_t i = 0; i < size; i += j, addr += j) {
1692 if (O->is64Bit())
1693 outs() << format("%016" PRIx64, addr) << "\t";
1694 else
1695 outs() << format("%08" PRIx64, addr) << "\t";
1696 for (j = 0; j < 4 * sizeof(int32_t) && i + j < size;
1697 j += sizeof(int32_t)) {
1698 if (i + j + sizeof(int32_t) <= size) {
1699 uint32_t long_word;
1700 memcpy(&long_word, sect + i + j, sizeof(int32_t));
1701 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1702 sys::swapByteOrder(long_word);
1703 outs() << format("%08" PRIx32, long_word) << " ";
1704 } else {
1705 for (uint32_t k = 0; i + j + k < size; k++) {
1706 uint8_t byte_word = *(sect + i + j + k);
1707 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
1708 }
1709 }
1710 }
1711 outs() << "\n";
1712 }
1713 }
1714 }
1715
1716 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
1717 StringRef DisSegName, StringRef DisSectName);
1718 static void DumpProtocolSection(MachOObjectFile *O, const char *sect,
1719 uint32_t size, uint32_t addr);
1720 #ifdef LLVM_HAVE_LIBXAR
1721 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect,
1722 uint32_t size, bool verbose,
1723 bool PrintXarHeader, bool PrintXarFileHeaders,
1724 std::string XarMemberName);
1725 #endif // defined(LLVM_HAVE_LIBXAR)
1726
DumpSectionContents(StringRef Filename,MachOObjectFile * O,bool verbose)1727 static void DumpSectionContents(StringRef Filename, MachOObjectFile *O,
1728 bool verbose) {
1729 SymbolAddressMap AddrMap;
1730 if (verbose)
1731 CreateSymbolAddressMap(O, &AddrMap);
1732
1733 for (unsigned i = 0; i < FilterSections.size(); ++i) {
1734 StringRef DumpSection = FilterSections[i];
1735 std::pair<StringRef, StringRef> DumpSegSectName;
1736 DumpSegSectName = DumpSection.split(',');
1737 StringRef DumpSegName, DumpSectName;
1738 if (!DumpSegSectName.second.empty()) {
1739 DumpSegName = DumpSegSectName.first;
1740 DumpSectName = DumpSegSectName.second;
1741 } else {
1742 DumpSegName = "";
1743 DumpSectName = DumpSegSectName.first;
1744 }
1745 for (const SectionRef &Section : O->sections()) {
1746 StringRef SectName;
1747 Expected<StringRef> SecNameOrErr = Section.getName();
1748 if (SecNameOrErr)
1749 SectName = *SecNameOrErr;
1750 else
1751 consumeError(SecNameOrErr.takeError());
1752
1753 if (!DumpSection.empty())
1754 FoundSectionSet.insert(DumpSection);
1755
1756 DataRefImpl Ref = Section.getRawDataRefImpl();
1757 StringRef SegName = O->getSectionFinalSegmentName(Ref);
1758 if ((DumpSegName.empty() || SegName == DumpSegName) &&
1759 (SectName == DumpSectName)) {
1760
1761 uint32_t section_flags;
1762 if (O->is64Bit()) {
1763 const MachO::section_64 Sec = O->getSection64(Ref);
1764 section_flags = Sec.flags;
1765
1766 } else {
1767 const MachO::section Sec = O->getSection(Ref);
1768 section_flags = Sec.flags;
1769 }
1770 uint32_t section_type = section_flags & MachO::SECTION_TYPE;
1771
1772 StringRef BytesStr =
1773 unwrapOrError(Section.getContents(), O->getFileName());
1774 const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1775 uint32_t sect_size = BytesStr.size();
1776 uint64_t sect_addr = Section.getAddress();
1777
1778 if (LeadingHeaders)
1779 outs() << "Contents of (" << SegName << "," << SectName
1780 << ") section\n";
1781
1782 if (verbose) {
1783 if ((section_flags & MachO::S_ATTR_PURE_INSTRUCTIONS) ||
1784 (section_flags & MachO::S_ATTR_SOME_INSTRUCTIONS)) {
1785 DisassembleMachO(Filename, O, SegName, SectName);
1786 continue;
1787 }
1788 if (SegName == "__TEXT" && SectName == "__info_plist") {
1789 outs() << sect;
1790 continue;
1791 }
1792 if (SegName == "__OBJC" && SectName == "__protocol") {
1793 DumpProtocolSection(O, sect, sect_size, sect_addr);
1794 continue;
1795 }
1796 #ifdef LLVM_HAVE_LIBXAR
1797 if (SegName == "__LLVM" && SectName == "__bundle") {
1798 DumpBitcodeSection(O, sect, sect_size, verbose, SymbolicOperands,
1799 ArchiveHeaders, "");
1800 continue;
1801 }
1802 #endif // defined(LLVM_HAVE_LIBXAR)
1803 switch (section_type) {
1804 case MachO::S_REGULAR:
1805 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1806 break;
1807 case MachO::S_ZEROFILL:
1808 outs() << "zerofill section and has no contents in the file\n";
1809 break;
1810 case MachO::S_CSTRING_LITERALS:
1811 DumpCstringSection(O, sect, sect_size, sect_addr, LeadingAddr);
1812 break;
1813 case MachO::S_4BYTE_LITERALS:
1814 DumpLiteral4Section(O, sect, sect_size, sect_addr, LeadingAddr);
1815 break;
1816 case MachO::S_8BYTE_LITERALS:
1817 DumpLiteral8Section(O, sect, sect_size, sect_addr, LeadingAddr);
1818 break;
1819 case MachO::S_16BYTE_LITERALS:
1820 DumpLiteral16Section(O, sect, sect_size, sect_addr, LeadingAddr);
1821 break;
1822 case MachO::S_LITERAL_POINTERS:
1823 DumpLiteralPointerSection(O, Section, sect, sect_size, sect_addr,
1824 LeadingAddr);
1825 break;
1826 case MachO::S_MOD_INIT_FUNC_POINTERS:
1827 case MachO::S_MOD_TERM_FUNC_POINTERS:
1828 DumpInitTermPointerSection(O, Section, sect, sect_size, sect_addr,
1829 &AddrMap, verbose);
1830 break;
1831 default:
1832 outs() << "Unknown section type ("
1833 << format("0x%08" PRIx32, section_type) << ")\n";
1834 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1835 break;
1836 }
1837 } else {
1838 if (section_type == MachO::S_ZEROFILL)
1839 outs() << "zerofill section and has no contents in the file\n";
1840 else
1841 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1842 }
1843 }
1844 }
1845 }
1846 }
1847
DumpInfoPlistSectionContents(StringRef Filename,MachOObjectFile * O)1848 static void DumpInfoPlistSectionContents(StringRef Filename,
1849 MachOObjectFile *O) {
1850 for (const SectionRef &Section : O->sections()) {
1851 StringRef SectName;
1852 Expected<StringRef> SecNameOrErr = Section.getName();
1853 if (SecNameOrErr)
1854 SectName = *SecNameOrErr;
1855 else
1856 consumeError(SecNameOrErr.takeError());
1857
1858 DataRefImpl Ref = Section.getRawDataRefImpl();
1859 StringRef SegName = O->getSectionFinalSegmentName(Ref);
1860 if (SegName == "__TEXT" && SectName == "__info_plist") {
1861 if (LeadingHeaders)
1862 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
1863 StringRef BytesStr =
1864 unwrapOrError(Section.getContents(), O->getFileName());
1865 const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1866 outs() << format("%.*s", BytesStr.size(), sect) << "\n";
1867 return;
1868 }
1869 }
1870 }
1871
1872 // checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file
1873 // and if it is and there is a list of architecture flags is specified then
1874 // check to make sure this Mach-O file is one of those architectures or all
1875 // architectures were specified. If not then an error is generated and this
1876 // routine returns false. Else it returns true.
checkMachOAndArchFlags(ObjectFile * O,StringRef Filename)1877 static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) {
1878 auto *MachO = dyn_cast<MachOObjectFile>(O);
1879
1880 if (!MachO || ArchAll || ArchFlags.empty())
1881 return true;
1882
1883 MachO::mach_header H;
1884 MachO::mach_header_64 H_64;
1885 Triple T;
1886 const char *McpuDefault, *ArchFlag;
1887 if (MachO->is64Bit()) {
1888 H_64 = MachO->MachOObjectFile::getHeader64();
1889 T = MachOObjectFile::getArchTriple(H_64.cputype, H_64.cpusubtype,
1890 &McpuDefault, &ArchFlag);
1891 } else {
1892 H = MachO->MachOObjectFile::getHeader();
1893 T = MachOObjectFile::getArchTriple(H.cputype, H.cpusubtype,
1894 &McpuDefault, &ArchFlag);
1895 }
1896 const std::string ArchFlagName(ArchFlag);
1897 if (!llvm::is_contained(ArchFlags, ArchFlagName)) {
1898 WithColor::error(errs(), "llvm-objdump")
1899 << Filename << ": no architecture specified.\n";
1900 return false;
1901 }
1902 return true;
1903 }
1904
1905 static void printObjcMetaData(MachOObjectFile *O, bool verbose);
1906
1907 // ProcessMachO() is passed a single opened Mach-O file, which may be an
1908 // archive member and or in a slice of a universal file. It prints the
1909 // the file name and header info and then processes it according to the
1910 // command line options.
ProcessMachO(StringRef Name,MachOObjectFile * MachOOF,StringRef ArchiveMemberName=StringRef (),StringRef ArchitectureName=StringRef ())1911 static void ProcessMachO(StringRef Name, MachOObjectFile *MachOOF,
1912 StringRef ArchiveMemberName = StringRef(),
1913 StringRef ArchitectureName = StringRef()) {
1914 // If we are doing some processing here on the Mach-O file print the header
1915 // info. And don't print it otherwise like in the case of printing the
1916 // UniversalHeaders or ArchiveHeaders.
1917 if (Disassemble || Relocations || PrivateHeaders || ExportsTrie || Rebase ||
1918 Bind || SymbolTable || LazyBind || WeakBind || IndirectSymbols ||
1919 DataInCode || FunctionStarts || LinkOptHints || DyldInfo || DylibsUsed ||
1920 DylibId || Rpaths || ObjcMetaData || (!FilterSections.empty())) {
1921 if (LeadingHeaders) {
1922 outs() << Name;
1923 if (!ArchiveMemberName.empty())
1924 outs() << '(' << ArchiveMemberName << ')';
1925 if (!ArchitectureName.empty())
1926 outs() << " (architecture " << ArchitectureName << ")";
1927 outs() << ":\n";
1928 }
1929 }
1930 // To use the report_error() form with an ArchiveName and FileName set
1931 // these up based on what is passed for Name and ArchiveMemberName.
1932 StringRef ArchiveName;
1933 StringRef FileName;
1934 if (!ArchiveMemberName.empty()) {
1935 ArchiveName = Name;
1936 FileName = ArchiveMemberName;
1937 } else {
1938 ArchiveName = StringRef();
1939 FileName = Name;
1940 }
1941
1942 // If we need the symbol table to do the operation then check it here to
1943 // produce a good error message as to where the Mach-O file comes from in
1944 // the error message.
1945 if (Disassemble || IndirectSymbols || !FilterSections.empty() || UnwindInfo)
1946 if (Error Err = MachOOF->checkSymbolTable())
1947 reportError(std::move(Err), FileName, ArchiveName, ArchitectureName);
1948
1949 if (DisassembleAll) {
1950 for (const SectionRef &Section : MachOOF->sections()) {
1951 StringRef SectName;
1952 if (Expected<StringRef> NameOrErr = Section.getName())
1953 SectName = *NameOrErr;
1954 else
1955 consumeError(NameOrErr.takeError());
1956
1957 if (SectName.equals("__text")) {
1958 DataRefImpl Ref = Section.getRawDataRefImpl();
1959 StringRef SegName = MachOOF->getSectionFinalSegmentName(Ref);
1960 DisassembleMachO(FileName, MachOOF, SegName, SectName);
1961 }
1962 }
1963 }
1964 else if (Disassemble) {
1965 if (MachOOF->getHeader().filetype == MachO::MH_KEXT_BUNDLE &&
1966 MachOOF->getHeader().cputype == MachO::CPU_TYPE_ARM64)
1967 DisassembleMachO(FileName, MachOOF, "__TEXT_EXEC", "__text");
1968 else
1969 DisassembleMachO(FileName, MachOOF, "__TEXT", "__text");
1970 }
1971 if (IndirectSymbols)
1972 PrintIndirectSymbols(MachOOF, Verbose);
1973 if (DataInCode)
1974 PrintDataInCodeTable(MachOOF, Verbose);
1975 if (FunctionStarts)
1976 PrintFunctionStarts(MachOOF);
1977 if (LinkOptHints)
1978 PrintLinkOptHints(MachOOF);
1979 if (Relocations)
1980 PrintRelocations(MachOOF, Verbose);
1981 if (SectionHeaders)
1982 printSectionHeaders(*MachOOF);
1983 if (SectionContents)
1984 printSectionContents(MachOOF);
1985 if (!FilterSections.empty())
1986 DumpSectionContents(FileName, MachOOF, Verbose);
1987 if (InfoPlist)
1988 DumpInfoPlistSectionContents(FileName, MachOOF);
1989 if (DyldInfo)
1990 PrintDyldInfo(MachOOF);
1991 if (DylibsUsed)
1992 PrintDylibs(MachOOF, false);
1993 if (DylibId)
1994 PrintDylibs(MachOOF, true);
1995 if (SymbolTable)
1996 printSymbolTable(*MachOOF, ArchiveName, ArchitectureName);
1997 if (UnwindInfo)
1998 printMachOUnwindInfo(MachOOF);
1999 if (PrivateHeaders) {
2000 printMachOFileHeader(MachOOF);
2001 printMachOLoadCommands(MachOOF);
2002 }
2003 if (FirstPrivateHeader)
2004 printMachOFileHeader(MachOOF);
2005 if (ObjcMetaData)
2006 printObjcMetaData(MachOOF, Verbose);
2007 if (ExportsTrie)
2008 printExportsTrie(MachOOF);
2009 if (Rebase)
2010 printRebaseTable(MachOOF);
2011 if (Rpaths)
2012 printRpaths(MachOOF);
2013 if (Bind)
2014 printBindTable(MachOOF);
2015 if (LazyBind)
2016 printLazyBindTable(MachOOF);
2017 if (WeakBind)
2018 printWeakBindTable(MachOOF);
2019
2020 if (DwarfDumpType != DIDT_Null) {
2021 std::unique_ptr<DIContext> DICtx = DWARFContext::create(*MachOOF);
2022 // Dump the complete DWARF structure.
2023 DIDumpOptions DumpOpts;
2024 DumpOpts.DumpType = DwarfDumpType;
2025 DICtx->dump(outs(), DumpOpts);
2026 }
2027 }
2028
2029 // printUnknownCPUType() helps print_fat_headers for unknown CPU's.
printUnknownCPUType(uint32_t cputype,uint32_t cpusubtype)2030 static void printUnknownCPUType(uint32_t cputype, uint32_t cpusubtype) {
2031 outs() << " cputype (" << cputype << ")\n";
2032 outs() << " cpusubtype (" << cpusubtype << ")\n";
2033 }
2034
2035 // printCPUType() helps print_fat_headers by printing the cputype and
2036 // pusubtype (symbolically for the one's it knows about).
printCPUType(uint32_t cputype,uint32_t cpusubtype)2037 static void printCPUType(uint32_t cputype, uint32_t cpusubtype) {
2038 switch (cputype) {
2039 case MachO::CPU_TYPE_I386:
2040 switch (cpusubtype) {
2041 case MachO::CPU_SUBTYPE_I386_ALL:
2042 outs() << " cputype CPU_TYPE_I386\n";
2043 outs() << " cpusubtype CPU_SUBTYPE_I386_ALL\n";
2044 break;
2045 default:
2046 printUnknownCPUType(cputype, cpusubtype);
2047 break;
2048 }
2049 break;
2050 case MachO::CPU_TYPE_X86_64:
2051 switch (cpusubtype) {
2052 case MachO::CPU_SUBTYPE_X86_64_ALL:
2053 outs() << " cputype CPU_TYPE_X86_64\n";
2054 outs() << " cpusubtype CPU_SUBTYPE_X86_64_ALL\n";
2055 break;
2056 case MachO::CPU_SUBTYPE_X86_64_H:
2057 outs() << " cputype CPU_TYPE_X86_64\n";
2058 outs() << " cpusubtype CPU_SUBTYPE_X86_64_H\n";
2059 break;
2060 default:
2061 printUnknownCPUType(cputype, cpusubtype);
2062 break;
2063 }
2064 break;
2065 case MachO::CPU_TYPE_ARM:
2066 switch (cpusubtype) {
2067 case MachO::CPU_SUBTYPE_ARM_ALL:
2068 outs() << " cputype CPU_TYPE_ARM\n";
2069 outs() << " cpusubtype CPU_SUBTYPE_ARM_ALL\n";
2070 break;
2071 case MachO::CPU_SUBTYPE_ARM_V4T:
2072 outs() << " cputype CPU_TYPE_ARM\n";
2073 outs() << " cpusubtype CPU_SUBTYPE_ARM_V4T\n";
2074 break;
2075 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
2076 outs() << " cputype CPU_TYPE_ARM\n";
2077 outs() << " cpusubtype CPU_SUBTYPE_ARM_V5TEJ\n";
2078 break;
2079 case MachO::CPU_SUBTYPE_ARM_XSCALE:
2080 outs() << " cputype CPU_TYPE_ARM\n";
2081 outs() << " cpusubtype CPU_SUBTYPE_ARM_XSCALE\n";
2082 break;
2083 case MachO::CPU_SUBTYPE_ARM_V6:
2084 outs() << " cputype CPU_TYPE_ARM\n";
2085 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6\n";
2086 break;
2087 case MachO::CPU_SUBTYPE_ARM_V6M:
2088 outs() << " cputype CPU_TYPE_ARM\n";
2089 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6M\n";
2090 break;
2091 case MachO::CPU_SUBTYPE_ARM_V7:
2092 outs() << " cputype CPU_TYPE_ARM\n";
2093 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7\n";
2094 break;
2095 case MachO::CPU_SUBTYPE_ARM_V7EM:
2096 outs() << " cputype CPU_TYPE_ARM\n";
2097 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7EM\n";
2098 break;
2099 case MachO::CPU_SUBTYPE_ARM_V7K:
2100 outs() << " cputype CPU_TYPE_ARM\n";
2101 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7K\n";
2102 break;
2103 case MachO::CPU_SUBTYPE_ARM_V7M:
2104 outs() << " cputype CPU_TYPE_ARM\n";
2105 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7M\n";
2106 break;
2107 case MachO::CPU_SUBTYPE_ARM_V7S:
2108 outs() << " cputype CPU_TYPE_ARM\n";
2109 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7S\n";
2110 break;
2111 default:
2112 printUnknownCPUType(cputype, cpusubtype);
2113 break;
2114 }
2115 break;
2116 case MachO::CPU_TYPE_ARM64:
2117 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2118 case MachO::CPU_SUBTYPE_ARM64_ALL:
2119 outs() << " cputype CPU_TYPE_ARM64\n";
2120 outs() << " cpusubtype CPU_SUBTYPE_ARM64_ALL\n";
2121 break;
2122 case MachO::CPU_SUBTYPE_ARM64_V8:
2123 outs() << " cputype CPU_TYPE_ARM64\n";
2124 outs() << " cpusubtype CPU_SUBTYPE_ARM64_V8\n";
2125 break;
2126 case MachO::CPU_SUBTYPE_ARM64E:
2127 outs() << " cputype CPU_TYPE_ARM64\n";
2128 outs() << " cpusubtype CPU_SUBTYPE_ARM64E\n";
2129 break;
2130 default:
2131 printUnknownCPUType(cputype, cpusubtype);
2132 break;
2133 }
2134 break;
2135 case MachO::CPU_TYPE_ARM64_32:
2136 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2137 case MachO::CPU_SUBTYPE_ARM64_32_V8:
2138 outs() << " cputype CPU_TYPE_ARM64_32\n";
2139 outs() << " cpusubtype CPU_SUBTYPE_ARM64_32_V8\n";
2140 break;
2141 default:
2142 printUnknownCPUType(cputype, cpusubtype);
2143 break;
2144 }
2145 break;
2146 default:
2147 printUnknownCPUType(cputype, cpusubtype);
2148 break;
2149 }
2150 }
2151
printMachOUniversalHeaders(const object::MachOUniversalBinary * UB,bool verbose)2152 static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB,
2153 bool verbose) {
2154 outs() << "Fat headers\n";
2155 if (verbose) {
2156 if (UB->getMagic() == MachO::FAT_MAGIC)
2157 outs() << "fat_magic FAT_MAGIC\n";
2158 else // UB->getMagic() == MachO::FAT_MAGIC_64
2159 outs() << "fat_magic FAT_MAGIC_64\n";
2160 } else
2161 outs() << "fat_magic " << format("0x%" PRIx32, MachO::FAT_MAGIC) << "\n";
2162
2163 uint32_t nfat_arch = UB->getNumberOfObjects();
2164 StringRef Buf = UB->getData();
2165 uint64_t size = Buf.size();
2166 uint64_t big_size = sizeof(struct MachO::fat_header) +
2167 nfat_arch * sizeof(struct MachO::fat_arch);
2168 outs() << "nfat_arch " << UB->getNumberOfObjects();
2169 if (nfat_arch == 0)
2170 outs() << " (malformed, contains zero architecture types)\n";
2171 else if (big_size > size)
2172 outs() << " (malformed, architectures past end of file)\n";
2173 else
2174 outs() << "\n";
2175
2176 for (uint32_t i = 0; i < nfat_arch; ++i) {
2177 MachOUniversalBinary::ObjectForArch OFA(UB, i);
2178 uint32_t cputype = OFA.getCPUType();
2179 uint32_t cpusubtype = OFA.getCPUSubType();
2180 outs() << "architecture ";
2181 for (uint32_t j = 0; i != 0 && j <= i - 1; j++) {
2182 MachOUniversalBinary::ObjectForArch other_OFA(UB, j);
2183 uint32_t other_cputype = other_OFA.getCPUType();
2184 uint32_t other_cpusubtype = other_OFA.getCPUSubType();
2185 if (cputype != 0 && cpusubtype != 0 && cputype == other_cputype &&
2186 (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) ==
2187 (other_cpusubtype & ~MachO::CPU_SUBTYPE_MASK)) {
2188 outs() << "(illegal duplicate architecture) ";
2189 break;
2190 }
2191 }
2192 if (verbose) {
2193 outs() << OFA.getArchFlagName() << "\n";
2194 printCPUType(cputype, cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2195 } else {
2196 outs() << i << "\n";
2197 outs() << " cputype " << cputype << "\n";
2198 outs() << " cpusubtype " << (cpusubtype & ~MachO::CPU_SUBTYPE_MASK)
2199 << "\n";
2200 }
2201 if (verbose &&
2202 (cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64)
2203 outs() << " capabilities CPU_SUBTYPE_LIB64\n";
2204 else
2205 outs() << " capabilities "
2206 << format("0x%" PRIx32,
2207 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24) << "\n";
2208 outs() << " offset " << OFA.getOffset();
2209 if (OFA.getOffset() > size)
2210 outs() << " (past end of file)";
2211 if (OFA.getOffset() % (1ull << OFA.getAlign()) != 0)
2212 outs() << " (not aligned on it's alignment (2^" << OFA.getAlign() << ")";
2213 outs() << "\n";
2214 outs() << " size " << OFA.getSize();
2215 big_size = OFA.getOffset() + OFA.getSize();
2216 if (big_size > size)
2217 outs() << " (past end of file)";
2218 outs() << "\n";
2219 outs() << " align 2^" << OFA.getAlign() << " (" << (1 << OFA.getAlign())
2220 << ")\n";
2221 }
2222 }
2223
printArchiveChild(StringRef Filename,const Archive::Child & C,size_t ChildIndex,bool verbose,bool print_offset,StringRef ArchitectureName=StringRef ())2224 static void printArchiveChild(StringRef Filename, const Archive::Child &C,
2225 size_t ChildIndex, bool verbose,
2226 bool print_offset,
2227 StringRef ArchitectureName = StringRef()) {
2228 if (print_offset)
2229 outs() << C.getChildOffset() << "\t";
2230 sys::fs::perms Mode =
2231 unwrapOrError(C.getAccessMode(), getFileNameForError(C, ChildIndex),
2232 Filename, ArchitectureName);
2233 if (verbose) {
2234 // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG.
2235 // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG.
2236 outs() << "-";
2237 outs() << ((Mode & sys::fs::owner_read) ? "r" : "-");
2238 outs() << ((Mode & sys::fs::owner_write) ? "w" : "-");
2239 outs() << ((Mode & sys::fs::owner_exe) ? "x" : "-");
2240 outs() << ((Mode & sys::fs::group_read) ? "r" : "-");
2241 outs() << ((Mode & sys::fs::group_write) ? "w" : "-");
2242 outs() << ((Mode & sys::fs::group_exe) ? "x" : "-");
2243 outs() << ((Mode & sys::fs::others_read) ? "r" : "-");
2244 outs() << ((Mode & sys::fs::others_write) ? "w" : "-");
2245 outs() << ((Mode & sys::fs::others_exe) ? "x" : "-");
2246 } else {
2247 outs() << format("0%o ", Mode);
2248 }
2249
2250 outs() << format("%3d/%-3d %5" PRId64 " ",
2251 unwrapOrError(C.getUID(), getFileNameForError(C, ChildIndex),
2252 Filename, ArchitectureName),
2253 unwrapOrError(C.getGID(), getFileNameForError(C, ChildIndex),
2254 Filename, ArchitectureName),
2255 unwrapOrError(C.getRawSize(),
2256 getFileNameForError(C, ChildIndex), Filename,
2257 ArchitectureName));
2258
2259 StringRef RawLastModified = C.getRawLastModified();
2260 if (verbose) {
2261 unsigned Seconds;
2262 if (RawLastModified.getAsInteger(10, Seconds))
2263 outs() << "(date: \"" << RawLastModified
2264 << "\" contains non-decimal chars) ";
2265 else {
2266 // Since cime(3) returns a 26 character string of the form:
2267 // "Sun Sep 16 01:03:52 1973\n\0"
2268 // just print 24 characters.
2269 time_t t = Seconds;
2270 outs() << format("%.24s ", ctime(&t));
2271 }
2272 } else {
2273 outs() << RawLastModified << " ";
2274 }
2275
2276 if (verbose) {
2277 Expected<StringRef> NameOrErr = C.getName();
2278 if (!NameOrErr) {
2279 consumeError(NameOrErr.takeError());
2280 outs() << unwrapOrError(C.getRawName(),
2281 getFileNameForError(C, ChildIndex), Filename,
2282 ArchitectureName)
2283 << "\n";
2284 } else {
2285 StringRef Name = NameOrErr.get();
2286 outs() << Name << "\n";
2287 }
2288 } else {
2289 outs() << unwrapOrError(C.getRawName(), getFileNameForError(C, ChildIndex),
2290 Filename, ArchitectureName)
2291 << "\n";
2292 }
2293 }
2294
printArchiveHeaders(StringRef Filename,Archive * A,bool verbose,bool print_offset,StringRef ArchitectureName=StringRef ())2295 static void printArchiveHeaders(StringRef Filename, Archive *A, bool verbose,
2296 bool print_offset,
2297 StringRef ArchitectureName = StringRef()) {
2298 Error Err = Error::success();
2299 size_t I = 0;
2300 for (const auto &C : A->children(Err, false))
2301 printArchiveChild(Filename, C, I++, verbose, print_offset,
2302 ArchitectureName);
2303
2304 if (Err)
2305 reportError(std::move(Err), Filename, "", ArchitectureName);
2306 }
2307
ValidateArchFlags()2308 static bool ValidateArchFlags() {
2309 // Check for -arch all and verifiy the -arch flags are valid.
2310 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
2311 if (ArchFlags[i] == "all") {
2312 ArchAll = true;
2313 } else {
2314 if (!MachOObjectFile::isValidArch(ArchFlags[i])) {
2315 WithColor::error(errs(), "llvm-objdump")
2316 << "unknown architecture named '" + ArchFlags[i] +
2317 "'for the -arch option\n";
2318 return false;
2319 }
2320 }
2321 }
2322 return true;
2323 }
2324
2325 // ParseInputMachO() parses the named Mach-O file in Filename and handles the
2326 // -arch flags selecting just those slices as specified by them and also parses
2327 // archive files. Then for each individual Mach-O file ProcessMachO() is
2328 // called to process the file based on the command line options.
parseInputMachO(StringRef Filename)2329 void objdump::parseInputMachO(StringRef Filename) {
2330 if (!ValidateArchFlags())
2331 return;
2332
2333 // Attempt to open the binary.
2334 Expected<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename);
2335 if (!BinaryOrErr) {
2336 if (Error E = isNotObjectErrorInvalidFileType(BinaryOrErr.takeError()))
2337 reportError(std::move(E), Filename);
2338 else
2339 outs() << Filename << ": is not an object file\n";
2340 return;
2341 }
2342 Binary &Bin = *BinaryOrErr.get().getBinary();
2343
2344 if (Archive *A = dyn_cast<Archive>(&Bin)) {
2345 outs() << "Archive : " << Filename << "\n";
2346 if (ArchiveHeaders)
2347 printArchiveHeaders(Filename, A, Verbose, ArchiveMemberOffsets);
2348
2349 Error Err = Error::success();
2350 unsigned I = -1;
2351 for (auto &C : A->children(Err)) {
2352 ++I;
2353 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2354 if (!ChildOrErr) {
2355 if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2356 reportError(std::move(E), getFileNameForError(C, I), Filename);
2357 continue;
2358 }
2359 if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
2360 if (!checkMachOAndArchFlags(O, Filename))
2361 return;
2362 ProcessMachO(Filename, O, O->getFileName());
2363 }
2364 }
2365 if (Err)
2366 reportError(std::move(Err), Filename);
2367 return;
2368 }
2369 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) {
2370 parseInputMachO(UB);
2371 return;
2372 }
2373 if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) {
2374 if (!checkMachOAndArchFlags(O, Filename))
2375 return;
2376 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O))
2377 ProcessMachO(Filename, MachOOF);
2378 else
2379 WithColor::error(errs(), "llvm-objdump")
2380 << Filename << "': "
2381 << "object is not a Mach-O file type.\n";
2382 return;
2383 }
2384 llvm_unreachable("Input object can't be invalid at this point");
2385 }
2386
parseInputMachO(MachOUniversalBinary * UB)2387 void objdump::parseInputMachO(MachOUniversalBinary *UB) {
2388 if (!ValidateArchFlags())
2389 return;
2390
2391 auto Filename = UB->getFileName();
2392
2393 if (UniversalHeaders)
2394 printMachOUniversalHeaders(UB, Verbose);
2395
2396 // If we have a list of architecture flags specified dump only those.
2397 if (!ArchAll && !ArchFlags.empty()) {
2398 // Look for a slice in the universal binary that matches each ArchFlag.
2399 bool ArchFound;
2400 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
2401 ArchFound = false;
2402 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2403 E = UB->end_objects();
2404 I != E; ++I) {
2405 if (ArchFlags[i] == I->getArchFlagName()) {
2406 ArchFound = true;
2407 Expected<std::unique_ptr<ObjectFile>> ObjOrErr =
2408 I->getAsObjectFile();
2409 std::string ArchitectureName;
2410 if (ArchFlags.size() > 1)
2411 ArchitectureName = I->getArchFlagName();
2412 if (ObjOrErr) {
2413 ObjectFile &O = *ObjOrErr.get();
2414 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
2415 ProcessMachO(Filename, MachOOF, "", ArchitectureName);
2416 } else if (Error E = isNotObjectErrorInvalidFileType(
2417 ObjOrErr.takeError())) {
2418 reportError(std::move(E), "", Filename, ArchitectureName);
2419 continue;
2420 } else if (Expected<std::unique_ptr<Archive>> AOrErr =
2421 I->getAsArchive()) {
2422 std::unique_ptr<Archive> &A = *AOrErr;
2423 outs() << "Archive : " << Filename;
2424 if (!ArchitectureName.empty())
2425 outs() << " (architecture " << ArchitectureName << ")";
2426 outs() << "\n";
2427 if (ArchiveHeaders)
2428 printArchiveHeaders(Filename, A.get(), Verbose,
2429 ArchiveMemberOffsets, ArchitectureName);
2430 Error Err = Error::success();
2431 unsigned I = -1;
2432 for (auto &C : A->children(Err)) {
2433 ++I;
2434 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2435 if (!ChildOrErr) {
2436 if (Error E =
2437 isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2438 reportError(std::move(E), getFileNameForError(C, I), Filename,
2439 ArchitectureName);
2440 continue;
2441 }
2442 if (MachOObjectFile *O =
2443 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
2444 ProcessMachO(Filename, O, O->getFileName(), ArchitectureName);
2445 }
2446 if (Err)
2447 reportError(std::move(Err), Filename);
2448 } else {
2449 consumeError(AOrErr.takeError());
2450 reportError(Filename,
2451 "Mach-O universal file for architecture " +
2452 StringRef(I->getArchFlagName()) +
2453 " is not a Mach-O file or an archive file");
2454 }
2455 }
2456 }
2457 if (!ArchFound) {
2458 WithColor::error(errs(), "llvm-objdump")
2459 << "file: " + Filename + " does not contain "
2460 << "architecture: " + ArchFlags[i] + "\n";
2461 return;
2462 }
2463 }
2464 return;
2465 }
2466 // No architecture flags were specified so if this contains a slice that
2467 // matches the host architecture dump only that.
2468 if (!ArchAll) {
2469 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2470 E = UB->end_objects();
2471 I != E; ++I) {
2472 if (MachOObjectFile::getHostArch().getArchName() ==
2473 I->getArchFlagName()) {
2474 Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
2475 std::string ArchiveName;
2476 ArchiveName.clear();
2477 if (ObjOrErr) {
2478 ObjectFile &O = *ObjOrErr.get();
2479 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
2480 ProcessMachO(Filename, MachOOF);
2481 } else if (Error E =
2482 isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) {
2483 reportError(std::move(E), Filename);
2484 } else if (Expected<std::unique_ptr<Archive>> AOrErr =
2485 I->getAsArchive()) {
2486 std::unique_ptr<Archive> &A = *AOrErr;
2487 outs() << "Archive : " << Filename << "\n";
2488 if (ArchiveHeaders)
2489 printArchiveHeaders(Filename, A.get(), Verbose,
2490 ArchiveMemberOffsets);
2491 Error Err = Error::success();
2492 unsigned I = -1;
2493 for (auto &C : A->children(Err)) {
2494 ++I;
2495 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2496 if (!ChildOrErr) {
2497 if (Error E =
2498 isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2499 reportError(std::move(E), getFileNameForError(C, I), Filename);
2500 continue;
2501 }
2502 if (MachOObjectFile *O =
2503 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
2504 ProcessMachO(Filename, O, O->getFileName());
2505 }
2506 if (Err)
2507 reportError(std::move(Err), Filename);
2508 } else {
2509 consumeError(AOrErr.takeError());
2510 reportError(Filename, "Mach-O universal file for architecture " +
2511 StringRef(I->getArchFlagName()) +
2512 " is not a Mach-O file or an archive file");
2513 }
2514 return;
2515 }
2516 }
2517 }
2518 // Either all architectures have been specified or none have been specified
2519 // and this does not contain the host architecture so dump all the slices.
2520 bool moreThanOneArch = UB->getNumberOfObjects() > 1;
2521 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2522 E = UB->end_objects();
2523 I != E; ++I) {
2524 Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
2525 std::string ArchitectureName;
2526 if (moreThanOneArch)
2527 ArchitectureName = I->getArchFlagName();
2528 if (ObjOrErr) {
2529 ObjectFile &Obj = *ObjOrErr.get();
2530 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj))
2531 ProcessMachO(Filename, MachOOF, "", ArchitectureName);
2532 } else if (Error E =
2533 isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) {
2534 reportError(std::move(E), Filename, "", ArchitectureName);
2535 } else if (Expected<std::unique_ptr<Archive>> AOrErr = I->getAsArchive()) {
2536 std::unique_ptr<Archive> &A = *AOrErr;
2537 outs() << "Archive : " << Filename;
2538 if (!ArchitectureName.empty())
2539 outs() << " (architecture " << ArchitectureName << ")";
2540 outs() << "\n";
2541 if (ArchiveHeaders)
2542 printArchiveHeaders(Filename, A.get(), Verbose, ArchiveMemberOffsets,
2543 ArchitectureName);
2544 Error Err = Error::success();
2545 unsigned I = -1;
2546 for (auto &C : A->children(Err)) {
2547 ++I;
2548 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2549 if (!ChildOrErr) {
2550 if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2551 reportError(std::move(E), getFileNameForError(C, I), Filename,
2552 ArchitectureName);
2553 continue;
2554 }
2555 if (MachOObjectFile *O =
2556 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
2557 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O))
2558 ProcessMachO(Filename, MachOOF, MachOOF->getFileName(),
2559 ArchitectureName);
2560 }
2561 }
2562 if (Err)
2563 reportError(std::move(Err), Filename);
2564 } else {
2565 consumeError(AOrErr.takeError());
2566 reportError(Filename, "Mach-O universal file for architecture " +
2567 StringRef(I->getArchFlagName()) +
2568 " is not a Mach-O file or an archive file");
2569 }
2570 }
2571 }
2572
2573 namespace {
2574 // The block of info used by the Symbolizer call backs.
2575 struct DisassembleInfo {
DisassembleInfo__anon4a9557200611::DisassembleInfo2576 DisassembleInfo(MachOObjectFile *O, SymbolAddressMap *AddrMap,
2577 std::vector<SectionRef> *Sections, bool verbose)
2578 : verbose(verbose), O(O), AddrMap(AddrMap), Sections(Sections) {}
2579 bool verbose;
2580 MachOObjectFile *O;
2581 SectionRef S;
2582 SymbolAddressMap *AddrMap;
2583 std::vector<SectionRef> *Sections;
2584 const char *class_name = nullptr;
2585 const char *selector_name = nullptr;
2586 std::unique_ptr<char[]> method = nullptr;
2587 char *demangled_name = nullptr;
2588 uint64_t adrp_addr = 0;
2589 uint32_t adrp_inst = 0;
2590 std::unique_ptr<SymbolAddressMap> bindtable;
2591 uint32_t depth = 0;
2592 };
2593 } // namespace
2594
2595 // SymbolizerGetOpInfo() is the operand information call back function.
2596 // This is called to get the symbolic information for operand(s) of an
2597 // instruction when it is being done. This routine does this from
2598 // the relocation information, symbol table, etc. That block of information
2599 // is a pointer to the struct DisassembleInfo that was passed when the
2600 // disassembler context was created and passed to back to here when
2601 // called back by the disassembler for instruction operands that could have
2602 // relocation information. The address of the instruction containing operand is
2603 // at the Pc parameter. The immediate value the operand has is passed in
2604 // op_info->Value and is at Offset past the start of the instruction and has a
2605 // byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the
2606 // LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol
2607 // names and addends of the symbolic expression to add for the operand. The
2608 // value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic
2609 // information is returned then this function returns 1 else it returns 0.
SymbolizerGetOpInfo(void * DisInfo,uint64_t Pc,uint64_t Offset,uint64_t OpSize,uint64_t InstSize,int TagType,void * TagBuf)2610 static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset,
2611 uint64_t OpSize, uint64_t InstSize, int TagType,
2612 void *TagBuf) {
2613 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
2614 struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf;
2615 uint64_t value = op_info->Value;
2616
2617 // Make sure all fields returned are zero if we don't set them.
2618 memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1));
2619 op_info->Value = value;
2620
2621 // If the TagType is not the value 1 which it code knows about or if no
2622 // verbose symbolic information is wanted then just return 0, indicating no
2623 // information is being returned.
2624 if (TagType != 1 || !info->verbose)
2625 return 0;
2626
2627 unsigned int Arch = info->O->getArch();
2628 if (Arch == Triple::x86) {
2629 if (OpSize != 1 && OpSize != 2 && OpSize != 4 && OpSize != 0)
2630 return 0;
2631 if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2632 // TODO:
2633 // Search the external relocation entries of a fully linked image
2634 // (if any) for an entry that matches this segment offset.
2635 // uint32_t seg_offset = (Pc + Offset);
2636 return 0;
2637 }
2638 // In MH_OBJECT filetypes search the section's relocation entries (if any)
2639 // for an entry for this section offset.
2640 uint32_t sect_addr = info->S.getAddress();
2641 uint32_t sect_offset = (Pc + Offset) - sect_addr;
2642 bool reloc_found = false;
2643 DataRefImpl Rel;
2644 MachO::any_relocation_info RE;
2645 bool isExtern = false;
2646 SymbolRef Symbol;
2647 bool r_scattered = false;
2648 uint32_t r_value, pair_r_value, r_type;
2649 for (const RelocationRef &Reloc : info->S.relocations()) {
2650 uint64_t RelocOffset = Reloc.getOffset();
2651 if (RelocOffset == sect_offset) {
2652 Rel = Reloc.getRawDataRefImpl();
2653 RE = info->O->getRelocation(Rel);
2654 r_type = info->O->getAnyRelocationType(RE);
2655 r_scattered = info->O->isRelocationScattered(RE);
2656 if (r_scattered) {
2657 r_value = info->O->getScatteredRelocationValue(RE);
2658 if (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
2659 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) {
2660 DataRefImpl RelNext = Rel;
2661 info->O->moveRelocationNext(RelNext);
2662 MachO::any_relocation_info RENext;
2663 RENext = info->O->getRelocation(RelNext);
2664 if (info->O->isRelocationScattered(RENext))
2665 pair_r_value = info->O->getScatteredRelocationValue(RENext);
2666 else
2667 return 0;
2668 }
2669 } else {
2670 isExtern = info->O->getPlainRelocationExternal(RE);
2671 if (isExtern) {
2672 symbol_iterator RelocSym = Reloc.getSymbol();
2673 Symbol = *RelocSym;
2674 }
2675 }
2676 reloc_found = true;
2677 break;
2678 }
2679 }
2680 if (reloc_found && isExtern) {
2681 op_info->AddSymbol.Present = 1;
2682 op_info->AddSymbol.Name =
2683 unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2684 // For i386 extern relocation entries the value in the instruction is
2685 // the offset from the symbol, and value is already set in op_info->Value.
2686 return 1;
2687 }
2688 if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
2689 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) {
2690 const char *add = GuessSymbolName(r_value, info->AddrMap);
2691 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
2692 uint32_t offset = value - (r_value - pair_r_value);
2693 op_info->AddSymbol.Present = 1;
2694 if (add != nullptr)
2695 op_info->AddSymbol.Name = add;
2696 else
2697 op_info->AddSymbol.Value = r_value;
2698 op_info->SubtractSymbol.Present = 1;
2699 if (sub != nullptr)
2700 op_info->SubtractSymbol.Name = sub;
2701 else
2702 op_info->SubtractSymbol.Value = pair_r_value;
2703 op_info->Value = offset;
2704 return 1;
2705 }
2706 return 0;
2707 }
2708 if (Arch == Triple::x86_64) {
2709 if (OpSize != 1 && OpSize != 2 && OpSize != 4 && OpSize != 0)
2710 return 0;
2711 // For non MH_OBJECT types, like MH_KEXT_BUNDLE, Search the external
2712 // relocation entries of a linked image (if any) for an entry that matches
2713 // this segment offset.
2714 if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2715 uint64_t seg_offset = Pc + Offset;
2716 bool reloc_found = false;
2717 DataRefImpl Rel;
2718 MachO::any_relocation_info RE;
2719 bool isExtern = false;
2720 SymbolRef Symbol;
2721 for (const RelocationRef &Reloc : info->O->external_relocations()) {
2722 uint64_t RelocOffset = Reloc.getOffset();
2723 if (RelocOffset == seg_offset) {
2724 Rel = Reloc.getRawDataRefImpl();
2725 RE = info->O->getRelocation(Rel);
2726 // external relocation entries should always be external.
2727 isExtern = info->O->getPlainRelocationExternal(RE);
2728 if (isExtern) {
2729 symbol_iterator RelocSym = Reloc.getSymbol();
2730 Symbol = *RelocSym;
2731 }
2732 reloc_found = true;
2733 break;
2734 }
2735 }
2736 if (reloc_found && isExtern) {
2737 // The Value passed in will be adjusted by the Pc if the instruction
2738 // adds the Pc. But for x86_64 external relocation entries the Value
2739 // is the offset from the external symbol.
2740 if (info->O->getAnyRelocationPCRel(RE))
2741 op_info->Value -= Pc + InstSize;
2742 const char *name =
2743 unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2744 op_info->AddSymbol.Present = 1;
2745 op_info->AddSymbol.Name = name;
2746 return 1;
2747 }
2748 return 0;
2749 }
2750 // In MH_OBJECT filetypes search the section's relocation entries (if any)
2751 // for an entry for this section offset.
2752 uint64_t sect_addr = info->S.getAddress();
2753 uint64_t sect_offset = (Pc + Offset) - sect_addr;
2754 bool reloc_found = false;
2755 DataRefImpl Rel;
2756 MachO::any_relocation_info RE;
2757 bool isExtern = false;
2758 SymbolRef Symbol;
2759 for (const RelocationRef &Reloc : info->S.relocations()) {
2760 uint64_t RelocOffset = Reloc.getOffset();
2761 if (RelocOffset == sect_offset) {
2762 Rel = Reloc.getRawDataRefImpl();
2763 RE = info->O->getRelocation(Rel);
2764 // NOTE: Scattered relocations don't exist on x86_64.
2765 isExtern = info->O->getPlainRelocationExternal(RE);
2766 if (isExtern) {
2767 symbol_iterator RelocSym = Reloc.getSymbol();
2768 Symbol = *RelocSym;
2769 }
2770 reloc_found = true;
2771 break;
2772 }
2773 }
2774 if (reloc_found && isExtern) {
2775 // The Value passed in will be adjusted by the Pc if the instruction
2776 // adds the Pc. But for x86_64 external relocation entries the Value
2777 // is the offset from the external symbol.
2778 if (info->O->getAnyRelocationPCRel(RE))
2779 op_info->Value -= Pc + InstSize;
2780 const char *name =
2781 unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2782 unsigned Type = info->O->getAnyRelocationType(RE);
2783 if (Type == MachO::X86_64_RELOC_SUBTRACTOR) {
2784 DataRefImpl RelNext = Rel;
2785 info->O->moveRelocationNext(RelNext);
2786 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2787 unsigned TypeNext = info->O->getAnyRelocationType(RENext);
2788 bool isExternNext = info->O->getPlainRelocationExternal(RENext);
2789 unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext);
2790 if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) {
2791 op_info->SubtractSymbol.Present = 1;
2792 op_info->SubtractSymbol.Name = name;
2793 symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum);
2794 Symbol = *RelocSymNext;
2795 name = unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2796 }
2797 }
2798 // TODO: add the VariantKinds to op_info->VariantKind for relocation types
2799 // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT.
2800 op_info->AddSymbol.Present = 1;
2801 op_info->AddSymbol.Name = name;
2802 return 1;
2803 }
2804 return 0;
2805 }
2806 if (Arch == Triple::arm) {
2807 if (Offset != 0 || (InstSize != 4 && InstSize != 2))
2808 return 0;
2809 if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2810 // TODO:
2811 // Search the external relocation entries of a fully linked image
2812 // (if any) for an entry that matches this segment offset.
2813 // uint32_t seg_offset = (Pc + Offset);
2814 return 0;
2815 }
2816 // In MH_OBJECT filetypes search the section's relocation entries (if any)
2817 // for an entry for this section offset.
2818 uint32_t sect_addr = info->S.getAddress();
2819 uint32_t sect_offset = (Pc + Offset) - sect_addr;
2820 DataRefImpl Rel;
2821 MachO::any_relocation_info RE;
2822 bool isExtern = false;
2823 SymbolRef Symbol;
2824 bool r_scattered = false;
2825 uint32_t r_value, pair_r_value, r_type, r_length, other_half;
2826 auto Reloc =
2827 find_if(info->S.relocations(), [&](const RelocationRef &Reloc) {
2828 uint64_t RelocOffset = Reloc.getOffset();
2829 return RelocOffset == sect_offset;
2830 });
2831
2832 if (Reloc == info->S.relocations().end())
2833 return 0;
2834
2835 Rel = Reloc->getRawDataRefImpl();
2836 RE = info->O->getRelocation(Rel);
2837 r_length = info->O->getAnyRelocationLength(RE);
2838 r_scattered = info->O->isRelocationScattered(RE);
2839 if (r_scattered) {
2840 r_value = info->O->getScatteredRelocationValue(RE);
2841 r_type = info->O->getScatteredRelocationType(RE);
2842 } else {
2843 r_type = info->O->getAnyRelocationType(RE);
2844 isExtern = info->O->getPlainRelocationExternal(RE);
2845 if (isExtern) {
2846 symbol_iterator RelocSym = Reloc->getSymbol();
2847 Symbol = *RelocSym;
2848 }
2849 }
2850 if (r_type == MachO::ARM_RELOC_HALF ||
2851 r_type == MachO::ARM_RELOC_SECTDIFF ||
2852 r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
2853 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2854 DataRefImpl RelNext = Rel;
2855 info->O->moveRelocationNext(RelNext);
2856 MachO::any_relocation_info RENext;
2857 RENext = info->O->getRelocation(RelNext);
2858 other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff;
2859 if (info->O->isRelocationScattered(RENext))
2860 pair_r_value = info->O->getScatteredRelocationValue(RENext);
2861 }
2862
2863 if (isExtern) {
2864 const char *name =
2865 unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2866 op_info->AddSymbol.Present = 1;
2867 op_info->AddSymbol.Name = name;
2868 switch (r_type) {
2869 case MachO::ARM_RELOC_HALF:
2870 if ((r_length & 0x1) == 1) {
2871 op_info->Value = value << 16 | other_half;
2872 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2873 } else {
2874 op_info->Value = other_half << 16 | value;
2875 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2876 }
2877 break;
2878 default:
2879 break;
2880 }
2881 return 1;
2882 }
2883 // If we have a branch that is not an external relocation entry then
2884 // return 0 so the code in tryAddingSymbolicOperand() can use the
2885 // SymbolLookUp call back with the branch target address to look up the
2886 // symbol and possibility add an annotation for a symbol stub.
2887 if (isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 ||
2888 r_type == MachO::ARM_THUMB_RELOC_BR22))
2889 return 0;
2890
2891 uint32_t offset = 0;
2892 if (r_type == MachO::ARM_RELOC_HALF ||
2893 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2894 if ((r_length & 0x1) == 1)
2895 value = value << 16 | other_half;
2896 else
2897 value = other_half << 16 | value;
2898 }
2899 if (r_scattered && (r_type != MachO::ARM_RELOC_HALF &&
2900 r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) {
2901 offset = value - r_value;
2902 value = r_value;
2903 }
2904
2905 if (r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2906 if ((r_length & 0x1) == 1)
2907 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2908 else
2909 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2910 const char *add = GuessSymbolName(r_value, info->AddrMap);
2911 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
2912 int32_t offset = value - (r_value - pair_r_value);
2913 op_info->AddSymbol.Present = 1;
2914 if (add != nullptr)
2915 op_info->AddSymbol.Name = add;
2916 else
2917 op_info->AddSymbol.Value = r_value;
2918 op_info->SubtractSymbol.Present = 1;
2919 if (sub != nullptr)
2920 op_info->SubtractSymbol.Name = sub;
2921 else
2922 op_info->SubtractSymbol.Value = pair_r_value;
2923 op_info->Value = offset;
2924 return 1;
2925 }
2926
2927 op_info->AddSymbol.Present = 1;
2928 op_info->Value = offset;
2929 if (r_type == MachO::ARM_RELOC_HALF) {
2930 if ((r_length & 0x1) == 1)
2931 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2932 else
2933 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2934 }
2935 const char *add = GuessSymbolName(value, info->AddrMap);
2936 if (add != nullptr) {
2937 op_info->AddSymbol.Name = add;
2938 return 1;
2939 }
2940 op_info->AddSymbol.Value = value;
2941 return 1;
2942 }
2943 if (Arch == Triple::aarch64) {
2944 if (Offset != 0 || InstSize != 4)
2945 return 0;
2946 if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2947 // TODO:
2948 // Search the external relocation entries of a fully linked image
2949 // (if any) for an entry that matches this segment offset.
2950 // uint64_t seg_offset = (Pc + Offset);
2951 return 0;
2952 }
2953 // In MH_OBJECT filetypes search the section's relocation entries (if any)
2954 // for an entry for this section offset.
2955 uint64_t sect_addr = info->S.getAddress();
2956 uint64_t sect_offset = (Pc + Offset) - sect_addr;
2957 auto Reloc =
2958 find_if(info->S.relocations(), [&](const RelocationRef &Reloc) {
2959 uint64_t RelocOffset = Reloc.getOffset();
2960 return RelocOffset == sect_offset;
2961 });
2962
2963 if (Reloc == info->S.relocations().end())
2964 return 0;
2965
2966 DataRefImpl Rel = Reloc->getRawDataRefImpl();
2967 MachO::any_relocation_info RE = info->O->getRelocation(Rel);
2968 uint32_t r_type = info->O->getAnyRelocationType(RE);
2969 if (r_type == MachO::ARM64_RELOC_ADDEND) {
2970 DataRefImpl RelNext = Rel;
2971 info->O->moveRelocationNext(RelNext);
2972 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2973 if (value == 0) {
2974 value = info->O->getPlainRelocationSymbolNum(RENext);
2975 op_info->Value = value;
2976 }
2977 }
2978 // NOTE: Scattered relocations don't exist on arm64.
2979 if (!info->O->getPlainRelocationExternal(RE))
2980 return 0;
2981 const char *name =
2982 unwrapOrError(Reloc->getSymbol()->getName(), info->O->getFileName())
2983 .data();
2984 op_info->AddSymbol.Present = 1;
2985 op_info->AddSymbol.Name = name;
2986
2987 switch (r_type) {
2988 case MachO::ARM64_RELOC_PAGE21:
2989 /* @page */
2990 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE;
2991 break;
2992 case MachO::ARM64_RELOC_PAGEOFF12:
2993 /* @pageoff */
2994 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF;
2995 break;
2996 case MachO::ARM64_RELOC_GOT_LOAD_PAGE21:
2997 /* @gotpage */
2998 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE;
2999 break;
3000 case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12:
3001 /* @gotpageoff */
3002 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF;
3003 break;
3004 case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21:
3005 /* @tvlppage is not implemented in llvm-mc */
3006 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP;
3007 break;
3008 case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12:
3009 /* @tvlppageoff is not implemented in llvm-mc */
3010 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF;
3011 break;
3012 default:
3013 case MachO::ARM64_RELOC_BRANCH26:
3014 op_info->VariantKind = LLVMDisassembler_VariantKind_None;
3015 break;
3016 }
3017 return 1;
3018 }
3019 return 0;
3020 }
3021
3022 // GuessCstringPointer is passed the address of what might be a pointer to a
3023 // literal string in a cstring section. If that address is in a cstring section
3024 // it returns a pointer to that string. Else it returns nullptr.
GuessCstringPointer(uint64_t ReferenceValue,struct DisassembleInfo * info)3025 static const char *GuessCstringPointer(uint64_t ReferenceValue,
3026 struct DisassembleInfo *info) {
3027 for (const auto &Load : info->O->load_commands()) {
3028 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3029 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3030 for (unsigned J = 0; J < Seg.nsects; ++J) {
3031 MachO::section_64 Sec = info->O->getSection64(Load, J);
3032 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3033 if (section_type == MachO::S_CSTRING_LITERALS &&
3034 ReferenceValue >= Sec.addr &&
3035 ReferenceValue < Sec.addr + Sec.size) {
3036 uint64_t sect_offset = ReferenceValue - Sec.addr;
3037 uint64_t object_offset = Sec.offset + sect_offset;
3038 StringRef MachOContents = info->O->getData();
3039 uint64_t object_size = MachOContents.size();
3040 const char *object_addr = (const char *)MachOContents.data();
3041 if (object_offset < object_size) {
3042 const char *name = object_addr + object_offset;
3043 return name;
3044 } else {
3045 return nullptr;
3046 }
3047 }
3048 }
3049 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
3050 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
3051 for (unsigned J = 0; J < Seg.nsects; ++J) {
3052 MachO::section Sec = info->O->getSection(Load, J);
3053 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3054 if (section_type == MachO::S_CSTRING_LITERALS &&
3055 ReferenceValue >= Sec.addr &&
3056 ReferenceValue < Sec.addr + Sec.size) {
3057 uint64_t sect_offset = ReferenceValue - Sec.addr;
3058 uint64_t object_offset = Sec.offset + sect_offset;
3059 StringRef MachOContents = info->O->getData();
3060 uint64_t object_size = MachOContents.size();
3061 const char *object_addr = (const char *)MachOContents.data();
3062 if (object_offset < object_size) {
3063 const char *name = object_addr + object_offset;
3064 return name;
3065 } else {
3066 return nullptr;
3067 }
3068 }
3069 }
3070 }
3071 }
3072 return nullptr;
3073 }
3074
3075 // GuessIndirectSymbol returns the name of the indirect symbol for the
3076 // ReferenceValue passed in or nullptr. This is used when ReferenceValue maybe
3077 // an address of a symbol stub or a lazy or non-lazy pointer to associate the
3078 // symbol name being referenced by the stub or pointer.
GuessIndirectSymbol(uint64_t ReferenceValue,struct DisassembleInfo * info)3079 static const char *GuessIndirectSymbol(uint64_t ReferenceValue,
3080 struct DisassembleInfo *info) {
3081 MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand();
3082 MachO::symtab_command Symtab = info->O->getSymtabLoadCommand();
3083 for (const auto &Load : info->O->load_commands()) {
3084 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3085 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3086 for (unsigned J = 0; J < Seg.nsects; ++J) {
3087 MachO::section_64 Sec = info->O->getSection64(Load, J);
3088 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3089 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3090 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3091 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3092 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
3093 section_type == MachO::S_SYMBOL_STUBS) &&
3094 ReferenceValue >= Sec.addr &&
3095 ReferenceValue < Sec.addr + Sec.size) {
3096 uint32_t stride;
3097 if (section_type == MachO::S_SYMBOL_STUBS)
3098 stride = Sec.reserved2;
3099 else
3100 stride = 8;
3101 if (stride == 0)
3102 return nullptr;
3103 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
3104 if (index < Dysymtab.nindirectsyms) {
3105 uint32_t indirect_symbol =
3106 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
3107 if (indirect_symbol < Symtab.nsyms) {
3108 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
3109 return unwrapOrError(Sym->getName(), info->O->getFileName())
3110 .data();
3111 }
3112 }
3113 }
3114 }
3115 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
3116 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
3117 for (unsigned J = 0; J < Seg.nsects; ++J) {
3118 MachO::section Sec = info->O->getSection(Load, J);
3119 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3120 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3121 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3122 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3123 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
3124 section_type == MachO::S_SYMBOL_STUBS) &&
3125 ReferenceValue >= Sec.addr &&
3126 ReferenceValue < Sec.addr + Sec.size) {
3127 uint32_t stride;
3128 if (section_type == MachO::S_SYMBOL_STUBS)
3129 stride = Sec.reserved2;
3130 else
3131 stride = 4;
3132 if (stride == 0)
3133 return nullptr;
3134 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
3135 if (index < Dysymtab.nindirectsyms) {
3136 uint32_t indirect_symbol =
3137 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
3138 if (indirect_symbol < Symtab.nsyms) {
3139 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
3140 return unwrapOrError(Sym->getName(), info->O->getFileName())
3141 .data();
3142 }
3143 }
3144 }
3145 }
3146 }
3147 }
3148 return nullptr;
3149 }
3150
3151 // method_reference() is called passing it the ReferenceName that might be
3152 // a reference it to an Objective-C method call. If so then it allocates and
3153 // assembles a method call string with the values last seen and saved in
3154 // the DisassembleInfo's class_name and selector_name fields. This is saved
3155 // into the method field of the info and any previous string is free'ed.
3156 // Then the class_name field in the info is set to nullptr. The method call
3157 // string is set into ReferenceName and ReferenceType is set to
3158 // LLVMDisassembler_ReferenceType_Out_Objc_Message. If this not a method call
3159 // then both ReferenceType and ReferenceName are left unchanged.
method_reference(struct DisassembleInfo * info,uint64_t * ReferenceType,const char ** ReferenceName)3160 static void method_reference(struct DisassembleInfo *info,
3161 uint64_t *ReferenceType,
3162 const char **ReferenceName) {
3163 unsigned int Arch = info->O->getArch();
3164 if (*ReferenceName != nullptr) {
3165 if (strcmp(*ReferenceName, "_objc_msgSend") == 0) {
3166 if (info->selector_name != nullptr) {
3167 if (info->class_name != nullptr) {
3168 info->method = std::make_unique<char[]>(
3169 5 + strlen(info->class_name) + strlen(info->selector_name));
3170 char *method = info->method.get();
3171 if (method != nullptr) {
3172 strcpy(method, "+[");
3173 strcat(method, info->class_name);
3174 strcat(method, " ");
3175 strcat(method, info->selector_name);
3176 strcat(method, "]");
3177 *ReferenceName = method;
3178 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3179 }
3180 } else {
3181 info->method =
3182 std::make_unique<char[]>(9 + strlen(info->selector_name));
3183 char *method = info->method.get();
3184 if (method != nullptr) {
3185 if (Arch == Triple::x86_64)
3186 strcpy(method, "-[%rdi ");
3187 else if (Arch == Triple::aarch64)
3188 strcpy(method, "-[x0 ");
3189 else
3190 strcpy(method, "-[r? ");
3191 strcat(method, info->selector_name);
3192 strcat(method, "]");
3193 *ReferenceName = method;
3194 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3195 }
3196 }
3197 info->class_name = nullptr;
3198 }
3199 } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) {
3200 if (info->selector_name != nullptr) {
3201 info->method =
3202 std::make_unique<char[]>(17 + strlen(info->selector_name));
3203 char *method = info->method.get();
3204 if (method != nullptr) {
3205 if (Arch == Triple::x86_64)
3206 strcpy(method, "-[[%rdi super] ");
3207 else if (Arch == Triple::aarch64)
3208 strcpy(method, "-[[x0 super] ");
3209 else
3210 strcpy(method, "-[[r? super] ");
3211 strcat(method, info->selector_name);
3212 strcat(method, "]");
3213 *ReferenceName = method;
3214 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3215 }
3216 info->class_name = nullptr;
3217 }
3218 }
3219 }
3220 }
3221
3222 // GuessPointerPointer() is passed the address of what might be a pointer to
3223 // a reference to an Objective-C class, selector, message ref or cfstring.
3224 // If so the value of the pointer is returned and one of the booleans are set
3225 // to true. If not zero is returned and all the booleans are set to false.
GuessPointerPointer(uint64_t ReferenceValue,struct DisassembleInfo * info,bool & classref,bool & selref,bool & msgref,bool & cfstring)3226 static uint64_t GuessPointerPointer(uint64_t ReferenceValue,
3227 struct DisassembleInfo *info,
3228 bool &classref, bool &selref, bool &msgref,
3229 bool &cfstring) {
3230 classref = false;
3231 selref = false;
3232 msgref = false;
3233 cfstring = false;
3234 for (const auto &Load : info->O->load_commands()) {
3235 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3236 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3237 for (unsigned J = 0; J < Seg.nsects; ++J) {
3238 MachO::section_64 Sec = info->O->getSection64(Load, J);
3239 if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 ||
3240 strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
3241 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 ||
3242 strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 ||
3243 strncmp(Sec.sectname, "__cfstring", 16) == 0) &&
3244 ReferenceValue >= Sec.addr &&
3245 ReferenceValue < Sec.addr + Sec.size) {
3246 uint64_t sect_offset = ReferenceValue - Sec.addr;
3247 uint64_t object_offset = Sec.offset + sect_offset;
3248 StringRef MachOContents = info->O->getData();
3249 uint64_t object_size = MachOContents.size();
3250 const char *object_addr = (const char *)MachOContents.data();
3251 if (object_offset < object_size) {
3252 uint64_t pointer_value;
3253 memcpy(&pointer_value, object_addr + object_offset,
3254 sizeof(uint64_t));
3255 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
3256 sys::swapByteOrder(pointer_value);
3257 if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0)
3258 selref = true;
3259 else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
3260 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0)
3261 classref = true;
3262 else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 &&
3263 ReferenceValue + 8 < Sec.addr + Sec.size) {
3264 msgref = true;
3265 memcpy(&pointer_value, object_addr + object_offset + 8,
3266 sizeof(uint64_t));
3267 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
3268 sys::swapByteOrder(pointer_value);
3269 } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0)
3270 cfstring = true;
3271 return pointer_value;
3272 } else {
3273 return 0;
3274 }
3275 }
3276 }
3277 }
3278 // TODO: Look for LC_SEGMENT for 32-bit Mach-O files.
3279 }
3280 return 0;
3281 }
3282
3283 // get_pointer_64 returns a pointer to the bytes in the object file at the
3284 // Address from a section in the Mach-O file. And indirectly returns the
3285 // offset into the section, number of bytes left in the section past the offset
3286 // and which section is was being referenced. If the Address is not in a
3287 // section nullptr is returned.
get_pointer_64(uint64_t Address,uint32_t & offset,uint32_t & left,SectionRef & S,DisassembleInfo * info,bool objc_only=false)3288 static const char *get_pointer_64(uint64_t Address, uint32_t &offset,
3289 uint32_t &left, SectionRef &S,
3290 DisassembleInfo *info,
3291 bool objc_only = false) {
3292 offset = 0;
3293 left = 0;
3294 S = SectionRef();
3295 for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) {
3296 uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress();
3297 uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize();
3298 if (SectSize == 0)
3299 continue;
3300 if (objc_only) {
3301 StringRef SectName;
3302 Expected<StringRef> SecNameOrErr =
3303 ((*(info->Sections))[SectIdx]).getName();
3304 if (SecNameOrErr)
3305 SectName = *SecNameOrErr;
3306 else
3307 consumeError(SecNameOrErr.takeError());
3308
3309 DataRefImpl Ref = ((*(info->Sections))[SectIdx]).getRawDataRefImpl();
3310 StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
3311 if (SegName != "__OBJC" && SectName != "__cstring")
3312 continue;
3313 }
3314 if (Address >= SectAddress && Address < SectAddress + SectSize) {
3315 S = (*(info->Sections))[SectIdx];
3316 offset = Address - SectAddress;
3317 left = SectSize - offset;
3318 StringRef SectContents = unwrapOrError(
3319 ((*(info->Sections))[SectIdx]).getContents(), info->O->getFileName());
3320 return SectContents.data() + offset;
3321 }
3322 }
3323 return nullptr;
3324 }
3325
get_pointer_32(uint32_t Address,uint32_t & offset,uint32_t & left,SectionRef & S,DisassembleInfo * info,bool objc_only=false)3326 static const char *get_pointer_32(uint32_t Address, uint32_t &offset,
3327 uint32_t &left, SectionRef &S,
3328 DisassembleInfo *info,
3329 bool objc_only = false) {
3330 return get_pointer_64(Address, offset, left, S, info, objc_only);
3331 }
3332
3333 // get_symbol_64() returns the name of a symbol (or nullptr) and the address of
3334 // the symbol indirectly through n_value. Based on the relocation information
3335 // for the specified section offset in the specified section reference.
3336 // If no relocation information is found and a non-zero ReferenceValue for the
3337 // symbol is passed, look up that address in the info's AddrMap.
get_symbol_64(uint32_t sect_offset,SectionRef S,DisassembleInfo * info,uint64_t & n_value,uint64_t ReferenceValue=0)3338 static const char *get_symbol_64(uint32_t sect_offset, SectionRef S,
3339 DisassembleInfo *info, uint64_t &n_value,
3340 uint64_t ReferenceValue = 0) {
3341 n_value = 0;
3342 if (!info->verbose)
3343 return nullptr;
3344
3345 // See if there is an external relocation entry at the sect_offset.
3346 bool reloc_found = false;
3347 DataRefImpl Rel;
3348 MachO::any_relocation_info RE;
3349 bool isExtern = false;
3350 SymbolRef Symbol;
3351 for (const RelocationRef &Reloc : S.relocations()) {
3352 uint64_t RelocOffset = Reloc.getOffset();
3353 if (RelocOffset == sect_offset) {
3354 Rel = Reloc.getRawDataRefImpl();
3355 RE = info->O->getRelocation(Rel);
3356 if (info->O->isRelocationScattered(RE))
3357 continue;
3358 isExtern = info->O->getPlainRelocationExternal(RE);
3359 if (isExtern) {
3360 symbol_iterator RelocSym = Reloc.getSymbol();
3361 Symbol = *RelocSym;
3362 }
3363 reloc_found = true;
3364 break;
3365 }
3366 }
3367 // If there is an external relocation entry for a symbol in this section
3368 // at this section_offset then use that symbol's value for the n_value
3369 // and return its name.
3370 const char *SymbolName = nullptr;
3371 if (reloc_found && isExtern) {
3372 n_value = cantFail(Symbol.getValue());
3373 StringRef Name = unwrapOrError(Symbol.getName(), info->O->getFileName());
3374 if (!Name.empty()) {
3375 SymbolName = Name.data();
3376 return SymbolName;
3377 }
3378 }
3379
3380 // TODO: For fully linked images, look through the external relocation
3381 // entries off the dynamic symtab command. For these the r_offset is from the
3382 // start of the first writeable segment in the Mach-O file. So the offset
3383 // to this section from that segment is passed to this routine by the caller,
3384 // as the database_offset. Which is the difference of the section's starting
3385 // address and the first writable segment.
3386 //
3387 // NOTE: need add passing the database_offset to this routine.
3388
3389 // We did not find an external relocation entry so look up the ReferenceValue
3390 // as an address of a symbol and if found return that symbol's name.
3391 SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
3392
3393 return SymbolName;
3394 }
3395
get_symbol_32(uint32_t sect_offset,SectionRef S,DisassembleInfo * info,uint32_t ReferenceValue)3396 static const char *get_symbol_32(uint32_t sect_offset, SectionRef S,
3397 DisassembleInfo *info,
3398 uint32_t ReferenceValue) {
3399 uint64_t n_value64;
3400 return get_symbol_64(sect_offset, S, info, n_value64, ReferenceValue);
3401 }
3402
3403 namespace {
3404
3405 // These are structs in the Objective-C meta data and read to produce the
3406 // comments for disassembly. While these are part of the ABI they are no
3407 // public defintions. So the are here not in include/llvm/BinaryFormat/MachO.h
3408 // .
3409
3410 // The cfstring object in a 64-bit Mach-O file.
3411 struct cfstring64_t {
3412 uint64_t isa; // class64_t * (64-bit pointer)
3413 uint64_t flags; // flag bits
3414 uint64_t characters; // char * (64-bit pointer)
3415 uint64_t length; // number of non-NULL characters in above
3416 };
3417
3418 // The class object in a 64-bit Mach-O file.
3419 struct class64_t {
3420 uint64_t isa; // class64_t * (64-bit pointer)
3421 uint64_t superclass; // class64_t * (64-bit pointer)
3422 uint64_t cache; // Cache (64-bit pointer)
3423 uint64_t vtable; // IMP * (64-bit pointer)
3424 uint64_t data; // class_ro64_t * (64-bit pointer)
3425 };
3426
3427 struct class32_t {
3428 uint32_t isa; /* class32_t * (32-bit pointer) */
3429 uint32_t superclass; /* class32_t * (32-bit pointer) */
3430 uint32_t cache; /* Cache (32-bit pointer) */
3431 uint32_t vtable; /* IMP * (32-bit pointer) */
3432 uint32_t data; /* class_ro32_t * (32-bit pointer) */
3433 };
3434
3435 struct class_ro64_t {
3436 uint32_t flags;
3437 uint32_t instanceStart;
3438 uint32_t instanceSize;
3439 uint32_t reserved;
3440 uint64_t ivarLayout; // const uint8_t * (64-bit pointer)
3441 uint64_t name; // const char * (64-bit pointer)
3442 uint64_t baseMethods; // const method_list_t * (64-bit pointer)
3443 uint64_t baseProtocols; // const protocol_list_t * (64-bit pointer)
3444 uint64_t ivars; // const ivar_list_t * (64-bit pointer)
3445 uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer)
3446 uint64_t baseProperties; // const struct objc_property_list (64-bit pointer)
3447 };
3448
3449 struct class_ro32_t {
3450 uint32_t flags;
3451 uint32_t instanceStart;
3452 uint32_t instanceSize;
3453 uint32_t ivarLayout; /* const uint8_t * (32-bit pointer) */
3454 uint32_t name; /* const char * (32-bit pointer) */
3455 uint32_t baseMethods; /* const method_list_t * (32-bit pointer) */
3456 uint32_t baseProtocols; /* const protocol_list_t * (32-bit pointer) */
3457 uint32_t ivars; /* const ivar_list_t * (32-bit pointer) */
3458 uint32_t weakIvarLayout; /* const uint8_t * (32-bit pointer) */
3459 uint32_t baseProperties; /* const struct objc_property_list *
3460 (32-bit pointer) */
3461 };
3462
3463 /* Values for class_ro{64,32}_t->flags */
3464 #define RO_META (1 << 0)
3465 #define RO_ROOT (1 << 1)
3466 #define RO_HAS_CXX_STRUCTORS (1 << 2)
3467
3468 struct method_list64_t {
3469 uint32_t entsize;
3470 uint32_t count;
3471 /* struct method64_t first; These structures follow inline */
3472 };
3473
3474 struct method_list32_t {
3475 uint32_t entsize;
3476 uint32_t count;
3477 /* struct method32_t first; These structures follow inline */
3478 };
3479
3480 struct method64_t {
3481 uint64_t name; /* SEL (64-bit pointer) */
3482 uint64_t types; /* const char * (64-bit pointer) */
3483 uint64_t imp; /* IMP (64-bit pointer) */
3484 };
3485
3486 struct method32_t {
3487 uint32_t name; /* SEL (32-bit pointer) */
3488 uint32_t types; /* const char * (32-bit pointer) */
3489 uint32_t imp; /* IMP (32-bit pointer) */
3490 };
3491
3492 struct protocol_list64_t {
3493 uint64_t count; /* uintptr_t (a 64-bit value) */
3494 /* struct protocol64_t * list[0]; These pointers follow inline */
3495 };
3496
3497 struct protocol_list32_t {
3498 uint32_t count; /* uintptr_t (a 32-bit value) */
3499 /* struct protocol32_t * list[0]; These pointers follow inline */
3500 };
3501
3502 struct protocol64_t {
3503 uint64_t isa; /* id * (64-bit pointer) */
3504 uint64_t name; /* const char * (64-bit pointer) */
3505 uint64_t protocols; /* struct protocol_list64_t *
3506 (64-bit pointer) */
3507 uint64_t instanceMethods; /* method_list_t * (64-bit pointer) */
3508 uint64_t classMethods; /* method_list_t * (64-bit pointer) */
3509 uint64_t optionalInstanceMethods; /* method_list_t * (64-bit pointer) */
3510 uint64_t optionalClassMethods; /* method_list_t * (64-bit pointer) */
3511 uint64_t instanceProperties; /* struct objc_property_list *
3512 (64-bit pointer) */
3513 };
3514
3515 struct protocol32_t {
3516 uint32_t isa; /* id * (32-bit pointer) */
3517 uint32_t name; /* const char * (32-bit pointer) */
3518 uint32_t protocols; /* struct protocol_list_t *
3519 (32-bit pointer) */
3520 uint32_t instanceMethods; /* method_list_t * (32-bit pointer) */
3521 uint32_t classMethods; /* method_list_t * (32-bit pointer) */
3522 uint32_t optionalInstanceMethods; /* method_list_t * (32-bit pointer) */
3523 uint32_t optionalClassMethods; /* method_list_t * (32-bit pointer) */
3524 uint32_t instanceProperties; /* struct objc_property_list *
3525 (32-bit pointer) */
3526 };
3527
3528 struct ivar_list64_t {
3529 uint32_t entsize;
3530 uint32_t count;
3531 /* struct ivar64_t first; These structures follow inline */
3532 };
3533
3534 struct ivar_list32_t {
3535 uint32_t entsize;
3536 uint32_t count;
3537 /* struct ivar32_t first; These structures follow inline */
3538 };
3539
3540 struct ivar64_t {
3541 uint64_t offset; /* uintptr_t * (64-bit pointer) */
3542 uint64_t name; /* const char * (64-bit pointer) */
3543 uint64_t type; /* const char * (64-bit pointer) */
3544 uint32_t alignment;
3545 uint32_t size;
3546 };
3547
3548 struct ivar32_t {
3549 uint32_t offset; /* uintptr_t * (32-bit pointer) */
3550 uint32_t name; /* const char * (32-bit pointer) */
3551 uint32_t type; /* const char * (32-bit pointer) */
3552 uint32_t alignment;
3553 uint32_t size;
3554 };
3555
3556 struct objc_property_list64 {
3557 uint32_t entsize;
3558 uint32_t count;
3559 /* struct objc_property64 first; These structures follow inline */
3560 };
3561
3562 struct objc_property_list32 {
3563 uint32_t entsize;
3564 uint32_t count;
3565 /* struct objc_property32 first; These structures follow inline */
3566 };
3567
3568 struct objc_property64 {
3569 uint64_t name; /* const char * (64-bit pointer) */
3570 uint64_t attributes; /* const char * (64-bit pointer) */
3571 };
3572
3573 struct objc_property32 {
3574 uint32_t name; /* const char * (32-bit pointer) */
3575 uint32_t attributes; /* const char * (32-bit pointer) */
3576 };
3577
3578 struct category64_t {
3579 uint64_t name; /* const char * (64-bit pointer) */
3580 uint64_t cls; /* struct class_t * (64-bit pointer) */
3581 uint64_t instanceMethods; /* struct method_list_t * (64-bit pointer) */
3582 uint64_t classMethods; /* struct method_list_t * (64-bit pointer) */
3583 uint64_t protocols; /* struct protocol_list_t * (64-bit pointer) */
3584 uint64_t instanceProperties; /* struct objc_property_list *
3585 (64-bit pointer) */
3586 };
3587
3588 struct category32_t {
3589 uint32_t name; /* const char * (32-bit pointer) */
3590 uint32_t cls; /* struct class_t * (32-bit pointer) */
3591 uint32_t instanceMethods; /* struct method_list_t * (32-bit pointer) */
3592 uint32_t classMethods; /* struct method_list_t * (32-bit pointer) */
3593 uint32_t protocols; /* struct protocol_list_t * (32-bit pointer) */
3594 uint32_t instanceProperties; /* struct objc_property_list *
3595 (32-bit pointer) */
3596 };
3597
3598 struct objc_image_info64 {
3599 uint32_t version;
3600 uint32_t flags;
3601 };
3602 struct objc_image_info32 {
3603 uint32_t version;
3604 uint32_t flags;
3605 };
3606 struct imageInfo_t {
3607 uint32_t version;
3608 uint32_t flags;
3609 };
3610 /* masks for objc_image_info.flags */
3611 #define OBJC_IMAGE_IS_REPLACEMENT (1 << 0)
3612 #define OBJC_IMAGE_SUPPORTS_GC (1 << 1)
3613 #define OBJC_IMAGE_IS_SIMULATED (1 << 5)
3614 #define OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES (1 << 6)
3615
3616 struct message_ref64 {
3617 uint64_t imp; /* IMP (64-bit pointer) */
3618 uint64_t sel; /* SEL (64-bit pointer) */
3619 };
3620
3621 struct message_ref32 {
3622 uint32_t imp; /* IMP (32-bit pointer) */
3623 uint32_t sel; /* SEL (32-bit pointer) */
3624 };
3625
3626 // Objective-C 1 (32-bit only) meta data structs.
3627
3628 struct objc_module_t {
3629 uint32_t version;
3630 uint32_t size;
3631 uint32_t name; /* char * (32-bit pointer) */
3632 uint32_t symtab; /* struct objc_symtab * (32-bit pointer) */
3633 };
3634
3635 struct objc_symtab_t {
3636 uint32_t sel_ref_cnt;
3637 uint32_t refs; /* SEL * (32-bit pointer) */
3638 uint16_t cls_def_cnt;
3639 uint16_t cat_def_cnt;
3640 // uint32_t defs[1]; /* void * (32-bit pointer) variable size */
3641 };
3642
3643 struct objc_class_t {
3644 uint32_t isa; /* struct objc_class * (32-bit pointer) */
3645 uint32_t super_class; /* struct objc_class * (32-bit pointer) */
3646 uint32_t name; /* const char * (32-bit pointer) */
3647 int32_t version;
3648 int32_t info;
3649 int32_t instance_size;
3650 uint32_t ivars; /* struct objc_ivar_list * (32-bit pointer) */
3651 uint32_t methodLists; /* struct objc_method_list ** (32-bit pointer) */
3652 uint32_t cache; /* struct objc_cache * (32-bit pointer) */
3653 uint32_t protocols; /* struct objc_protocol_list * (32-bit pointer) */
3654 };
3655
3656 #define CLS_GETINFO(cls, infomask) ((cls)->info & (infomask))
3657 // class is not a metaclass
3658 #define CLS_CLASS 0x1
3659 // class is a metaclass
3660 #define CLS_META 0x2
3661
3662 struct objc_category_t {
3663 uint32_t category_name; /* char * (32-bit pointer) */
3664 uint32_t class_name; /* char * (32-bit pointer) */
3665 uint32_t instance_methods; /* struct objc_method_list * (32-bit pointer) */
3666 uint32_t class_methods; /* struct objc_method_list * (32-bit pointer) */
3667 uint32_t protocols; /* struct objc_protocol_list * (32-bit ptr) */
3668 };
3669
3670 struct objc_ivar_t {
3671 uint32_t ivar_name; /* char * (32-bit pointer) */
3672 uint32_t ivar_type; /* char * (32-bit pointer) */
3673 int32_t ivar_offset;
3674 };
3675
3676 struct objc_ivar_list_t {
3677 int32_t ivar_count;
3678 // struct objc_ivar_t ivar_list[1]; /* variable length structure */
3679 };
3680
3681 struct objc_method_list_t {
3682 uint32_t obsolete; /* struct objc_method_list * (32-bit pointer) */
3683 int32_t method_count;
3684 // struct objc_method_t method_list[1]; /* variable length structure */
3685 };
3686
3687 struct objc_method_t {
3688 uint32_t method_name; /* SEL, aka struct objc_selector * (32-bit pointer) */
3689 uint32_t method_types; /* char * (32-bit pointer) */
3690 uint32_t method_imp; /* IMP, aka function pointer, (*IMP)(id, SEL, ...)
3691 (32-bit pointer) */
3692 };
3693
3694 struct objc_protocol_list_t {
3695 uint32_t next; /* struct objc_protocol_list * (32-bit pointer) */
3696 int32_t count;
3697 // uint32_t list[1]; /* Protocol *, aka struct objc_protocol_t *
3698 // (32-bit pointer) */
3699 };
3700
3701 struct objc_protocol_t {
3702 uint32_t isa; /* struct objc_class * (32-bit pointer) */
3703 uint32_t protocol_name; /* char * (32-bit pointer) */
3704 uint32_t protocol_list; /* struct objc_protocol_list * (32-bit pointer) */
3705 uint32_t instance_methods; /* struct objc_method_description_list *
3706 (32-bit pointer) */
3707 uint32_t class_methods; /* struct objc_method_description_list *
3708 (32-bit pointer) */
3709 };
3710
3711 struct objc_method_description_list_t {
3712 int32_t count;
3713 // struct objc_method_description_t list[1];
3714 };
3715
3716 struct objc_method_description_t {
3717 uint32_t name; /* SEL, aka struct objc_selector * (32-bit pointer) */
3718 uint32_t types; /* char * (32-bit pointer) */
3719 };
3720
swapStruct(struct cfstring64_t & cfs)3721 inline void swapStruct(struct cfstring64_t &cfs) {
3722 sys::swapByteOrder(cfs.isa);
3723 sys::swapByteOrder(cfs.flags);
3724 sys::swapByteOrder(cfs.characters);
3725 sys::swapByteOrder(cfs.length);
3726 }
3727
swapStruct(struct class64_t & c)3728 inline void swapStruct(struct class64_t &c) {
3729 sys::swapByteOrder(c.isa);
3730 sys::swapByteOrder(c.superclass);
3731 sys::swapByteOrder(c.cache);
3732 sys::swapByteOrder(c.vtable);
3733 sys::swapByteOrder(c.data);
3734 }
3735
swapStruct(struct class32_t & c)3736 inline void swapStruct(struct class32_t &c) {
3737 sys::swapByteOrder(c.isa);
3738 sys::swapByteOrder(c.superclass);
3739 sys::swapByteOrder(c.cache);
3740 sys::swapByteOrder(c.vtable);
3741 sys::swapByteOrder(c.data);
3742 }
3743
swapStruct(struct class_ro64_t & cro)3744 inline void swapStruct(struct class_ro64_t &cro) {
3745 sys::swapByteOrder(cro.flags);
3746 sys::swapByteOrder(cro.instanceStart);
3747 sys::swapByteOrder(cro.instanceSize);
3748 sys::swapByteOrder(cro.reserved);
3749 sys::swapByteOrder(cro.ivarLayout);
3750 sys::swapByteOrder(cro.name);
3751 sys::swapByteOrder(cro.baseMethods);
3752 sys::swapByteOrder(cro.baseProtocols);
3753 sys::swapByteOrder(cro.ivars);
3754 sys::swapByteOrder(cro.weakIvarLayout);
3755 sys::swapByteOrder(cro.baseProperties);
3756 }
3757
swapStruct(struct class_ro32_t & cro)3758 inline void swapStruct(struct class_ro32_t &cro) {
3759 sys::swapByteOrder(cro.flags);
3760 sys::swapByteOrder(cro.instanceStart);
3761 sys::swapByteOrder(cro.instanceSize);
3762 sys::swapByteOrder(cro.ivarLayout);
3763 sys::swapByteOrder(cro.name);
3764 sys::swapByteOrder(cro.baseMethods);
3765 sys::swapByteOrder(cro.baseProtocols);
3766 sys::swapByteOrder(cro.ivars);
3767 sys::swapByteOrder(cro.weakIvarLayout);
3768 sys::swapByteOrder(cro.baseProperties);
3769 }
3770
swapStruct(struct method_list64_t & ml)3771 inline void swapStruct(struct method_list64_t &ml) {
3772 sys::swapByteOrder(ml.entsize);
3773 sys::swapByteOrder(ml.count);
3774 }
3775
swapStruct(struct method_list32_t & ml)3776 inline void swapStruct(struct method_list32_t &ml) {
3777 sys::swapByteOrder(ml.entsize);
3778 sys::swapByteOrder(ml.count);
3779 }
3780
swapStruct(struct method64_t & m)3781 inline void swapStruct(struct method64_t &m) {
3782 sys::swapByteOrder(m.name);
3783 sys::swapByteOrder(m.types);
3784 sys::swapByteOrder(m.imp);
3785 }
3786
swapStruct(struct method32_t & m)3787 inline void swapStruct(struct method32_t &m) {
3788 sys::swapByteOrder(m.name);
3789 sys::swapByteOrder(m.types);
3790 sys::swapByteOrder(m.imp);
3791 }
3792
swapStruct(struct protocol_list64_t & pl)3793 inline void swapStruct(struct protocol_list64_t &pl) {
3794 sys::swapByteOrder(pl.count);
3795 }
3796
swapStruct(struct protocol_list32_t & pl)3797 inline void swapStruct(struct protocol_list32_t &pl) {
3798 sys::swapByteOrder(pl.count);
3799 }
3800
swapStruct(struct protocol64_t & p)3801 inline void swapStruct(struct protocol64_t &p) {
3802 sys::swapByteOrder(p.isa);
3803 sys::swapByteOrder(p.name);
3804 sys::swapByteOrder(p.protocols);
3805 sys::swapByteOrder(p.instanceMethods);
3806 sys::swapByteOrder(p.classMethods);
3807 sys::swapByteOrder(p.optionalInstanceMethods);
3808 sys::swapByteOrder(p.optionalClassMethods);
3809 sys::swapByteOrder(p.instanceProperties);
3810 }
3811
swapStruct(struct protocol32_t & p)3812 inline void swapStruct(struct protocol32_t &p) {
3813 sys::swapByteOrder(p.isa);
3814 sys::swapByteOrder(p.name);
3815 sys::swapByteOrder(p.protocols);
3816 sys::swapByteOrder(p.instanceMethods);
3817 sys::swapByteOrder(p.classMethods);
3818 sys::swapByteOrder(p.optionalInstanceMethods);
3819 sys::swapByteOrder(p.optionalClassMethods);
3820 sys::swapByteOrder(p.instanceProperties);
3821 }
3822
swapStruct(struct ivar_list64_t & il)3823 inline void swapStruct(struct ivar_list64_t &il) {
3824 sys::swapByteOrder(il.entsize);
3825 sys::swapByteOrder(il.count);
3826 }
3827
swapStruct(struct ivar_list32_t & il)3828 inline void swapStruct(struct ivar_list32_t &il) {
3829 sys::swapByteOrder(il.entsize);
3830 sys::swapByteOrder(il.count);
3831 }
3832
swapStruct(struct ivar64_t & i)3833 inline void swapStruct(struct ivar64_t &i) {
3834 sys::swapByteOrder(i.offset);
3835 sys::swapByteOrder(i.name);
3836 sys::swapByteOrder(i.type);
3837 sys::swapByteOrder(i.alignment);
3838 sys::swapByteOrder(i.size);
3839 }
3840
swapStruct(struct ivar32_t & i)3841 inline void swapStruct(struct ivar32_t &i) {
3842 sys::swapByteOrder(i.offset);
3843 sys::swapByteOrder(i.name);
3844 sys::swapByteOrder(i.type);
3845 sys::swapByteOrder(i.alignment);
3846 sys::swapByteOrder(i.size);
3847 }
3848
swapStruct(struct objc_property_list64 & pl)3849 inline void swapStruct(struct objc_property_list64 &pl) {
3850 sys::swapByteOrder(pl.entsize);
3851 sys::swapByteOrder(pl.count);
3852 }
3853
swapStruct(struct objc_property_list32 & pl)3854 inline void swapStruct(struct objc_property_list32 &pl) {
3855 sys::swapByteOrder(pl.entsize);
3856 sys::swapByteOrder(pl.count);
3857 }
3858
swapStruct(struct objc_property64 & op)3859 inline void swapStruct(struct objc_property64 &op) {
3860 sys::swapByteOrder(op.name);
3861 sys::swapByteOrder(op.attributes);
3862 }
3863
swapStruct(struct objc_property32 & op)3864 inline void swapStruct(struct objc_property32 &op) {
3865 sys::swapByteOrder(op.name);
3866 sys::swapByteOrder(op.attributes);
3867 }
3868
swapStruct(struct category64_t & c)3869 inline void swapStruct(struct category64_t &c) {
3870 sys::swapByteOrder(c.name);
3871 sys::swapByteOrder(c.cls);
3872 sys::swapByteOrder(c.instanceMethods);
3873 sys::swapByteOrder(c.classMethods);
3874 sys::swapByteOrder(c.protocols);
3875 sys::swapByteOrder(c.instanceProperties);
3876 }
3877
swapStruct(struct category32_t & c)3878 inline void swapStruct(struct category32_t &c) {
3879 sys::swapByteOrder(c.name);
3880 sys::swapByteOrder(c.cls);
3881 sys::swapByteOrder(c.instanceMethods);
3882 sys::swapByteOrder(c.classMethods);
3883 sys::swapByteOrder(c.protocols);
3884 sys::swapByteOrder(c.instanceProperties);
3885 }
3886
swapStruct(struct objc_image_info64 & o)3887 inline void swapStruct(struct objc_image_info64 &o) {
3888 sys::swapByteOrder(o.version);
3889 sys::swapByteOrder(o.flags);
3890 }
3891
swapStruct(struct objc_image_info32 & o)3892 inline void swapStruct(struct objc_image_info32 &o) {
3893 sys::swapByteOrder(o.version);
3894 sys::swapByteOrder(o.flags);
3895 }
3896
swapStruct(struct imageInfo_t & o)3897 inline void swapStruct(struct imageInfo_t &o) {
3898 sys::swapByteOrder(o.version);
3899 sys::swapByteOrder(o.flags);
3900 }
3901
swapStruct(struct message_ref64 & mr)3902 inline void swapStruct(struct message_ref64 &mr) {
3903 sys::swapByteOrder(mr.imp);
3904 sys::swapByteOrder(mr.sel);
3905 }
3906
swapStruct(struct message_ref32 & mr)3907 inline void swapStruct(struct message_ref32 &mr) {
3908 sys::swapByteOrder(mr.imp);
3909 sys::swapByteOrder(mr.sel);
3910 }
3911
swapStruct(struct objc_module_t & module)3912 inline void swapStruct(struct objc_module_t &module) {
3913 sys::swapByteOrder(module.version);
3914 sys::swapByteOrder(module.size);
3915 sys::swapByteOrder(module.name);
3916 sys::swapByteOrder(module.symtab);
3917 }
3918
swapStruct(struct objc_symtab_t & symtab)3919 inline void swapStruct(struct objc_symtab_t &symtab) {
3920 sys::swapByteOrder(symtab.sel_ref_cnt);
3921 sys::swapByteOrder(symtab.refs);
3922 sys::swapByteOrder(symtab.cls_def_cnt);
3923 sys::swapByteOrder(symtab.cat_def_cnt);
3924 }
3925
swapStruct(struct objc_class_t & objc_class)3926 inline void swapStruct(struct objc_class_t &objc_class) {
3927 sys::swapByteOrder(objc_class.isa);
3928 sys::swapByteOrder(objc_class.super_class);
3929 sys::swapByteOrder(objc_class.name);
3930 sys::swapByteOrder(objc_class.version);
3931 sys::swapByteOrder(objc_class.info);
3932 sys::swapByteOrder(objc_class.instance_size);
3933 sys::swapByteOrder(objc_class.ivars);
3934 sys::swapByteOrder(objc_class.methodLists);
3935 sys::swapByteOrder(objc_class.cache);
3936 sys::swapByteOrder(objc_class.protocols);
3937 }
3938
swapStruct(struct objc_category_t & objc_category)3939 inline void swapStruct(struct objc_category_t &objc_category) {
3940 sys::swapByteOrder(objc_category.category_name);
3941 sys::swapByteOrder(objc_category.class_name);
3942 sys::swapByteOrder(objc_category.instance_methods);
3943 sys::swapByteOrder(objc_category.class_methods);
3944 sys::swapByteOrder(objc_category.protocols);
3945 }
3946
swapStruct(struct objc_ivar_list_t & objc_ivar_list)3947 inline void swapStruct(struct objc_ivar_list_t &objc_ivar_list) {
3948 sys::swapByteOrder(objc_ivar_list.ivar_count);
3949 }
3950
swapStruct(struct objc_ivar_t & objc_ivar)3951 inline void swapStruct(struct objc_ivar_t &objc_ivar) {
3952 sys::swapByteOrder(objc_ivar.ivar_name);
3953 sys::swapByteOrder(objc_ivar.ivar_type);
3954 sys::swapByteOrder(objc_ivar.ivar_offset);
3955 }
3956
swapStruct(struct objc_method_list_t & method_list)3957 inline void swapStruct(struct objc_method_list_t &method_list) {
3958 sys::swapByteOrder(method_list.obsolete);
3959 sys::swapByteOrder(method_list.method_count);
3960 }
3961
swapStruct(struct objc_method_t & method)3962 inline void swapStruct(struct objc_method_t &method) {
3963 sys::swapByteOrder(method.method_name);
3964 sys::swapByteOrder(method.method_types);
3965 sys::swapByteOrder(method.method_imp);
3966 }
3967
swapStruct(struct objc_protocol_list_t & protocol_list)3968 inline void swapStruct(struct objc_protocol_list_t &protocol_list) {
3969 sys::swapByteOrder(protocol_list.next);
3970 sys::swapByteOrder(protocol_list.count);
3971 }
3972
swapStruct(struct objc_protocol_t & protocol)3973 inline void swapStruct(struct objc_protocol_t &protocol) {
3974 sys::swapByteOrder(protocol.isa);
3975 sys::swapByteOrder(protocol.protocol_name);
3976 sys::swapByteOrder(protocol.protocol_list);
3977 sys::swapByteOrder(protocol.instance_methods);
3978 sys::swapByteOrder(protocol.class_methods);
3979 }
3980
swapStruct(struct objc_method_description_list_t & mdl)3981 inline void swapStruct(struct objc_method_description_list_t &mdl) {
3982 sys::swapByteOrder(mdl.count);
3983 }
3984
swapStruct(struct objc_method_description_t & md)3985 inline void swapStruct(struct objc_method_description_t &md) {
3986 sys::swapByteOrder(md.name);
3987 sys::swapByteOrder(md.types);
3988 }
3989
3990 } // namespace
3991
3992 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
3993 struct DisassembleInfo *info);
3994
3995 // get_objc2_64bit_class_name() is used for disassembly and is passed a pointer
3996 // to an Objective-C class and returns the class name. It is also passed the
3997 // address of the pointer, so when the pointer is zero as it can be in an .o
3998 // file, that is used to look for an external relocation entry with a symbol
3999 // name.
get_objc2_64bit_class_name(uint64_t pointer_value,uint64_t ReferenceValue,struct DisassembleInfo * info)4000 static const char *get_objc2_64bit_class_name(uint64_t pointer_value,
4001 uint64_t ReferenceValue,
4002 struct DisassembleInfo *info) {
4003 const char *r;
4004 uint32_t offset, left;
4005 SectionRef S;
4006
4007 // The pointer_value can be 0 in an object file and have a relocation
4008 // entry for the class symbol at the ReferenceValue (the address of the
4009 // pointer).
4010 if (pointer_value == 0) {
4011 r = get_pointer_64(ReferenceValue, offset, left, S, info);
4012 if (r == nullptr || left < sizeof(uint64_t))
4013 return nullptr;
4014 uint64_t n_value;
4015 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
4016 if (symbol_name == nullptr)
4017 return nullptr;
4018 const char *class_name = strrchr(symbol_name, '$');
4019 if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0')
4020 return class_name + 2;
4021 else
4022 return nullptr;
4023 }
4024
4025 // The case were the pointer_value is non-zero and points to a class defined
4026 // in this Mach-O file.
4027 r = get_pointer_64(pointer_value, offset, left, S, info);
4028 if (r == nullptr || left < sizeof(struct class64_t))
4029 return nullptr;
4030 struct class64_t c;
4031 memcpy(&c, r, sizeof(struct class64_t));
4032 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4033 swapStruct(c);
4034 if (c.data == 0)
4035 return nullptr;
4036 r = get_pointer_64(c.data, offset, left, S, info);
4037 if (r == nullptr || left < sizeof(struct class_ro64_t))
4038 return nullptr;
4039 struct class_ro64_t cro;
4040 memcpy(&cro, r, sizeof(struct class_ro64_t));
4041 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4042 swapStruct(cro);
4043 if (cro.name == 0)
4044 return nullptr;
4045 const char *name = get_pointer_64(cro.name, offset, left, S, info);
4046 return name;
4047 }
4048
4049 // get_objc2_64bit_cfstring_name is used for disassembly and is passed a
4050 // pointer to a cfstring and returns its name or nullptr.
get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,struct DisassembleInfo * info)4051 static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,
4052 struct DisassembleInfo *info) {
4053 const char *r, *name;
4054 uint32_t offset, left;
4055 SectionRef S;
4056 struct cfstring64_t cfs;
4057 uint64_t cfs_characters;
4058
4059 r = get_pointer_64(ReferenceValue, offset, left, S, info);
4060 if (r == nullptr || left < sizeof(struct cfstring64_t))
4061 return nullptr;
4062 memcpy(&cfs, r, sizeof(struct cfstring64_t));
4063 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4064 swapStruct(cfs);
4065 if (cfs.characters == 0) {
4066 uint64_t n_value;
4067 const char *symbol_name = get_symbol_64(
4068 offset + offsetof(struct cfstring64_t, characters), S, info, n_value);
4069 if (symbol_name == nullptr)
4070 return nullptr;
4071 cfs_characters = n_value;
4072 } else
4073 cfs_characters = cfs.characters;
4074 name = get_pointer_64(cfs_characters, offset, left, S, info);
4075
4076 return name;
4077 }
4078
4079 // get_objc2_64bit_selref() is used for disassembly and is passed a the address
4080 // of a pointer to an Objective-C selector reference when the pointer value is
4081 // zero as in a .o file and is likely to have a external relocation entry with
4082 // who's symbol's n_value is the real pointer to the selector name. If that is
4083 // the case the real pointer to the selector name is returned else 0 is
4084 // returned
get_objc2_64bit_selref(uint64_t ReferenceValue,struct DisassembleInfo * info)4085 static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue,
4086 struct DisassembleInfo *info) {
4087 uint32_t offset, left;
4088 SectionRef S;
4089
4090 const char *r = get_pointer_64(ReferenceValue, offset, left, S, info);
4091 if (r == nullptr || left < sizeof(uint64_t))
4092 return 0;
4093 uint64_t n_value;
4094 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
4095 if (symbol_name == nullptr)
4096 return 0;
4097 return n_value;
4098 }
4099
get_section(MachOObjectFile * O,const char * segname,const char * sectname)4100 static const SectionRef get_section(MachOObjectFile *O, const char *segname,
4101 const char *sectname) {
4102 for (const SectionRef &Section : O->sections()) {
4103 StringRef SectName;
4104 Expected<StringRef> SecNameOrErr = Section.getName();
4105 if (SecNameOrErr)
4106 SectName = *SecNameOrErr;
4107 else
4108 consumeError(SecNameOrErr.takeError());
4109
4110 DataRefImpl Ref = Section.getRawDataRefImpl();
4111 StringRef SegName = O->getSectionFinalSegmentName(Ref);
4112 if (SegName == segname && SectName == sectname)
4113 return Section;
4114 }
4115 return SectionRef();
4116 }
4117
4118 static void
walk_pointer_list_64(const char * listname,const SectionRef S,MachOObjectFile * O,struct DisassembleInfo * info,void (* func)(uint64_t,struct DisassembleInfo * info))4119 walk_pointer_list_64(const char *listname, const SectionRef S,
4120 MachOObjectFile *O, struct DisassembleInfo *info,
4121 void (*func)(uint64_t, struct DisassembleInfo *info)) {
4122 if (S == SectionRef())
4123 return;
4124
4125 StringRef SectName;
4126 Expected<StringRef> SecNameOrErr = S.getName();
4127 if (SecNameOrErr)
4128 SectName = *SecNameOrErr;
4129 else
4130 consumeError(SecNameOrErr.takeError());
4131
4132 DataRefImpl Ref = S.getRawDataRefImpl();
4133 StringRef SegName = O->getSectionFinalSegmentName(Ref);
4134 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
4135
4136 StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName());
4137 const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
4138
4139 for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint64_t)) {
4140 uint32_t left = S.getSize() - i;
4141 uint32_t size = left < sizeof(uint64_t) ? left : sizeof(uint64_t);
4142 uint64_t p = 0;
4143 memcpy(&p, Contents + i, size);
4144 if (i + sizeof(uint64_t) > S.getSize())
4145 outs() << listname << " list pointer extends past end of (" << SegName
4146 << "," << SectName << ") section\n";
4147 outs() << format("%016" PRIx64, S.getAddress() + i) << " ";
4148
4149 if (O->isLittleEndian() != sys::IsLittleEndianHost)
4150 sys::swapByteOrder(p);
4151
4152 uint64_t n_value = 0;
4153 const char *name = get_symbol_64(i, S, info, n_value, p);
4154 if (name == nullptr)
4155 name = get_dyld_bind_info_symbolname(S.getAddress() + i, info);
4156
4157 if (n_value != 0) {
4158 outs() << format("0x%" PRIx64, n_value);
4159 if (p != 0)
4160 outs() << " + " << format("0x%" PRIx64, p);
4161 } else
4162 outs() << format("0x%" PRIx64, p);
4163 if (name != nullptr)
4164 outs() << " " << name;
4165 outs() << "\n";
4166
4167 p += n_value;
4168 if (func)
4169 func(p, info);
4170 }
4171 }
4172
4173 static void
walk_pointer_list_32(const char * listname,const SectionRef S,MachOObjectFile * O,struct DisassembleInfo * info,void (* func)(uint32_t,struct DisassembleInfo * info))4174 walk_pointer_list_32(const char *listname, const SectionRef S,
4175 MachOObjectFile *O, struct DisassembleInfo *info,
4176 void (*func)(uint32_t, struct DisassembleInfo *info)) {
4177 if (S == SectionRef())
4178 return;
4179
4180 StringRef SectName = unwrapOrError(S.getName(), O->getFileName());
4181 DataRefImpl Ref = S.getRawDataRefImpl();
4182 StringRef SegName = O->getSectionFinalSegmentName(Ref);
4183 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
4184
4185 StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName());
4186 const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
4187
4188 for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint32_t)) {
4189 uint32_t left = S.getSize() - i;
4190 uint32_t size = left < sizeof(uint32_t) ? left : sizeof(uint32_t);
4191 uint32_t p = 0;
4192 memcpy(&p, Contents + i, size);
4193 if (i + sizeof(uint32_t) > S.getSize())
4194 outs() << listname << " list pointer extends past end of (" << SegName
4195 << "," << SectName << ") section\n";
4196 uint32_t Address = S.getAddress() + i;
4197 outs() << format("%08" PRIx32, Address) << " ";
4198
4199 if (O->isLittleEndian() != sys::IsLittleEndianHost)
4200 sys::swapByteOrder(p);
4201 outs() << format("0x%" PRIx32, p);
4202
4203 const char *name = get_symbol_32(i, S, info, p);
4204 if (name != nullptr)
4205 outs() << " " << name;
4206 outs() << "\n";
4207
4208 if (func)
4209 func(p, info);
4210 }
4211 }
4212
print_layout_map(const char * layout_map,uint32_t left)4213 static void print_layout_map(const char *layout_map, uint32_t left) {
4214 if (layout_map == nullptr)
4215 return;
4216 outs() << " layout map: ";
4217 do {
4218 outs() << format("0x%02" PRIx32, (*layout_map) & 0xff) << " ";
4219 left--;
4220 layout_map++;
4221 } while (*layout_map != '\0' && left != 0);
4222 outs() << "\n";
4223 }
4224
print_layout_map64(uint64_t p,struct DisassembleInfo * info)4225 static void print_layout_map64(uint64_t p, struct DisassembleInfo *info) {
4226 uint32_t offset, left;
4227 SectionRef S;
4228 const char *layout_map;
4229
4230 if (p == 0)
4231 return;
4232 layout_map = get_pointer_64(p, offset, left, S, info);
4233 print_layout_map(layout_map, left);
4234 }
4235
print_layout_map32(uint32_t p,struct DisassembleInfo * info)4236 static void print_layout_map32(uint32_t p, struct DisassembleInfo *info) {
4237 uint32_t offset, left;
4238 SectionRef S;
4239 const char *layout_map;
4240
4241 if (p == 0)
4242 return;
4243 layout_map = get_pointer_32(p, offset, left, S, info);
4244 print_layout_map(layout_map, left);
4245 }
4246
print_method_list64_t(uint64_t p,struct DisassembleInfo * info,const char * indent)4247 static void print_method_list64_t(uint64_t p, struct DisassembleInfo *info,
4248 const char *indent) {
4249 struct method_list64_t ml;
4250 struct method64_t m;
4251 const char *r;
4252 uint32_t offset, xoffset, left, i;
4253 SectionRef S, xS;
4254 const char *name, *sym_name;
4255 uint64_t n_value;
4256
4257 r = get_pointer_64(p, offset, left, S, info);
4258 if (r == nullptr)
4259 return;
4260 memset(&ml, '\0', sizeof(struct method_list64_t));
4261 if (left < sizeof(struct method_list64_t)) {
4262 memcpy(&ml, r, left);
4263 outs() << " (method_list_t entends past the end of the section)\n";
4264 } else
4265 memcpy(&ml, r, sizeof(struct method_list64_t));
4266 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4267 swapStruct(ml);
4268 outs() << indent << "\t\t entsize " << ml.entsize << "\n";
4269 outs() << indent << "\t\t count " << ml.count << "\n";
4270
4271 p += sizeof(struct method_list64_t);
4272 offset += sizeof(struct method_list64_t);
4273 for (i = 0; i < ml.count; i++) {
4274 r = get_pointer_64(p, offset, left, S, info);
4275 if (r == nullptr)
4276 return;
4277 memset(&m, '\0', sizeof(struct method64_t));
4278 if (left < sizeof(struct method64_t)) {
4279 memcpy(&m, r, left);
4280 outs() << indent << " (method_t extends past the end of the section)\n";
4281 } else
4282 memcpy(&m, r, sizeof(struct method64_t));
4283 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4284 swapStruct(m);
4285
4286 outs() << indent << "\t\t name ";
4287 sym_name = get_symbol_64(offset + offsetof(struct method64_t, name), S,
4288 info, n_value, m.name);
4289 if (n_value != 0) {
4290 if (info->verbose && sym_name != nullptr)
4291 outs() << sym_name;
4292 else
4293 outs() << format("0x%" PRIx64, n_value);
4294 if (m.name != 0)
4295 outs() << " + " << format("0x%" PRIx64, m.name);
4296 } else
4297 outs() << format("0x%" PRIx64, m.name);
4298 name = get_pointer_64(m.name + n_value, xoffset, left, xS, info);
4299 if (name != nullptr)
4300 outs() << format(" %.*s", left, name);
4301 outs() << "\n";
4302
4303 outs() << indent << "\t\t types ";
4304 sym_name = get_symbol_64(offset + offsetof(struct method64_t, types), S,
4305 info, n_value, m.types);
4306 if (n_value != 0) {
4307 if (info->verbose && sym_name != nullptr)
4308 outs() << sym_name;
4309 else
4310 outs() << format("0x%" PRIx64, n_value);
4311 if (m.types != 0)
4312 outs() << " + " << format("0x%" PRIx64, m.types);
4313 } else
4314 outs() << format("0x%" PRIx64, m.types);
4315 name = get_pointer_64(m.types + n_value, xoffset, left, xS, info);
4316 if (name != nullptr)
4317 outs() << format(" %.*s", left, name);
4318 outs() << "\n";
4319
4320 outs() << indent << "\t\t imp ";
4321 name = get_symbol_64(offset + offsetof(struct method64_t, imp), S, info,
4322 n_value, m.imp);
4323 if (info->verbose && name == nullptr) {
4324 if (n_value != 0) {
4325 outs() << format("0x%" PRIx64, n_value) << " ";
4326 if (m.imp != 0)
4327 outs() << "+ " << format("0x%" PRIx64, m.imp) << " ";
4328 } else
4329 outs() << format("0x%" PRIx64, m.imp) << " ";
4330 }
4331 if (name != nullptr)
4332 outs() << name;
4333 outs() << "\n";
4334
4335 p += sizeof(struct method64_t);
4336 offset += sizeof(struct method64_t);
4337 }
4338 }
4339
print_method_list32_t(uint64_t p,struct DisassembleInfo * info,const char * indent)4340 static void print_method_list32_t(uint64_t p, struct DisassembleInfo *info,
4341 const char *indent) {
4342 struct method_list32_t ml;
4343 struct method32_t m;
4344 const char *r, *name;
4345 uint32_t offset, xoffset, left, i;
4346 SectionRef S, xS;
4347
4348 r = get_pointer_32(p, offset, left, S, info);
4349 if (r == nullptr)
4350 return;
4351 memset(&ml, '\0', sizeof(struct method_list32_t));
4352 if (left < sizeof(struct method_list32_t)) {
4353 memcpy(&ml, r, left);
4354 outs() << " (method_list_t entends past the end of the section)\n";
4355 } else
4356 memcpy(&ml, r, sizeof(struct method_list32_t));
4357 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4358 swapStruct(ml);
4359 outs() << indent << "\t\t entsize " << ml.entsize << "\n";
4360 outs() << indent << "\t\t count " << ml.count << "\n";
4361
4362 p += sizeof(struct method_list32_t);
4363 offset += sizeof(struct method_list32_t);
4364 for (i = 0; i < ml.count; i++) {
4365 r = get_pointer_32(p, offset, left, S, info);
4366 if (r == nullptr)
4367 return;
4368 memset(&m, '\0', sizeof(struct method32_t));
4369 if (left < sizeof(struct method32_t)) {
4370 memcpy(&ml, r, left);
4371 outs() << indent << " (method_t entends past the end of the section)\n";
4372 } else
4373 memcpy(&m, r, sizeof(struct method32_t));
4374 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4375 swapStruct(m);
4376
4377 outs() << indent << "\t\t name " << format("0x%" PRIx32, m.name);
4378 name = get_pointer_32(m.name, xoffset, left, xS, info);
4379 if (name != nullptr)
4380 outs() << format(" %.*s", left, name);
4381 outs() << "\n";
4382
4383 outs() << indent << "\t\t types " << format("0x%" PRIx32, m.types);
4384 name = get_pointer_32(m.types, xoffset, left, xS, info);
4385 if (name != nullptr)
4386 outs() << format(" %.*s", left, name);
4387 outs() << "\n";
4388
4389 outs() << indent << "\t\t imp " << format("0x%" PRIx32, m.imp);
4390 name = get_symbol_32(offset + offsetof(struct method32_t, imp), S, info,
4391 m.imp);
4392 if (name != nullptr)
4393 outs() << " " << name;
4394 outs() << "\n";
4395
4396 p += sizeof(struct method32_t);
4397 offset += sizeof(struct method32_t);
4398 }
4399 }
4400
print_method_list(uint32_t p,struct DisassembleInfo * info)4401 static bool print_method_list(uint32_t p, struct DisassembleInfo *info) {
4402 uint32_t offset, left, xleft;
4403 SectionRef S;
4404 struct objc_method_list_t method_list;
4405 struct objc_method_t method;
4406 const char *r, *methods, *name, *SymbolName;
4407 int32_t i;
4408
4409 r = get_pointer_32(p, offset, left, S, info, true);
4410 if (r == nullptr)
4411 return true;
4412
4413 outs() << "\n";
4414 if (left > sizeof(struct objc_method_list_t)) {
4415 memcpy(&method_list, r, sizeof(struct objc_method_list_t));
4416 } else {
4417 outs() << "\t\t objc_method_list extends past end of the section\n";
4418 memset(&method_list, '\0', sizeof(struct objc_method_list_t));
4419 memcpy(&method_list, r, left);
4420 }
4421 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4422 swapStruct(method_list);
4423
4424 outs() << "\t\t obsolete "
4425 << format("0x%08" PRIx32, method_list.obsolete) << "\n";
4426 outs() << "\t\t method_count " << method_list.method_count << "\n";
4427
4428 methods = r + sizeof(struct objc_method_list_t);
4429 for (i = 0; i < method_list.method_count; i++) {
4430 if ((i + 1) * sizeof(struct objc_method_t) > left) {
4431 outs() << "\t\t remaining method's extend past the of the section\n";
4432 break;
4433 }
4434 memcpy(&method, methods + i * sizeof(struct objc_method_t),
4435 sizeof(struct objc_method_t));
4436 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4437 swapStruct(method);
4438
4439 outs() << "\t\t method_name "
4440 << format("0x%08" PRIx32, method.method_name);
4441 if (info->verbose) {
4442 name = get_pointer_32(method.method_name, offset, xleft, S, info, true);
4443 if (name != nullptr)
4444 outs() << format(" %.*s", xleft, name);
4445 else
4446 outs() << " (not in an __OBJC section)";
4447 }
4448 outs() << "\n";
4449
4450 outs() << "\t\t method_types "
4451 << format("0x%08" PRIx32, method.method_types);
4452 if (info->verbose) {
4453 name = get_pointer_32(method.method_types, offset, xleft, S, info, true);
4454 if (name != nullptr)
4455 outs() << format(" %.*s", xleft, name);
4456 else
4457 outs() << " (not in an __OBJC section)";
4458 }
4459 outs() << "\n";
4460
4461 outs() << "\t\t method_imp "
4462 << format("0x%08" PRIx32, method.method_imp) << " ";
4463 if (info->verbose) {
4464 SymbolName = GuessSymbolName(method.method_imp, info->AddrMap);
4465 if (SymbolName != nullptr)
4466 outs() << SymbolName;
4467 }
4468 outs() << "\n";
4469 }
4470 return false;
4471 }
4472
print_protocol_list64_t(uint64_t p,struct DisassembleInfo * info)4473 static void print_protocol_list64_t(uint64_t p, struct DisassembleInfo *info) {
4474 struct protocol_list64_t pl;
4475 uint64_t q, n_value;
4476 struct protocol64_t pc;
4477 const char *r;
4478 uint32_t offset, xoffset, left, i;
4479 SectionRef S, xS;
4480 const char *name, *sym_name;
4481
4482 r = get_pointer_64(p, offset, left, S, info);
4483 if (r == nullptr)
4484 return;
4485 memset(&pl, '\0', sizeof(struct protocol_list64_t));
4486 if (left < sizeof(struct protocol_list64_t)) {
4487 memcpy(&pl, r, left);
4488 outs() << " (protocol_list_t entends past the end of the section)\n";
4489 } else
4490 memcpy(&pl, r, sizeof(struct protocol_list64_t));
4491 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4492 swapStruct(pl);
4493 outs() << " count " << pl.count << "\n";
4494
4495 p += sizeof(struct protocol_list64_t);
4496 offset += sizeof(struct protocol_list64_t);
4497 for (i = 0; i < pl.count; i++) {
4498 r = get_pointer_64(p, offset, left, S, info);
4499 if (r == nullptr)
4500 return;
4501 q = 0;
4502 if (left < sizeof(uint64_t)) {
4503 memcpy(&q, r, left);
4504 outs() << " (protocol_t * entends past the end of the section)\n";
4505 } else
4506 memcpy(&q, r, sizeof(uint64_t));
4507 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4508 sys::swapByteOrder(q);
4509
4510 outs() << "\t\t list[" << i << "] ";
4511 sym_name = get_symbol_64(offset, S, info, n_value, q);
4512 if (n_value != 0) {
4513 if (info->verbose && sym_name != nullptr)
4514 outs() << sym_name;
4515 else
4516 outs() << format("0x%" PRIx64, n_value);
4517 if (q != 0)
4518 outs() << " + " << format("0x%" PRIx64, q);
4519 } else
4520 outs() << format("0x%" PRIx64, q);
4521 outs() << " (struct protocol_t *)\n";
4522
4523 r = get_pointer_64(q + n_value, offset, left, S, info);
4524 if (r == nullptr)
4525 return;
4526 memset(&pc, '\0', sizeof(struct protocol64_t));
4527 if (left < sizeof(struct protocol64_t)) {
4528 memcpy(&pc, r, left);
4529 outs() << " (protocol_t entends past the end of the section)\n";
4530 } else
4531 memcpy(&pc, r, sizeof(struct protocol64_t));
4532 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4533 swapStruct(pc);
4534
4535 outs() << "\t\t\t isa " << format("0x%" PRIx64, pc.isa) << "\n";
4536
4537 outs() << "\t\t\t name ";
4538 sym_name = get_symbol_64(offset + offsetof(struct protocol64_t, name), S,
4539 info, n_value, pc.name);
4540 if (n_value != 0) {
4541 if (info->verbose && sym_name != nullptr)
4542 outs() << sym_name;
4543 else
4544 outs() << format("0x%" PRIx64, n_value);
4545 if (pc.name != 0)
4546 outs() << " + " << format("0x%" PRIx64, pc.name);
4547 } else
4548 outs() << format("0x%" PRIx64, pc.name);
4549 name = get_pointer_64(pc.name + n_value, xoffset, left, xS, info);
4550 if (name != nullptr)
4551 outs() << format(" %.*s", left, name);
4552 outs() << "\n";
4553
4554 outs() << "\t\t\tprotocols " << format("0x%" PRIx64, pc.protocols) << "\n";
4555
4556 outs() << "\t\t instanceMethods ";
4557 sym_name =
4558 get_symbol_64(offset + offsetof(struct protocol64_t, instanceMethods),
4559 S, info, n_value, pc.instanceMethods);
4560 if (n_value != 0) {
4561 if (info->verbose && sym_name != nullptr)
4562 outs() << sym_name;
4563 else
4564 outs() << format("0x%" PRIx64, n_value);
4565 if (pc.instanceMethods != 0)
4566 outs() << " + " << format("0x%" PRIx64, pc.instanceMethods);
4567 } else
4568 outs() << format("0x%" PRIx64, pc.instanceMethods);
4569 outs() << " (struct method_list_t *)\n";
4570 if (pc.instanceMethods + n_value != 0)
4571 print_method_list64_t(pc.instanceMethods + n_value, info, "\t");
4572
4573 outs() << "\t\t classMethods ";
4574 sym_name =
4575 get_symbol_64(offset + offsetof(struct protocol64_t, classMethods), S,
4576 info, n_value, pc.classMethods);
4577 if (n_value != 0) {
4578 if (info->verbose && sym_name != nullptr)
4579 outs() << sym_name;
4580 else
4581 outs() << format("0x%" PRIx64, n_value);
4582 if (pc.classMethods != 0)
4583 outs() << " + " << format("0x%" PRIx64, pc.classMethods);
4584 } else
4585 outs() << format("0x%" PRIx64, pc.classMethods);
4586 outs() << " (struct method_list_t *)\n";
4587 if (pc.classMethods + n_value != 0)
4588 print_method_list64_t(pc.classMethods + n_value, info, "\t");
4589
4590 outs() << "\t optionalInstanceMethods "
4591 << format("0x%" PRIx64, pc.optionalInstanceMethods) << "\n";
4592 outs() << "\t optionalClassMethods "
4593 << format("0x%" PRIx64, pc.optionalClassMethods) << "\n";
4594 outs() << "\t instanceProperties "
4595 << format("0x%" PRIx64, pc.instanceProperties) << "\n";
4596
4597 p += sizeof(uint64_t);
4598 offset += sizeof(uint64_t);
4599 }
4600 }
4601
print_protocol_list32_t(uint32_t p,struct DisassembleInfo * info)4602 static void print_protocol_list32_t(uint32_t p, struct DisassembleInfo *info) {
4603 struct protocol_list32_t pl;
4604 uint32_t q;
4605 struct protocol32_t pc;
4606 const char *r;
4607 uint32_t offset, xoffset, left, i;
4608 SectionRef S, xS;
4609 const char *name;
4610
4611 r = get_pointer_32(p, offset, left, S, info);
4612 if (r == nullptr)
4613 return;
4614 memset(&pl, '\0', sizeof(struct protocol_list32_t));
4615 if (left < sizeof(struct protocol_list32_t)) {
4616 memcpy(&pl, r, left);
4617 outs() << " (protocol_list_t entends past the end of the section)\n";
4618 } else
4619 memcpy(&pl, r, sizeof(struct protocol_list32_t));
4620 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4621 swapStruct(pl);
4622 outs() << " count " << pl.count << "\n";
4623
4624 p += sizeof(struct protocol_list32_t);
4625 offset += sizeof(struct protocol_list32_t);
4626 for (i = 0; i < pl.count; i++) {
4627 r = get_pointer_32(p, offset, left, S, info);
4628 if (r == nullptr)
4629 return;
4630 q = 0;
4631 if (left < sizeof(uint32_t)) {
4632 memcpy(&q, r, left);
4633 outs() << " (protocol_t * entends past the end of the section)\n";
4634 } else
4635 memcpy(&q, r, sizeof(uint32_t));
4636 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4637 sys::swapByteOrder(q);
4638 outs() << "\t\t list[" << i << "] " << format("0x%" PRIx32, q)
4639 << " (struct protocol_t *)\n";
4640 r = get_pointer_32(q, offset, left, S, info);
4641 if (r == nullptr)
4642 return;
4643 memset(&pc, '\0', sizeof(struct protocol32_t));
4644 if (left < sizeof(struct protocol32_t)) {
4645 memcpy(&pc, r, left);
4646 outs() << " (protocol_t entends past the end of the section)\n";
4647 } else
4648 memcpy(&pc, r, sizeof(struct protocol32_t));
4649 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4650 swapStruct(pc);
4651 outs() << "\t\t\t isa " << format("0x%" PRIx32, pc.isa) << "\n";
4652 outs() << "\t\t\t name " << format("0x%" PRIx32, pc.name);
4653 name = get_pointer_32(pc.name, xoffset, left, xS, info);
4654 if (name != nullptr)
4655 outs() << format(" %.*s", left, name);
4656 outs() << "\n";
4657 outs() << "\t\t\tprotocols " << format("0x%" PRIx32, pc.protocols) << "\n";
4658 outs() << "\t\t instanceMethods "
4659 << format("0x%" PRIx32, pc.instanceMethods)
4660 << " (struct method_list_t *)\n";
4661 if (pc.instanceMethods != 0)
4662 print_method_list32_t(pc.instanceMethods, info, "\t");
4663 outs() << "\t\t classMethods " << format("0x%" PRIx32, pc.classMethods)
4664 << " (struct method_list_t *)\n";
4665 if (pc.classMethods != 0)
4666 print_method_list32_t(pc.classMethods, info, "\t");
4667 outs() << "\t optionalInstanceMethods "
4668 << format("0x%" PRIx32, pc.optionalInstanceMethods) << "\n";
4669 outs() << "\t optionalClassMethods "
4670 << format("0x%" PRIx32, pc.optionalClassMethods) << "\n";
4671 outs() << "\t instanceProperties "
4672 << format("0x%" PRIx32, pc.instanceProperties) << "\n";
4673 p += sizeof(uint32_t);
4674 offset += sizeof(uint32_t);
4675 }
4676 }
4677
print_indent(uint32_t indent)4678 static void print_indent(uint32_t indent) {
4679 for (uint32_t i = 0; i < indent;) {
4680 if (indent - i >= 8) {
4681 outs() << "\t";
4682 i += 8;
4683 } else {
4684 for (uint32_t j = i; j < indent; j++)
4685 outs() << " ";
4686 return;
4687 }
4688 }
4689 }
4690
print_method_description_list(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4691 static bool print_method_description_list(uint32_t p, uint32_t indent,
4692 struct DisassembleInfo *info) {
4693 uint32_t offset, left, xleft;
4694 SectionRef S;
4695 struct objc_method_description_list_t mdl;
4696 struct objc_method_description_t md;
4697 const char *r, *list, *name;
4698 int32_t i;
4699
4700 r = get_pointer_32(p, offset, left, S, info, true);
4701 if (r == nullptr)
4702 return true;
4703
4704 outs() << "\n";
4705 if (left > sizeof(struct objc_method_description_list_t)) {
4706 memcpy(&mdl, r, sizeof(struct objc_method_description_list_t));
4707 } else {
4708 print_indent(indent);
4709 outs() << " objc_method_description_list extends past end of the section\n";
4710 memset(&mdl, '\0', sizeof(struct objc_method_description_list_t));
4711 memcpy(&mdl, r, left);
4712 }
4713 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4714 swapStruct(mdl);
4715
4716 print_indent(indent);
4717 outs() << " count " << mdl.count << "\n";
4718
4719 list = r + sizeof(struct objc_method_description_list_t);
4720 for (i = 0; i < mdl.count; i++) {
4721 if ((i + 1) * sizeof(struct objc_method_description_t) > left) {
4722 print_indent(indent);
4723 outs() << " remaining list entries extend past the of the section\n";
4724 break;
4725 }
4726 print_indent(indent);
4727 outs() << " list[" << i << "]\n";
4728 memcpy(&md, list + i * sizeof(struct objc_method_description_t),
4729 sizeof(struct objc_method_description_t));
4730 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4731 swapStruct(md);
4732
4733 print_indent(indent);
4734 outs() << " name " << format("0x%08" PRIx32, md.name);
4735 if (info->verbose) {
4736 name = get_pointer_32(md.name, offset, xleft, S, info, true);
4737 if (name != nullptr)
4738 outs() << format(" %.*s", xleft, name);
4739 else
4740 outs() << " (not in an __OBJC section)";
4741 }
4742 outs() << "\n";
4743
4744 print_indent(indent);
4745 outs() << " types " << format("0x%08" PRIx32, md.types);
4746 if (info->verbose) {
4747 name = get_pointer_32(md.types, offset, xleft, S, info, true);
4748 if (name != nullptr)
4749 outs() << format(" %.*s", xleft, name);
4750 else
4751 outs() << " (not in an __OBJC section)";
4752 }
4753 outs() << "\n";
4754 }
4755 return false;
4756 }
4757
4758 static bool print_protocol_list(uint32_t p, uint32_t indent,
4759 struct DisassembleInfo *info);
4760
print_protocol(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4761 static bool print_protocol(uint32_t p, uint32_t indent,
4762 struct DisassembleInfo *info) {
4763 uint32_t offset, left;
4764 SectionRef S;
4765 struct objc_protocol_t protocol;
4766 const char *r, *name;
4767
4768 r = get_pointer_32(p, offset, left, S, info, true);
4769 if (r == nullptr)
4770 return true;
4771
4772 outs() << "\n";
4773 if (left >= sizeof(struct objc_protocol_t)) {
4774 memcpy(&protocol, r, sizeof(struct objc_protocol_t));
4775 } else {
4776 print_indent(indent);
4777 outs() << " Protocol extends past end of the section\n";
4778 memset(&protocol, '\0', sizeof(struct objc_protocol_t));
4779 memcpy(&protocol, r, left);
4780 }
4781 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4782 swapStruct(protocol);
4783
4784 print_indent(indent);
4785 outs() << " isa " << format("0x%08" PRIx32, protocol.isa)
4786 << "\n";
4787
4788 print_indent(indent);
4789 outs() << " protocol_name "
4790 << format("0x%08" PRIx32, protocol.protocol_name);
4791 if (info->verbose) {
4792 name = get_pointer_32(protocol.protocol_name, offset, left, S, info, true);
4793 if (name != nullptr)
4794 outs() << format(" %.*s", left, name);
4795 else
4796 outs() << " (not in an __OBJC section)";
4797 }
4798 outs() << "\n";
4799
4800 print_indent(indent);
4801 outs() << " protocol_list "
4802 << format("0x%08" PRIx32, protocol.protocol_list);
4803 if (print_protocol_list(protocol.protocol_list, indent + 4, info))
4804 outs() << " (not in an __OBJC section)\n";
4805
4806 print_indent(indent);
4807 outs() << " instance_methods "
4808 << format("0x%08" PRIx32, protocol.instance_methods);
4809 if (print_method_description_list(protocol.instance_methods, indent, info))
4810 outs() << " (not in an __OBJC section)\n";
4811
4812 print_indent(indent);
4813 outs() << " class_methods "
4814 << format("0x%08" PRIx32, protocol.class_methods);
4815 if (print_method_description_list(protocol.class_methods, indent, info))
4816 outs() << " (not in an __OBJC section)\n";
4817
4818 return false;
4819 }
4820
print_protocol_list(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4821 static bool print_protocol_list(uint32_t p, uint32_t indent,
4822 struct DisassembleInfo *info) {
4823 uint32_t offset, left, l;
4824 SectionRef S;
4825 struct objc_protocol_list_t protocol_list;
4826 const char *r, *list;
4827 int32_t i;
4828
4829 r = get_pointer_32(p, offset, left, S, info, true);
4830 if (r == nullptr)
4831 return true;
4832
4833 outs() << "\n";
4834 if (left > sizeof(struct objc_protocol_list_t)) {
4835 memcpy(&protocol_list, r, sizeof(struct objc_protocol_list_t));
4836 } else {
4837 outs() << "\t\t objc_protocol_list_t extends past end of the section\n";
4838 memset(&protocol_list, '\0', sizeof(struct objc_protocol_list_t));
4839 memcpy(&protocol_list, r, left);
4840 }
4841 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4842 swapStruct(protocol_list);
4843
4844 print_indent(indent);
4845 outs() << " next " << format("0x%08" PRIx32, protocol_list.next)
4846 << "\n";
4847 print_indent(indent);
4848 outs() << " count " << protocol_list.count << "\n";
4849
4850 list = r + sizeof(struct objc_protocol_list_t);
4851 for (i = 0; i < protocol_list.count; i++) {
4852 if ((i + 1) * sizeof(uint32_t) > left) {
4853 outs() << "\t\t remaining list entries extend past the of the section\n";
4854 break;
4855 }
4856 memcpy(&l, list + i * sizeof(uint32_t), sizeof(uint32_t));
4857 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4858 sys::swapByteOrder(l);
4859
4860 print_indent(indent);
4861 outs() << " list[" << i << "] " << format("0x%08" PRIx32, l);
4862 if (print_protocol(l, indent, info))
4863 outs() << "(not in an __OBJC section)\n";
4864 }
4865 return false;
4866 }
4867
print_ivar_list64_t(uint64_t p,struct DisassembleInfo * info)4868 static void print_ivar_list64_t(uint64_t p, struct DisassembleInfo *info) {
4869 struct ivar_list64_t il;
4870 struct ivar64_t i;
4871 const char *r;
4872 uint32_t offset, xoffset, left, j;
4873 SectionRef S, xS;
4874 const char *name, *sym_name, *ivar_offset_p;
4875 uint64_t ivar_offset, n_value;
4876
4877 r = get_pointer_64(p, offset, left, S, info);
4878 if (r == nullptr)
4879 return;
4880 memset(&il, '\0', sizeof(struct ivar_list64_t));
4881 if (left < sizeof(struct ivar_list64_t)) {
4882 memcpy(&il, r, left);
4883 outs() << " (ivar_list_t entends past the end of the section)\n";
4884 } else
4885 memcpy(&il, r, sizeof(struct ivar_list64_t));
4886 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4887 swapStruct(il);
4888 outs() << " entsize " << il.entsize << "\n";
4889 outs() << " count " << il.count << "\n";
4890
4891 p += sizeof(struct ivar_list64_t);
4892 offset += sizeof(struct ivar_list64_t);
4893 for (j = 0; j < il.count; j++) {
4894 r = get_pointer_64(p, offset, left, S, info);
4895 if (r == nullptr)
4896 return;
4897 memset(&i, '\0', sizeof(struct ivar64_t));
4898 if (left < sizeof(struct ivar64_t)) {
4899 memcpy(&i, r, left);
4900 outs() << " (ivar_t entends past the end of the section)\n";
4901 } else
4902 memcpy(&i, r, sizeof(struct ivar64_t));
4903 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4904 swapStruct(i);
4905
4906 outs() << "\t\t\t offset ";
4907 sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, offset), S,
4908 info, n_value, i.offset);
4909 if (n_value != 0) {
4910 if (info->verbose && sym_name != nullptr)
4911 outs() << sym_name;
4912 else
4913 outs() << format("0x%" PRIx64, n_value);
4914 if (i.offset != 0)
4915 outs() << " + " << format("0x%" PRIx64, i.offset);
4916 } else
4917 outs() << format("0x%" PRIx64, i.offset);
4918 ivar_offset_p = get_pointer_64(i.offset + n_value, xoffset, left, xS, info);
4919 if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) {
4920 memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset));
4921 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4922 sys::swapByteOrder(ivar_offset);
4923 outs() << " " << ivar_offset << "\n";
4924 } else
4925 outs() << "\n";
4926
4927 outs() << "\t\t\t name ";
4928 sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, name), S, info,
4929 n_value, i.name);
4930 if (n_value != 0) {
4931 if (info->verbose && sym_name != nullptr)
4932 outs() << sym_name;
4933 else
4934 outs() << format("0x%" PRIx64, n_value);
4935 if (i.name != 0)
4936 outs() << " + " << format("0x%" PRIx64, i.name);
4937 } else
4938 outs() << format("0x%" PRIx64, i.name);
4939 name = get_pointer_64(i.name + n_value, xoffset, left, xS, info);
4940 if (name != nullptr)
4941 outs() << format(" %.*s", left, name);
4942 outs() << "\n";
4943
4944 outs() << "\t\t\t type ";
4945 sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, type), S, info,
4946 n_value, i.name);
4947 name = get_pointer_64(i.type + n_value, xoffset, left, xS, info);
4948 if (n_value != 0) {
4949 if (info->verbose && sym_name != nullptr)
4950 outs() << sym_name;
4951 else
4952 outs() << format("0x%" PRIx64, n_value);
4953 if (i.type != 0)
4954 outs() << " + " << format("0x%" PRIx64, i.type);
4955 } else
4956 outs() << format("0x%" PRIx64, i.type);
4957 if (name != nullptr)
4958 outs() << format(" %.*s", left, name);
4959 outs() << "\n";
4960
4961 outs() << "\t\t\talignment " << i.alignment << "\n";
4962 outs() << "\t\t\t size " << i.size << "\n";
4963
4964 p += sizeof(struct ivar64_t);
4965 offset += sizeof(struct ivar64_t);
4966 }
4967 }
4968
print_ivar_list32_t(uint32_t p,struct DisassembleInfo * info)4969 static void print_ivar_list32_t(uint32_t p, struct DisassembleInfo *info) {
4970 struct ivar_list32_t il;
4971 struct ivar32_t i;
4972 const char *r;
4973 uint32_t offset, xoffset, left, j;
4974 SectionRef S, xS;
4975 const char *name, *ivar_offset_p;
4976 uint32_t ivar_offset;
4977
4978 r = get_pointer_32(p, offset, left, S, info);
4979 if (r == nullptr)
4980 return;
4981 memset(&il, '\0', sizeof(struct ivar_list32_t));
4982 if (left < sizeof(struct ivar_list32_t)) {
4983 memcpy(&il, r, left);
4984 outs() << " (ivar_list_t entends past the end of the section)\n";
4985 } else
4986 memcpy(&il, r, sizeof(struct ivar_list32_t));
4987 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4988 swapStruct(il);
4989 outs() << " entsize " << il.entsize << "\n";
4990 outs() << " count " << il.count << "\n";
4991
4992 p += sizeof(struct ivar_list32_t);
4993 offset += sizeof(struct ivar_list32_t);
4994 for (j = 0; j < il.count; j++) {
4995 r = get_pointer_32(p, offset, left, S, info);
4996 if (r == nullptr)
4997 return;
4998 memset(&i, '\0', sizeof(struct ivar32_t));
4999 if (left < sizeof(struct ivar32_t)) {
5000 memcpy(&i, r, left);
5001 outs() << " (ivar_t entends past the end of the section)\n";
5002 } else
5003 memcpy(&i, r, sizeof(struct ivar32_t));
5004 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5005 swapStruct(i);
5006
5007 outs() << "\t\t\t offset " << format("0x%" PRIx32, i.offset);
5008 ivar_offset_p = get_pointer_32(i.offset, xoffset, left, xS, info);
5009 if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) {
5010 memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset));
5011 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5012 sys::swapByteOrder(ivar_offset);
5013 outs() << " " << ivar_offset << "\n";
5014 } else
5015 outs() << "\n";
5016
5017 outs() << "\t\t\t name " << format("0x%" PRIx32, i.name);
5018 name = get_pointer_32(i.name, xoffset, left, xS, info);
5019 if (name != nullptr)
5020 outs() << format(" %.*s", left, name);
5021 outs() << "\n";
5022
5023 outs() << "\t\t\t type " << format("0x%" PRIx32, i.type);
5024 name = get_pointer_32(i.type, xoffset, left, xS, info);
5025 if (name != nullptr)
5026 outs() << format(" %.*s", left, name);
5027 outs() << "\n";
5028
5029 outs() << "\t\t\talignment " << i.alignment << "\n";
5030 outs() << "\t\t\t size " << i.size << "\n";
5031
5032 p += sizeof(struct ivar32_t);
5033 offset += sizeof(struct ivar32_t);
5034 }
5035 }
5036
print_objc_property_list64(uint64_t p,struct DisassembleInfo * info)5037 static void print_objc_property_list64(uint64_t p,
5038 struct DisassembleInfo *info) {
5039 struct objc_property_list64 opl;
5040 struct objc_property64 op;
5041 const char *r;
5042 uint32_t offset, xoffset, left, j;
5043 SectionRef S, xS;
5044 const char *name, *sym_name;
5045 uint64_t n_value;
5046
5047 r = get_pointer_64(p, offset, left, S, info);
5048 if (r == nullptr)
5049 return;
5050 memset(&opl, '\0', sizeof(struct objc_property_list64));
5051 if (left < sizeof(struct objc_property_list64)) {
5052 memcpy(&opl, r, left);
5053 outs() << " (objc_property_list entends past the end of the section)\n";
5054 } else
5055 memcpy(&opl, r, sizeof(struct objc_property_list64));
5056 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5057 swapStruct(opl);
5058 outs() << " entsize " << opl.entsize << "\n";
5059 outs() << " count " << opl.count << "\n";
5060
5061 p += sizeof(struct objc_property_list64);
5062 offset += sizeof(struct objc_property_list64);
5063 for (j = 0; j < opl.count; j++) {
5064 r = get_pointer_64(p, offset, left, S, info);
5065 if (r == nullptr)
5066 return;
5067 memset(&op, '\0', sizeof(struct objc_property64));
5068 if (left < sizeof(struct objc_property64)) {
5069 memcpy(&op, r, left);
5070 outs() << " (objc_property entends past the end of the section)\n";
5071 } else
5072 memcpy(&op, r, sizeof(struct objc_property64));
5073 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5074 swapStruct(op);
5075
5076 outs() << "\t\t\t name ";
5077 sym_name = get_symbol_64(offset + offsetof(struct objc_property64, name), S,
5078 info, n_value, op.name);
5079 if (n_value != 0) {
5080 if (info->verbose && sym_name != nullptr)
5081 outs() << sym_name;
5082 else
5083 outs() << format("0x%" PRIx64, n_value);
5084 if (op.name != 0)
5085 outs() << " + " << format("0x%" PRIx64, op.name);
5086 } else
5087 outs() << format("0x%" PRIx64, op.name);
5088 name = get_pointer_64(op.name + n_value, xoffset, left, xS, info);
5089 if (name != nullptr)
5090 outs() << format(" %.*s", left, name);
5091 outs() << "\n";
5092
5093 outs() << "\t\t\tattributes ";
5094 sym_name =
5095 get_symbol_64(offset + offsetof(struct objc_property64, attributes), S,
5096 info, n_value, op.attributes);
5097 if (n_value != 0) {
5098 if (info->verbose && sym_name != nullptr)
5099 outs() << sym_name;
5100 else
5101 outs() << format("0x%" PRIx64, n_value);
5102 if (op.attributes != 0)
5103 outs() << " + " << format("0x%" PRIx64, op.attributes);
5104 } else
5105 outs() << format("0x%" PRIx64, op.attributes);
5106 name = get_pointer_64(op.attributes + n_value, xoffset, left, xS, info);
5107 if (name != nullptr)
5108 outs() << format(" %.*s", left, name);
5109 outs() << "\n";
5110
5111 p += sizeof(struct objc_property64);
5112 offset += sizeof(struct objc_property64);
5113 }
5114 }
5115
print_objc_property_list32(uint32_t p,struct DisassembleInfo * info)5116 static void print_objc_property_list32(uint32_t p,
5117 struct DisassembleInfo *info) {
5118 struct objc_property_list32 opl;
5119 struct objc_property32 op;
5120 const char *r;
5121 uint32_t offset, xoffset, left, j;
5122 SectionRef S, xS;
5123 const char *name;
5124
5125 r = get_pointer_32(p, offset, left, S, info);
5126 if (r == nullptr)
5127 return;
5128 memset(&opl, '\0', sizeof(struct objc_property_list32));
5129 if (left < sizeof(struct objc_property_list32)) {
5130 memcpy(&opl, r, left);
5131 outs() << " (objc_property_list entends past the end of the section)\n";
5132 } else
5133 memcpy(&opl, r, sizeof(struct objc_property_list32));
5134 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5135 swapStruct(opl);
5136 outs() << " entsize " << opl.entsize << "\n";
5137 outs() << " count " << opl.count << "\n";
5138
5139 p += sizeof(struct objc_property_list32);
5140 offset += sizeof(struct objc_property_list32);
5141 for (j = 0; j < opl.count; j++) {
5142 r = get_pointer_32(p, offset, left, S, info);
5143 if (r == nullptr)
5144 return;
5145 memset(&op, '\0', sizeof(struct objc_property32));
5146 if (left < sizeof(struct objc_property32)) {
5147 memcpy(&op, r, left);
5148 outs() << " (objc_property entends past the end of the section)\n";
5149 } else
5150 memcpy(&op, r, sizeof(struct objc_property32));
5151 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5152 swapStruct(op);
5153
5154 outs() << "\t\t\t name " << format("0x%" PRIx32, op.name);
5155 name = get_pointer_32(op.name, xoffset, left, xS, info);
5156 if (name != nullptr)
5157 outs() << format(" %.*s", left, name);
5158 outs() << "\n";
5159
5160 outs() << "\t\t\tattributes " << format("0x%" PRIx32, op.attributes);
5161 name = get_pointer_32(op.attributes, xoffset, left, xS, info);
5162 if (name != nullptr)
5163 outs() << format(" %.*s", left, name);
5164 outs() << "\n";
5165
5166 p += sizeof(struct objc_property32);
5167 offset += sizeof(struct objc_property32);
5168 }
5169 }
5170
print_class_ro64_t(uint64_t p,struct DisassembleInfo * info,bool & is_meta_class)5171 static bool print_class_ro64_t(uint64_t p, struct DisassembleInfo *info,
5172 bool &is_meta_class) {
5173 struct class_ro64_t cro;
5174 const char *r;
5175 uint32_t offset, xoffset, left;
5176 SectionRef S, xS;
5177 const char *name, *sym_name;
5178 uint64_t n_value;
5179
5180 r = get_pointer_64(p, offset, left, S, info);
5181 if (r == nullptr || left < sizeof(struct class_ro64_t))
5182 return false;
5183 memcpy(&cro, r, sizeof(struct class_ro64_t));
5184 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5185 swapStruct(cro);
5186 outs() << " flags " << format("0x%" PRIx32, cro.flags);
5187 if (cro.flags & RO_META)
5188 outs() << " RO_META";
5189 if (cro.flags & RO_ROOT)
5190 outs() << " RO_ROOT";
5191 if (cro.flags & RO_HAS_CXX_STRUCTORS)
5192 outs() << " RO_HAS_CXX_STRUCTORS";
5193 outs() << "\n";
5194 outs() << " instanceStart " << cro.instanceStart << "\n";
5195 outs() << " instanceSize " << cro.instanceSize << "\n";
5196 outs() << " reserved " << format("0x%" PRIx32, cro.reserved)
5197 << "\n";
5198 outs() << " ivarLayout " << format("0x%" PRIx64, cro.ivarLayout)
5199 << "\n";
5200 print_layout_map64(cro.ivarLayout, info);
5201
5202 outs() << " name ";
5203 sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, name), S,
5204 info, n_value, cro.name);
5205 if (n_value != 0) {
5206 if (info->verbose && sym_name != nullptr)
5207 outs() << sym_name;
5208 else
5209 outs() << format("0x%" PRIx64, n_value);
5210 if (cro.name != 0)
5211 outs() << " + " << format("0x%" PRIx64, cro.name);
5212 } else
5213 outs() << format("0x%" PRIx64, cro.name);
5214 name = get_pointer_64(cro.name + n_value, xoffset, left, xS, info);
5215 if (name != nullptr)
5216 outs() << format(" %.*s", left, name);
5217 outs() << "\n";
5218
5219 outs() << " baseMethods ";
5220 sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, baseMethods),
5221 S, info, n_value, cro.baseMethods);
5222 if (n_value != 0) {
5223 if (info->verbose && sym_name != nullptr)
5224 outs() << sym_name;
5225 else
5226 outs() << format("0x%" PRIx64, n_value);
5227 if (cro.baseMethods != 0)
5228 outs() << " + " << format("0x%" PRIx64, cro.baseMethods);
5229 } else
5230 outs() << format("0x%" PRIx64, cro.baseMethods);
5231 outs() << " (struct method_list_t *)\n";
5232 if (cro.baseMethods + n_value != 0)
5233 print_method_list64_t(cro.baseMethods + n_value, info, "");
5234
5235 outs() << " baseProtocols ";
5236 sym_name =
5237 get_symbol_64(offset + offsetof(struct class_ro64_t, baseProtocols), S,
5238 info, n_value, cro.baseProtocols);
5239 if (n_value != 0) {
5240 if (info->verbose && sym_name != nullptr)
5241 outs() << sym_name;
5242 else
5243 outs() << format("0x%" PRIx64, n_value);
5244 if (cro.baseProtocols != 0)
5245 outs() << " + " << format("0x%" PRIx64, cro.baseProtocols);
5246 } else
5247 outs() << format("0x%" PRIx64, cro.baseProtocols);
5248 outs() << "\n";
5249 if (cro.baseProtocols + n_value != 0)
5250 print_protocol_list64_t(cro.baseProtocols + n_value, info);
5251
5252 outs() << " ivars ";
5253 sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, ivars), S,
5254 info, n_value, cro.ivars);
5255 if (n_value != 0) {
5256 if (info->verbose && sym_name != nullptr)
5257 outs() << sym_name;
5258 else
5259 outs() << format("0x%" PRIx64, n_value);
5260 if (cro.ivars != 0)
5261 outs() << " + " << format("0x%" PRIx64, cro.ivars);
5262 } else
5263 outs() << format("0x%" PRIx64, cro.ivars);
5264 outs() << "\n";
5265 if (cro.ivars + n_value != 0)
5266 print_ivar_list64_t(cro.ivars + n_value, info);
5267
5268 outs() << " weakIvarLayout ";
5269 sym_name =
5270 get_symbol_64(offset + offsetof(struct class_ro64_t, weakIvarLayout), S,
5271 info, n_value, cro.weakIvarLayout);
5272 if (n_value != 0) {
5273 if (info->verbose && sym_name != nullptr)
5274 outs() << sym_name;
5275 else
5276 outs() << format("0x%" PRIx64, n_value);
5277 if (cro.weakIvarLayout != 0)
5278 outs() << " + " << format("0x%" PRIx64, cro.weakIvarLayout);
5279 } else
5280 outs() << format("0x%" PRIx64, cro.weakIvarLayout);
5281 outs() << "\n";
5282 print_layout_map64(cro.weakIvarLayout + n_value, info);
5283
5284 outs() << " baseProperties ";
5285 sym_name =
5286 get_symbol_64(offset + offsetof(struct class_ro64_t, baseProperties), S,
5287 info, n_value, cro.baseProperties);
5288 if (n_value != 0) {
5289 if (info->verbose && sym_name != nullptr)
5290 outs() << sym_name;
5291 else
5292 outs() << format("0x%" PRIx64, n_value);
5293 if (cro.baseProperties != 0)
5294 outs() << " + " << format("0x%" PRIx64, cro.baseProperties);
5295 } else
5296 outs() << format("0x%" PRIx64, cro.baseProperties);
5297 outs() << "\n";
5298 if (cro.baseProperties + n_value != 0)
5299 print_objc_property_list64(cro.baseProperties + n_value, info);
5300
5301 is_meta_class = (cro.flags & RO_META) != 0;
5302 return true;
5303 }
5304
print_class_ro32_t(uint32_t p,struct DisassembleInfo * info,bool & is_meta_class)5305 static bool print_class_ro32_t(uint32_t p, struct DisassembleInfo *info,
5306 bool &is_meta_class) {
5307 struct class_ro32_t cro;
5308 const char *r;
5309 uint32_t offset, xoffset, left;
5310 SectionRef S, xS;
5311 const char *name;
5312
5313 r = get_pointer_32(p, offset, left, S, info);
5314 if (r == nullptr)
5315 return false;
5316 memset(&cro, '\0', sizeof(struct class_ro32_t));
5317 if (left < sizeof(struct class_ro32_t)) {
5318 memcpy(&cro, r, left);
5319 outs() << " (class_ro_t entends past the end of the section)\n";
5320 } else
5321 memcpy(&cro, r, sizeof(struct class_ro32_t));
5322 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5323 swapStruct(cro);
5324 outs() << " flags " << format("0x%" PRIx32, cro.flags);
5325 if (cro.flags & RO_META)
5326 outs() << " RO_META";
5327 if (cro.flags & RO_ROOT)
5328 outs() << " RO_ROOT";
5329 if (cro.flags & RO_HAS_CXX_STRUCTORS)
5330 outs() << " RO_HAS_CXX_STRUCTORS";
5331 outs() << "\n";
5332 outs() << " instanceStart " << cro.instanceStart << "\n";
5333 outs() << " instanceSize " << cro.instanceSize << "\n";
5334 outs() << " ivarLayout " << format("0x%" PRIx32, cro.ivarLayout)
5335 << "\n";
5336 print_layout_map32(cro.ivarLayout, info);
5337
5338 outs() << " name " << format("0x%" PRIx32, cro.name);
5339 name = get_pointer_32(cro.name, xoffset, left, xS, info);
5340 if (name != nullptr)
5341 outs() << format(" %.*s", left, name);
5342 outs() << "\n";
5343
5344 outs() << " baseMethods "
5345 << format("0x%" PRIx32, cro.baseMethods)
5346 << " (struct method_list_t *)\n";
5347 if (cro.baseMethods != 0)
5348 print_method_list32_t(cro.baseMethods, info, "");
5349
5350 outs() << " baseProtocols "
5351 << format("0x%" PRIx32, cro.baseProtocols) << "\n";
5352 if (cro.baseProtocols != 0)
5353 print_protocol_list32_t(cro.baseProtocols, info);
5354 outs() << " ivars " << format("0x%" PRIx32, cro.ivars)
5355 << "\n";
5356 if (cro.ivars != 0)
5357 print_ivar_list32_t(cro.ivars, info);
5358 outs() << " weakIvarLayout "
5359 << format("0x%" PRIx32, cro.weakIvarLayout) << "\n";
5360 print_layout_map32(cro.weakIvarLayout, info);
5361 outs() << " baseProperties "
5362 << format("0x%" PRIx32, cro.baseProperties) << "\n";
5363 if (cro.baseProperties != 0)
5364 print_objc_property_list32(cro.baseProperties, info);
5365 is_meta_class = (cro.flags & RO_META) != 0;
5366 return true;
5367 }
5368
print_class64_t(uint64_t p,struct DisassembleInfo * info)5369 static void print_class64_t(uint64_t p, struct DisassembleInfo *info) {
5370 struct class64_t c;
5371 const char *r;
5372 uint32_t offset, left;
5373 SectionRef S;
5374 const char *name;
5375 uint64_t isa_n_value, n_value;
5376
5377 r = get_pointer_64(p, offset, left, S, info);
5378 if (r == nullptr || left < sizeof(struct class64_t))
5379 return;
5380 memcpy(&c, r, sizeof(struct class64_t));
5381 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5382 swapStruct(c);
5383
5384 outs() << " isa " << format("0x%" PRIx64, c.isa);
5385 name = get_symbol_64(offset + offsetof(struct class64_t, isa), S, info,
5386 isa_n_value, c.isa);
5387 if (name != nullptr)
5388 outs() << " " << name;
5389 outs() << "\n";
5390
5391 outs() << " superclass " << format("0x%" PRIx64, c.superclass);
5392 name = get_symbol_64(offset + offsetof(struct class64_t, superclass), S, info,
5393 n_value, c.superclass);
5394 if (name != nullptr)
5395 outs() << " " << name;
5396 else {
5397 name = get_dyld_bind_info_symbolname(S.getAddress() +
5398 offset + offsetof(struct class64_t, superclass), info);
5399 if (name != nullptr)
5400 outs() << " " << name;
5401 }
5402 outs() << "\n";
5403
5404 outs() << " cache " << format("0x%" PRIx64, c.cache);
5405 name = get_symbol_64(offset + offsetof(struct class64_t, cache), S, info,
5406 n_value, c.cache);
5407 if (name != nullptr)
5408 outs() << " " << name;
5409 outs() << "\n";
5410
5411 outs() << " vtable " << format("0x%" PRIx64, c.vtable);
5412 name = get_symbol_64(offset + offsetof(struct class64_t, vtable), S, info,
5413 n_value, c.vtable);
5414 if (name != nullptr)
5415 outs() << " " << name;
5416 outs() << "\n";
5417
5418 name = get_symbol_64(offset + offsetof(struct class64_t, data), S, info,
5419 n_value, c.data);
5420 outs() << " data ";
5421 if (n_value != 0) {
5422 if (info->verbose && name != nullptr)
5423 outs() << name;
5424 else
5425 outs() << format("0x%" PRIx64, n_value);
5426 if (c.data != 0)
5427 outs() << " + " << format("0x%" PRIx64, c.data);
5428 } else
5429 outs() << format("0x%" PRIx64, c.data);
5430 outs() << " (struct class_ro_t *)";
5431
5432 // This is a Swift class if some of the low bits of the pointer are set.
5433 if ((c.data + n_value) & 0x7)
5434 outs() << " Swift class";
5435 outs() << "\n";
5436 bool is_meta_class;
5437 if (!print_class_ro64_t((c.data + n_value) & ~0x7, info, is_meta_class))
5438 return;
5439
5440 if (!is_meta_class &&
5441 c.isa + isa_n_value != p &&
5442 c.isa + isa_n_value != 0 &&
5443 info->depth < 100) {
5444 info->depth++;
5445 outs() << "Meta Class\n";
5446 print_class64_t(c.isa + isa_n_value, info);
5447 }
5448 }
5449
print_class32_t(uint32_t p,struct DisassembleInfo * info)5450 static void print_class32_t(uint32_t p, struct DisassembleInfo *info) {
5451 struct class32_t c;
5452 const char *r;
5453 uint32_t offset, left;
5454 SectionRef S;
5455 const char *name;
5456
5457 r = get_pointer_32(p, offset, left, S, info);
5458 if (r == nullptr)
5459 return;
5460 memset(&c, '\0', sizeof(struct class32_t));
5461 if (left < sizeof(struct class32_t)) {
5462 memcpy(&c, r, left);
5463 outs() << " (class_t entends past the end of the section)\n";
5464 } else
5465 memcpy(&c, r, sizeof(struct class32_t));
5466 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5467 swapStruct(c);
5468
5469 outs() << " isa " << format("0x%" PRIx32, c.isa);
5470 name =
5471 get_symbol_32(offset + offsetof(struct class32_t, isa), S, info, c.isa);
5472 if (name != nullptr)
5473 outs() << " " << name;
5474 outs() << "\n";
5475
5476 outs() << " superclass " << format("0x%" PRIx32, c.superclass);
5477 name = get_symbol_32(offset + offsetof(struct class32_t, superclass), S, info,
5478 c.superclass);
5479 if (name != nullptr)
5480 outs() << " " << name;
5481 outs() << "\n";
5482
5483 outs() << " cache " << format("0x%" PRIx32, c.cache);
5484 name = get_symbol_32(offset + offsetof(struct class32_t, cache), S, info,
5485 c.cache);
5486 if (name != nullptr)
5487 outs() << " " << name;
5488 outs() << "\n";
5489
5490 outs() << " vtable " << format("0x%" PRIx32, c.vtable);
5491 name = get_symbol_32(offset + offsetof(struct class32_t, vtable), S, info,
5492 c.vtable);
5493 if (name != nullptr)
5494 outs() << " " << name;
5495 outs() << "\n";
5496
5497 name =
5498 get_symbol_32(offset + offsetof(struct class32_t, data), S, info, c.data);
5499 outs() << " data " << format("0x%" PRIx32, c.data)
5500 << " (struct class_ro_t *)";
5501
5502 // This is a Swift class if some of the low bits of the pointer are set.
5503 if (c.data & 0x3)
5504 outs() << " Swift class";
5505 outs() << "\n";
5506 bool is_meta_class;
5507 if (!print_class_ro32_t(c.data & ~0x3, info, is_meta_class))
5508 return;
5509
5510 if (!is_meta_class) {
5511 outs() << "Meta Class\n";
5512 print_class32_t(c.isa, info);
5513 }
5514 }
5515
print_objc_class_t(struct objc_class_t * objc_class,struct DisassembleInfo * info)5516 static void print_objc_class_t(struct objc_class_t *objc_class,
5517 struct DisassembleInfo *info) {
5518 uint32_t offset, left, xleft;
5519 const char *name, *p, *ivar_list;
5520 SectionRef S;
5521 int32_t i;
5522 struct objc_ivar_list_t objc_ivar_list;
5523 struct objc_ivar_t ivar;
5524
5525 outs() << "\t\t isa " << format("0x%08" PRIx32, objc_class->isa);
5526 if (info->verbose && CLS_GETINFO(objc_class, CLS_META)) {
5527 name = get_pointer_32(objc_class->isa, offset, left, S, info, true);
5528 if (name != nullptr)
5529 outs() << format(" %.*s", left, name);
5530 else
5531 outs() << " (not in an __OBJC section)";
5532 }
5533 outs() << "\n";
5534
5535 outs() << "\t super_class "
5536 << format("0x%08" PRIx32, objc_class->super_class);
5537 if (info->verbose) {
5538 name = get_pointer_32(objc_class->super_class, offset, left, S, info, true);
5539 if (name != nullptr)
5540 outs() << format(" %.*s", left, name);
5541 else
5542 outs() << " (not in an __OBJC section)";
5543 }
5544 outs() << "\n";
5545
5546 outs() << "\t\t name " << format("0x%08" PRIx32, objc_class->name);
5547 if (info->verbose) {
5548 name = get_pointer_32(objc_class->name, offset, left, S, info, true);
5549 if (name != nullptr)
5550 outs() << format(" %.*s", left, name);
5551 else
5552 outs() << " (not in an __OBJC section)";
5553 }
5554 outs() << "\n";
5555
5556 outs() << "\t\t version " << format("0x%08" PRIx32, objc_class->version)
5557 << "\n";
5558
5559 outs() << "\t\t info " << format("0x%08" PRIx32, objc_class->info);
5560 if (info->verbose) {
5561 if (CLS_GETINFO(objc_class, CLS_CLASS))
5562 outs() << " CLS_CLASS";
5563 else if (CLS_GETINFO(objc_class, CLS_META))
5564 outs() << " CLS_META";
5565 }
5566 outs() << "\n";
5567
5568 outs() << "\t instance_size "
5569 << format("0x%08" PRIx32, objc_class->instance_size) << "\n";
5570
5571 p = get_pointer_32(objc_class->ivars, offset, left, S, info, true);
5572 outs() << "\t\t ivars " << format("0x%08" PRIx32, objc_class->ivars);
5573 if (p != nullptr) {
5574 if (left > sizeof(struct objc_ivar_list_t)) {
5575 outs() << "\n";
5576 memcpy(&objc_ivar_list, p, sizeof(struct objc_ivar_list_t));
5577 } else {
5578 outs() << " (entends past the end of the section)\n";
5579 memset(&objc_ivar_list, '\0', sizeof(struct objc_ivar_list_t));
5580 memcpy(&objc_ivar_list, p, left);
5581 }
5582 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5583 swapStruct(objc_ivar_list);
5584 outs() << "\t\t ivar_count " << objc_ivar_list.ivar_count << "\n";
5585 ivar_list = p + sizeof(struct objc_ivar_list_t);
5586 for (i = 0; i < objc_ivar_list.ivar_count; i++) {
5587 if ((i + 1) * sizeof(struct objc_ivar_t) > left) {
5588 outs() << "\t\t remaining ivar's extend past the of the section\n";
5589 break;
5590 }
5591 memcpy(&ivar, ivar_list + i * sizeof(struct objc_ivar_t),
5592 sizeof(struct objc_ivar_t));
5593 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5594 swapStruct(ivar);
5595
5596 outs() << "\t\t\tivar_name " << format("0x%08" PRIx32, ivar.ivar_name);
5597 if (info->verbose) {
5598 name = get_pointer_32(ivar.ivar_name, offset, xleft, S, info, true);
5599 if (name != nullptr)
5600 outs() << format(" %.*s", xleft, name);
5601 else
5602 outs() << " (not in an __OBJC section)";
5603 }
5604 outs() << "\n";
5605
5606 outs() << "\t\t\tivar_type " << format("0x%08" PRIx32, ivar.ivar_type);
5607 if (info->verbose) {
5608 name = get_pointer_32(ivar.ivar_type, offset, xleft, S, info, true);
5609 if (name != nullptr)
5610 outs() << format(" %.*s", xleft, name);
5611 else
5612 outs() << " (not in an __OBJC section)";
5613 }
5614 outs() << "\n";
5615
5616 outs() << "\t\t ivar_offset "
5617 << format("0x%08" PRIx32, ivar.ivar_offset) << "\n";
5618 }
5619 } else {
5620 outs() << " (not in an __OBJC section)\n";
5621 }
5622
5623 outs() << "\t\t methods " << format("0x%08" PRIx32, objc_class->methodLists);
5624 if (print_method_list(objc_class->methodLists, info))
5625 outs() << " (not in an __OBJC section)\n";
5626
5627 outs() << "\t\t cache " << format("0x%08" PRIx32, objc_class->cache)
5628 << "\n";
5629
5630 outs() << "\t\tprotocols " << format("0x%08" PRIx32, objc_class->protocols);
5631 if (print_protocol_list(objc_class->protocols, 16, info))
5632 outs() << " (not in an __OBJC section)\n";
5633 }
5634
print_objc_objc_category_t(struct objc_category_t * objc_category,struct DisassembleInfo * info)5635 static void print_objc_objc_category_t(struct objc_category_t *objc_category,
5636 struct DisassembleInfo *info) {
5637 uint32_t offset, left;
5638 const char *name;
5639 SectionRef S;
5640
5641 outs() << "\t category name "
5642 << format("0x%08" PRIx32, objc_category->category_name);
5643 if (info->verbose) {
5644 name = get_pointer_32(objc_category->category_name, offset, left, S, info,
5645 true);
5646 if (name != nullptr)
5647 outs() << format(" %.*s", left, name);
5648 else
5649 outs() << " (not in an __OBJC section)";
5650 }
5651 outs() << "\n";
5652
5653 outs() << "\t\t class name "
5654 << format("0x%08" PRIx32, objc_category->class_name);
5655 if (info->verbose) {
5656 name =
5657 get_pointer_32(objc_category->class_name, offset, left, S, info, true);
5658 if (name != nullptr)
5659 outs() << format(" %.*s", left, name);
5660 else
5661 outs() << " (not in an __OBJC section)";
5662 }
5663 outs() << "\n";
5664
5665 outs() << "\t instance methods "
5666 << format("0x%08" PRIx32, objc_category->instance_methods);
5667 if (print_method_list(objc_category->instance_methods, info))
5668 outs() << " (not in an __OBJC section)\n";
5669
5670 outs() << "\t class methods "
5671 << format("0x%08" PRIx32, objc_category->class_methods);
5672 if (print_method_list(objc_category->class_methods, info))
5673 outs() << " (not in an __OBJC section)\n";
5674 }
5675
print_category64_t(uint64_t p,struct DisassembleInfo * info)5676 static void print_category64_t(uint64_t p, struct DisassembleInfo *info) {
5677 struct category64_t c;
5678 const char *r;
5679 uint32_t offset, xoffset, left;
5680 SectionRef S, xS;
5681 const char *name, *sym_name;
5682 uint64_t n_value;
5683
5684 r = get_pointer_64(p, offset, left, S, info);
5685 if (r == nullptr)
5686 return;
5687 memset(&c, '\0', sizeof(struct category64_t));
5688 if (left < sizeof(struct category64_t)) {
5689 memcpy(&c, r, left);
5690 outs() << " (category_t entends past the end of the section)\n";
5691 } else
5692 memcpy(&c, r, sizeof(struct category64_t));
5693 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5694 swapStruct(c);
5695
5696 outs() << " name ";
5697 sym_name = get_symbol_64(offset + offsetof(struct category64_t, name), S,
5698 info, n_value, c.name);
5699 if (n_value != 0) {
5700 if (info->verbose && sym_name != nullptr)
5701 outs() << sym_name;
5702 else
5703 outs() << format("0x%" PRIx64, n_value);
5704 if (c.name != 0)
5705 outs() << " + " << format("0x%" PRIx64, c.name);
5706 } else
5707 outs() << format("0x%" PRIx64, c.name);
5708 name = get_pointer_64(c.name + n_value, xoffset, left, xS, info);
5709 if (name != nullptr)
5710 outs() << format(" %.*s", left, name);
5711 outs() << "\n";
5712
5713 outs() << " cls ";
5714 sym_name = get_symbol_64(offset + offsetof(struct category64_t, cls), S, info,
5715 n_value, c.cls);
5716 if (n_value != 0) {
5717 if (info->verbose && sym_name != nullptr)
5718 outs() << sym_name;
5719 else
5720 outs() << format("0x%" PRIx64, n_value);
5721 if (c.cls != 0)
5722 outs() << " + " << format("0x%" PRIx64, c.cls);
5723 } else
5724 outs() << format("0x%" PRIx64, c.cls);
5725 outs() << "\n";
5726 if (c.cls + n_value != 0)
5727 print_class64_t(c.cls + n_value, info);
5728
5729 outs() << " instanceMethods ";
5730 sym_name =
5731 get_symbol_64(offset + offsetof(struct category64_t, instanceMethods), S,
5732 info, n_value, c.instanceMethods);
5733 if (n_value != 0) {
5734 if (info->verbose && sym_name != nullptr)
5735 outs() << sym_name;
5736 else
5737 outs() << format("0x%" PRIx64, n_value);
5738 if (c.instanceMethods != 0)
5739 outs() << " + " << format("0x%" PRIx64, c.instanceMethods);
5740 } else
5741 outs() << format("0x%" PRIx64, c.instanceMethods);
5742 outs() << "\n";
5743 if (c.instanceMethods + n_value != 0)
5744 print_method_list64_t(c.instanceMethods + n_value, info, "");
5745
5746 outs() << " classMethods ";
5747 sym_name = get_symbol_64(offset + offsetof(struct category64_t, classMethods),
5748 S, info, n_value, c.classMethods);
5749 if (n_value != 0) {
5750 if (info->verbose && sym_name != nullptr)
5751 outs() << sym_name;
5752 else
5753 outs() << format("0x%" PRIx64, n_value);
5754 if (c.classMethods != 0)
5755 outs() << " + " << format("0x%" PRIx64, c.classMethods);
5756 } else
5757 outs() << format("0x%" PRIx64, c.classMethods);
5758 outs() << "\n";
5759 if (c.classMethods + n_value != 0)
5760 print_method_list64_t(c.classMethods + n_value, info, "");
5761
5762 outs() << " protocols ";
5763 sym_name = get_symbol_64(offset + offsetof(struct category64_t, protocols), S,
5764 info, n_value, c.protocols);
5765 if (n_value != 0) {
5766 if (info->verbose && sym_name != nullptr)
5767 outs() << sym_name;
5768 else
5769 outs() << format("0x%" PRIx64, n_value);
5770 if (c.protocols != 0)
5771 outs() << " + " << format("0x%" PRIx64, c.protocols);
5772 } else
5773 outs() << format("0x%" PRIx64, c.protocols);
5774 outs() << "\n";
5775 if (c.protocols + n_value != 0)
5776 print_protocol_list64_t(c.protocols + n_value, info);
5777
5778 outs() << "instanceProperties ";
5779 sym_name =
5780 get_symbol_64(offset + offsetof(struct category64_t, instanceProperties),
5781 S, info, n_value, c.instanceProperties);
5782 if (n_value != 0) {
5783 if (info->verbose && sym_name != nullptr)
5784 outs() << sym_name;
5785 else
5786 outs() << format("0x%" PRIx64, n_value);
5787 if (c.instanceProperties != 0)
5788 outs() << " + " << format("0x%" PRIx64, c.instanceProperties);
5789 } else
5790 outs() << format("0x%" PRIx64, c.instanceProperties);
5791 outs() << "\n";
5792 if (c.instanceProperties + n_value != 0)
5793 print_objc_property_list64(c.instanceProperties + n_value, info);
5794 }
5795
print_category32_t(uint32_t p,struct DisassembleInfo * info)5796 static void print_category32_t(uint32_t p, struct DisassembleInfo *info) {
5797 struct category32_t c;
5798 const char *r;
5799 uint32_t offset, left;
5800 SectionRef S, xS;
5801 const char *name;
5802
5803 r = get_pointer_32(p, offset, left, S, info);
5804 if (r == nullptr)
5805 return;
5806 memset(&c, '\0', sizeof(struct category32_t));
5807 if (left < sizeof(struct category32_t)) {
5808 memcpy(&c, r, left);
5809 outs() << " (category_t entends past the end of the section)\n";
5810 } else
5811 memcpy(&c, r, sizeof(struct category32_t));
5812 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5813 swapStruct(c);
5814
5815 outs() << " name " << format("0x%" PRIx32, c.name);
5816 name = get_symbol_32(offset + offsetof(struct category32_t, name), S, info,
5817 c.name);
5818 if (name)
5819 outs() << " " << name;
5820 outs() << "\n";
5821
5822 outs() << " cls " << format("0x%" PRIx32, c.cls) << "\n";
5823 if (c.cls != 0)
5824 print_class32_t(c.cls, info);
5825 outs() << " instanceMethods " << format("0x%" PRIx32, c.instanceMethods)
5826 << "\n";
5827 if (c.instanceMethods != 0)
5828 print_method_list32_t(c.instanceMethods, info, "");
5829 outs() << " classMethods " << format("0x%" PRIx32, c.classMethods)
5830 << "\n";
5831 if (c.classMethods != 0)
5832 print_method_list32_t(c.classMethods, info, "");
5833 outs() << " protocols " << format("0x%" PRIx32, c.protocols) << "\n";
5834 if (c.protocols != 0)
5835 print_protocol_list32_t(c.protocols, info);
5836 outs() << "instanceProperties " << format("0x%" PRIx32, c.instanceProperties)
5837 << "\n";
5838 if (c.instanceProperties != 0)
5839 print_objc_property_list32(c.instanceProperties, info);
5840 }
5841
print_message_refs64(SectionRef S,struct DisassembleInfo * info)5842 static void print_message_refs64(SectionRef S, struct DisassembleInfo *info) {
5843 uint32_t i, left, offset, xoffset;
5844 uint64_t p, n_value;
5845 struct message_ref64 mr;
5846 const char *name, *sym_name;
5847 const char *r;
5848 SectionRef xS;
5849
5850 if (S == SectionRef())
5851 return;
5852
5853 StringRef SectName;
5854 Expected<StringRef> SecNameOrErr = S.getName();
5855 if (SecNameOrErr)
5856 SectName = *SecNameOrErr;
5857 else
5858 consumeError(SecNameOrErr.takeError());
5859
5860 DataRefImpl Ref = S.getRawDataRefImpl();
5861 StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5862 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5863 offset = 0;
5864 for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) {
5865 p = S.getAddress() + i;
5866 r = get_pointer_64(p, offset, left, S, info);
5867 if (r == nullptr)
5868 return;
5869 memset(&mr, '\0', sizeof(struct message_ref64));
5870 if (left < sizeof(struct message_ref64)) {
5871 memcpy(&mr, r, left);
5872 outs() << " (message_ref entends past the end of the section)\n";
5873 } else
5874 memcpy(&mr, r, sizeof(struct message_ref64));
5875 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5876 swapStruct(mr);
5877
5878 outs() << " imp ";
5879 name = get_symbol_64(offset + offsetof(struct message_ref64, imp), S, info,
5880 n_value, mr.imp);
5881 if (n_value != 0) {
5882 outs() << format("0x%" PRIx64, n_value) << " ";
5883 if (mr.imp != 0)
5884 outs() << "+ " << format("0x%" PRIx64, mr.imp) << " ";
5885 } else
5886 outs() << format("0x%" PRIx64, mr.imp) << " ";
5887 if (name != nullptr)
5888 outs() << " " << name;
5889 outs() << "\n";
5890
5891 outs() << " sel ";
5892 sym_name = get_symbol_64(offset + offsetof(struct message_ref64, sel), S,
5893 info, n_value, mr.sel);
5894 if (n_value != 0) {
5895 if (info->verbose && sym_name != nullptr)
5896 outs() << sym_name;
5897 else
5898 outs() << format("0x%" PRIx64, n_value);
5899 if (mr.sel != 0)
5900 outs() << " + " << format("0x%" PRIx64, mr.sel);
5901 } else
5902 outs() << format("0x%" PRIx64, mr.sel);
5903 name = get_pointer_64(mr.sel + n_value, xoffset, left, xS, info);
5904 if (name != nullptr)
5905 outs() << format(" %.*s", left, name);
5906 outs() << "\n";
5907
5908 offset += sizeof(struct message_ref64);
5909 }
5910 }
5911
print_message_refs32(SectionRef S,struct DisassembleInfo * info)5912 static void print_message_refs32(SectionRef S, struct DisassembleInfo *info) {
5913 uint32_t i, left, offset, xoffset, p;
5914 struct message_ref32 mr;
5915 const char *name, *r;
5916 SectionRef xS;
5917
5918 if (S == SectionRef())
5919 return;
5920
5921 StringRef SectName;
5922 Expected<StringRef> SecNameOrErr = S.getName();
5923 if (SecNameOrErr)
5924 SectName = *SecNameOrErr;
5925 else
5926 consumeError(SecNameOrErr.takeError());
5927
5928 DataRefImpl Ref = S.getRawDataRefImpl();
5929 StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5930 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5931 offset = 0;
5932 for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) {
5933 p = S.getAddress() + i;
5934 r = get_pointer_32(p, offset, left, S, info);
5935 if (r == nullptr)
5936 return;
5937 memset(&mr, '\0', sizeof(struct message_ref32));
5938 if (left < sizeof(struct message_ref32)) {
5939 memcpy(&mr, r, left);
5940 outs() << " (message_ref entends past the end of the section)\n";
5941 } else
5942 memcpy(&mr, r, sizeof(struct message_ref32));
5943 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5944 swapStruct(mr);
5945
5946 outs() << " imp " << format("0x%" PRIx32, mr.imp);
5947 name = get_symbol_32(offset + offsetof(struct message_ref32, imp), S, info,
5948 mr.imp);
5949 if (name != nullptr)
5950 outs() << " " << name;
5951 outs() << "\n";
5952
5953 outs() << " sel " << format("0x%" PRIx32, mr.sel);
5954 name = get_pointer_32(mr.sel, xoffset, left, xS, info);
5955 if (name != nullptr)
5956 outs() << " " << name;
5957 outs() << "\n";
5958
5959 offset += sizeof(struct message_ref32);
5960 }
5961 }
5962
print_image_info64(SectionRef S,struct DisassembleInfo * info)5963 static void print_image_info64(SectionRef S, struct DisassembleInfo *info) {
5964 uint32_t left, offset, swift_version;
5965 uint64_t p;
5966 struct objc_image_info64 o;
5967 const char *r;
5968
5969 if (S == SectionRef())
5970 return;
5971
5972 StringRef SectName;
5973 Expected<StringRef> SecNameOrErr = S.getName();
5974 if (SecNameOrErr)
5975 SectName = *SecNameOrErr;
5976 else
5977 consumeError(SecNameOrErr.takeError());
5978
5979 DataRefImpl Ref = S.getRawDataRefImpl();
5980 StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5981 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5982 p = S.getAddress();
5983 r = get_pointer_64(p, offset, left, S, info);
5984 if (r == nullptr)
5985 return;
5986 memset(&o, '\0', sizeof(struct objc_image_info64));
5987 if (left < sizeof(struct objc_image_info64)) {
5988 memcpy(&o, r, left);
5989 outs() << " (objc_image_info entends past the end of the section)\n";
5990 } else
5991 memcpy(&o, r, sizeof(struct objc_image_info64));
5992 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5993 swapStruct(o);
5994 outs() << " version " << o.version << "\n";
5995 outs() << " flags " << format("0x%" PRIx32, o.flags);
5996 if (o.flags & OBJC_IMAGE_IS_REPLACEMENT)
5997 outs() << " OBJC_IMAGE_IS_REPLACEMENT";
5998 if (o.flags & OBJC_IMAGE_SUPPORTS_GC)
5999 outs() << " OBJC_IMAGE_SUPPORTS_GC";
6000 if (o.flags & OBJC_IMAGE_IS_SIMULATED)
6001 outs() << " OBJC_IMAGE_IS_SIMULATED";
6002 if (o.flags & OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES)
6003 outs() << " OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES";
6004 swift_version = (o.flags >> 8) & 0xff;
6005 if (swift_version != 0) {
6006 if (swift_version == 1)
6007 outs() << " Swift 1.0";
6008 else if (swift_version == 2)
6009 outs() << " Swift 1.1";
6010 else if(swift_version == 3)
6011 outs() << " Swift 2.0";
6012 else if(swift_version == 4)
6013 outs() << " Swift 3.0";
6014 else if(swift_version == 5)
6015 outs() << " Swift 4.0";
6016 else if(swift_version == 6)
6017 outs() << " Swift 4.1/Swift 4.2";
6018 else if(swift_version == 7)
6019 outs() << " Swift 5 or later";
6020 else
6021 outs() << " unknown future Swift version (" << swift_version << ")";
6022 }
6023 outs() << "\n";
6024 }
6025
print_image_info32(SectionRef S,struct DisassembleInfo * info)6026 static void print_image_info32(SectionRef S, struct DisassembleInfo *info) {
6027 uint32_t left, offset, swift_version, p;
6028 struct objc_image_info32 o;
6029 const char *r;
6030
6031 if (S == SectionRef())
6032 return;
6033
6034 StringRef SectName;
6035 Expected<StringRef> SecNameOrErr = S.getName();
6036 if (SecNameOrErr)
6037 SectName = *SecNameOrErr;
6038 else
6039 consumeError(SecNameOrErr.takeError());
6040
6041 DataRefImpl Ref = S.getRawDataRefImpl();
6042 StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
6043 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
6044 p = S.getAddress();
6045 r = get_pointer_32(p, offset, left, S, info);
6046 if (r == nullptr)
6047 return;
6048 memset(&o, '\0', sizeof(struct objc_image_info32));
6049 if (left < sizeof(struct objc_image_info32)) {
6050 memcpy(&o, r, left);
6051 outs() << " (objc_image_info entends past the end of the section)\n";
6052 } else
6053 memcpy(&o, r, sizeof(struct objc_image_info32));
6054 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
6055 swapStruct(o);
6056 outs() << " version " << o.version << "\n";
6057 outs() << " flags " << format("0x%" PRIx32, o.flags);
6058 if (o.flags & OBJC_IMAGE_IS_REPLACEMENT)
6059 outs() << " OBJC_IMAGE_IS_REPLACEMENT";
6060 if (o.flags & OBJC_IMAGE_SUPPORTS_GC)
6061 outs() << " OBJC_IMAGE_SUPPORTS_GC";
6062 swift_version = (o.flags >> 8) & 0xff;
6063 if (swift_version != 0) {
6064 if (swift_version == 1)
6065 outs() << " Swift 1.0";
6066 else if (swift_version == 2)
6067 outs() << " Swift 1.1";
6068 else if(swift_version == 3)
6069 outs() << " Swift 2.0";
6070 else if(swift_version == 4)
6071 outs() << " Swift 3.0";
6072 else if(swift_version == 5)
6073 outs() << " Swift 4.0";
6074 else if(swift_version == 6)
6075 outs() << " Swift 4.1/Swift 4.2";
6076 else if(swift_version == 7)
6077 outs() << " Swift 5 or later";
6078 else
6079 outs() << " unknown future Swift version (" << swift_version << ")";
6080 }
6081 outs() << "\n";
6082 }
6083
print_image_info(SectionRef S,struct DisassembleInfo * info)6084 static void print_image_info(SectionRef S, struct DisassembleInfo *info) {
6085 uint32_t left, offset, p;
6086 struct imageInfo_t o;
6087 const char *r;
6088
6089 StringRef SectName;
6090 Expected<StringRef> SecNameOrErr = S.getName();
6091 if (SecNameOrErr)
6092 SectName = *SecNameOrErr;
6093 else
6094 consumeError(SecNameOrErr.takeError());
6095
6096 DataRefImpl Ref = S.getRawDataRefImpl();
6097 StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
6098 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
6099 p = S.getAddress();
6100 r = get_pointer_32(p, offset, left, S, info);
6101 if (r == nullptr)
6102 return;
6103 memset(&o, '\0', sizeof(struct imageInfo_t));
6104 if (left < sizeof(struct imageInfo_t)) {
6105 memcpy(&o, r, left);
6106 outs() << " (imageInfo entends past the end of the section)\n";
6107 } else
6108 memcpy(&o, r, sizeof(struct imageInfo_t));
6109 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
6110 swapStruct(o);
6111 outs() << " version " << o.version << "\n";
6112 outs() << " flags " << format("0x%" PRIx32, o.flags);
6113 if (o.flags & 0x1)
6114 outs() << " F&C";
6115 if (o.flags & 0x2)
6116 outs() << " GC";
6117 if (o.flags & 0x4)
6118 outs() << " GC-only";
6119 else
6120 outs() << " RR";
6121 outs() << "\n";
6122 }
6123
printObjc2_64bit_MetaData(MachOObjectFile * O,bool verbose)6124 static void printObjc2_64bit_MetaData(MachOObjectFile *O, bool verbose) {
6125 SymbolAddressMap AddrMap;
6126 if (verbose)
6127 CreateSymbolAddressMap(O, &AddrMap);
6128
6129 std::vector<SectionRef> Sections;
6130 append_range(Sections, O->sections());
6131
6132 struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6133
6134 SectionRef CL = get_section(O, "__OBJC2", "__class_list");
6135 if (CL == SectionRef())
6136 CL = get_section(O, "__DATA", "__objc_classlist");
6137 if (CL == SectionRef())
6138 CL = get_section(O, "__DATA_CONST", "__objc_classlist");
6139 if (CL == SectionRef())
6140 CL = get_section(O, "__DATA_DIRTY", "__objc_classlist");
6141 info.S = CL;
6142 walk_pointer_list_64("class", CL, O, &info, print_class64_t);
6143
6144 SectionRef CR = get_section(O, "__OBJC2", "__class_refs");
6145 if (CR == SectionRef())
6146 CR = get_section(O, "__DATA", "__objc_classrefs");
6147 if (CR == SectionRef())
6148 CR = get_section(O, "__DATA_CONST", "__objc_classrefs");
6149 if (CR == SectionRef())
6150 CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs");
6151 info.S = CR;
6152 walk_pointer_list_64("class refs", CR, O, &info, nullptr);
6153
6154 SectionRef SR = get_section(O, "__OBJC2", "__super_refs");
6155 if (SR == SectionRef())
6156 SR = get_section(O, "__DATA", "__objc_superrefs");
6157 if (SR == SectionRef())
6158 SR = get_section(O, "__DATA_CONST", "__objc_superrefs");
6159 if (SR == SectionRef())
6160 SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs");
6161 info.S = SR;
6162 walk_pointer_list_64("super refs", SR, O, &info, nullptr);
6163
6164 SectionRef CA = get_section(O, "__OBJC2", "__category_list");
6165 if (CA == SectionRef())
6166 CA = get_section(O, "__DATA", "__objc_catlist");
6167 if (CA == SectionRef())
6168 CA = get_section(O, "__DATA_CONST", "__objc_catlist");
6169 if (CA == SectionRef())
6170 CA = get_section(O, "__DATA_DIRTY", "__objc_catlist");
6171 info.S = CA;
6172 walk_pointer_list_64("category", CA, O, &info, print_category64_t);
6173
6174 SectionRef PL = get_section(O, "__OBJC2", "__protocol_list");
6175 if (PL == SectionRef())
6176 PL = get_section(O, "__DATA", "__objc_protolist");
6177 if (PL == SectionRef())
6178 PL = get_section(O, "__DATA_CONST", "__objc_protolist");
6179 if (PL == SectionRef())
6180 PL = get_section(O, "__DATA_DIRTY", "__objc_protolist");
6181 info.S = PL;
6182 walk_pointer_list_64("protocol", PL, O, &info, nullptr);
6183
6184 SectionRef MR = get_section(O, "__OBJC2", "__message_refs");
6185 if (MR == SectionRef())
6186 MR = get_section(O, "__DATA", "__objc_msgrefs");
6187 if (MR == SectionRef())
6188 MR = get_section(O, "__DATA_CONST", "__objc_msgrefs");
6189 if (MR == SectionRef())
6190 MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs");
6191 info.S = MR;
6192 print_message_refs64(MR, &info);
6193
6194 SectionRef II = get_section(O, "__OBJC2", "__image_info");
6195 if (II == SectionRef())
6196 II = get_section(O, "__DATA", "__objc_imageinfo");
6197 if (II == SectionRef())
6198 II = get_section(O, "__DATA_CONST", "__objc_imageinfo");
6199 if (II == SectionRef())
6200 II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo");
6201 info.S = II;
6202 print_image_info64(II, &info);
6203 }
6204
printObjc2_32bit_MetaData(MachOObjectFile * O,bool verbose)6205 static void printObjc2_32bit_MetaData(MachOObjectFile *O, bool verbose) {
6206 SymbolAddressMap AddrMap;
6207 if (verbose)
6208 CreateSymbolAddressMap(O, &AddrMap);
6209
6210 std::vector<SectionRef> Sections;
6211 append_range(Sections, O->sections());
6212
6213 struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6214
6215 SectionRef CL = get_section(O, "__OBJC2", "__class_list");
6216 if (CL == SectionRef())
6217 CL = get_section(O, "__DATA", "__objc_classlist");
6218 if (CL == SectionRef())
6219 CL = get_section(O, "__DATA_CONST", "__objc_classlist");
6220 if (CL == SectionRef())
6221 CL = get_section(O, "__DATA_DIRTY", "__objc_classlist");
6222 info.S = CL;
6223 walk_pointer_list_32("class", CL, O, &info, print_class32_t);
6224
6225 SectionRef CR = get_section(O, "__OBJC2", "__class_refs");
6226 if (CR == SectionRef())
6227 CR = get_section(O, "__DATA", "__objc_classrefs");
6228 if (CR == SectionRef())
6229 CR = get_section(O, "__DATA_CONST", "__objc_classrefs");
6230 if (CR == SectionRef())
6231 CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs");
6232 info.S = CR;
6233 walk_pointer_list_32("class refs", CR, O, &info, nullptr);
6234
6235 SectionRef SR = get_section(O, "__OBJC2", "__super_refs");
6236 if (SR == SectionRef())
6237 SR = get_section(O, "__DATA", "__objc_superrefs");
6238 if (SR == SectionRef())
6239 SR = get_section(O, "__DATA_CONST", "__objc_superrefs");
6240 if (SR == SectionRef())
6241 SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs");
6242 info.S = SR;
6243 walk_pointer_list_32("super refs", SR, O, &info, nullptr);
6244
6245 SectionRef CA = get_section(O, "__OBJC2", "__category_list");
6246 if (CA == SectionRef())
6247 CA = get_section(O, "__DATA", "__objc_catlist");
6248 if (CA == SectionRef())
6249 CA = get_section(O, "__DATA_CONST", "__objc_catlist");
6250 if (CA == SectionRef())
6251 CA = get_section(O, "__DATA_DIRTY", "__objc_catlist");
6252 info.S = CA;
6253 walk_pointer_list_32("category", CA, O, &info, print_category32_t);
6254
6255 SectionRef PL = get_section(O, "__OBJC2", "__protocol_list");
6256 if (PL == SectionRef())
6257 PL = get_section(O, "__DATA", "__objc_protolist");
6258 if (PL == SectionRef())
6259 PL = get_section(O, "__DATA_CONST", "__objc_protolist");
6260 if (PL == SectionRef())
6261 PL = get_section(O, "__DATA_DIRTY", "__objc_protolist");
6262 info.S = PL;
6263 walk_pointer_list_32("protocol", PL, O, &info, nullptr);
6264
6265 SectionRef MR = get_section(O, "__OBJC2", "__message_refs");
6266 if (MR == SectionRef())
6267 MR = get_section(O, "__DATA", "__objc_msgrefs");
6268 if (MR == SectionRef())
6269 MR = get_section(O, "__DATA_CONST", "__objc_msgrefs");
6270 if (MR == SectionRef())
6271 MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs");
6272 info.S = MR;
6273 print_message_refs32(MR, &info);
6274
6275 SectionRef II = get_section(O, "__OBJC2", "__image_info");
6276 if (II == SectionRef())
6277 II = get_section(O, "__DATA", "__objc_imageinfo");
6278 if (II == SectionRef())
6279 II = get_section(O, "__DATA_CONST", "__objc_imageinfo");
6280 if (II == SectionRef())
6281 II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo");
6282 info.S = II;
6283 print_image_info32(II, &info);
6284 }
6285
printObjc1_32bit_MetaData(MachOObjectFile * O,bool verbose)6286 static bool printObjc1_32bit_MetaData(MachOObjectFile *O, bool verbose) {
6287 uint32_t i, j, p, offset, xoffset, left, defs_left, def;
6288 const char *r, *name, *defs;
6289 struct objc_module_t module;
6290 SectionRef S, xS;
6291 struct objc_symtab_t symtab;
6292 struct objc_class_t objc_class;
6293 struct objc_category_t objc_category;
6294
6295 outs() << "Objective-C segment\n";
6296 S = get_section(O, "__OBJC", "__module_info");
6297 if (S == SectionRef())
6298 return false;
6299
6300 SymbolAddressMap AddrMap;
6301 if (verbose)
6302 CreateSymbolAddressMap(O, &AddrMap);
6303
6304 std::vector<SectionRef> Sections;
6305 append_range(Sections, O->sections());
6306
6307 struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6308
6309 for (i = 0; i < S.getSize(); i += sizeof(struct objc_module_t)) {
6310 p = S.getAddress() + i;
6311 r = get_pointer_32(p, offset, left, S, &info, true);
6312 if (r == nullptr)
6313 return true;
6314 memset(&module, '\0', sizeof(struct objc_module_t));
6315 if (left < sizeof(struct objc_module_t)) {
6316 memcpy(&module, r, left);
6317 outs() << " (module extends past end of __module_info section)\n";
6318 } else
6319 memcpy(&module, r, sizeof(struct objc_module_t));
6320 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6321 swapStruct(module);
6322
6323 outs() << "Module " << format("0x%" PRIx32, p) << "\n";
6324 outs() << " version " << module.version << "\n";
6325 outs() << " size " << module.size << "\n";
6326 outs() << " name ";
6327 name = get_pointer_32(module.name, xoffset, left, xS, &info, true);
6328 if (name != nullptr)
6329 outs() << format("%.*s", left, name);
6330 else
6331 outs() << format("0x%08" PRIx32, module.name)
6332 << "(not in an __OBJC section)";
6333 outs() << "\n";
6334
6335 r = get_pointer_32(module.symtab, xoffset, left, xS, &info, true);
6336 if (module.symtab == 0 || r == nullptr) {
6337 outs() << " symtab " << format("0x%08" PRIx32, module.symtab)
6338 << " (not in an __OBJC section)\n";
6339 continue;
6340 }
6341 outs() << " symtab " << format("0x%08" PRIx32, module.symtab) << "\n";
6342 memset(&symtab, '\0', sizeof(struct objc_symtab_t));
6343 defs_left = 0;
6344 defs = nullptr;
6345 if (left < sizeof(struct objc_symtab_t)) {
6346 memcpy(&symtab, r, left);
6347 outs() << "\tsymtab extends past end of an __OBJC section)\n";
6348 } else {
6349 memcpy(&symtab, r, sizeof(struct objc_symtab_t));
6350 if (left > sizeof(struct objc_symtab_t)) {
6351 defs_left = left - sizeof(struct objc_symtab_t);
6352 defs = r + sizeof(struct objc_symtab_t);
6353 }
6354 }
6355 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6356 swapStruct(symtab);
6357
6358 outs() << "\tsel_ref_cnt " << symtab.sel_ref_cnt << "\n";
6359 r = get_pointer_32(symtab.refs, xoffset, left, xS, &info, true);
6360 outs() << "\trefs " << format("0x%08" PRIx32, symtab.refs);
6361 if (r == nullptr)
6362 outs() << " (not in an __OBJC section)";
6363 outs() << "\n";
6364 outs() << "\tcls_def_cnt " << symtab.cls_def_cnt << "\n";
6365 outs() << "\tcat_def_cnt " << symtab.cat_def_cnt << "\n";
6366 if (symtab.cls_def_cnt > 0)
6367 outs() << "\tClass Definitions\n";
6368 for (j = 0; j < symtab.cls_def_cnt; j++) {
6369 if ((j + 1) * sizeof(uint32_t) > defs_left) {
6370 outs() << "\t(remaining class defs entries entends past the end of the "
6371 << "section)\n";
6372 break;
6373 }
6374 memcpy(&def, defs + j * sizeof(uint32_t), sizeof(uint32_t));
6375 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6376 sys::swapByteOrder(def);
6377
6378 r = get_pointer_32(def, xoffset, left, xS, &info, true);
6379 outs() << "\tdefs[" << j << "] " << format("0x%08" PRIx32, def);
6380 if (r != nullptr) {
6381 if (left > sizeof(struct objc_class_t)) {
6382 outs() << "\n";
6383 memcpy(&objc_class, r, sizeof(struct objc_class_t));
6384 } else {
6385 outs() << " (entends past the end of the section)\n";
6386 memset(&objc_class, '\0', sizeof(struct objc_class_t));
6387 memcpy(&objc_class, r, left);
6388 }
6389 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6390 swapStruct(objc_class);
6391 print_objc_class_t(&objc_class, &info);
6392 } else {
6393 outs() << "(not in an __OBJC section)\n";
6394 }
6395
6396 if (CLS_GETINFO(&objc_class, CLS_CLASS)) {
6397 outs() << "\tMeta Class";
6398 r = get_pointer_32(objc_class.isa, xoffset, left, xS, &info, true);
6399 if (r != nullptr) {
6400 if (left > sizeof(struct objc_class_t)) {
6401 outs() << "\n";
6402 memcpy(&objc_class, r, sizeof(struct objc_class_t));
6403 } else {
6404 outs() << " (entends past the end of the section)\n";
6405 memset(&objc_class, '\0', sizeof(struct objc_class_t));
6406 memcpy(&objc_class, r, left);
6407 }
6408 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6409 swapStruct(objc_class);
6410 print_objc_class_t(&objc_class, &info);
6411 } else {
6412 outs() << "(not in an __OBJC section)\n";
6413 }
6414 }
6415 }
6416 if (symtab.cat_def_cnt > 0)
6417 outs() << "\tCategory Definitions\n";
6418 for (j = 0; j < symtab.cat_def_cnt; j++) {
6419 if ((j + symtab.cls_def_cnt + 1) * sizeof(uint32_t) > defs_left) {
6420 outs() << "\t(remaining category defs entries entends past the end of "
6421 << "the section)\n";
6422 break;
6423 }
6424 memcpy(&def, defs + (j + symtab.cls_def_cnt) * sizeof(uint32_t),
6425 sizeof(uint32_t));
6426 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6427 sys::swapByteOrder(def);
6428
6429 r = get_pointer_32(def, xoffset, left, xS, &info, true);
6430 outs() << "\tdefs[" << j + symtab.cls_def_cnt << "] "
6431 << format("0x%08" PRIx32, def);
6432 if (r != nullptr) {
6433 if (left > sizeof(struct objc_category_t)) {
6434 outs() << "\n";
6435 memcpy(&objc_category, r, sizeof(struct objc_category_t));
6436 } else {
6437 outs() << " (entends past the end of the section)\n";
6438 memset(&objc_category, '\0', sizeof(struct objc_category_t));
6439 memcpy(&objc_category, r, left);
6440 }
6441 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6442 swapStruct(objc_category);
6443 print_objc_objc_category_t(&objc_category, &info);
6444 } else {
6445 outs() << "(not in an __OBJC section)\n";
6446 }
6447 }
6448 }
6449 const SectionRef II = get_section(O, "__OBJC", "__image_info");
6450 if (II != SectionRef())
6451 print_image_info(II, &info);
6452
6453 return true;
6454 }
6455
DumpProtocolSection(MachOObjectFile * O,const char * sect,uint32_t size,uint32_t addr)6456 static void DumpProtocolSection(MachOObjectFile *O, const char *sect,
6457 uint32_t size, uint32_t addr) {
6458 SymbolAddressMap AddrMap;
6459 CreateSymbolAddressMap(O, &AddrMap);
6460
6461 std::vector<SectionRef> Sections;
6462 append_range(Sections, O->sections());
6463
6464 struct DisassembleInfo info(O, &AddrMap, &Sections, true);
6465
6466 const char *p;
6467 struct objc_protocol_t protocol;
6468 uint32_t left, paddr;
6469 for (p = sect; p < sect + size; p += sizeof(struct objc_protocol_t)) {
6470 memset(&protocol, '\0', sizeof(struct objc_protocol_t));
6471 left = size - (p - sect);
6472 if (left < sizeof(struct objc_protocol_t)) {
6473 outs() << "Protocol extends past end of __protocol section\n";
6474 memcpy(&protocol, p, left);
6475 } else
6476 memcpy(&protocol, p, sizeof(struct objc_protocol_t));
6477 if (O->isLittleEndian() != sys::IsLittleEndianHost)
6478 swapStruct(protocol);
6479 paddr = addr + (p - sect);
6480 outs() << "Protocol " << format("0x%" PRIx32, paddr);
6481 if (print_protocol(paddr, 0, &info))
6482 outs() << "(not in an __OBJC section)\n";
6483 }
6484 }
6485
6486 #ifdef LLVM_HAVE_LIBXAR
swapStruct(struct xar_header & xar)6487 static inline void swapStruct(struct xar_header &xar) {
6488 sys::swapByteOrder(xar.magic);
6489 sys::swapByteOrder(xar.size);
6490 sys::swapByteOrder(xar.version);
6491 sys::swapByteOrder(xar.toc_length_compressed);
6492 sys::swapByteOrder(xar.toc_length_uncompressed);
6493 sys::swapByteOrder(xar.cksum_alg);
6494 }
6495
PrintModeVerbose(uint32_t mode)6496 static void PrintModeVerbose(uint32_t mode) {
6497 switch(mode & S_IFMT){
6498 case S_IFDIR:
6499 outs() << "d";
6500 break;
6501 case S_IFCHR:
6502 outs() << "c";
6503 break;
6504 case S_IFBLK:
6505 outs() << "b";
6506 break;
6507 case S_IFREG:
6508 outs() << "-";
6509 break;
6510 case S_IFLNK:
6511 outs() << "l";
6512 break;
6513 case S_IFSOCK:
6514 outs() << "s";
6515 break;
6516 default:
6517 outs() << "?";
6518 break;
6519 }
6520
6521 /* owner permissions */
6522 if(mode & S_IREAD)
6523 outs() << "r";
6524 else
6525 outs() << "-";
6526 if(mode & S_IWRITE)
6527 outs() << "w";
6528 else
6529 outs() << "-";
6530 if(mode & S_ISUID)
6531 outs() << "s";
6532 else if(mode & S_IEXEC)
6533 outs() << "x";
6534 else
6535 outs() << "-";
6536
6537 /* group permissions */
6538 if(mode & (S_IREAD >> 3))
6539 outs() << "r";
6540 else
6541 outs() << "-";
6542 if(mode & (S_IWRITE >> 3))
6543 outs() << "w";
6544 else
6545 outs() << "-";
6546 if(mode & S_ISGID)
6547 outs() << "s";
6548 else if(mode & (S_IEXEC >> 3))
6549 outs() << "x";
6550 else
6551 outs() << "-";
6552
6553 /* other permissions */
6554 if(mode & (S_IREAD >> 6))
6555 outs() << "r";
6556 else
6557 outs() << "-";
6558 if(mode & (S_IWRITE >> 6))
6559 outs() << "w";
6560 else
6561 outs() << "-";
6562 if(mode & S_ISVTX)
6563 outs() << "t";
6564 else if(mode & (S_IEXEC >> 6))
6565 outs() << "x";
6566 else
6567 outs() << "-";
6568 }
6569
PrintXarFilesSummary(const char * XarFilename,xar_t xar)6570 static void PrintXarFilesSummary(const char *XarFilename, xar_t xar) {
6571 xar_file_t xf;
6572 const char *key, *type, *mode, *user, *group, *size, *mtime, *name, *m;
6573 char *endp;
6574 uint32_t mode_value;
6575
6576 ScopedXarIter xi;
6577 if (!xi) {
6578 WithColor::error(errs(), "llvm-objdump")
6579 << "can't obtain an xar iterator for xar archive " << XarFilename
6580 << "\n";
6581 return;
6582 }
6583
6584 // Go through the xar's files.
6585 for (xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)) {
6586 ScopedXarIter xp;
6587 if(!xp){
6588 WithColor::error(errs(), "llvm-objdump")
6589 << "can't obtain an xar iterator for xar archive " << XarFilename
6590 << "\n";
6591 return;
6592 }
6593 type = nullptr;
6594 mode = nullptr;
6595 user = nullptr;
6596 group = nullptr;
6597 size = nullptr;
6598 mtime = nullptr;
6599 name = nullptr;
6600 for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){
6601 const char *val = nullptr;
6602 xar_prop_get(xf, key, &val);
6603 #if 0 // Useful for debugging.
6604 outs() << "key: " << key << " value: " << val << "\n";
6605 #endif
6606 if(strcmp(key, "type") == 0)
6607 type = val;
6608 if(strcmp(key, "mode") == 0)
6609 mode = val;
6610 if(strcmp(key, "user") == 0)
6611 user = val;
6612 if(strcmp(key, "group") == 0)
6613 group = val;
6614 if(strcmp(key, "data/size") == 0)
6615 size = val;
6616 if(strcmp(key, "mtime") == 0)
6617 mtime = val;
6618 if(strcmp(key, "name") == 0)
6619 name = val;
6620 }
6621 if(mode != nullptr){
6622 mode_value = strtoul(mode, &endp, 8);
6623 if(*endp != '\0')
6624 outs() << "(mode: \"" << mode << "\" contains non-octal chars) ";
6625 if(strcmp(type, "file") == 0)
6626 mode_value |= S_IFREG;
6627 PrintModeVerbose(mode_value);
6628 outs() << " ";
6629 }
6630 if(user != nullptr)
6631 outs() << format("%10s/", user);
6632 if(group != nullptr)
6633 outs() << format("%-10s ", group);
6634 if(size != nullptr)
6635 outs() << format("%7s ", size);
6636 if(mtime != nullptr){
6637 for(m = mtime; *m != 'T' && *m != '\0'; m++)
6638 outs() << *m;
6639 if(*m == 'T')
6640 m++;
6641 outs() << " ";
6642 for( ; *m != 'Z' && *m != '\0'; m++)
6643 outs() << *m;
6644 outs() << " ";
6645 }
6646 if(name != nullptr)
6647 outs() << name;
6648 outs() << "\n";
6649 }
6650 }
6651
DumpBitcodeSection(MachOObjectFile * O,const char * sect,uint32_t size,bool verbose,bool PrintXarHeader,bool PrintXarFileHeaders,std::string XarMemberName)6652 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect,
6653 uint32_t size, bool verbose,
6654 bool PrintXarHeader, bool PrintXarFileHeaders,
6655 std::string XarMemberName) {
6656 if(size < sizeof(struct xar_header)) {
6657 outs() << "size of (__LLVM,__bundle) section too small (smaller than size "
6658 "of struct xar_header)\n";
6659 return;
6660 }
6661 struct xar_header XarHeader;
6662 memcpy(&XarHeader, sect, sizeof(struct xar_header));
6663 if (sys::IsLittleEndianHost)
6664 swapStruct(XarHeader);
6665 if (PrintXarHeader) {
6666 if (!XarMemberName.empty())
6667 outs() << "In xar member " << XarMemberName << ": ";
6668 else
6669 outs() << "For (__LLVM,__bundle) section: ";
6670 outs() << "xar header\n";
6671 if (XarHeader.magic == XAR_HEADER_MAGIC)
6672 outs() << " magic XAR_HEADER_MAGIC\n";
6673 else
6674 outs() << " magic "
6675 << format_hex(XarHeader.magic, 10, true)
6676 << " (not XAR_HEADER_MAGIC)\n";
6677 outs() << " size " << XarHeader.size << "\n";
6678 outs() << " version " << XarHeader.version << "\n";
6679 outs() << " toc_length_compressed " << XarHeader.toc_length_compressed
6680 << "\n";
6681 outs() << "toc_length_uncompressed " << XarHeader.toc_length_uncompressed
6682 << "\n";
6683 outs() << " cksum_alg ";
6684 switch (XarHeader.cksum_alg) {
6685 case XAR_CKSUM_NONE:
6686 outs() << "XAR_CKSUM_NONE\n";
6687 break;
6688 case XAR_CKSUM_SHA1:
6689 outs() << "XAR_CKSUM_SHA1\n";
6690 break;
6691 case XAR_CKSUM_MD5:
6692 outs() << "XAR_CKSUM_MD5\n";
6693 break;
6694 #ifdef XAR_CKSUM_SHA256
6695 case XAR_CKSUM_SHA256:
6696 outs() << "XAR_CKSUM_SHA256\n";
6697 break;
6698 #endif
6699 #ifdef XAR_CKSUM_SHA512
6700 case XAR_CKSUM_SHA512:
6701 outs() << "XAR_CKSUM_SHA512\n";
6702 break;
6703 #endif
6704 default:
6705 outs() << XarHeader.cksum_alg << "\n";
6706 }
6707 }
6708
6709 SmallString<128> XarFilename;
6710 int FD;
6711 std::error_code XarEC =
6712 sys::fs::createTemporaryFile("llvm-objdump", "xar", FD, XarFilename);
6713 if (XarEC) {
6714 WithColor::error(errs(), "llvm-objdump") << XarEC.message() << "\n";
6715 return;
6716 }
6717 ToolOutputFile XarFile(XarFilename, FD);
6718 raw_fd_ostream &XarOut = XarFile.os();
6719 StringRef XarContents(sect, size);
6720 XarOut << XarContents;
6721 XarOut.close();
6722 if (XarOut.has_error())
6723 return;
6724
6725 ScopedXarFile xar(XarFilename.c_str(), READ);
6726 if (!xar) {
6727 WithColor::error(errs(), "llvm-objdump")
6728 << "can't create temporary xar archive " << XarFilename << "\n";
6729 return;
6730 }
6731
6732 SmallString<128> TocFilename;
6733 std::error_code TocEC =
6734 sys::fs::createTemporaryFile("llvm-objdump", "toc", TocFilename);
6735 if (TocEC) {
6736 WithColor::error(errs(), "llvm-objdump") << TocEC.message() << "\n";
6737 return;
6738 }
6739 xar_serialize(xar, TocFilename.c_str());
6740
6741 if (PrintXarFileHeaders) {
6742 if (!XarMemberName.empty())
6743 outs() << "In xar member " << XarMemberName << ": ";
6744 else
6745 outs() << "For (__LLVM,__bundle) section: ";
6746 outs() << "xar archive files:\n";
6747 PrintXarFilesSummary(XarFilename.c_str(), xar);
6748 }
6749
6750 ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr =
6751 MemoryBuffer::getFileOrSTDIN(TocFilename.c_str());
6752 if (std::error_code EC = FileOrErr.getError()) {
6753 WithColor::error(errs(), "llvm-objdump") << EC.message() << "\n";
6754 return;
6755 }
6756 std::unique_ptr<MemoryBuffer> &Buffer = FileOrErr.get();
6757
6758 if (!XarMemberName.empty())
6759 outs() << "In xar member " << XarMemberName << ": ";
6760 else
6761 outs() << "For (__LLVM,__bundle) section: ";
6762 outs() << "xar table of contents:\n";
6763 outs() << Buffer->getBuffer() << "\n";
6764
6765 // TODO: Go through the xar's files.
6766 ScopedXarIter xi;
6767 if(!xi){
6768 WithColor::error(errs(), "llvm-objdump")
6769 << "can't obtain an xar iterator for xar archive "
6770 << XarFilename.c_str() << "\n";
6771 return;
6772 }
6773 for(xar_file_t xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)){
6774 const char *key;
6775 const char *member_name, *member_type, *member_size_string;
6776 size_t member_size;
6777
6778 ScopedXarIter xp;
6779 if(!xp){
6780 WithColor::error(errs(), "llvm-objdump")
6781 << "can't obtain an xar iterator for xar archive "
6782 << XarFilename.c_str() << "\n";
6783 return;
6784 }
6785 member_name = NULL;
6786 member_type = NULL;
6787 member_size_string = NULL;
6788 for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){
6789 const char *val = nullptr;
6790 xar_prop_get(xf, key, &val);
6791 #if 0 // Useful for debugging.
6792 outs() << "key: " << key << " value: " << val << "\n";
6793 #endif
6794 if (strcmp(key, "name") == 0)
6795 member_name = val;
6796 if (strcmp(key, "type") == 0)
6797 member_type = val;
6798 if (strcmp(key, "data/size") == 0)
6799 member_size_string = val;
6800 }
6801 /*
6802 * If we find a file with a name, date/size and type properties
6803 * and with the type being "file" see if that is a xar file.
6804 */
6805 if (member_name != NULL && member_type != NULL &&
6806 strcmp(member_type, "file") == 0 &&
6807 member_size_string != NULL){
6808 // Extract the file into a buffer.
6809 char *endptr;
6810 member_size = strtoul(member_size_string, &endptr, 10);
6811 if (*endptr == '\0' && member_size != 0) {
6812 char *buffer;
6813 if (xar_extract_tobuffersz(xar, xf, &buffer, &member_size) == 0) {
6814 #if 0 // Useful for debugging.
6815 outs() << "xar member: " << member_name << " extracted\n";
6816 #endif
6817 // Set the XarMemberName we want to see printed in the header.
6818 std::string OldXarMemberName;
6819 // If XarMemberName is already set this is nested. So
6820 // save the old name and create the nested name.
6821 if (!XarMemberName.empty()) {
6822 OldXarMemberName = XarMemberName;
6823 XarMemberName =
6824 (Twine("[") + XarMemberName + "]" + member_name).str();
6825 } else {
6826 OldXarMemberName = "";
6827 XarMemberName = member_name;
6828 }
6829 // See if this is could be a xar file (nested).
6830 if (member_size >= sizeof(struct xar_header)) {
6831 #if 0 // Useful for debugging.
6832 outs() << "could be a xar file: " << member_name << "\n";
6833 #endif
6834 memcpy((char *)&XarHeader, buffer, sizeof(struct xar_header));
6835 if (sys::IsLittleEndianHost)
6836 swapStruct(XarHeader);
6837 if (XarHeader.magic == XAR_HEADER_MAGIC)
6838 DumpBitcodeSection(O, buffer, member_size, verbose,
6839 PrintXarHeader, PrintXarFileHeaders,
6840 XarMemberName);
6841 }
6842 XarMemberName = OldXarMemberName;
6843 delete buffer;
6844 }
6845 }
6846 }
6847 }
6848 }
6849 #endif // defined(LLVM_HAVE_LIBXAR)
6850
printObjcMetaData(MachOObjectFile * O,bool verbose)6851 static void printObjcMetaData(MachOObjectFile *O, bool verbose) {
6852 if (O->is64Bit())
6853 printObjc2_64bit_MetaData(O, verbose);
6854 else {
6855 MachO::mach_header H;
6856 H = O->getHeader();
6857 if (H.cputype == MachO::CPU_TYPE_ARM)
6858 printObjc2_32bit_MetaData(O, verbose);
6859 else {
6860 // This is the 32-bit non-arm cputype case. Which is normally
6861 // the first Objective-C ABI. But it may be the case of a
6862 // binary for the iOS simulator which is the second Objective-C
6863 // ABI. In that case printObjc1_32bit_MetaData() will determine that
6864 // and return false.
6865 if (!printObjc1_32bit_MetaData(O, verbose))
6866 printObjc2_32bit_MetaData(O, verbose);
6867 }
6868 }
6869 }
6870
6871 // GuessLiteralPointer returns a string which for the item in the Mach-O file
6872 // for the address passed in as ReferenceValue for printing as a comment with
6873 // the instruction and also returns the corresponding type of that item
6874 // indirectly through ReferenceType.
6875 //
6876 // If ReferenceValue is an address of literal cstring then a pointer to the
6877 // cstring is returned and ReferenceType is set to
6878 // LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr .
6879 //
6880 // If ReferenceValue is an address of an Objective-C CFString, Selector ref or
6881 // Class ref that name is returned and the ReferenceType is set accordingly.
6882 //
6883 // Lastly, literals which are Symbol address in a literal pool are looked for
6884 // and if found the symbol name is returned and ReferenceType is set to
6885 // LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr .
6886 //
6887 // If there is no item in the Mach-O file for the address passed in as
6888 // ReferenceValue nullptr is returned and ReferenceType is unchanged.
GuessLiteralPointer(uint64_t ReferenceValue,uint64_t ReferencePC,uint64_t * ReferenceType,struct DisassembleInfo * info)6889 static const char *GuessLiteralPointer(uint64_t ReferenceValue,
6890 uint64_t ReferencePC,
6891 uint64_t *ReferenceType,
6892 struct DisassembleInfo *info) {
6893 // First see if there is an external relocation entry at the ReferencePC.
6894 if (info->O->getHeader().filetype == MachO::MH_OBJECT) {
6895 uint64_t sect_addr = info->S.getAddress();
6896 uint64_t sect_offset = ReferencePC - sect_addr;
6897 bool reloc_found = false;
6898 DataRefImpl Rel;
6899 MachO::any_relocation_info RE;
6900 bool isExtern = false;
6901 SymbolRef Symbol;
6902 for (const RelocationRef &Reloc : info->S.relocations()) {
6903 uint64_t RelocOffset = Reloc.getOffset();
6904 if (RelocOffset == sect_offset) {
6905 Rel = Reloc.getRawDataRefImpl();
6906 RE = info->O->getRelocation(Rel);
6907 if (info->O->isRelocationScattered(RE))
6908 continue;
6909 isExtern = info->O->getPlainRelocationExternal(RE);
6910 if (isExtern) {
6911 symbol_iterator RelocSym = Reloc.getSymbol();
6912 Symbol = *RelocSym;
6913 }
6914 reloc_found = true;
6915 break;
6916 }
6917 }
6918 // If there is an external relocation entry for a symbol in a section
6919 // then used that symbol's value for the value of the reference.
6920 if (reloc_found && isExtern) {
6921 if (info->O->getAnyRelocationPCRel(RE)) {
6922 unsigned Type = info->O->getAnyRelocationType(RE);
6923 if (Type == MachO::X86_64_RELOC_SIGNED) {
6924 ReferenceValue = cantFail(Symbol.getValue());
6925 }
6926 }
6927 }
6928 }
6929
6930 // Look for literals such as Objective-C CFStrings refs, Selector refs,
6931 // Message refs and Class refs.
6932 bool classref, selref, msgref, cfstring;
6933 uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref,
6934 selref, msgref, cfstring);
6935 if (classref && pointer_value == 0) {
6936 // Note the ReferenceValue is a pointer into the __objc_classrefs section.
6937 // And the pointer_value in that section is typically zero as it will be
6938 // set by dyld as part of the "bind information".
6939 const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info);
6940 if (name != nullptr) {
6941 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
6942 const char *class_name = strrchr(name, '$');
6943 if (class_name != nullptr && class_name[1] == '_' &&
6944 class_name[2] != '\0') {
6945 info->class_name = class_name + 2;
6946 return name;
6947 }
6948 }
6949 }
6950
6951 if (classref) {
6952 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
6953 const char *name =
6954 get_objc2_64bit_class_name(pointer_value, ReferenceValue, info);
6955 if (name != nullptr)
6956 info->class_name = name;
6957 else
6958 name = "bad class ref";
6959 return name;
6960 }
6961
6962 if (cfstring) {
6963 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref;
6964 const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info);
6965 return name;
6966 }
6967
6968 if (selref && pointer_value == 0)
6969 pointer_value = get_objc2_64bit_selref(ReferenceValue, info);
6970
6971 if (pointer_value != 0)
6972 ReferenceValue = pointer_value;
6973
6974 const char *name = GuessCstringPointer(ReferenceValue, info);
6975 if (name) {
6976 if (pointer_value != 0 && selref) {
6977 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref;
6978 info->selector_name = name;
6979 } else if (pointer_value != 0 && msgref) {
6980 info->class_name = nullptr;
6981 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref;
6982 info->selector_name = name;
6983 } else
6984 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr;
6985 return name;
6986 }
6987
6988 // Lastly look for an indirect symbol with this ReferenceValue which is in
6989 // a literal pool. If found return that symbol name.
6990 name = GuessIndirectSymbol(ReferenceValue, info);
6991 if (name) {
6992 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr;
6993 return name;
6994 }
6995
6996 return nullptr;
6997 }
6998
6999 // SymbolizerSymbolLookUp is the symbol lookup function passed when creating
7000 // the Symbolizer. It looks up the ReferenceValue using the info passed via the
7001 // pointer to the struct DisassembleInfo that was passed when MCSymbolizer
7002 // is created and returns the symbol name that matches the ReferenceValue or
7003 // nullptr if none. The ReferenceType is passed in for the IN type of
7004 // reference the instruction is making from the values in defined in the header
7005 // "llvm-c/Disassembler.h". On return the ReferenceType can set to a specific
7006 // Out type and the ReferenceName will also be set which is added as a comment
7007 // to the disassembled instruction.
7008 //
7009 // If the symbol name is a C++ mangled name then the demangled name is
7010 // returned through ReferenceName and ReferenceType is set to
7011 // LLVMDisassembler_ReferenceType_DeMangled_Name .
7012 //
7013 // When this is called to get a symbol name for a branch target then the
7014 // ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then
7015 // SymbolValue will be looked for in the indirect symbol table to determine if
7016 // it is an address for a symbol stub. If so then the symbol name for that
7017 // stub is returned indirectly through ReferenceName and then ReferenceType is
7018 // set to LLVMDisassembler_ReferenceType_Out_SymbolStub.
7019 //
7020 // When this is called with an value loaded via a PC relative load then
7021 // ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the
7022 // SymbolValue is checked to be an address of literal pointer, symbol pointer,
7023 // or an Objective-C meta data reference. If so the output ReferenceType is
7024 // set to correspond to that as well as setting the ReferenceName.
SymbolizerSymbolLookUp(void * DisInfo,uint64_t ReferenceValue,uint64_t * ReferenceType,uint64_t ReferencePC,const char ** ReferenceName)7025 static const char *SymbolizerSymbolLookUp(void *DisInfo,
7026 uint64_t ReferenceValue,
7027 uint64_t *ReferenceType,
7028 uint64_t ReferencePC,
7029 const char **ReferenceName) {
7030 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
7031 // If no verbose symbolic information is wanted then just return nullptr.
7032 if (!info->verbose) {
7033 *ReferenceName = nullptr;
7034 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7035 return nullptr;
7036 }
7037
7038 const char *SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
7039
7040 if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) {
7041 *ReferenceName = GuessIndirectSymbol(ReferenceValue, info);
7042 if (*ReferenceName != nullptr) {
7043 method_reference(info, ReferenceType, ReferenceName);
7044 if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message)
7045 *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub;
7046 } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
7047 if (info->demangled_name != nullptr)
7048 free(info->demangled_name);
7049 int status;
7050 info->demangled_name =
7051 itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status);
7052 if (info->demangled_name != nullptr) {
7053 *ReferenceName = info->demangled_name;
7054 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
7055 } else
7056 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7057 } else
7058 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7059 } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) {
7060 *ReferenceName =
7061 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7062 if (*ReferenceName)
7063 method_reference(info, ReferenceType, ReferenceName);
7064 else
7065 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7066 // If this is arm64 and the reference is an adrp instruction save the
7067 // instruction, passed in ReferenceValue and the address of the instruction
7068 // for use later if we see and add immediate instruction.
7069 } else if (info->O->getArch() == Triple::aarch64 &&
7070 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
7071 info->adrp_inst = ReferenceValue;
7072 info->adrp_addr = ReferencePC;
7073 SymbolName = nullptr;
7074 *ReferenceName = nullptr;
7075 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7076 // If this is arm64 and reference is an add immediate instruction and we
7077 // have
7078 // seen an adrp instruction just before it and the adrp's Xd register
7079 // matches
7080 // this add's Xn register reconstruct the value being referenced and look to
7081 // see if it is a literal pointer. Note the add immediate instruction is
7082 // passed in ReferenceValue.
7083 } else if (info->O->getArch() == Triple::aarch64 &&
7084 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri &&
7085 ReferencePC - 4 == info->adrp_addr &&
7086 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
7087 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
7088 uint32_t addxri_inst;
7089 uint64_t adrp_imm, addxri_imm;
7090
7091 adrp_imm =
7092 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
7093 if (info->adrp_inst & 0x0200000)
7094 adrp_imm |= 0xfffffffffc000000LL;
7095
7096 addxri_inst = ReferenceValue;
7097 addxri_imm = (addxri_inst >> 10) & 0xfff;
7098 if (((addxri_inst >> 22) & 0x3) == 1)
7099 addxri_imm <<= 12;
7100
7101 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
7102 (adrp_imm << 12) + addxri_imm;
7103
7104 *ReferenceName =
7105 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7106 if (*ReferenceName == nullptr)
7107 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7108 // If this is arm64 and the reference is a load register instruction and we
7109 // have seen an adrp instruction just before it and the adrp's Xd register
7110 // matches this add's Xn register reconstruct the value being referenced and
7111 // look to see if it is a literal pointer. Note the load register
7112 // instruction is passed in ReferenceValue.
7113 } else if (info->O->getArch() == Triple::aarch64 &&
7114 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui &&
7115 ReferencePC - 4 == info->adrp_addr &&
7116 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
7117 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
7118 uint32_t ldrxui_inst;
7119 uint64_t adrp_imm, ldrxui_imm;
7120
7121 adrp_imm =
7122 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
7123 if (info->adrp_inst & 0x0200000)
7124 adrp_imm |= 0xfffffffffc000000LL;
7125
7126 ldrxui_inst = ReferenceValue;
7127 ldrxui_imm = (ldrxui_inst >> 10) & 0xfff;
7128
7129 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
7130 (adrp_imm << 12) + (ldrxui_imm << 3);
7131
7132 *ReferenceName =
7133 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7134 if (*ReferenceName == nullptr)
7135 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7136 }
7137 // If this arm64 and is an load register (PC-relative) instruction the
7138 // ReferenceValue is the PC plus the immediate value.
7139 else if (info->O->getArch() == Triple::aarch64 &&
7140 (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl ||
7141 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) {
7142 *ReferenceName =
7143 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7144 if (*ReferenceName == nullptr)
7145 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7146 } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
7147 if (info->demangled_name != nullptr)
7148 free(info->demangled_name);
7149 int status;
7150 info->demangled_name =
7151 itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status);
7152 if (info->demangled_name != nullptr) {
7153 *ReferenceName = info->demangled_name;
7154 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
7155 }
7156 }
7157 else {
7158 *ReferenceName = nullptr;
7159 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7160 }
7161
7162 return SymbolName;
7163 }
7164
7165 /// Emits the comments that are stored in the CommentStream.
7166 /// Each comment in the CommentStream must end with a newline.
emitComments(raw_svector_ostream & CommentStream,SmallString<128> & CommentsToEmit,formatted_raw_ostream & FormattedOS,const MCAsmInfo & MAI)7167 static void emitComments(raw_svector_ostream &CommentStream,
7168 SmallString<128> &CommentsToEmit,
7169 formatted_raw_ostream &FormattedOS,
7170 const MCAsmInfo &MAI) {
7171 // Flush the stream before taking its content.
7172 StringRef Comments = CommentsToEmit.str();
7173 // Get the default information for printing a comment.
7174 StringRef CommentBegin = MAI.getCommentString();
7175 unsigned CommentColumn = MAI.getCommentColumn();
7176 ListSeparator LS("\n");
7177 while (!Comments.empty()) {
7178 FormattedOS << LS;
7179 // Emit a line of comments.
7180 FormattedOS.PadToColumn(CommentColumn);
7181 size_t Position = Comments.find('\n');
7182 FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position);
7183 // Move after the newline character.
7184 Comments = Comments.substr(Position + 1);
7185 }
7186 FormattedOS.flush();
7187
7188 // Tell the comment stream that the vector changed underneath it.
7189 CommentsToEmit.clear();
7190 }
7191
DisassembleMachO(StringRef Filename,MachOObjectFile * MachOOF,StringRef DisSegName,StringRef DisSectName)7192 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
7193 StringRef DisSegName, StringRef DisSectName) {
7194 const char *McpuDefault = nullptr;
7195 const Target *ThumbTarget = nullptr;
7196 const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget);
7197 if (!TheTarget) {
7198 // GetTarget prints out stuff.
7199 return;
7200 }
7201 std::string MachOMCPU;
7202 if (MCPU.empty() && McpuDefault)
7203 MachOMCPU = McpuDefault;
7204 else
7205 MachOMCPU = MCPU;
7206
7207 #define CHECK_TARGET_INFO_CREATION(NAME) \
7208 do { \
7209 if (!NAME) { \
7210 WithColor::error(errs(), "llvm-objdump") \
7211 << "couldn't initialize disassembler for target " << TripleName \
7212 << '\n'; \
7213 return; \
7214 } \
7215 } while (false)
7216 #define CHECK_THUMB_TARGET_INFO_CREATION(NAME) \
7217 do { \
7218 if (!NAME) { \
7219 WithColor::error(errs(), "llvm-objdump") \
7220 << "couldn't initialize disassembler for target " << ThumbTripleName \
7221 << '\n'; \
7222 return; \
7223 } \
7224 } while (false)
7225
7226 std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
7227 CHECK_TARGET_INFO_CREATION(InstrInfo);
7228 std::unique_ptr<const MCInstrInfo> ThumbInstrInfo;
7229 if (ThumbTarget) {
7230 ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo());
7231 CHECK_THUMB_TARGET_INFO_CREATION(ThumbInstrInfo);
7232 }
7233
7234 // Package up features to be passed to target/subtarget
7235 std::string FeaturesStr;
7236 if (!MAttrs.empty()) {
7237 SubtargetFeatures Features;
7238 for (unsigned i = 0; i != MAttrs.size(); ++i)
7239 Features.AddFeature(MAttrs[i]);
7240 FeaturesStr = Features.getString();
7241 }
7242
7243 MCTargetOptions MCOptions;
7244 // Set up disassembler.
7245 std::unique_ptr<const MCRegisterInfo> MRI(
7246 TheTarget->createMCRegInfo(TripleName));
7247 CHECK_TARGET_INFO_CREATION(MRI);
7248 std::unique_ptr<const MCAsmInfo> AsmInfo(
7249 TheTarget->createMCAsmInfo(*MRI, TripleName, MCOptions));
7250 CHECK_TARGET_INFO_CREATION(AsmInfo);
7251 std::unique_ptr<const MCSubtargetInfo> STI(
7252 TheTarget->createMCSubtargetInfo(TripleName, MachOMCPU, FeaturesStr));
7253 CHECK_TARGET_INFO_CREATION(STI);
7254 MCContext Ctx(Triple(TripleName), AsmInfo.get(), MRI.get(), STI.get());
7255 std::unique_ptr<MCDisassembler> DisAsm(
7256 TheTarget->createMCDisassembler(*STI, Ctx));
7257 CHECK_TARGET_INFO_CREATION(DisAsm);
7258 std::unique_ptr<MCSymbolizer> Symbolizer;
7259 struct DisassembleInfo SymbolizerInfo(nullptr, nullptr, nullptr, false);
7260 std::unique_ptr<MCRelocationInfo> RelInfo(
7261 TheTarget->createMCRelocationInfo(TripleName, Ctx));
7262 if (RelInfo) {
7263 Symbolizer.reset(TheTarget->createMCSymbolizer(
7264 TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
7265 &SymbolizerInfo, &Ctx, std::move(RelInfo)));
7266 DisAsm->setSymbolizer(std::move(Symbolizer));
7267 }
7268 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
7269 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
7270 Triple(TripleName), AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI));
7271 CHECK_TARGET_INFO_CREATION(IP);
7272 // Set the display preference for hex vs. decimal immediates.
7273 IP->setPrintImmHex(PrintImmHex);
7274 // Comment stream and backing vector.
7275 SmallString<128> CommentsToEmit;
7276 raw_svector_ostream CommentStream(CommentsToEmit);
7277 // FIXME: Setting the CommentStream in the InstPrinter is problematic in that
7278 // if it is done then arm64 comments for string literals don't get printed
7279 // and some constant get printed instead and not setting it causes intel
7280 // (32-bit and 64-bit) comments printed with different spacing before the
7281 // comment causing different diffs with the 'C' disassembler library API.
7282 // IP->setCommentStream(CommentStream);
7283
7284 // Set up separate thumb disassembler if needed.
7285 std::unique_ptr<const MCRegisterInfo> ThumbMRI;
7286 std::unique_ptr<const MCAsmInfo> ThumbAsmInfo;
7287 std::unique_ptr<const MCSubtargetInfo> ThumbSTI;
7288 std::unique_ptr<MCDisassembler> ThumbDisAsm;
7289 std::unique_ptr<MCInstPrinter> ThumbIP;
7290 std::unique_ptr<MCContext> ThumbCtx;
7291 std::unique_ptr<MCSymbolizer> ThumbSymbolizer;
7292 struct DisassembleInfo ThumbSymbolizerInfo(nullptr, nullptr, nullptr, false);
7293 std::unique_ptr<MCRelocationInfo> ThumbRelInfo;
7294 if (ThumbTarget) {
7295 ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName));
7296 CHECK_THUMB_TARGET_INFO_CREATION(ThumbMRI);
7297 ThumbAsmInfo.reset(
7298 ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName, MCOptions));
7299 CHECK_THUMB_TARGET_INFO_CREATION(ThumbAsmInfo);
7300 ThumbSTI.reset(
7301 ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MachOMCPU,
7302 FeaturesStr));
7303 CHECK_THUMB_TARGET_INFO_CREATION(ThumbSTI);
7304 ThumbCtx.reset(new MCContext(Triple(ThumbTripleName), ThumbAsmInfo.get(),
7305 ThumbMRI.get(), ThumbSTI.get()));
7306 ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx));
7307 CHECK_THUMB_TARGET_INFO_CREATION(ThumbDisAsm);
7308 MCContext *PtrThumbCtx = ThumbCtx.get();
7309 ThumbRelInfo.reset(
7310 ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx));
7311 if (ThumbRelInfo) {
7312 ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer(
7313 ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
7314 &ThumbSymbolizerInfo, PtrThumbCtx, std::move(ThumbRelInfo)));
7315 ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer));
7316 }
7317 int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect();
7318 ThumbIP.reset(ThumbTarget->createMCInstPrinter(
7319 Triple(ThumbTripleName), ThumbAsmPrinterVariant, *ThumbAsmInfo,
7320 *ThumbInstrInfo, *ThumbMRI));
7321 CHECK_THUMB_TARGET_INFO_CREATION(ThumbIP);
7322 // Set the display preference for hex vs. decimal immediates.
7323 ThumbIP->setPrintImmHex(PrintImmHex);
7324 }
7325
7326 #undef CHECK_TARGET_INFO_CREATION
7327 #undef CHECK_THUMB_TARGET_INFO_CREATION
7328
7329 MachO::mach_header Header = MachOOF->getHeader();
7330
7331 // FIXME: Using the -cfg command line option, this code used to be able to
7332 // annotate relocations with the referenced symbol's name, and if this was
7333 // inside a __[cf]string section, the data it points to. This is now replaced
7334 // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
7335 std::vector<SectionRef> Sections;
7336 std::vector<SymbolRef> Symbols;
7337 SmallVector<uint64_t, 8> FoundFns;
7338 uint64_t BaseSegmentAddress = 0;
7339
7340 getSectionsAndSymbols(MachOOF, Sections, Symbols, FoundFns,
7341 BaseSegmentAddress);
7342
7343 // Sort the symbols by address, just in case they didn't come in that way.
7344 llvm::stable_sort(Symbols, SymbolSorter());
7345
7346 // Build a data in code table that is sorted on by the address of each entry.
7347 uint64_t BaseAddress = 0;
7348 if (Header.filetype == MachO::MH_OBJECT)
7349 BaseAddress = Sections[0].getAddress();
7350 else
7351 BaseAddress = BaseSegmentAddress;
7352 DiceTable Dices;
7353 for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
7354 DI != DE; ++DI) {
7355 uint32_t Offset;
7356 DI->getOffset(Offset);
7357 Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
7358 }
7359 array_pod_sort(Dices.begin(), Dices.end());
7360
7361 // Try to find debug info and set up the DIContext for it.
7362 std::unique_ptr<DIContext> diContext;
7363 std::unique_ptr<Binary> DSYMBinary;
7364 std::unique_ptr<MemoryBuffer> DSYMBuf;
7365 if (UseDbg) {
7366 ObjectFile *DbgObj = MachOOF;
7367
7368 // A separate DSym file path was specified, parse it as a macho file,
7369 // get the sections and supply it to the section name parsing machinery.
7370 if (!DSYMFile.empty()) {
7371 std::string DSYMPath(DSYMFile);
7372
7373 // If DSYMPath is a .dSYM directory, append the Mach-O file.
7374 if (llvm::sys::fs::is_directory(DSYMPath) &&
7375 llvm::sys::path::extension(DSYMPath) == ".dSYM") {
7376 SmallString<128> ShortName(llvm::sys::path::filename(DSYMPath));
7377 llvm::sys::path::replace_extension(ShortName, "");
7378 SmallString<1024> FullPath(DSYMPath);
7379 llvm::sys::path::append(FullPath, "Contents", "Resources", "DWARF",
7380 ShortName);
7381 DSYMPath = std::string(FullPath.str());
7382 }
7383
7384 // Load the file.
7385 ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr =
7386 MemoryBuffer::getFileOrSTDIN(DSYMPath);
7387 if (std::error_code EC = BufOrErr.getError()) {
7388 reportError(errorCodeToError(EC), DSYMPath);
7389 return;
7390 }
7391
7392 // We need to keep the file alive, because we're replacing DbgObj with it.
7393 DSYMBuf = std::move(BufOrErr.get());
7394
7395 Expected<std::unique_ptr<Binary>> BinaryOrErr =
7396 createBinary(DSYMBuf.get()->getMemBufferRef());
7397 if (!BinaryOrErr) {
7398 reportError(BinaryOrErr.takeError(), DSYMPath);
7399 return;
7400 }
7401
7402 // We need to keep the Binary alive with the buffer
7403 DSYMBinary = std::move(BinaryOrErr.get());
7404 if (ObjectFile *O = dyn_cast<ObjectFile>(DSYMBinary.get())) {
7405 // this is a Mach-O object file, use it
7406 if (MachOObjectFile *MachDSYM = dyn_cast<MachOObjectFile>(&*O)) {
7407 DbgObj = MachDSYM;
7408 }
7409 else {
7410 WithColor::error(errs(), "llvm-objdump")
7411 << DSYMPath << " is not a Mach-O file type.\n";
7412 return;
7413 }
7414 }
7415 else if (auto UB = dyn_cast<MachOUniversalBinary>(DSYMBinary.get())){
7416 // this is a Universal Binary, find a Mach-O for this architecture
7417 uint32_t CPUType, CPUSubType;
7418 const char *ArchFlag;
7419 if (MachOOF->is64Bit()) {
7420 const MachO::mach_header_64 H_64 = MachOOF->getHeader64();
7421 CPUType = H_64.cputype;
7422 CPUSubType = H_64.cpusubtype;
7423 } else {
7424 const MachO::mach_header H = MachOOF->getHeader();
7425 CPUType = H.cputype;
7426 CPUSubType = H.cpusubtype;
7427 }
7428 Triple T = MachOObjectFile::getArchTriple(CPUType, CPUSubType, nullptr,
7429 &ArchFlag);
7430 Expected<std::unique_ptr<MachOObjectFile>> MachDSYM =
7431 UB->getMachOObjectForArch(ArchFlag);
7432 if (!MachDSYM) {
7433 reportError(MachDSYM.takeError(), DSYMPath);
7434 return;
7435 }
7436
7437 // We need to keep the Binary alive with the buffer
7438 DbgObj = &*MachDSYM.get();
7439 DSYMBinary = std::move(*MachDSYM);
7440 }
7441 else {
7442 WithColor::error(errs(), "llvm-objdump")
7443 << DSYMPath << " is not a Mach-O or Universal file type.\n";
7444 return;
7445 }
7446 }
7447
7448 // Setup the DIContext
7449 diContext = DWARFContext::create(*DbgObj);
7450 }
7451
7452 if (FilterSections.empty())
7453 outs() << "(" << DisSegName << "," << DisSectName << ") section\n";
7454
7455 for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
7456 Expected<StringRef> SecNameOrErr = Sections[SectIdx].getName();
7457 if (!SecNameOrErr) {
7458 consumeError(SecNameOrErr.takeError());
7459 continue;
7460 }
7461 if (*SecNameOrErr != DisSectName)
7462 continue;
7463
7464 DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
7465
7466 StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
7467 if (SegmentName != DisSegName)
7468 continue;
7469
7470 StringRef BytesStr =
7471 unwrapOrError(Sections[SectIdx].getContents(), Filename);
7472 ArrayRef<uint8_t> Bytes = arrayRefFromStringRef(BytesStr);
7473 uint64_t SectAddress = Sections[SectIdx].getAddress();
7474
7475 bool symbolTableWorked = false;
7476
7477 // Create a map of symbol addresses to symbol names for use by
7478 // the SymbolizerSymbolLookUp() routine.
7479 SymbolAddressMap AddrMap;
7480 bool DisSymNameFound = false;
7481 for (const SymbolRef &Symbol : MachOOF->symbols()) {
7482 SymbolRef::Type ST =
7483 unwrapOrError(Symbol.getType(), MachOOF->getFileName());
7484 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
7485 ST == SymbolRef::ST_Other) {
7486 uint64_t Address = cantFail(Symbol.getValue());
7487 StringRef SymName =
7488 unwrapOrError(Symbol.getName(), MachOOF->getFileName());
7489 AddrMap[Address] = SymName;
7490 if (!DisSymName.empty() && DisSymName == SymName)
7491 DisSymNameFound = true;
7492 }
7493 }
7494 if (!DisSymName.empty() && !DisSymNameFound) {
7495 outs() << "Can't find -dis-symname: " << DisSymName << "\n";
7496 return;
7497 }
7498 // Set up the block of info used by the Symbolizer call backs.
7499 SymbolizerInfo.verbose = SymbolicOperands;
7500 SymbolizerInfo.O = MachOOF;
7501 SymbolizerInfo.S = Sections[SectIdx];
7502 SymbolizerInfo.AddrMap = &AddrMap;
7503 SymbolizerInfo.Sections = &Sections;
7504 // Same for the ThumbSymbolizer
7505 ThumbSymbolizerInfo.verbose = SymbolicOperands;
7506 ThumbSymbolizerInfo.O = MachOOF;
7507 ThumbSymbolizerInfo.S = Sections[SectIdx];
7508 ThumbSymbolizerInfo.AddrMap = &AddrMap;
7509 ThumbSymbolizerInfo.Sections = &Sections;
7510
7511 unsigned int Arch = MachOOF->getArch();
7512
7513 // Skip all symbols if this is a stubs file.
7514 if (Bytes.empty())
7515 return;
7516
7517 // If the section has symbols but no symbol at the start of the section
7518 // these are used to make sure the bytes before the first symbol are
7519 // disassembled.
7520 bool FirstSymbol = true;
7521 bool FirstSymbolAtSectionStart = true;
7522
7523 // Disassemble symbol by symbol.
7524 for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
7525 StringRef SymName =
7526 unwrapOrError(Symbols[SymIdx].getName(), MachOOF->getFileName());
7527 SymbolRef::Type ST =
7528 unwrapOrError(Symbols[SymIdx].getType(), MachOOF->getFileName());
7529 if (ST != SymbolRef::ST_Function && ST != SymbolRef::ST_Data)
7530 continue;
7531
7532 // Make sure the symbol is defined in this section.
7533 bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]);
7534 if (!containsSym) {
7535 if (!DisSymName.empty() && DisSymName == SymName) {
7536 outs() << "-dis-symname: " << DisSymName << " not in the section\n";
7537 return;
7538 }
7539 continue;
7540 }
7541 // The __mh_execute_header is special and we need to deal with that fact
7542 // this symbol is before the start of the (__TEXT,__text) section and at the
7543 // address of the start of the __TEXT segment. This is because this symbol
7544 // is an N_SECT symbol in the (__TEXT,__text) but its address is before the
7545 // start of the section in a standard MH_EXECUTE filetype.
7546 if (!DisSymName.empty() && DisSymName == "__mh_execute_header") {
7547 outs() << "-dis-symname: __mh_execute_header not in any section\n";
7548 return;
7549 }
7550 // When this code is trying to disassemble a symbol at a time and in the
7551 // case there is only the __mh_execute_header symbol left as in a stripped
7552 // executable, we need to deal with this by ignoring this symbol so the
7553 // whole section is disassembled and this symbol is then not displayed.
7554 if (SymName == "__mh_execute_header" || SymName == "__mh_dylib_header" ||
7555 SymName == "__mh_bundle_header" || SymName == "__mh_object_header" ||
7556 SymName == "__mh_preload_header" || SymName == "__mh_dylinker_header")
7557 continue;
7558
7559 // If we are only disassembling one symbol see if this is that symbol.
7560 if (!DisSymName.empty() && DisSymName != SymName)
7561 continue;
7562
7563 // Start at the address of the symbol relative to the section's address.
7564 uint64_t SectSize = Sections[SectIdx].getSize();
7565 uint64_t Start = cantFail(Symbols[SymIdx].getValue());
7566 uint64_t SectionAddress = Sections[SectIdx].getAddress();
7567 Start -= SectionAddress;
7568
7569 if (Start > SectSize) {
7570 outs() << "section data ends, " << SymName
7571 << " lies outside valid range\n";
7572 return;
7573 }
7574
7575 // Stop disassembling either at the beginning of the next symbol or at
7576 // the end of the section.
7577 bool containsNextSym = false;
7578 uint64_t NextSym = 0;
7579 uint64_t NextSymIdx = SymIdx + 1;
7580 while (Symbols.size() > NextSymIdx) {
7581 SymbolRef::Type NextSymType = unwrapOrError(
7582 Symbols[NextSymIdx].getType(), MachOOF->getFileName());
7583 if (NextSymType == SymbolRef::ST_Function) {
7584 containsNextSym =
7585 Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]);
7586 NextSym = cantFail(Symbols[NextSymIdx].getValue());
7587 NextSym -= SectionAddress;
7588 break;
7589 }
7590 ++NextSymIdx;
7591 }
7592
7593 uint64_t End = containsNextSym ? std::min(NextSym, SectSize) : SectSize;
7594 uint64_t Size;
7595
7596 symbolTableWorked = true;
7597
7598 DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl();
7599 uint32_t SymbolFlags = cantFail(MachOOF->getSymbolFlags(Symb));
7600 bool IsThumb = SymbolFlags & SymbolRef::SF_Thumb;
7601
7602 // We only need the dedicated Thumb target if there's a real choice
7603 // (i.e. we're not targeting M-class) and the function is Thumb.
7604 bool UseThumbTarget = IsThumb && ThumbTarget;
7605
7606 // If we are not specifying a symbol to start disassembly with and this
7607 // is the first symbol in the section but not at the start of the section
7608 // then move the disassembly index to the start of the section and
7609 // don't print the symbol name just yet. This is so the bytes before the
7610 // first symbol are disassembled.
7611 uint64_t SymbolStart = Start;
7612 if (DisSymName.empty() && FirstSymbol && Start != 0) {
7613 FirstSymbolAtSectionStart = false;
7614 Start = 0;
7615 }
7616 else
7617 outs() << SymName << ":\n";
7618
7619 DILineInfo lastLine;
7620 for (uint64_t Index = Start; Index < End; Index += Size) {
7621 MCInst Inst;
7622
7623 // If this is the first symbol in the section and it was not at the
7624 // start of the section, see if we are at its Index now and if so print
7625 // the symbol name.
7626 if (FirstSymbol && !FirstSymbolAtSectionStart && Index == SymbolStart)
7627 outs() << SymName << ":\n";
7628
7629 uint64_t PC = SectAddress + Index;
7630 if (LeadingAddr) {
7631 if (FullLeadingAddr) {
7632 if (MachOOF->is64Bit())
7633 outs() << format("%016" PRIx64, PC);
7634 else
7635 outs() << format("%08" PRIx64, PC);
7636 } else {
7637 outs() << format("%8" PRIx64 ":", PC);
7638 }
7639 }
7640 if (ShowRawInsn || Arch == Triple::arm)
7641 outs() << "\t";
7642
7643 if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, Size))
7644 continue;
7645
7646 SmallVector<char, 64> AnnotationsBytes;
7647 raw_svector_ostream Annotations(AnnotationsBytes);
7648
7649 bool gotInst;
7650 if (UseThumbTarget)
7651 gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index),
7652 PC, Annotations);
7653 else
7654 gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC,
7655 Annotations);
7656 if (gotInst) {
7657 if (ShowRawInsn || Arch == Triple::arm) {
7658 dumpBytes(makeArrayRef(Bytes.data() + Index, Size), outs());
7659 }
7660 formatted_raw_ostream FormattedOS(outs());
7661 StringRef AnnotationsStr = Annotations.str();
7662 if (UseThumbTarget)
7663 ThumbIP->printInst(&Inst, PC, AnnotationsStr, *ThumbSTI,
7664 FormattedOS);
7665 else
7666 IP->printInst(&Inst, PC, AnnotationsStr, *STI, FormattedOS);
7667 emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo);
7668
7669 // Print debug info.
7670 if (diContext) {
7671 DILineInfo dli = diContext->getLineInfoForAddress({PC, SectIdx});
7672 // Print valid line info if it changed.
7673 if (dli != lastLine && dli.Line != 0)
7674 outs() << "\t## " << dli.FileName << ':' << dli.Line << ':'
7675 << dli.Column;
7676 lastLine = dli;
7677 }
7678 outs() << "\n";
7679 } else {
7680 if (MachOOF->getArchTriple().isX86()) {
7681 outs() << format("\t.byte 0x%02x #bad opcode\n",
7682 *(Bytes.data() + Index) & 0xff);
7683 Size = 1; // skip exactly one illegible byte and move on.
7684 } else if (Arch == Triple::aarch64 ||
7685 (Arch == Triple::arm && !IsThumb)) {
7686 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
7687 (*(Bytes.data() + Index + 1) & 0xff) << 8 |
7688 (*(Bytes.data() + Index + 2) & 0xff) << 16 |
7689 (*(Bytes.data() + Index + 3) & 0xff) << 24;
7690 outs() << format("\t.long\t0x%08x\n", opcode);
7691 Size = 4;
7692 } else if (Arch == Triple::arm) {
7693 assert(IsThumb && "ARM mode should have been dealt with above");
7694 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
7695 (*(Bytes.data() + Index + 1) & 0xff) << 8;
7696 outs() << format("\t.short\t0x%04x\n", opcode);
7697 Size = 2;
7698 } else{
7699 WithColor::warning(errs(), "llvm-objdump")
7700 << "invalid instruction encoding\n";
7701 if (Size == 0)
7702 Size = 1; // skip illegible bytes
7703 }
7704 }
7705 }
7706 // Now that we are done disassembled the first symbol set the bool that
7707 // were doing this to false.
7708 FirstSymbol = false;
7709 }
7710 if (!symbolTableWorked) {
7711 // Reading the symbol table didn't work, disassemble the whole section.
7712 uint64_t SectAddress = Sections[SectIdx].getAddress();
7713 uint64_t SectSize = Sections[SectIdx].getSize();
7714 uint64_t InstSize;
7715 for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
7716 MCInst Inst;
7717
7718 uint64_t PC = SectAddress + Index;
7719
7720 if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, InstSize))
7721 continue;
7722
7723 SmallVector<char, 64> AnnotationsBytes;
7724 raw_svector_ostream Annotations(AnnotationsBytes);
7725 if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC,
7726 Annotations)) {
7727 if (LeadingAddr) {
7728 if (FullLeadingAddr) {
7729 if (MachOOF->is64Bit())
7730 outs() << format("%016" PRIx64, PC);
7731 else
7732 outs() << format("%08" PRIx64, PC);
7733 } else {
7734 outs() << format("%8" PRIx64 ":", PC);
7735 }
7736 }
7737 if (ShowRawInsn || Arch == Triple::arm) {
7738 outs() << "\t";
7739 dumpBytes(makeArrayRef(Bytes.data() + Index, InstSize), outs());
7740 }
7741 StringRef AnnotationsStr = Annotations.str();
7742 IP->printInst(&Inst, PC, AnnotationsStr, *STI, outs());
7743 outs() << "\n";
7744 } else {
7745 if (MachOOF->getArchTriple().isX86()) {
7746 outs() << format("\t.byte 0x%02x #bad opcode\n",
7747 *(Bytes.data() + Index) & 0xff);
7748 InstSize = 1; // skip exactly one illegible byte and move on.
7749 } else {
7750 WithColor::warning(errs(), "llvm-objdump")
7751 << "invalid instruction encoding\n";
7752 if (InstSize == 0)
7753 InstSize = 1; // skip illegible bytes
7754 }
7755 }
7756 }
7757 }
7758 // The TripleName's need to be reset if we are called again for a different
7759 // architecture.
7760 TripleName = "";
7761 ThumbTripleName = "";
7762
7763 if (SymbolizerInfo.demangled_name != nullptr)
7764 free(SymbolizerInfo.demangled_name);
7765 if (ThumbSymbolizerInfo.demangled_name != nullptr)
7766 free(ThumbSymbolizerInfo.demangled_name);
7767 }
7768 }
7769
7770 //===----------------------------------------------------------------------===//
7771 // __compact_unwind section dumping
7772 //===----------------------------------------------------------------------===//
7773
7774 namespace {
7775
7776 template <typename T>
read(StringRef Contents,ptrdiff_t Offset)7777 static uint64_t read(StringRef Contents, ptrdiff_t Offset) {
7778 using llvm::support::little;
7779 using llvm::support::unaligned;
7780
7781 if (Offset + sizeof(T) > Contents.size()) {
7782 outs() << "warning: attempt to read past end of buffer\n";
7783 return T();
7784 }
7785
7786 uint64_t Val =
7787 support::endian::read<T, little, unaligned>(Contents.data() + Offset);
7788 return Val;
7789 }
7790
7791 template <typename T>
readNext(StringRef Contents,ptrdiff_t & Offset)7792 static uint64_t readNext(StringRef Contents, ptrdiff_t &Offset) {
7793 T Val = read<T>(Contents, Offset);
7794 Offset += sizeof(T);
7795 return Val;
7796 }
7797
7798 struct CompactUnwindEntry {
7799 uint32_t OffsetInSection;
7800
7801 uint64_t FunctionAddr;
7802 uint32_t Length;
7803 uint32_t CompactEncoding;
7804 uint64_t PersonalityAddr;
7805 uint64_t LSDAAddr;
7806
7807 RelocationRef FunctionReloc;
7808 RelocationRef PersonalityReloc;
7809 RelocationRef LSDAReloc;
7810
CompactUnwindEntry__anon4a9557200a11::CompactUnwindEntry7811 CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64)
7812 : OffsetInSection(Offset) {
7813 if (Is64)
7814 read<uint64_t>(Contents, Offset);
7815 else
7816 read<uint32_t>(Contents, Offset);
7817 }
7818
7819 private:
read__anon4a9557200a11::CompactUnwindEntry7820 template <typename UIntPtr> void read(StringRef Contents, ptrdiff_t Offset) {
7821 FunctionAddr = readNext<UIntPtr>(Contents, Offset);
7822 Length = readNext<uint32_t>(Contents, Offset);
7823 CompactEncoding = readNext<uint32_t>(Contents, Offset);
7824 PersonalityAddr = readNext<UIntPtr>(Contents, Offset);
7825 LSDAAddr = readNext<UIntPtr>(Contents, Offset);
7826 }
7827 };
7828 }
7829
7830 /// Given a relocation from __compact_unwind, consisting of the RelocationRef
7831 /// and data being relocated, determine the best base Name and Addend to use for
7832 /// display purposes.
7833 ///
7834 /// 1. An Extern relocation will directly reference a symbol (and the data is
7835 /// then already an addend), so use that.
7836 /// 2. Otherwise the data is an offset in the object file's layout; try to find
7837 // a symbol before it in the same section, and use the offset from there.
7838 /// 3. Finally, if all that fails, fall back to an offset from the start of the
7839 /// referenced section.
findUnwindRelocNameAddend(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const RelocationRef & Reloc,uint64_t Addr,StringRef & Name,uint64_t & Addend)7840 static void findUnwindRelocNameAddend(const MachOObjectFile *Obj,
7841 std::map<uint64_t, SymbolRef> &Symbols,
7842 const RelocationRef &Reloc, uint64_t Addr,
7843 StringRef &Name, uint64_t &Addend) {
7844 if (Reloc.getSymbol() != Obj->symbol_end()) {
7845 Name = unwrapOrError(Reloc.getSymbol()->getName(), Obj->getFileName());
7846 Addend = Addr;
7847 return;
7848 }
7849
7850 auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl());
7851 SectionRef RelocSection = Obj->getAnyRelocationSection(RE);
7852
7853 uint64_t SectionAddr = RelocSection.getAddress();
7854
7855 auto Sym = Symbols.upper_bound(Addr);
7856 if (Sym == Symbols.begin()) {
7857 // The first symbol in the object is after this reference, the best we can
7858 // do is section-relative notation.
7859 if (Expected<StringRef> NameOrErr = RelocSection.getName())
7860 Name = *NameOrErr;
7861 else
7862 consumeError(NameOrErr.takeError());
7863
7864 Addend = Addr - SectionAddr;
7865 return;
7866 }
7867
7868 // Go back one so that SymbolAddress <= Addr.
7869 --Sym;
7870
7871 section_iterator SymSection =
7872 unwrapOrError(Sym->second.getSection(), Obj->getFileName());
7873 if (RelocSection == *SymSection) {
7874 // There's a valid symbol in the same section before this reference.
7875 Name = unwrapOrError(Sym->second.getName(), Obj->getFileName());
7876 Addend = Addr - Sym->first;
7877 return;
7878 }
7879
7880 // There is a symbol before this reference, but it's in a different
7881 // section. Probably not helpful to mention it, so use the section name.
7882 if (Expected<StringRef> NameOrErr = RelocSection.getName())
7883 Name = *NameOrErr;
7884 else
7885 consumeError(NameOrErr.takeError());
7886
7887 Addend = Addr - SectionAddr;
7888 }
7889
printUnwindRelocDest(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const RelocationRef & Reloc,uint64_t Addr)7890 static void printUnwindRelocDest(const MachOObjectFile *Obj,
7891 std::map<uint64_t, SymbolRef> &Symbols,
7892 const RelocationRef &Reloc, uint64_t Addr) {
7893 StringRef Name;
7894 uint64_t Addend;
7895
7896 if (!Reloc.getObject())
7897 return;
7898
7899 findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend);
7900
7901 outs() << Name;
7902 if (Addend)
7903 outs() << " + " << format("0x%" PRIx64, Addend);
7904 }
7905
7906 static void
printMachOCompactUnwindSection(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const SectionRef & CompactUnwind)7907 printMachOCompactUnwindSection(const MachOObjectFile *Obj,
7908 std::map<uint64_t, SymbolRef> &Symbols,
7909 const SectionRef &CompactUnwind) {
7910
7911 if (!Obj->isLittleEndian()) {
7912 outs() << "Skipping big-endian __compact_unwind section\n";
7913 return;
7914 }
7915
7916 bool Is64 = Obj->is64Bit();
7917 uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t);
7918 uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t);
7919
7920 StringRef Contents =
7921 unwrapOrError(CompactUnwind.getContents(), Obj->getFileName());
7922 SmallVector<CompactUnwindEntry, 4> CompactUnwinds;
7923
7924 // First populate the initial raw offsets, encodings and so on from the entry.
7925 for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) {
7926 CompactUnwindEntry Entry(Contents, Offset, Is64);
7927 CompactUnwinds.push_back(Entry);
7928 }
7929
7930 // Next we need to look at the relocations to find out what objects are
7931 // actually being referred to.
7932 for (const RelocationRef &Reloc : CompactUnwind.relocations()) {
7933 uint64_t RelocAddress = Reloc.getOffset();
7934
7935 uint32_t EntryIdx = RelocAddress / EntrySize;
7936 uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize;
7937 CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx];
7938
7939 if (OffsetInEntry == 0)
7940 Entry.FunctionReloc = Reloc;
7941 else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t))
7942 Entry.PersonalityReloc = Reloc;
7943 else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t))
7944 Entry.LSDAReloc = Reloc;
7945 else {
7946 outs() << "Invalid relocation in __compact_unwind section\n";
7947 return;
7948 }
7949 }
7950
7951 // Finally, we're ready to print the data we've gathered.
7952 outs() << "Contents of __compact_unwind section:\n";
7953 for (auto &Entry : CompactUnwinds) {
7954 outs() << " Entry at offset "
7955 << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n";
7956
7957 // 1. Start of the region this entry applies to.
7958 outs() << " start: " << format("0x%" PRIx64,
7959 Entry.FunctionAddr) << ' ';
7960 printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr);
7961 outs() << '\n';
7962
7963 // 2. Length of the region this entry applies to.
7964 outs() << " length: " << format("0x%" PRIx32, Entry.Length)
7965 << '\n';
7966 // 3. The 32-bit compact encoding.
7967 outs() << " compact encoding: "
7968 << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n';
7969
7970 // 4. The personality function, if present.
7971 if (Entry.PersonalityReloc.getObject()) {
7972 outs() << " personality function: "
7973 << format("0x%" PRIx64, Entry.PersonalityAddr) << ' ';
7974 printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc,
7975 Entry.PersonalityAddr);
7976 outs() << '\n';
7977 }
7978
7979 // 5. This entry's language-specific data area.
7980 if (Entry.LSDAReloc.getObject()) {
7981 outs() << " LSDA: " << format("0x%" PRIx64,
7982 Entry.LSDAAddr) << ' ';
7983 printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr);
7984 outs() << '\n';
7985 }
7986 }
7987 }
7988
7989 //===----------------------------------------------------------------------===//
7990 // __unwind_info section dumping
7991 //===----------------------------------------------------------------------===//
7992
printRegularSecondLevelUnwindPage(StringRef PageData)7993 static void printRegularSecondLevelUnwindPage(StringRef PageData) {
7994 ptrdiff_t Pos = 0;
7995 uint32_t Kind = readNext<uint32_t>(PageData, Pos);
7996 (void)Kind;
7997 assert(Kind == 2 && "kind for a regular 2nd level index should be 2");
7998
7999 uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos);
8000 uint16_t NumEntries = readNext<uint16_t>(PageData, Pos);
8001
8002 Pos = EntriesStart;
8003 for (unsigned i = 0; i < NumEntries; ++i) {
8004 uint32_t FunctionOffset = readNext<uint32_t>(PageData, Pos);
8005 uint32_t Encoding = readNext<uint32_t>(PageData, Pos);
8006
8007 outs() << " [" << i << "]: "
8008 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8009 << ", "
8010 << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n';
8011 }
8012 }
8013
printCompressedSecondLevelUnwindPage(StringRef PageData,uint32_t FunctionBase,const SmallVectorImpl<uint32_t> & CommonEncodings)8014 static void printCompressedSecondLevelUnwindPage(
8015 StringRef PageData, uint32_t FunctionBase,
8016 const SmallVectorImpl<uint32_t> &CommonEncodings) {
8017 ptrdiff_t Pos = 0;
8018 uint32_t Kind = readNext<uint32_t>(PageData, Pos);
8019 (void)Kind;
8020 assert(Kind == 3 && "kind for a compressed 2nd level index should be 3");
8021
8022 uint32_t NumCommonEncodings = CommonEncodings.size();
8023 uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos);
8024 uint16_t NumEntries = readNext<uint16_t>(PageData, Pos);
8025
8026 uint16_t PageEncodingsStart = readNext<uint16_t>(PageData, Pos);
8027 uint16_t NumPageEncodings = readNext<uint16_t>(PageData, Pos);
8028 SmallVector<uint32_t, 64> PageEncodings;
8029 if (NumPageEncodings) {
8030 outs() << " Page encodings: (count = " << NumPageEncodings << ")\n";
8031 Pos = PageEncodingsStart;
8032 for (unsigned i = 0; i < NumPageEncodings; ++i) {
8033 uint32_t Encoding = readNext<uint32_t>(PageData, Pos);
8034 PageEncodings.push_back(Encoding);
8035 outs() << " encoding[" << (i + NumCommonEncodings)
8036 << "]: " << format("0x%08" PRIx32, Encoding) << '\n';
8037 }
8038 }
8039
8040 Pos = EntriesStart;
8041 for (unsigned i = 0; i < NumEntries; ++i) {
8042 uint32_t Entry = readNext<uint32_t>(PageData, Pos);
8043 uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff);
8044 uint32_t EncodingIdx = Entry >> 24;
8045
8046 uint32_t Encoding;
8047 if (EncodingIdx < NumCommonEncodings)
8048 Encoding = CommonEncodings[EncodingIdx];
8049 else
8050 Encoding = PageEncodings[EncodingIdx - NumCommonEncodings];
8051
8052 outs() << " [" << i << "]: "
8053 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8054 << ", "
8055 << "encoding[" << EncodingIdx
8056 << "]=" << format("0x%08" PRIx32, Encoding) << '\n';
8057 }
8058 }
8059
printMachOUnwindInfoSection(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const SectionRef & UnwindInfo)8060 static void printMachOUnwindInfoSection(const MachOObjectFile *Obj,
8061 std::map<uint64_t, SymbolRef> &Symbols,
8062 const SectionRef &UnwindInfo) {
8063
8064 if (!Obj->isLittleEndian()) {
8065 outs() << "Skipping big-endian __unwind_info section\n";
8066 return;
8067 }
8068
8069 outs() << "Contents of __unwind_info section:\n";
8070
8071 StringRef Contents =
8072 unwrapOrError(UnwindInfo.getContents(), Obj->getFileName());
8073 ptrdiff_t Pos = 0;
8074
8075 //===----------------------------------
8076 // Section header
8077 //===----------------------------------
8078
8079 uint32_t Version = readNext<uint32_t>(Contents, Pos);
8080 outs() << " Version: "
8081 << format("0x%" PRIx32, Version) << '\n';
8082 if (Version != 1) {
8083 outs() << " Skipping section with unknown version\n";
8084 return;
8085 }
8086
8087 uint32_t CommonEncodingsStart = readNext<uint32_t>(Contents, Pos);
8088 outs() << " Common encodings array section offset: "
8089 << format("0x%" PRIx32, CommonEncodingsStart) << '\n';
8090 uint32_t NumCommonEncodings = readNext<uint32_t>(Contents, Pos);
8091 outs() << " Number of common encodings in array: "
8092 << format("0x%" PRIx32, NumCommonEncodings) << '\n';
8093
8094 uint32_t PersonalitiesStart = readNext<uint32_t>(Contents, Pos);
8095 outs() << " Personality function array section offset: "
8096 << format("0x%" PRIx32, PersonalitiesStart) << '\n';
8097 uint32_t NumPersonalities = readNext<uint32_t>(Contents, Pos);
8098 outs() << " Number of personality functions in array: "
8099 << format("0x%" PRIx32, NumPersonalities) << '\n';
8100
8101 uint32_t IndicesStart = readNext<uint32_t>(Contents, Pos);
8102 outs() << " Index array section offset: "
8103 << format("0x%" PRIx32, IndicesStart) << '\n';
8104 uint32_t NumIndices = readNext<uint32_t>(Contents, Pos);
8105 outs() << " Number of indices in array: "
8106 << format("0x%" PRIx32, NumIndices) << '\n';
8107
8108 //===----------------------------------
8109 // A shared list of common encodings
8110 //===----------------------------------
8111
8112 // These occupy indices in the range [0, N] whenever an encoding is referenced
8113 // from a compressed 2nd level index table. In practice the linker only
8114 // creates ~128 of these, so that indices are available to embed encodings in
8115 // the 2nd level index.
8116
8117 SmallVector<uint32_t, 64> CommonEncodings;
8118 outs() << " Common encodings: (count = " << NumCommonEncodings << ")\n";
8119 Pos = CommonEncodingsStart;
8120 for (unsigned i = 0; i < NumCommonEncodings; ++i) {
8121 uint32_t Encoding = readNext<uint32_t>(Contents, Pos);
8122 CommonEncodings.push_back(Encoding);
8123
8124 outs() << " encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding)
8125 << '\n';
8126 }
8127
8128 //===----------------------------------
8129 // Personality functions used in this executable
8130 //===----------------------------------
8131
8132 // There should be only a handful of these (one per source language,
8133 // roughly). Particularly since they only get 2 bits in the compact encoding.
8134
8135 outs() << " Personality functions: (count = " << NumPersonalities << ")\n";
8136 Pos = PersonalitiesStart;
8137 for (unsigned i = 0; i < NumPersonalities; ++i) {
8138 uint32_t PersonalityFn = readNext<uint32_t>(Contents, Pos);
8139 outs() << " personality[" << i + 1
8140 << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n';
8141 }
8142
8143 //===----------------------------------
8144 // The level 1 index entries
8145 //===----------------------------------
8146
8147 // These specify an approximate place to start searching for the more detailed
8148 // information, sorted by PC.
8149
8150 struct IndexEntry {
8151 uint32_t FunctionOffset;
8152 uint32_t SecondLevelPageStart;
8153 uint32_t LSDAStart;
8154 };
8155
8156 SmallVector<IndexEntry, 4> IndexEntries;
8157
8158 outs() << " Top level indices: (count = " << NumIndices << ")\n";
8159 Pos = IndicesStart;
8160 for (unsigned i = 0; i < NumIndices; ++i) {
8161 IndexEntry Entry;
8162
8163 Entry.FunctionOffset = readNext<uint32_t>(Contents, Pos);
8164 Entry.SecondLevelPageStart = readNext<uint32_t>(Contents, Pos);
8165 Entry.LSDAStart = readNext<uint32_t>(Contents, Pos);
8166 IndexEntries.push_back(Entry);
8167
8168 outs() << " [" << i << "]: "
8169 << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset)
8170 << ", "
8171 << "2nd level page offset="
8172 << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", "
8173 << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n';
8174 }
8175
8176 //===----------------------------------
8177 // Next come the LSDA tables
8178 //===----------------------------------
8179
8180 // The LSDA layout is rather implicit: it's a contiguous array of entries from
8181 // the first top-level index's LSDAOffset to the last (sentinel).
8182
8183 outs() << " LSDA descriptors:\n";
8184 Pos = IndexEntries[0].LSDAStart;
8185 const uint32_t LSDASize = 2 * sizeof(uint32_t);
8186 int NumLSDAs =
8187 (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) / LSDASize;
8188
8189 for (int i = 0; i < NumLSDAs; ++i) {
8190 uint32_t FunctionOffset = readNext<uint32_t>(Contents, Pos);
8191 uint32_t LSDAOffset = readNext<uint32_t>(Contents, Pos);
8192 outs() << " [" << i << "]: "
8193 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8194 << ", "
8195 << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n';
8196 }
8197
8198 //===----------------------------------
8199 // Finally, the 2nd level indices
8200 //===----------------------------------
8201
8202 // Generally these are 4K in size, and have 2 possible forms:
8203 // + Regular stores up to 511 entries with disparate encodings
8204 // + Compressed stores up to 1021 entries if few enough compact encoding
8205 // values are used.
8206 outs() << " Second level indices:\n";
8207 for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) {
8208 // The final sentinel top-level index has no associated 2nd level page
8209 if (IndexEntries[i].SecondLevelPageStart == 0)
8210 break;
8211
8212 outs() << " Second level index[" << i << "]: "
8213 << "offset in section="
8214 << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart)
8215 << ", "
8216 << "base function offset="
8217 << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n';
8218
8219 Pos = IndexEntries[i].SecondLevelPageStart;
8220 if (Pos + sizeof(uint32_t) > Contents.size()) {
8221 outs() << "warning: invalid offset for second level page: " << Pos << '\n';
8222 continue;
8223 }
8224
8225 uint32_t Kind =
8226 *reinterpret_cast<const support::ulittle32_t *>(Contents.data() + Pos);
8227 if (Kind == 2)
8228 printRegularSecondLevelUnwindPage(Contents.substr(Pos, 4096));
8229 else if (Kind == 3)
8230 printCompressedSecondLevelUnwindPage(Contents.substr(Pos, 4096),
8231 IndexEntries[i].FunctionOffset,
8232 CommonEncodings);
8233 else
8234 outs() << " Skipping 2nd level page with unknown kind " << Kind
8235 << '\n';
8236 }
8237 }
8238
printMachOUnwindInfo(const MachOObjectFile * Obj)8239 void objdump::printMachOUnwindInfo(const MachOObjectFile *Obj) {
8240 std::map<uint64_t, SymbolRef> Symbols;
8241 for (const SymbolRef &SymRef : Obj->symbols()) {
8242 // Discard any undefined or absolute symbols. They're not going to take part
8243 // in the convenience lookup for unwind info and just take up resources.
8244 auto SectOrErr = SymRef.getSection();
8245 if (!SectOrErr) {
8246 // TODO: Actually report errors helpfully.
8247 consumeError(SectOrErr.takeError());
8248 continue;
8249 }
8250 section_iterator Section = *SectOrErr;
8251 if (Section == Obj->section_end())
8252 continue;
8253
8254 uint64_t Addr = cantFail(SymRef.getValue());
8255 Symbols.insert(std::make_pair(Addr, SymRef));
8256 }
8257
8258 for (const SectionRef &Section : Obj->sections()) {
8259 StringRef SectName;
8260 if (Expected<StringRef> NameOrErr = Section.getName())
8261 SectName = *NameOrErr;
8262 else
8263 consumeError(NameOrErr.takeError());
8264
8265 if (SectName == "__compact_unwind")
8266 printMachOCompactUnwindSection(Obj, Symbols, Section);
8267 else if (SectName == "__unwind_info")
8268 printMachOUnwindInfoSection(Obj, Symbols, Section);
8269 }
8270 }
8271
PrintMachHeader(uint32_t magic,uint32_t cputype,uint32_t cpusubtype,uint32_t filetype,uint32_t ncmds,uint32_t sizeofcmds,uint32_t flags,bool verbose)8272 static void PrintMachHeader(uint32_t magic, uint32_t cputype,
8273 uint32_t cpusubtype, uint32_t filetype,
8274 uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags,
8275 bool verbose) {
8276 outs() << "Mach header\n";
8277 outs() << " magic cputype cpusubtype caps filetype ncmds "
8278 "sizeofcmds flags\n";
8279 if (verbose) {
8280 if (magic == MachO::MH_MAGIC)
8281 outs() << " MH_MAGIC";
8282 else if (magic == MachO::MH_MAGIC_64)
8283 outs() << "MH_MAGIC_64";
8284 else
8285 outs() << format(" 0x%08" PRIx32, magic);
8286 switch (cputype) {
8287 case MachO::CPU_TYPE_I386:
8288 outs() << " I386";
8289 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8290 case MachO::CPU_SUBTYPE_I386_ALL:
8291 outs() << " ALL";
8292 break;
8293 default:
8294 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8295 break;
8296 }
8297 break;
8298 case MachO::CPU_TYPE_X86_64:
8299 outs() << " X86_64";
8300 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8301 case MachO::CPU_SUBTYPE_X86_64_ALL:
8302 outs() << " ALL";
8303 break;
8304 case MachO::CPU_SUBTYPE_X86_64_H:
8305 outs() << " Haswell";
8306 break;
8307 default:
8308 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8309 break;
8310 }
8311 break;
8312 case MachO::CPU_TYPE_ARM:
8313 outs() << " ARM";
8314 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8315 case MachO::CPU_SUBTYPE_ARM_ALL:
8316 outs() << " ALL";
8317 break;
8318 case MachO::CPU_SUBTYPE_ARM_V4T:
8319 outs() << " V4T";
8320 break;
8321 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
8322 outs() << " V5TEJ";
8323 break;
8324 case MachO::CPU_SUBTYPE_ARM_XSCALE:
8325 outs() << " XSCALE";
8326 break;
8327 case MachO::CPU_SUBTYPE_ARM_V6:
8328 outs() << " V6";
8329 break;
8330 case MachO::CPU_SUBTYPE_ARM_V6M:
8331 outs() << " V6M";
8332 break;
8333 case MachO::CPU_SUBTYPE_ARM_V7:
8334 outs() << " V7";
8335 break;
8336 case MachO::CPU_SUBTYPE_ARM_V7EM:
8337 outs() << " V7EM";
8338 break;
8339 case MachO::CPU_SUBTYPE_ARM_V7K:
8340 outs() << " V7K";
8341 break;
8342 case MachO::CPU_SUBTYPE_ARM_V7M:
8343 outs() << " V7M";
8344 break;
8345 case MachO::CPU_SUBTYPE_ARM_V7S:
8346 outs() << " V7S";
8347 break;
8348 default:
8349 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8350 break;
8351 }
8352 break;
8353 case MachO::CPU_TYPE_ARM64:
8354 outs() << " ARM64";
8355 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8356 case MachO::CPU_SUBTYPE_ARM64_ALL:
8357 outs() << " ALL";
8358 break;
8359 case MachO::CPU_SUBTYPE_ARM64_V8:
8360 outs() << " V8";
8361 break;
8362 case MachO::CPU_SUBTYPE_ARM64E:
8363 outs() << " E";
8364 break;
8365 default:
8366 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8367 break;
8368 }
8369 break;
8370 case MachO::CPU_TYPE_ARM64_32:
8371 outs() << " ARM64_32";
8372 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8373 case MachO::CPU_SUBTYPE_ARM64_32_V8:
8374 outs() << " V8";
8375 break;
8376 default:
8377 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8378 break;
8379 }
8380 break;
8381 case MachO::CPU_TYPE_POWERPC:
8382 outs() << " PPC";
8383 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8384 case MachO::CPU_SUBTYPE_POWERPC_ALL:
8385 outs() << " ALL";
8386 break;
8387 default:
8388 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8389 break;
8390 }
8391 break;
8392 case MachO::CPU_TYPE_POWERPC64:
8393 outs() << " PPC64";
8394 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8395 case MachO::CPU_SUBTYPE_POWERPC_ALL:
8396 outs() << " ALL";
8397 break;
8398 default:
8399 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8400 break;
8401 }
8402 break;
8403 default:
8404 outs() << format(" %7d", cputype);
8405 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8406 break;
8407 }
8408 if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) {
8409 outs() << " LIB64";
8410 } else {
8411 outs() << format(" 0x%02" PRIx32,
8412 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
8413 }
8414 switch (filetype) {
8415 case MachO::MH_OBJECT:
8416 outs() << " OBJECT";
8417 break;
8418 case MachO::MH_EXECUTE:
8419 outs() << " EXECUTE";
8420 break;
8421 case MachO::MH_FVMLIB:
8422 outs() << " FVMLIB";
8423 break;
8424 case MachO::MH_CORE:
8425 outs() << " CORE";
8426 break;
8427 case MachO::MH_PRELOAD:
8428 outs() << " PRELOAD";
8429 break;
8430 case MachO::MH_DYLIB:
8431 outs() << " DYLIB";
8432 break;
8433 case MachO::MH_DYLIB_STUB:
8434 outs() << " DYLIB_STUB";
8435 break;
8436 case MachO::MH_DYLINKER:
8437 outs() << " DYLINKER";
8438 break;
8439 case MachO::MH_BUNDLE:
8440 outs() << " BUNDLE";
8441 break;
8442 case MachO::MH_DSYM:
8443 outs() << " DSYM";
8444 break;
8445 case MachO::MH_KEXT_BUNDLE:
8446 outs() << " KEXTBUNDLE";
8447 break;
8448 default:
8449 outs() << format(" %10u", filetype);
8450 break;
8451 }
8452 outs() << format(" %5u", ncmds);
8453 outs() << format(" %10u", sizeofcmds);
8454 uint32_t f = flags;
8455 if (f & MachO::MH_NOUNDEFS) {
8456 outs() << " NOUNDEFS";
8457 f &= ~MachO::MH_NOUNDEFS;
8458 }
8459 if (f & MachO::MH_INCRLINK) {
8460 outs() << " INCRLINK";
8461 f &= ~MachO::MH_INCRLINK;
8462 }
8463 if (f & MachO::MH_DYLDLINK) {
8464 outs() << " DYLDLINK";
8465 f &= ~MachO::MH_DYLDLINK;
8466 }
8467 if (f & MachO::MH_BINDATLOAD) {
8468 outs() << " BINDATLOAD";
8469 f &= ~MachO::MH_BINDATLOAD;
8470 }
8471 if (f & MachO::MH_PREBOUND) {
8472 outs() << " PREBOUND";
8473 f &= ~MachO::MH_PREBOUND;
8474 }
8475 if (f & MachO::MH_SPLIT_SEGS) {
8476 outs() << " SPLIT_SEGS";
8477 f &= ~MachO::MH_SPLIT_SEGS;
8478 }
8479 if (f & MachO::MH_LAZY_INIT) {
8480 outs() << " LAZY_INIT";
8481 f &= ~MachO::MH_LAZY_INIT;
8482 }
8483 if (f & MachO::MH_TWOLEVEL) {
8484 outs() << " TWOLEVEL";
8485 f &= ~MachO::MH_TWOLEVEL;
8486 }
8487 if (f & MachO::MH_FORCE_FLAT) {
8488 outs() << " FORCE_FLAT";
8489 f &= ~MachO::MH_FORCE_FLAT;
8490 }
8491 if (f & MachO::MH_NOMULTIDEFS) {
8492 outs() << " NOMULTIDEFS";
8493 f &= ~MachO::MH_NOMULTIDEFS;
8494 }
8495 if (f & MachO::MH_NOFIXPREBINDING) {
8496 outs() << " NOFIXPREBINDING";
8497 f &= ~MachO::MH_NOFIXPREBINDING;
8498 }
8499 if (f & MachO::MH_PREBINDABLE) {
8500 outs() << " PREBINDABLE";
8501 f &= ~MachO::MH_PREBINDABLE;
8502 }
8503 if (f & MachO::MH_ALLMODSBOUND) {
8504 outs() << " ALLMODSBOUND";
8505 f &= ~MachO::MH_ALLMODSBOUND;
8506 }
8507 if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) {
8508 outs() << " SUBSECTIONS_VIA_SYMBOLS";
8509 f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
8510 }
8511 if (f & MachO::MH_CANONICAL) {
8512 outs() << " CANONICAL";
8513 f &= ~MachO::MH_CANONICAL;
8514 }
8515 if (f & MachO::MH_WEAK_DEFINES) {
8516 outs() << " WEAK_DEFINES";
8517 f &= ~MachO::MH_WEAK_DEFINES;
8518 }
8519 if (f & MachO::MH_BINDS_TO_WEAK) {
8520 outs() << " BINDS_TO_WEAK";
8521 f &= ~MachO::MH_BINDS_TO_WEAK;
8522 }
8523 if (f & MachO::MH_ALLOW_STACK_EXECUTION) {
8524 outs() << " ALLOW_STACK_EXECUTION";
8525 f &= ~MachO::MH_ALLOW_STACK_EXECUTION;
8526 }
8527 if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) {
8528 outs() << " DEAD_STRIPPABLE_DYLIB";
8529 f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB;
8530 }
8531 if (f & MachO::MH_PIE) {
8532 outs() << " PIE";
8533 f &= ~MachO::MH_PIE;
8534 }
8535 if (f & MachO::MH_NO_REEXPORTED_DYLIBS) {
8536 outs() << " NO_REEXPORTED_DYLIBS";
8537 f &= ~MachO::MH_NO_REEXPORTED_DYLIBS;
8538 }
8539 if (f & MachO::MH_HAS_TLV_DESCRIPTORS) {
8540 outs() << " MH_HAS_TLV_DESCRIPTORS";
8541 f &= ~MachO::MH_HAS_TLV_DESCRIPTORS;
8542 }
8543 if (f & MachO::MH_NO_HEAP_EXECUTION) {
8544 outs() << " MH_NO_HEAP_EXECUTION";
8545 f &= ~MachO::MH_NO_HEAP_EXECUTION;
8546 }
8547 if (f & MachO::MH_APP_EXTENSION_SAFE) {
8548 outs() << " APP_EXTENSION_SAFE";
8549 f &= ~MachO::MH_APP_EXTENSION_SAFE;
8550 }
8551 if (f & MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO) {
8552 outs() << " NLIST_OUTOFSYNC_WITH_DYLDINFO";
8553 f &= ~MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO;
8554 }
8555 if (f != 0 || flags == 0)
8556 outs() << format(" 0x%08" PRIx32, f);
8557 } else {
8558 outs() << format(" 0x%08" PRIx32, magic);
8559 outs() << format(" %7d", cputype);
8560 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8561 outs() << format(" 0x%02" PRIx32,
8562 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
8563 outs() << format(" %10u", filetype);
8564 outs() << format(" %5u", ncmds);
8565 outs() << format(" %10u", sizeofcmds);
8566 outs() << format(" 0x%08" PRIx32, flags);
8567 }
8568 outs() << "\n";
8569 }
8570
PrintSegmentCommand(uint32_t cmd,uint32_t cmdsize,StringRef SegName,uint64_t vmaddr,uint64_t vmsize,uint64_t fileoff,uint64_t filesize,uint32_t maxprot,uint32_t initprot,uint32_t nsects,uint32_t flags,uint32_t object_size,bool verbose)8571 static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize,
8572 StringRef SegName, uint64_t vmaddr,
8573 uint64_t vmsize, uint64_t fileoff,
8574 uint64_t filesize, uint32_t maxprot,
8575 uint32_t initprot, uint32_t nsects,
8576 uint32_t flags, uint32_t object_size,
8577 bool verbose) {
8578 uint64_t expected_cmdsize;
8579 if (cmd == MachO::LC_SEGMENT) {
8580 outs() << " cmd LC_SEGMENT\n";
8581 expected_cmdsize = nsects;
8582 expected_cmdsize *= sizeof(struct MachO::section);
8583 expected_cmdsize += sizeof(struct MachO::segment_command);
8584 } else {
8585 outs() << " cmd LC_SEGMENT_64\n";
8586 expected_cmdsize = nsects;
8587 expected_cmdsize *= sizeof(struct MachO::section_64);
8588 expected_cmdsize += sizeof(struct MachO::segment_command_64);
8589 }
8590 outs() << " cmdsize " << cmdsize;
8591 if (cmdsize != expected_cmdsize)
8592 outs() << " Inconsistent size\n";
8593 else
8594 outs() << "\n";
8595 outs() << " segname " << SegName << "\n";
8596 if (cmd == MachO::LC_SEGMENT_64) {
8597 outs() << " vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n";
8598 outs() << " vmsize " << format("0x%016" PRIx64, vmsize) << "\n";
8599 } else {
8600 outs() << " vmaddr " << format("0x%08" PRIx64, vmaddr) << "\n";
8601 outs() << " vmsize " << format("0x%08" PRIx64, vmsize) << "\n";
8602 }
8603 outs() << " fileoff " << fileoff;
8604 if (fileoff > object_size)
8605 outs() << " (past end of file)\n";
8606 else
8607 outs() << "\n";
8608 outs() << " filesize " << filesize;
8609 if (fileoff + filesize > object_size)
8610 outs() << " (past end of file)\n";
8611 else
8612 outs() << "\n";
8613 if (verbose) {
8614 if ((maxprot &
8615 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
8616 MachO::VM_PROT_EXECUTE)) != 0)
8617 outs() << " maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n";
8618 else {
8619 outs() << " maxprot ";
8620 outs() << ((maxprot & MachO::VM_PROT_READ) ? "r" : "-");
8621 outs() << ((maxprot & MachO::VM_PROT_WRITE) ? "w" : "-");
8622 outs() << ((maxprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n");
8623 }
8624 if ((initprot &
8625 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
8626 MachO::VM_PROT_EXECUTE)) != 0)
8627 outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n";
8628 else {
8629 outs() << " initprot ";
8630 outs() << ((initprot & MachO::VM_PROT_READ) ? "r" : "-");
8631 outs() << ((initprot & MachO::VM_PROT_WRITE) ? "w" : "-");
8632 outs() << ((initprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n");
8633 }
8634 } else {
8635 outs() << " maxprot " << format("0x%08" PRIx32, maxprot) << "\n";
8636 outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n";
8637 }
8638 outs() << " nsects " << nsects << "\n";
8639 if (verbose) {
8640 outs() << " flags";
8641 if (flags == 0)
8642 outs() << " (none)\n";
8643 else {
8644 if (flags & MachO::SG_HIGHVM) {
8645 outs() << " HIGHVM";
8646 flags &= ~MachO::SG_HIGHVM;
8647 }
8648 if (flags & MachO::SG_FVMLIB) {
8649 outs() << " FVMLIB";
8650 flags &= ~MachO::SG_FVMLIB;
8651 }
8652 if (flags & MachO::SG_NORELOC) {
8653 outs() << " NORELOC";
8654 flags &= ~MachO::SG_NORELOC;
8655 }
8656 if (flags & MachO::SG_PROTECTED_VERSION_1) {
8657 outs() << " PROTECTED_VERSION_1";
8658 flags &= ~MachO::SG_PROTECTED_VERSION_1;
8659 }
8660 if (flags)
8661 outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n";
8662 else
8663 outs() << "\n";
8664 }
8665 } else {
8666 outs() << " flags " << format("0x%" PRIx32, flags) << "\n";
8667 }
8668 }
8669
PrintSection(const char * sectname,const char * segname,uint64_t addr,uint64_t size,uint32_t offset,uint32_t align,uint32_t reloff,uint32_t nreloc,uint32_t flags,uint32_t reserved1,uint32_t reserved2,uint32_t cmd,const char * sg_segname,uint32_t filetype,uint32_t object_size,bool verbose)8670 static void PrintSection(const char *sectname, const char *segname,
8671 uint64_t addr, uint64_t size, uint32_t offset,
8672 uint32_t align, uint32_t reloff, uint32_t nreloc,
8673 uint32_t flags, uint32_t reserved1, uint32_t reserved2,
8674 uint32_t cmd, const char *sg_segname,
8675 uint32_t filetype, uint32_t object_size,
8676 bool verbose) {
8677 outs() << "Section\n";
8678 outs() << " sectname " << format("%.16s\n", sectname);
8679 outs() << " segname " << format("%.16s", segname);
8680 if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0)
8681 outs() << " (does not match segment)\n";
8682 else
8683 outs() << "\n";
8684 if (cmd == MachO::LC_SEGMENT_64) {
8685 outs() << " addr " << format("0x%016" PRIx64, addr) << "\n";
8686 outs() << " size " << format("0x%016" PRIx64, size);
8687 } else {
8688 outs() << " addr " << format("0x%08" PRIx64, addr) << "\n";
8689 outs() << " size " << format("0x%08" PRIx64, size);
8690 }
8691 if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size)
8692 outs() << " (past end of file)\n";
8693 else
8694 outs() << "\n";
8695 outs() << " offset " << offset;
8696 if (offset > object_size)
8697 outs() << " (past end of file)\n";
8698 else
8699 outs() << "\n";
8700 uint32_t align_shifted = 1 << align;
8701 outs() << " align 2^" << align << " (" << align_shifted << ")\n";
8702 outs() << " reloff " << reloff;
8703 if (reloff > object_size)
8704 outs() << " (past end of file)\n";
8705 else
8706 outs() << "\n";
8707 outs() << " nreloc " << nreloc;
8708 if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size)
8709 outs() << " (past end of file)\n";
8710 else
8711 outs() << "\n";
8712 uint32_t section_type = flags & MachO::SECTION_TYPE;
8713 if (verbose) {
8714 outs() << " type";
8715 if (section_type == MachO::S_REGULAR)
8716 outs() << " S_REGULAR\n";
8717 else if (section_type == MachO::S_ZEROFILL)
8718 outs() << " S_ZEROFILL\n";
8719 else if (section_type == MachO::S_CSTRING_LITERALS)
8720 outs() << " S_CSTRING_LITERALS\n";
8721 else if (section_type == MachO::S_4BYTE_LITERALS)
8722 outs() << " S_4BYTE_LITERALS\n";
8723 else if (section_type == MachO::S_8BYTE_LITERALS)
8724 outs() << " S_8BYTE_LITERALS\n";
8725 else if (section_type == MachO::S_16BYTE_LITERALS)
8726 outs() << " S_16BYTE_LITERALS\n";
8727 else if (section_type == MachO::S_LITERAL_POINTERS)
8728 outs() << " S_LITERAL_POINTERS\n";
8729 else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS)
8730 outs() << " S_NON_LAZY_SYMBOL_POINTERS\n";
8731 else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS)
8732 outs() << " S_LAZY_SYMBOL_POINTERS\n";
8733 else if (section_type == MachO::S_SYMBOL_STUBS)
8734 outs() << " S_SYMBOL_STUBS\n";
8735 else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS)
8736 outs() << " S_MOD_INIT_FUNC_POINTERS\n";
8737 else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS)
8738 outs() << " S_MOD_TERM_FUNC_POINTERS\n";
8739 else if (section_type == MachO::S_COALESCED)
8740 outs() << " S_COALESCED\n";
8741 else if (section_type == MachO::S_INTERPOSING)
8742 outs() << " S_INTERPOSING\n";
8743 else if (section_type == MachO::S_DTRACE_DOF)
8744 outs() << " S_DTRACE_DOF\n";
8745 else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS)
8746 outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n";
8747 else if (section_type == MachO::S_THREAD_LOCAL_REGULAR)
8748 outs() << " S_THREAD_LOCAL_REGULAR\n";
8749 else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL)
8750 outs() << " S_THREAD_LOCAL_ZEROFILL\n";
8751 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES)
8752 outs() << " S_THREAD_LOCAL_VARIABLES\n";
8753 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
8754 outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n";
8755 else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS)
8756 outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n";
8757 else
8758 outs() << format("0x%08" PRIx32, section_type) << "\n";
8759 outs() << "attributes";
8760 uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES;
8761 if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS)
8762 outs() << " PURE_INSTRUCTIONS";
8763 if (section_attributes & MachO::S_ATTR_NO_TOC)
8764 outs() << " NO_TOC";
8765 if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS)
8766 outs() << " STRIP_STATIC_SYMS";
8767 if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP)
8768 outs() << " NO_DEAD_STRIP";
8769 if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT)
8770 outs() << " LIVE_SUPPORT";
8771 if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE)
8772 outs() << " SELF_MODIFYING_CODE";
8773 if (section_attributes & MachO::S_ATTR_DEBUG)
8774 outs() << " DEBUG";
8775 if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS)
8776 outs() << " SOME_INSTRUCTIONS";
8777 if (section_attributes & MachO::S_ATTR_EXT_RELOC)
8778 outs() << " EXT_RELOC";
8779 if (section_attributes & MachO::S_ATTR_LOC_RELOC)
8780 outs() << " LOC_RELOC";
8781 if (section_attributes == 0)
8782 outs() << " (none)";
8783 outs() << "\n";
8784 } else
8785 outs() << " flags " << format("0x%08" PRIx32, flags) << "\n";
8786 outs() << " reserved1 " << reserved1;
8787 if (section_type == MachO::S_SYMBOL_STUBS ||
8788 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
8789 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
8790 section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
8791 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
8792 outs() << " (index into indirect symbol table)\n";
8793 else
8794 outs() << "\n";
8795 outs() << " reserved2 " << reserved2;
8796 if (section_type == MachO::S_SYMBOL_STUBS)
8797 outs() << " (size of stubs)\n";
8798 else
8799 outs() << "\n";
8800 }
8801
PrintSymtabLoadCommand(MachO::symtab_command st,bool Is64Bit,uint32_t object_size)8802 static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit,
8803 uint32_t object_size) {
8804 outs() << " cmd LC_SYMTAB\n";
8805 outs() << " cmdsize " << st.cmdsize;
8806 if (st.cmdsize != sizeof(struct MachO::symtab_command))
8807 outs() << " Incorrect size\n";
8808 else
8809 outs() << "\n";
8810 outs() << " symoff " << st.symoff;
8811 if (st.symoff > object_size)
8812 outs() << " (past end of file)\n";
8813 else
8814 outs() << "\n";
8815 outs() << " nsyms " << st.nsyms;
8816 uint64_t big_size;
8817 if (Is64Bit) {
8818 big_size = st.nsyms;
8819 big_size *= sizeof(struct MachO::nlist_64);
8820 big_size += st.symoff;
8821 if (big_size > object_size)
8822 outs() << " (past end of file)\n";
8823 else
8824 outs() << "\n";
8825 } else {
8826 big_size = st.nsyms;
8827 big_size *= sizeof(struct MachO::nlist);
8828 big_size += st.symoff;
8829 if (big_size > object_size)
8830 outs() << " (past end of file)\n";
8831 else
8832 outs() << "\n";
8833 }
8834 outs() << " stroff " << st.stroff;
8835 if (st.stroff > object_size)
8836 outs() << " (past end of file)\n";
8837 else
8838 outs() << "\n";
8839 outs() << " strsize " << st.strsize;
8840 big_size = st.stroff;
8841 big_size += st.strsize;
8842 if (big_size > object_size)
8843 outs() << " (past end of file)\n";
8844 else
8845 outs() << "\n";
8846 }
8847
PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,uint32_t nsyms,uint32_t object_size,bool Is64Bit)8848 static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,
8849 uint32_t nsyms, uint32_t object_size,
8850 bool Is64Bit) {
8851 outs() << " cmd LC_DYSYMTAB\n";
8852 outs() << " cmdsize " << dyst.cmdsize;
8853 if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command))
8854 outs() << " Incorrect size\n";
8855 else
8856 outs() << "\n";
8857 outs() << " ilocalsym " << dyst.ilocalsym;
8858 if (dyst.ilocalsym > nsyms)
8859 outs() << " (greater than the number of symbols)\n";
8860 else
8861 outs() << "\n";
8862 outs() << " nlocalsym " << dyst.nlocalsym;
8863 uint64_t big_size;
8864 big_size = dyst.ilocalsym;
8865 big_size += dyst.nlocalsym;
8866 if (big_size > nsyms)
8867 outs() << " (past the end of the symbol table)\n";
8868 else
8869 outs() << "\n";
8870 outs() << " iextdefsym " << dyst.iextdefsym;
8871 if (dyst.iextdefsym > nsyms)
8872 outs() << " (greater than the number of symbols)\n";
8873 else
8874 outs() << "\n";
8875 outs() << " nextdefsym " << dyst.nextdefsym;
8876 big_size = dyst.iextdefsym;
8877 big_size += dyst.nextdefsym;
8878 if (big_size > nsyms)
8879 outs() << " (past the end of the symbol table)\n";
8880 else
8881 outs() << "\n";
8882 outs() << " iundefsym " << dyst.iundefsym;
8883 if (dyst.iundefsym > nsyms)
8884 outs() << " (greater than the number of symbols)\n";
8885 else
8886 outs() << "\n";
8887 outs() << " nundefsym " << dyst.nundefsym;
8888 big_size = dyst.iundefsym;
8889 big_size += dyst.nundefsym;
8890 if (big_size > nsyms)
8891 outs() << " (past the end of the symbol table)\n";
8892 else
8893 outs() << "\n";
8894 outs() << " tocoff " << dyst.tocoff;
8895 if (dyst.tocoff > object_size)
8896 outs() << " (past end of file)\n";
8897 else
8898 outs() << "\n";
8899 outs() << " ntoc " << dyst.ntoc;
8900 big_size = dyst.ntoc;
8901 big_size *= sizeof(struct MachO::dylib_table_of_contents);
8902 big_size += dyst.tocoff;
8903 if (big_size > object_size)
8904 outs() << " (past end of file)\n";
8905 else
8906 outs() << "\n";
8907 outs() << " modtaboff " << dyst.modtaboff;
8908 if (dyst.modtaboff > object_size)
8909 outs() << " (past end of file)\n";
8910 else
8911 outs() << "\n";
8912 outs() << " nmodtab " << dyst.nmodtab;
8913 uint64_t modtabend;
8914 if (Is64Bit) {
8915 modtabend = dyst.nmodtab;
8916 modtabend *= sizeof(struct MachO::dylib_module_64);
8917 modtabend += dyst.modtaboff;
8918 } else {
8919 modtabend = dyst.nmodtab;
8920 modtabend *= sizeof(struct MachO::dylib_module);
8921 modtabend += dyst.modtaboff;
8922 }
8923 if (modtabend > object_size)
8924 outs() << " (past end of file)\n";
8925 else
8926 outs() << "\n";
8927 outs() << " extrefsymoff " << dyst.extrefsymoff;
8928 if (dyst.extrefsymoff > object_size)
8929 outs() << " (past end of file)\n";
8930 else
8931 outs() << "\n";
8932 outs() << " nextrefsyms " << dyst.nextrefsyms;
8933 big_size = dyst.nextrefsyms;
8934 big_size *= sizeof(struct MachO::dylib_reference);
8935 big_size += dyst.extrefsymoff;
8936 if (big_size > object_size)
8937 outs() << " (past end of file)\n";
8938 else
8939 outs() << "\n";
8940 outs() << " indirectsymoff " << dyst.indirectsymoff;
8941 if (dyst.indirectsymoff > object_size)
8942 outs() << " (past end of file)\n";
8943 else
8944 outs() << "\n";
8945 outs() << " nindirectsyms " << dyst.nindirectsyms;
8946 big_size = dyst.nindirectsyms;
8947 big_size *= sizeof(uint32_t);
8948 big_size += dyst.indirectsymoff;
8949 if (big_size > object_size)
8950 outs() << " (past end of file)\n";
8951 else
8952 outs() << "\n";
8953 outs() << " extreloff " << dyst.extreloff;
8954 if (dyst.extreloff > object_size)
8955 outs() << " (past end of file)\n";
8956 else
8957 outs() << "\n";
8958 outs() << " nextrel " << dyst.nextrel;
8959 big_size = dyst.nextrel;
8960 big_size *= sizeof(struct MachO::relocation_info);
8961 big_size += dyst.extreloff;
8962 if (big_size > object_size)
8963 outs() << " (past end of file)\n";
8964 else
8965 outs() << "\n";
8966 outs() << " locreloff " << dyst.locreloff;
8967 if (dyst.locreloff > object_size)
8968 outs() << " (past end of file)\n";
8969 else
8970 outs() << "\n";
8971 outs() << " nlocrel " << dyst.nlocrel;
8972 big_size = dyst.nlocrel;
8973 big_size *= sizeof(struct MachO::relocation_info);
8974 big_size += dyst.locreloff;
8975 if (big_size > object_size)
8976 outs() << " (past end of file)\n";
8977 else
8978 outs() << "\n";
8979 }
8980
PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,uint32_t object_size)8981 static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,
8982 uint32_t object_size) {
8983 if (dc.cmd == MachO::LC_DYLD_INFO)
8984 outs() << " cmd LC_DYLD_INFO\n";
8985 else
8986 outs() << " cmd LC_DYLD_INFO_ONLY\n";
8987 outs() << " cmdsize " << dc.cmdsize;
8988 if (dc.cmdsize != sizeof(struct MachO::dyld_info_command))
8989 outs() << " Incorrect size\n";
8990 else
8991 outs() << "\n";
8992 outs() << " rebase_off " << dc.rebase_off;
8993 if (dc.rebase_off > object_size)
8994 outs() << " (past end of file)\n";
8995 else
8996 outs() << "\n";
8997 outs() << " rebase_size " << dc.rebase_size;
8998 uint64_t big_size;
8999 big_size = dc.rebase_off;
9000 big_size += dc.rebase_size;
9001 if (big_size > object_size)
9002 outs() << " (past end of file)\n";
9003 else
9004 outs() << "\n";
9005 outs() << " bind_off " << dc.bind_off;
9006 if (dc.bind_off > object_size)
9007 outs() << " (past end of file)\n";
9008 else
9009 outs() << "\n";
9010 outs() << " bind_size " << dc.bind_size;
9011 big_size = dc.bind_off;
9012 big_size += dc.bind_size;
9013 if (big_size > object_size)
9014 outs() << " (past end of file)\n";
9015 else
9016 outs() << "\n";
9017 outs() << " weak_bind_off " << dc.weak_bind_off;
9018 if (dc.weak_bind_off > object_size)
9019 outs() << " (past end of file)\n";
9020 else
9021 outs() << "\n";
9022 outs() << " weak_bind_size " << dc.weak_bind_size;
9023 big_size = dc.weak_bind_off;
9024 big_size += dc.weak_bind_size;
9025 if (big_size > object_size)
9026 outs() << " (past end of file)\n";
9027 else
9028 outs() << "\n";
9029 outs() << " lazy_bind_off " << dc.lazy_bind_off;
9030 if (dc.lazy_bind_off > object_size)
9031 outs() << " (past end of file)\n";
9032 else
9033 outs() << "\n";
9034 outs() << " lazy_bind_size " << dc.lazy_bind_size;
9035 big_size = dc.lazy_bind_off;
9036 big_size += dc.lazy_bind_size;
9037 if (big_size > object_size)
9038 outs() << " (past end of file)\n";
9039 else
9040 outs() << "\n";
9041 outs() << " export_off " << dc.export_off;
9042 if (dc.export_off > object_size)
9043 outs() << " (past end of file)\n";
9044 else
9045 outs() << "\n";
9046 outs() << " export_size " << dc.export_size;
9047 big_size = dc.export_off;
9048 big_size += dc.export_size;
9049 if (big_size > object_size)
9050 outs() << " (past end of file)\n";
9051 else
9052 outs() << "\n";
9053 }
9054
PrintDyldLoadCommand(MachO::dylinker_command dyld,const char * Ptr)9055 static void PrintDyldLoadCommand(MachO::dylinker_command dyld,
9056 const char *Ptr) {
9057 if (dyld.cmd == MachO::LC_ID_DYLINKER)
9058 outs() << " cmd LC_ID_DYLINKER\n";
9059 else if (dyld.cmd == MachO::LC_LOAD_DYLINKER)
9060 outs() << " cmd LC_LOAD_DYLINKER\n";
9061 else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT)
9062 outs() << " cmd LC_DYLD_ENVIRONMENT\n";
9063 else
9064 outs() << " cmd ?(" << dyld.cmd << ")\n";
9065 outs() << " cmdsize " << dyld.cmdsize;
9066 if (dyld.cmdsize < sizeof(struct MachO::dylinker_command))
9067 outs() << " Incorrect size\n";
9068 else
9069 outs() << "\n";
9070 if (dyld.name >= dyld.cmdsize)
9071 outs() << " name ?(bad offset " << dyld.name << ")\n";
9072 else {
9073 const char *P = (const char *)(Ptr) + dyld.name;
9074 outs() << " name " << P << " (offset " << dyld.name << ")\n";
9075 }
9076 }
9077
PrintUuidLoadCommand(MachO::uuid_command uuid)9078 static void PrintUuidLoadCommand(MachO::uuid_command uuid) {
9079 outs() << " cmd LC_UUID\n";
9080 outs() << " cmdsize " << uuid.cmdsize;
9081 if (uuid.cmdsize != sizeof(struct MachO::uuid_command))
9082 outs() << " Incorrect size\n";
9083 else
9084 outs() << "\n";
9085 outs() << " uuid ";
9086 for (int i = 0; i < 16; ++i) {
9087 outs() << format("%02" PRIX32, uuid.uuid[i]);
9088 if (i == 3 || i == 5 || i == 7 || i == 9)
9089 outs() << "-";
9090 }
9091 outs() << "\n";
9092 }
9093
PrintRpathLoadCommand(MachO::rpath_command rpath,const char * Ptr)9094 static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) {
9095 outs() << " cmd LC_RPATH\n";
9096 outs() << " cmdsize " << rpath.cmdsize;
9097 if (rpath.cmdsize < sizeof(struct MachO::rpath_command))
9098 outs() << " Incorrect size\n";
9099 else
9100 outs() << "\n";
9101 if (rpath.path >= rpath.cmdsize)
9102 outs() << " path ?(bad offset " << rpath.path << ")\n";
9103 else {
9104 const char *P = (const char *)(Ptr) + rpath.path;
9105 outs() << " path " << P << " (offset " << rpath.path << ")\n";
9106 }
9107 }
9108
PrintVersionMinLoadCommand(MachO::version_min_command vd)9109 static void PrintVersionMinLoadCommand(MachO::version_min_command vd) {
9110 StringRef LoadCmdName;
9111 switch (vd.cmd) {
9112 case MachO::LC_VERSION_MIN_MACOSX:
9113 LoadCmdName = "LC_VERSION_MIN_MACOSX";
9114 break;
9115 case MachO::LC_VERSION_MIN_IPHONEOS:
9116 LoadCmdName = "LC_VERSION_MIN_IPHONEOS";
9117 break;
9118 case MachO::LC_VERSION_MIN_TVOS:
9119 LoadCmdName = "LC_VERSION_MIN_TVOS";
9120 break;
9121 case MachO::LC_VERSION_MIN_WATCHOS:
9122 LoadCmdName = "LC_VERSION_MIN_WATCHOS";
9123 break;
9124 default:
9125 llvm_unreachable("Unknown version min load command");
9126 }
9127
9128 outs() << " cmd " << LoadCmdName << '\n';
9129 outs() << " cmdsize " << vd.cmdsize;
9130 if (vd.cmdsize != sizeof(struct MachO::version_min_command))
9131 outs() << " Incorrect size\n";
9132 else
9133 outs() << "\n";
9134 outs() << " version "
9135 << MachOObjectFile::getVersionMinMajor(vd, false) << "."
9136 << MachOObjectFile::getVersionMinMinor(vd, false);
9137 uint32_t Update = MachOObjectFile::getVersionMinUpdate(vd, false);
9138 if (Update != 0)
9139 outs() << "." << Update;
9140 outs() << "\n";
9141 if (vd.sdk == 0)
9142 outs() << " sdk n/a";
9143 else {
9144 outs() << " sdk "
9145 << MachOObjectFile::getVersionMinMajor(vd, true) << "."
9146 << MachOObjectFile::getVersionMinMinor(vd, true);
9147 }
9148 Update = MachOObjectFile::getVersionMinUpdate(vd, true);
9149 if (Update != 0)
9150 outs() << "." << Update;
9151 outs() << "\n";
9152 }
9153
PrintNoteLoadCommand(MachO::note_command Nt)9154 static void PrintNoteLoadCommand(MachO::note_command Nt) {
9155 outs() << " cmd LC_NOTE\n";
9156 outs() << " cmdsize " << Nt.cmdsize;
9157 if (Nt.cmdsize != sizeof(struct MachO::note_command))
9158 outs() << " Incorrect size\n";
9159 else
9160 outs() << "\n";
9161 const char *d = Nt.data_owner;
9162 outs() << "data_owner " << format("%.16s\n", d);
9163 outs() << " offset " << Nt.offset << "\n";
9164 outs() << " size " << Nt.size << "\n";
9165 }
9166
PrintBuildToolVersion(MachO::build_tool_version bv,bool verbose)9167 static void PrintBuildToolVersion(MachO::build_tool_version bv, bool verbose) {
9168 outs() << " tool ";
9169 if (verbose)
9170 outs() << MachOObjectFile::getBuildTool(bv.tool);
9171 else
9172 outs() << bv.tool;
9173 outs() << "\n";
9174 outs() << " version " << MachOObjectFile::getVersionString(bv.version)
9175 << "\n";
9176 }
9177
PrintBuildVersionLoadCommand(const MachOObjectFile * obj,MachO::build_version_command bd,bool verbose)9178 static void PrintBuildVersionLoadCommand(const MachOObjectFile *obj,
9179 MachO::build_version_command bd,
9180 bool verbose) {
9181 outs() << " cmd LC_BUILD_VERSION\n";
9182 outs() << " cmdsize " << bd.cmdsize;
9183 if (bd.cmdsize !=
9184 sizeof(struct MachO::build_version_command) +
9185 bd.ntools * sizeof(struct MachO::build_tool_version))
9186 outs() << " Incorrect size\n";
9187 else
9188 outs() << "\n";
9189 outs() << " platform ";
9190 if (verbose)
9191 outs() << MachOObjectFile::getBuildPlatform(bd.platform);
9192 else
9193 outs() << bd.platform;
9194 outs() << "\n";
9195 if (bd.sdk)
9196 outs() << " sdk " << MachOObjectFile::getVersionString(bd.sdk)
9197 << "\n";
9198 else
9199 outs() << " sdk n/a\n";
9200 outs() << " minos " << MachOObjectFile::getVersionString(bd.minos)
9201 << "\n";
9202 outs() << " ntools " << bd.ntools << "\n";
9203 for (unsigned i = 0; i < bd.ntools; ++i) {
9204 MachO::build_tool_version bv = obj->getBuildToolVersion(i);
9205 PrintBuildToolVersion(bv, verbose);
9206 }
9207 }
9208
PrintSourceVersionCommand(MachO::source_version_command sd)9209 static void PrintSourceVersionCommand(MachO::source_version_command sd) {
9210 outs() << " cmd LC_SOURCE_VERSION\n";
9211 outs() << " cmdsize " << sd.cmdsize;
9212 if (sd.cmdsize != sizeof(struct MachO::source_version_command))
9213 outs() << " Incorrect size\n";
9214 else
9215 outs() << "\n";
9216 uint64_t a = (sd.version >> 40) & 0xffffff;
9217 uint64_t b = (sd.version >> 30) & 0x3ff;
9218 uint64_t c = (sd.version >> 20) & 0x3ff;
9219 uint64_t d = (sd.version >> 10) & 0x3ff;
9220 uint64_t e = sd.version & 0x3ff;
9221 outs() << " version " << a << "." << b;
9222 if (e != 0)
9223 outs() << "." << c << "." << d << "." << e;
9224 else if (d != 0)
9225 outs() << "." << c << "." << d;
9226 else if (c != 0)
9227 outs() << "." << c;
9228 outs() << "\n";
9229 }
9230
PrintEntryPointCommand(MachO::entry_point_command ep)9231 static void PrintEntryPointCommand(MachO::entry_point_command ep) {
9232 outs() << " cmd LC_MAIN\n";
9233 outs() << " cmdsize " << ep.cmdsize;
9234 if (ep.cmdsize != sizeof(struct MachO::entry_point_command))
9235 outs() << " Incorrect size\n";
9236 else
9237 outs() << "\n";
9238 outs() << " entryoff " << ep.entryoff << "\n";
9239 outs() << " stacksize " << ep.stacksize << "\n";
9240 }
9241
PrintEncryptionInfoCommand(MachO::encryption_info_command ec,uint32_t object_size)9242 static void PrintEncryptionInfoCommand(MachO::encryption_info_command ec,
9243 uint32_t object_size) {
9244 outs() << " cmd LC_ENCRYPTION_INFO\n";
9245 outs() << " cmdsize " << ec.cmdsize;
9246 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command))
9247 outs() << " Incorrect size\n";
9248 else
9249 outs() << "\n";
9250 outs() << " cryptoff " << ec.cryptoff;
9251 if (ec.cryptoff > object_size)
9252 outs() << " (past end of file)\n";
9253 else
9254 outs() << "\n";
9255 outs() << " cryptsize " << ec.cryptsize;
9256 if (ec.cryptsize > object_size)
9257 outs() << " (past end of file)\n";
9258 else
9259 outs() << "\n";
9260 outs() << " cryptid " << ec.cryptid << "\n";
9261 }
9262
PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,uint32_t object_size)9263 static void PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,
9264 uint32_t object_size) {
9265 outs() << " cmd LC_ENCRYPTION_INFO_64\n";
9266 outs() << " cmdsize " << ec.cmdsize;
9267 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command_64))
9268 outs() << " Incorrect size\n";
9269 else
9270 outs() << "\n";
9271 outs() << " cryptoff " << ec.cryptoff;
9272 if (ec.cryptoff > object_size)
9273 outs() << " (past end of file)\n";
9274 else
9275 outs() << "\n";
9276 outs() << " cryptsize " << ec.cryptsize;
9277 if (ec.cryptsize > object_size)
9278 outs() << " (past end of file)\n";
9279 else
9280 outs() << "\n";
9281 outs() << " cryptid " << ec.cryptid << "\n";
9282 outs() << " pad " << ec.pad << "\n";
9283 }
9284
PrintLinkerOptionCommand(MachO::linker_option_command lo,const char * Ptr)9285 static void PrintLinkerOptionCommand(MachO::linker_option_command lo,
9286 const char *Ptr) {
9287 outs() << " cmd LC_LINKER_OPTION\n";
9288 outs() << " cmdsize " << lo.cmdsize;
9289 if (lo.cmdsize < sizeof(struct MachO::linker_option_command))
9290 outs() << " Incorrect size\n";
9291 else
9292 outs() << "\n";
9293 outs() << " count " << lo.count << "\n";
9294 const char *string = Ptr + sizeof(struct MachO::linker_option_command);
9295 uint32_t left = lo.cmdsize - sizeof(struct MachO::linker_option_command);
9296 uint32_t i = 0;
9297 while (left > 0) {
9298 while (*string == '\0' && left > 0) {
9299 string++;
9300 left--;
9301 }
9302 if (left > 0) {
9303 i++;
9304 outs() << " string #" << i << " " << format("%.*s\n", left, string);
9305 uint32_t NullPos = StringRef(string, left).find('\0');
9306 uint32_t len = std::min(NullPos, left) + 1;
9307 string += len;
9308 left -= len;
9309 }
9310 }
9311 if (lo.count != i)
9312 outs() << " count " << lo.count << " does not match number of strings "
9313 << i << "\n";
9314 }
9315
PrintSubFrameworkCommand(MachO::sub_framework_command sub,const char * Ptr)9316 static void PrintSubFrameworkCommand(MachO::sub_framework_command sub,
9317 const char *Ptr) {
9318 outs() << " cmd LC_SUB_FRAMEWORK\n";
9319 outs() << " cmdsize " << sub.cmdsize;
9320 if (sub.cmdsize < sizeof(struct MachO::sub_framework_command))
9321 outs() << " Incorrect size\n";
9322 else
9323 outs() << "\n";
9324 if (sub.umbrella < sub.cmdsize) {
9325 const char *P = Ptr + sub.umbrella;
9326 outs() << " umbrella " << P << " (offset " << sub.umbrella << ")\n";
9327 } else {
9328 outs() << " umbrella ?(bad offset " << sub.umbrella << ")\n";
9329 }
9330 }
9331
PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,const char * Ptr)9332 static void PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,
9333 const char *Ptr) {
9334 outs() << " cmd LC_SUB_UMBRELLA\n";
9335 outs() << " cmdsize " << sub.cmdsize;
9336 if (sub.cmdsize < sizeof(struct MachO::sub_umbrella_command))
9337 outs() << " Incorrect size\n";
9338 else
9339 outs() << "\n";
9340 if (sub.sub_umbrella < sub.cmdsize) {
9341 const char *P = Ptr + sub.sub_umbrella;
9342 outs() << " sub_umbrella " << P << " (offset " << sub.sub_umbrella << ")\n";
9343 } else {
9344 outs() << " sub_umbrella ?(bad offset " << sub.sub_umbrella << ")\n";
9345 }
9346 }
9347
PrintSubLibraryCommand(MachO::sub_library_command sub,const char * Ptr)9348 static void PrintSubLibraryCommand(MachO::sub_library_command sub,
9349 const char *Ptr) {
9350 outs() << " cmd LC_SUB_LIBRARY\n";
9351 outs() << " cmdsize " << sub.cmdsize;
9352 if (sub.cmdsize < sizeof(struct MachO::sub_library_command))
9353 outs() << " Incorrect size\n";
9354 else
9355 outs() << "\n";
9356 if (sub.sub_library < sub.cmdsize) {
9357 const char *P = Ptr + sub.sub_library;
9358 outs() << " sub_library " << P << " (offset " << sub.sub_library << ")\n";
9359 } else {
9360 outs() << " sub_library ?(bad offset " << sub.sub_library << ")\n";
9361 }
9362 }
9363
PrintSubClientCommand(MachO::sub_client_command sub,const char * Ptr)9364 static void PrintSubClientCommand(MachO::sub_client_command sub,
9365 const char *Ptr) {
9366 outs() << " cmd LC_SUB_CLIENT\n";
9367 outs() << " cmdsize " << sub.cmdsize;
9368 if (sub.cmdsize < sizeof(struct MachO::sub_client_command))
9369 outs() << " Incorrect size\n";
9370 else
9371 outs() << "\n";
9372 if (sub.client < sub.cmdsize) {
9373 const char *P = Ptr + sub.client;
9374 outs() << " client " << P << " (offset " << sub.client << ")\n";
9375 } else {
9376 outs() << " client ?(bad offset " << sub.client << ")\n";
9377 }
9378 }
9379
PrintRoutinesCommand(MachO::routines_command r)9380 static void PrintRoutinesCommand(MachO::routines_command r) {
9381 outs() << " cmd LC_ROUTINES\n";
9382 outs() << " cmdsize " << r.cmdsize;
9383 if (r.cmdsize != sizeof(struct MachO::routines_command))
9384 outs() << " Incorrect size\n";
9385 else
9386 outs() << "\n";
9387 outs() << " init_address " << format("0x%08" PRIx32, r.init_address) << "\n";
9388 outs() << " init_module " << r.init_module << "\n";
9389 outs() << " reserved1 " << r.reserved1 << "\n";
9390 outs() << " reserved2 " << r.reserved2 << "\n";
9391 outs() << " reserved3 " << r.reserved3 << "\n";
9392 outs() << " reserved4 " << r.reserved4 << "\n";
9393 outs() << " reserved5 " << r.reserved5 << "\n";
9394 outs() << " reserved6 " << r.reserved6 << "\n";
9395 }
9396
PrintRoutinesCommand64(MachO::routines_command_64 r)9397 static void PrintRoutinesCommand64(MachO::routines_command_64 r) {
9398 outs() << " cmd LC_ROUTINES_64\n";
9399 outs() << " cmdsize " << r.cmdsize;
9400 if (r.cmdsize != sizeof(struct MachO::routines_command_64))
9401 outs() << " Incorrect size\n";
9402 else
9403 outs() << "\n";
9404 outs() << " init_address " << format("0x%016" PRIx64, r.init_address) << "\n";
9405 outs() << " init_module " << r.init_module << "\n";
9406 outs() << " reserved1 " << r.reserved1 << "\n";
9407 outs() << " reserved2 " << r.reserved2 << "\n";
9408 outs() << " reserved3 " << r.reserved3 << "\n";
9409 outs() << " reserved4 " << r.reserved4 << "\n";
9410 outs() << " reserved5 " << r.reserved5 << "\n";
9411 outs() << " reserved6 " << r.reserved6 << "\n";
9412 }
9413
Print_x86_thread_state32_t(MachO::x86_thread_state32_t & cpu32)9414 static void Print_x86_thread_state32_t(MachO::x86_thread_state32_t &cpu32) {
9415 outs() << "\t eax " << format("0x%08" PRIx32, cpu32.eax);
9416 outs() << " ebx " << format("0x%08" PRIx32, cpu32.ebx);
9417 outs() << " ecx " << format("0x%08" PRIx32, cpu32.ecx);
9418 outs() << " edx " << format("0x%08" PRIx32, cpu32.edx) << "\n";
9419 outs() << "\t edi " << format("0x%08" PRIx32, cpu32.edi);
9420 outs() << " esi " << format("0x%08" PRIx32, cpu32.esi);
9421 outs() << " ebp " << format("0x%08" PRIx32, cpu32.ebp);
9422 outs() << " esp " << format("0x%08" PRIx32, cpu32.esp) << "\n";
9423 outs() << "\t ss " << format("0x%08" PRIx32, cpu32.ss);
9424 outs() << " eflags " << format("0x%08" PRIx32, cpu32.eflags);
9425 outs() << " eip " << format("0x%08" PRIx32, cpu32.eip);
9426 outs() << " cs " << format("0x%08" PRIx32, cpu32.cs) << "\n";
9427 outs() << "\t ds " << format("0x%08" PRIx32, cpu32.ds);
9428 outs() << " es " << format("0x%08" PRIx32, cpu32.es);
9429 outs() << " fs " << format("0x%08" PRIx32, cpu32.fs);
9430 outs() << " gs " << format("0x%08" PRIx32, cpu32.gs) << "\n";
9431 }
9432
Print_x86_thread_state64_t(MachO::x86_thread_state64_t & cpu64)9433 static void Print_x86_thread_state64_t(MachO::x86_thread_state64_t &cpu64) {
9434 outs() << " rax " << format("0x%016" PRIx64, cpu64.rax);
9435 outs() << " rbx " << format("0x%016" PRIx64, cpu64.rbx);
9436 outs() << " rcx " << format("0x%016" PRIx64, cpu64.rcx) << "\n";
9437 outs() << " rdx " << format("0x%016" PRIx64, cpu64.rdx);
9438 outs() << " rdi " << format("0x%016" PRIx64, cpu64.rdi);
9439 outs() << " rsi " << format("0x%016" PRIx64, cpu64.rsi) << "\n";
9440 outs() << " rbp " << format("0x%016" PRIx64, cpu64.rbp);
9441 outs() << " rsp " << format("0x%016" PRIx64, cpu64.rsp);
9442 outs() << " r8 " << format("0x%016" PRIx64, cpu64.r8) << "\n";
9443 outs() << " r9 " << format("0x%016" PRIx64, cpu64.r9);
9444 outs() << " r10 " << format("0x%016" PRIx64, cpu64.r10);
9445 outs() << " r11 " << format("0x%016" PRIx64, cpu64.r11) << "\n";
9446 outs() << " r12 " << format("0x%016" PRIx64, cpu64.r12);
9447 outs() << " r13 " << format("0x%016" PRIx64, cpu64.r13);
9448 outs() << " r14 " << format("0x%016" PRIx64, cpu64.r14) << "\n";
9449 outs() << " r15 " << format("0x%016" PRIx64, cpu64.r15);
9450 outs() << " rip " << format("0x%016" PRIx64, cpu64.rip) << "\n";
9451 outs() << "rflags " << format("0x%016" PRIx64, cpu64.rflags);
9452 outs() << " cs " << format("0x%016" PRIx64, cpu64.cs);
9453 outs() << " fs " << format("0x%016" PRIx64, cpu64.fs) << "\n";
9454 outs() << " gs " << format("0x%016" PRIx64, cpu64.gs) << "\n";
9455 }
9456
Print_mmst_reg(MachO::mmst_reg_t & r)9457 static void Print_mmst_reg(MachO::mmst_reg_t &r) {
9458 uint32_t f;
9459 outs() << "\t mmst_reg ";
9460 for (f = 0; f < 10; f++)
9461 outs() << format("%02" PRIx32, (r.mmst_reg[f] & 0xff)) << " ";
9462 outs() << "\n";
9463 outs() << "\t mmst_rsrv ";
9464 for (f = 0; f < 6; f++)
9465 outs() << format("%02" PRIx32, (r.mmst_rsrv[f] & 0xff)) << " ";
9466 outs() << "\n";
9467 }
9468
Print_xmm_reg(MachO::xmm_reg_t & r)9469 static void Print_xmm_reg(MachO::xmm_reg_t &r) {
9470 uint32_t f;
9471 outs() << "\t xmm_reg ";
9472 for (f = 0; f < 16; f++)
9473 outs() << format("%02" PRIx32, (r.xmm_reg[f] & 0xff)) << " ";
9474 outs() << "\n";
9475 }
9476
Print_x86_float_state_t(MachO::x86_float_state64_t & fpu)9477 static void Print_x86_float_state_t(MachO::x86_float_state64_t &fpu) {
9478 outs() << "\t fpu_reserved[0] " << fpu.fpu_reserved[0];
9479 outs() << " fpu_reserved[1] " << fpu.fpu_reserved[1] << "\n";
9480 outs() << "\t control: invalid " << fpu.fpu_fcw.invalid;
9481 outs() << " denorm " << fpu.fpu_fcw.denorm;
9482 outs() << " zdiv " << fpu.fpu_fcw.zdiv;
9483 outs() << " ovrfl " << fpu.fpu_fcw.ovrfl;
9484 outs() << " undfl " << fpu.fpu_fcw.undfl;
9485 outs() << " precis " << fpu.fpu_fcw.precis << "\n";
9486 outs() << "\t\t pc ";
9487 if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_24B)
9488 outs() << "FP_PREC_24B ";
9489 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_53B)
9490 outs() << "FP_PREC_53B ";
9491 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_64B)
9492 outs() << "FP_PREC_64B ";
9493 else
9494 outs() << fpu.fpu_fcw.pc << " ";
9495 outs() << "rc ";
9496 if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_NEAR)
9497 outs() << "FP_RND_NEAR ";
9498 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_DOWN)
9499 outs() << "FP_RND_DOWN ";
9500 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_UP)
9501 outs() << "FP_RND_UP ";
9502 else if (fpu.fpu_fcw.rc == MachO::x86_FP_CHOP)
9503 outs() << "FP_CHOP ";
9504 outs() << "\n";
9505 outs() << "\t status: invalid " << fpu.fpu_fsw.invalid;
9506 outs() << " denorm " << fpu.fpu_fsw.denorm;
9507 outs() << " zdiv " << fpu.fpu_fsw.zdiv;
9508 outs() << " ovrfl " << fpu.fpu_fsw.ovrfl;
9509 outs() << " undfl " << fpu.fpu_fsw.undfl;
9510 outs() << " precis " << fpu.fpu_fsw.precis;
9511 outs() << " stkflt " << fpu.fpu_fsw.stkflt << "\n";
9512 outs() << "\t errsumm " << fpu.fpu_fsw.errsumm;
9513 outs() << " c0 " << fpu.fpu_fsw.c0;
9514 outs() << " c1 " << fpu.fpu_fsw.c1;
9515 outs() << " c2 " << fpu.fpu_fsw.c2;
9516 outs() << " tos " << fpu.fpu_fsw.tos;
9517 outs() << " c3 " << fpu.fpu_fsw.c3;
9518 outs() << " busy " << fpu.fpu_fsw.busy << "\n";
9519 outs() << "\t fpu_ftw " << format("0x%02" PRIx32, fpu.fpu_ftw);
9520 outs() << " fpu_rsrv1 " << format("0x%02" PRIx32, fpu.fpu_rsrv1);
9521 outs() << " fpu_fop " << format("0x%04" PRIx32, fpu.fpu_fop);
9522 outs() << " fpu_ip " << format("0x%08" PRIx32, fpu.fpu_ip) << "\n";
9523 outs() << "\t fpu_cs " << format("0x%04" PRIx32, fpu.fpu_cs);
9524 outs() << " fpu_rsrv2 " << format("0x%04" PRIx32, fpu.fpu_rsrv2);
9525 outs() << " fpu_dp " << format("0x%08" PRIx32, fpu.fpu_dp);
9526 outs() << " fpu_ds " << format("0x%04" PRIx32, fpu.fpu_ds) << "\n";
9527 outs() << "\t fpu_rsrv3 " << format("0x%04" PRIx32, fpu.fpu_rsrv3);
9528 outs() << " fpu_mxcsr " << format("0x%08" PRIx32, fpu.fpu_mxcsr);
9529 outs() << " fpu_mxcsrmask " << format("0x%08" PRIx32, fpu.fpu_mxcsrmask);
9530 outs() << "\n";
9531 outs() << "\t fpu_stmm0:\n";
9532 Print_mmst_reg(fpu.fpu_stmm0);
9533 outs() << "\t fpu_stmm1:\n";
9534 Print_mmst_reg(fpu.fpu_stmm1);
9535 outs() << "\t fpu_stmm2:\n";
9536 Print_mmst_reg(fpu.fpu_stmm2);
9537 outs() << "\t fpu_stmm3:\n";
9538 Print_mmst_reg(fpu.fpu_stmm3);
9539 outs() << "\t fpu_stmm4:\n";
9540 Print_mmst_reg(fpu.fpu_stmm4);
9541 outs() << "\t fpu_stmm5:\n";
9542 Print_mmst_reg(fpu.fpu_stmm5);
9543 outs() << "\t fpu_stmm6:\n";
9544 Print_mmst_reg(fpu.fpu_stmm6);
9545 outs() << "\t fpu_stmm7:\n";
9546 Print_mmst_reg(fpu.fpu_stmm7);
9547 outs() << "\t fpu_xmm0:\n";
9548 Print_xmm_reg(fpu.fpu_xmm0);
9549 outs() << "\t fpu_xmm1:\n";
9550 Print_xmm_reg(fpu.fpu_xmm1);
9551 outs() << "\t fpu_xmm2:\n";
9552 Print_xmm_reg(fpu.fpu_xmm2);
9553 outs() << "\t fpu_xmm3:\n";
9554 Print_xmm_reg(fpu.fpu_xmm3);
9555 outs() << "\t fpu_xmm4:\n";
9556 Print_xmm_reg(fpu.fpu_xmm4);
9557 outs() << "\t fpu_xmm5:\n";
9558 Print_xmm_reg(fpu.fpu_xmm5);
9559 outs() << "\t fpu_xmm6:\n";
9560 Print_xmm_reg(fpu.fpu_xmm6);
9561 outs() << "\t fpu_xmm7:\n";
9562 Print_xmm_reg(fpu.fpu_xmm7);
9563 outs() << "\t fpu_xmm8:\n";
9564 Print_xmm_reg(fpu.fpu_xmm8);
9565 outs() << "\t fpu_xmm9:\n";
9566 Print_xmm_reg(fpu.fpu_xmm9);
9567 outs() << "\t fpu_xmm10:\n";
9568 Print_xmm_reg(fpu.fpu_xmm10);
9569 outs() << "\t fpu_xmm11:\n";
9570 Print_xmm_reg(fpu.fpu_xmm11);
9571 outs() << "\t fpu_xmm12:\n";
9572 Print_xmm_reg(fpu.fpu_xmm12);
9573 outs() << "\t fpu_xmm13:\n";
9574 Print_xmm_reg(fpu.fpu_xmm13);
9575 outs() << "\t fpu_xmm14:\n";
9576 Print_xmm_reg(fpu.fpu_xmm14);
9577 outs() << "\t fpu_xmm15:\n";
9578 Print_xmm_reg(fpu.fpu_xmm15);
9579 outs() << "\t fpu_rsrv4:\n";
9580 for (uint32_t f = 0; f < 6; f++) {
9581 outs() << "\t ";
9582 for (uint32_t g = 0; g < 16; g++)
9583 outs() << format("%02" PRIx32, fpu.fpu_rsrv4[f * g]) << " ";
9584 outs() << "\n";
9585 }
9586 outs() << "\t fpu_reserved1 " << format("0x%08" PRIx32, fpu.fpu_reserved1);
9587 outs() << "\n";
9588 }
9589
Print_x86_exception_state_t(MachO::x86_exception_state64_t & exc64)9590 static void Print_x86_exception_state_t(MachO::x86_exception_state64_t &exc64) {
9591 outs() << "\t trapno " << format("0x%08" PRIx32, exc64.trapno);
9592 outs() << " err " << format("0x%08" PRIx32, exc64.err);
9593 outs() << " faultvaddr " << format("0x%016" PRIx64, exc64.faultvaddr) << "\n";
9594 }
9595
Print_arm_thread_state32_t(MachO::arm_thread_state32_t & cpu32)9596 static void Print_arm_thread_state32_t(MachO::arm_thread_state32_t &cpu32) {
9597 outs() << "\t r0 " << format("0x%08" PRIx32, cpu32.r[0]);
9598 outs() << " r1 " << format("0x%08" PRIx32, cpu32.r[1]);
9599 outs() << " r2 " << format("0x%08" PRIx32, cpu32.r[2]);
9600 outs() << " r3 " << format("0x%08" PRIx32, cpu32.r[3]) << "\n";
9601 outs() << "\t r4 " << format("0x%08" PRIx32, cpu32.r[4]);
9602 outs() << " r5 " << format("0x%08" PRIx32, cpu32.r[5]);
9603 outs() << " r6 " << format("0x%08" PRIx32, cpu32.r[6]);
9604 outs() << " r7 " << format("0x%08" PRIx32, cpu32.r[7]) << "\n";
9605 outs() << "\t r8 " << format("0x%08" PRIx32, cpu32.r[8]);
9606 outs() << " r9 " << format("0x%08" PRIx32, cpu32.r[9]);
9607 outs() << " r10 " << format("0x%08" PRIx32, cpu32.r[10]);
9608 outs() << " r11 " << format("0x%08" PRIx32, cpu32.r[11]) << "\n";
9609 outs() << "\t r12 " << format("0x%08" PRIx32, cpu32.r[12]);
9610 outs() << " sp " << format("0x%08" PRIx32, cpu32.sp);
9611 outs() << " lr " << format("0x%08" PRIx32, cpu32.lr);
9612 outs() << " pc " << format("0x%08" PRIx32, cpu32.pc) << "\n";
9613 outs() << "\t cpsr " << format("0x%08" PRIx32, cpu32.cpsr) << "\n";
9614 }
9615
Print_arm_thread_state64_t(MachO::arm_thread_state64_t & cpu64)9616 static void Print_arm_thread_state64_t(MachO::arm_thread_state64_t &cpu64) {
9617 outs() << "\t x0 " << format("0x%016" PRIx64, cpu64.x[0]);
9618 outs() << " x1 " << format("0x%016" PRIx64, cpu64.x[1]);
9619 outs() << " x2 " << format("0x%016" PRIx64, cpu64.x[2]) << "\n";
9620 outs() << "\t x3 " << format("0x%016" PRIx64, cpu64.x[3]);
9621 outs() << " x4 " << format("0x%016" PRIx64, cpu64.x[4]);
9622 outs() << " x5 " << format("0x%016" PRIx64, cpu64.x[5]) << "\n";
9623 outs() << "\t x6 " << format("0x%016" PRIx64, cpu64.x[6]);
9624 outs() << " x7 " << format("0x%016" PRIx64, cpu64.x[7]);
9625 outs() << " x8 " << format("0x%016" PRIx64, cpu64.x[8]) << "\n";
9626 outs() << "\t x9 " << format("0x%016" PRIx64, cpu64.x[9]);
9627 outs() << " x10 " << format("0x%016" PRIx64, cpu64.x[10]);
9628 outs() << " x11 " << format("0x%016" PRIx64, cpu64.x[11]) << "\n";
9629 outs() << "\t x12 " << format("0x%016" PRIx64, cpu64.x[12]);
9630 outs() << " x13 " << format("0x%016" PRIx64, cpu64.x[13]);
9631 outs() << " x14 " << format("0x%016" PRIx64, cpu64.x[14]) << "\n";
9632 outs() << "\t x15 " << format("0x%016" PRIx64, cpu64.x[15]);
9633 outs() << " x16 " << format("0x%016" PRIx64, cpu64.x[16]);
9634 outs() << " x17 " << format("0x%016" PRIx64, cpu64.x[17]) << "\n";
9635 outs() << "\t x18 " << format("0x%016" PRIx64, cpu64.x[18]);
9636 outs() << " x19 " << format("0x%016" PRIx64, cpu64.x[19]);
9637 outs() << " x20 " << format("0x%016" PRIx64, cpu64.x[20]) << "\n";
9638 outs() << "\t x21 " << format("0x%016" PRIx64, cpu64.x[21]);
9639 outs() << " x22 " << format("0x%016" PRIx64, cpu64.x[22]);
9640 outs() << " x23 " << format("0x%016" PRIx64, cpu64.x[23]) << "\n";
9641 outs() << "\t x24 " << format("0x%016" PRIx64, cpu64.x[24]);
9642 outs() << " x25 " << format("0x%016" PRIx64, cpu64.x[25]);
9643 outs() << " x26 " << format("0x%016" PRIx64, cpu64.x[26]) << "\n";
9644 outs() << "\t x27 " << format("0x%016" PRIx64, cpu64.x[27]);
9645 outs() << " x28 " << format("0x%016" PRIx64, cpu64.x[28]);
9646 outs() << " fp " << format("0x%016" PRIx64, cpu64.fp) << "\n";
9647 outs() << "\t lr " << format("0x%016" PRIx64, cpu64.lr);
9648 outs() << " sp " << format("0x%016" PRIx64, cpu64.sp);
9649 outs() << " pc " << format("0x%016" PRIx64, cpu64.pc) << "\n";
9650 outs() << "\t cpsr " << format("0x%08" PRIx32, cpu64.cpsr) << "\n";
9651 }
9652
PrintThreadCommand(MachO::thread_command t,const char * Ptr,bool isLittleEndian,uint32_t cputype)9653 static void PrintThreadCommand(MachO::thread_command t, const char *Ptr,
9654 bool isLittleEndian, uint32_t cputype) {
9655 if (t.cmd == MachO::LC_THREAD)
9656 outs() << " cmd LC_THREAD\n";
9657 else if (t.cmd == MachO::LC_UNIXTHREAD)
9658 outs() << " cmd LC_UNIXTHREAD\n";
9659 else
9660 outs() << " cmd " << t.cmd << " (unknown)\n";
9661 outs() << " cmdsize " << t.cmdsize;
9662 if (t.cmdsize < sizeof(struct MachO::thread_command) + 2 * sizeof(uint32_t))
9663 outs() << " Incorrect size\n";
9664 else
9665 outs() << "\n";
9666
9667 const char *begin = Ptr + sizeof(struct MachO::thread_command);
9668 const char *end = Ptr + t.cmdsize;
9669 uint32_t flavor, count, left;
9670 if (cputype == MachO::CPU_TYPE_I386) {
9671 while (begin < end) {
9672 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9673 memcpy((char *)&flavor, begin, sizeof(uint32_t));
9674 begin += sizeof(uint32_t);
9675 } else {
9676 flavor = 0;
9677 begin = end;
9678 }
9679 if (isLittleEndian != sys::IsLittleEndianHost)
9680 sys::swapByteOrder(flavor);
9681 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9682 memcpy((char *)&count, begin, sizeof(uint32_t));
9683 begin += sizeof(uint32_t);
9684 } else {
9685 count = 0;
9686 begin = end;
9687 }
9688 if (isLittleEndian != sys::IsLittleEndianHost)
9689 sys::swapByteOrder(count);
9690 if (flavor == MachO::x86_THREAD_STATE32) {
9691 outs() << " flavor i386_THREAD_STATE\n";
9692 if (count == MachO::x86_THREAD_STATE32_COUNT)
9693 outs() << " count i386_THREAD_STATE_COUNT\n";
9694 else
9695 outs() << " count " << count
9696 << " (not x86_THREAD_STATE32_COUNT)\n";
9697 MachO::x86_thread_state32_t cpu32;
9698 left = end - begin;
9699 if (left >= sizeof(MachO::x86_thread_state32_t)) {
9700 memcpy(&cpu32, begin, sizeof(MachO::x86_thread_state32_t));
9701 begin += sizeof(MachO::x86_thread_state32_t);
9702 } else {
9703 memset(&cpu32, '\0', sizeof(MachO::x86_thread_state32_t));
9704 memcpy(&cpu32, begin, left);
9705 begin += left;
9706 }
9707 if (isLittleEndian != sys::IsLittleEndianHost)
9708 swapStruct(cpu32);
9709 Print_x86_thread_state32_t(cpu32);
9710 } else if (flavor == MachO::x86_THREAD_STATE) {
9711 outs() << " flavor x86_THREAD_STATE\n";
9712 if (count == MachO::x86_THREAD_STATE_COUNT)
9713 outs() << " count x86_THREAD_STATE_COUNT\n";
9714 else
9715 outs() << " count " << count
9716 << " (not x86_THREAD_STATE_COUNT)\n";
9717 struct MachO::x86_thread_state_t ts;
9718 left = end - begin;
9719 if (left >= sizeof(MachO::x86_thread_state_t)) {
9720 memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
9721 begin += sizeof(MachO::x86_thread_state_t);
9722 } else {
9723 memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
9724 memcpy(&ts, begin, left);
9725 begin += left;
9726 }
9727 if (isLittleEndian != sys::IsLittleEndianHost)
9728 swapStruct(ts);
9729 if (ts.tsh.flavor == MachO::x86_THREAD_STATE32) {
9730 outs() << "\t tsh.flavor x86_THREAD_STATE32 ";
9731 if (ts.tsh.count == MachO::x86_THREAD_STATE32_COUNT)
9732 outs() << "tsh.count x86_THREAD_STATE32_COUNT\n";
9733 else
9734 outs() << "tsh.count " << ts.tsh.count
9735 << " (not x86_THREAD_STATE32_COUNT\n";
9736 Print_x86_thread_state32_t(ts.uts.ts32);
9737 } else {
9738 outs() << "\t tsh.flavor " << ts.tsh.flavor << " tsh.count "
9739 << ts.tsh.count << "\n";
9740 }
9741 } else {
9742 outs() << " flavor " << flavor << " (unknown)\n";
9743 outs() << " count " << count << "\n";
9744 outs() << " state (unknown)\n";
9745 begin += count * sizeof(uint32_t);
9746 }
9747 }
9748 } else if (cputype == MachO::CPU_TYPE_X86_64) {
9749 while (begin < end) {
9750 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9751 memcpy((char *)&flavor, begin, sizeof(uint32_t));
9752 begin += sizeof(uint32_t);
9753 } else {
9754 flavor = 0;
9755 begin = end;
9756 }
9757 if (isLittleEndian != sys::IsLittleEndianHost)
9758 sys::swapByteOrder(flavor);
9759 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9760 memcpy((char *)&count, begin, sizeof(uint32_t));
9761 begin += sizeof(uint32_t);
9762 } else {
9763 count = 0;
9764 begin = end;
9765 }
9766 if (isLittleEndian != sys::IsLittleEndianHost)
9767 sys::swapByteOrder(count);
9768 if (flavor == MachO::x86_THREAD_STATE64) {
9769 outs() << " flavor x86_THREAD_STATE64\n";
9770 if (count == MachO::x86_THREAD_STATE64_COUNT)
9771 outs() << " count x86_THREAD_STATE64_COUNT\n";
9772 else
9773 outs() << " count " << count
9774 << " (not x86_THREAD_STATE64_COUNT)\n";
9775 MachO::x86_thread_state64_t cpu64;
9776 left = end - begin;
9777 if (left >= sizeof(MachO::x86_thread_state64_t)) {
9778 memcpy(&cpu64, begin, sizeof(MachO::x86_thread_state64_t));
9779 begin += sizeof(MachO::x86_thread_state64_t);
9780 } else {
9781 memset(&cpu64, '\0', sizeof(MachO::x86_thread_state64_t));
9782 memcpy(&cpu64, begin, left);
9783 begin += left;
9784 }
9785 if (isLittleEndian != sys::IsLittleEndianHost)
9786 swapStruct(cpu64);
9787 Print_x86_thread_state64_t(cpu64);
9788 } else if (flavor == MachO::x86_THREAD_STATE) {
9789 outs() << " flavor x86_THREAD_STATE\n";
9790 if (count == MachO::x86_THREAD_STATE_COUNT)
9791 outs() << " count x86_THREAD_STATE_COUNT\n";
9792 else
9793 outs() << " count " << count
9794 << " (not x86_THREAD_STATE_COUNT)\n";
9795 struct MachO::x86_thread_state_t ts;
9796 left = end - begin;
9797 if (left >= sizeof(MachO::x86_thread_state_t)) {
9798 memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
9799 begin += sizeof(MachO::x86_thread_state_t);
9800 } else {
9801 memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
9802 memcpy(&ts, begin, left);
9803 begin += left;
9804 }
9805 if (isLittleEndian != sys::IsLittleEndianHost)
9806 swapStruct(ts);
9807 if (ts.tsh.flavor == MachO::x86_THREAD_STATE64) {
9808 outs() << "\t tsh.flavor x86_THREAD_STATE64 ";
9809 if (ts.tsh.count == MachO::x86_THREAD_STATE64_COUNT)
9810 outs() << "tsh.count x86_THREAD_STATE64_COUNT\n";
9811 else
9812 outs() << "tsh.count " << ts.tsh.count
9813 << " (not x86_THREAD_STATE64_COUNT\n";
9814 Print_x86_thread_state64_t(ts.uts.ts64);
9815 } else {
9816 outs() << "\t tsh.flavor " << ts.tsh.flavor << " tsh.count "
9817 << ts.tsh.count << "\n";
9818 }
9819 } else if (flavor == MachO::x86_FLOAT_STATE) {
9820 outs() << " flavor x86_FLOAT_STATE\n";
9821 if (count == MachO::x86_FLOAT_STATE_COUNT)
9822 outs() << " count x86_FLOAT_STATE_COUNT\n";
9823 else
9824 outs() << " count " << count << " (not x86_FLOAT_STATE_COUNT)\n";
9825 struct MachO::x86_float_state_t fs;
9826 left = end - begin;
9827 if (left >= sizeof(MachO::x86_float_state_t)) {
9828 memcpy(&fs, begin, sizeof(MachO::x86_float_state_t));
9829 begin += sizeof(MachO::x86_float_state_t);
9830 } else {
9831 memset(&fs, '\0', sizeof(MachO::x86_float_state_t));
9832 memcpy(&fs, begin, left);
9833 begin += left;
9834 }
9835 if (isLittleEndian != sys::IsLittleEndianHost)
9836 swapStruct(fs);
9837 if (fs.fsh.flavor == MachO::x86_FLOAT_STATE64) {
9838 outs() << "\t fsh.flavor x86_FLOAT_STATE64 ";
9839 if (fs.fsh.count == MachO::x86_FLOAT_STATE64_COUNT)
9840 outs() << "fsh.count x86_FLOAT_STATE64_COUNT\n";
9841 else
9842 outs() << "fsh.count " << fs.fsh.count
9843 << " (not x86_FLOAT_STATE64_COUNT\n";
9844 Print_x86_float_state_t(fs.ufs.fs64);
9845 } else {
9846 outs() << "\t fsh.flavor " << fs.fsh.flavor << " fsh.count "
9847 << fs.fsh.count << "\n";
9848 }
9849 } else if (flavor == MachO::x86_EXCEPTION_STATE) {
9850 outs() << " flavor x86_EXCEPTION_STATE\n";
9851 if (count == MachO::x86_EXCEPTION_STATE_COUNT)
9852 outs() << " count x86_EXCEPTION_STATE_COUNT\n";
9853 else
9854 outs() << " count " << count
9855 << " (not x86_EXCEPTION_STATE_COUNT)\n";
9856 struct MachO::x86_exception_state_t es;
9857 left = end - begin;
9858 if (left >= sizeof(MachO::x86_exception_state_t)) {
9859 memcpy(&es, begin, sizeof(MachO::x86_exception_state_t));
9860 begin += sizeof(MachO::x86_exception_state_t);
9861 } else {
9862 memset(&es, '\0', sizeof(MachO::x86_exception_state_t));
9863 memcpy(&es, begin, left);
9864 begin += left;
9865 }
9866 if (isLittleEndian != sys::IsLittleEndianHost)
9867 swapStruct(es);
9868 if (es.esh.flavor == MachO::x86_EXCEPTION_STATE64) {
9869 outs() << "\t esh.flavor x86_EXCEPTION_STATE64\n";
9870 if (es.esh.count == MachO::x86_EXCEPTION_STATE64_COUNT)
9871 outs() << "\t esh.count x86_EXCEPTION_STATE64_COUNT\n";
9872 else
9873 outs() << "\t esh.count " << es.esh.count
9874 << " (not x86_EXCEPTION_STATE64_COUNT\n";
9875 Print_x86_exception_state_t(es.ues.es64);
9876 } else {
9877 outs() << "\t esh.flavor " << es.esh.flavor << " esh.count "
9878 << es.esh.count << "\n";
9879 }
9880 } else if (flavor == MachO::x86_EXCEPTION_STATE64) {
9881 outs() << " flavor x86_EXCEPTION_STATE64\n";
9882 if (count == MachO::x86_EXCEPTION_STATE64_COUNT)
9883 outs() << " count x86_EXCEPTION_STATE64_COUNT\n";
9884 else
9885 outs() << " count " << count
9886 << " (not x86_EXCEPTION_STATE64_COUNT)\n";
9887 struct MachO::x86_exception_state64_t es64;
9888 left = end - begin;
9889 if (left >= sizeof(MachO::x86_exception_state64_t)) {
9890 memcpy(&es64, begin, sizeof(MachO::x86_exception_state64_t));
9891 begin += sizeof(MachO::x86_exception_state64_t);
9892 } else {
9893 memset(&es64, '\0', sizeof(MachO::x86_exception_state64_t));
9894 memcpy(&es64, begin, left);
9895 begin += left;
9896 }
9897 if (isLittleEndian != sys::IsLittleEndianHost)
9898 swapStruct(es64);
9899 Print_x86_exception_state_t(es64);
9900 } else {
9901 outs() << " flavor " << flavor << " (unknown)\n";
9902 outs() << " count " << count << "\n";
9903 outs() << " state (unknown)\n";
9904 begin += count * sizeof(uint32_t);
9905 }
9906 }
9907 } else if (cputype == MachO::CPU_TYPE_ARM) {
9908 while (begin < end) {
9909 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9910 memcpy((char *)&flavor, begin, sizeof(uint32_t));
9911 begin += sizeof(uint32_t);
9912 } else {
9913 flavor = 0;
9914 begin = end;
9915 }
9916 if (isLittleEndian != sys::IsLittleEndianHost)
9917 sys::swapByteOrder(flavor);
9918 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9919 memcpy((char *)&count, begin, sizeof(uint32_t));
9920 begin += sizeof(uint32_t);
9921 } else {
9922 count = 0;
9923 begin = end;
9924 }
9925 if (isLittleEndian != sys::IsLittleEndianHost)
9926 sys::swapByteOrder(count);
9927 if (flavor == MachO::ARM_THREAD_STATE) {
9928 outs() << " flavor ARM_THREAD_STATE\n";
9929 if (count == MachO::ARM_THREAD_STATE_COUNT)
9930 outs() << " count ARM_THREAD_STATE_COUNT\n";
9931 else
9932 outs() << " count " << count
9933 << " (not ARM_THREAD_STATE_COUNT)\n";
9934 MachO::arm_thread_state32_t cpu32;
9935 left = end - begin;
9936 if (left >= sizeof(MachO::arm_thread_state32_t)) {
9937 memcpy(&cpu32, begin, sizeof(MachO::arm_thread_state32_t));
9938 begin += sizeof(MachO::arm_thread_state32_t);
9939 } else {
9940 memset(&cpu32, '\0', sizeof(MachO::arm_thread_state32_t));
9941 memcpy(&cpu32, begin, left);
9942 begin += left;
9943 }
9944 if (isLittleEndian != sys::IsLittleEndianHost)
9945 swapStruct(cpu32);
9946 Print_arm_thread_state32_t(cpu32);
9947 } else {
9948 outs() << " flavor " << flavor << " (unknown)\n";
9949 outs() << " count " << count << "\n";
9950 outs() << " state (unknown)\n";
9951 begin += count * sizeof(uint32_t);
9952 }
9953 }
9954 } else if (cputype == MachO::CPU_TYPE_ARM64 ||
9955 cputype == MachO::CPU_TYPE_ARM64_32) {
9956 while (begin < end) {
9957 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9958 memcpy((char *)&flavor, begin, sizeof(uint32_t));
9959 begin += sizeof(uint32_t);
9960 } else {
9961 flavor = 0;
9962 begin = end;
9963 }
9964 if (isLittleEndian != sys::IsLittleEndianHost)
9965 sys::swapByteOrder(flavor);
9966 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9967 memcpy((char *)&count, begin, sizeof(uint32_t));
9968 begin += sizeof(uint32_t);
9969 } else {
9970 count = 0;
9971 begin = end;
9972 }
9973 if (isLittleEndian != sys::IsLittleEndianHost)
9974 sys::swapByteOrder(count);
9975 if (flavor == MachO::ARM_THREAD_STATE64) {
9976 outs() << " flavor ARM_THREAD_STATE64\n";
9977 if (count == MachO::ARM_THREAD_STATE64_COUNT)
9978 outs() << " count ARM_THREAD_STATE64_COUNT\n";
9979 else
9980 outs() << " count " << count
9981 << " (not ARM_THREAD_STATE64_COUNT)\n";
9982 MachO::arm_thread_state64_t cpu64;
9983 left = end - begin;
9984 if (left >= sizeof(MachO::arm_thread_state64_t)) {
9985 memcpy(&cpu64, begin, sizeof(MachO::arm_thread_state64_t));
9986 begin += sizeof(MachO::arm_thread_state64_t);
9987 } else {
9988 memset(&cpu64, '\0', sizeof(MachO::arm_thread_state64_t));
9989 memcpy(&cpu64, begin, left);
9990 begin += left;
9991 }
9992 if (isLittleEndian != sys::IsLittleEndianHost)
9993 swapStruct(cpu64);
9994 Print_arm_thread_state64_t(cpu64);
9995 } else {
9996 outs() << " flavor " << flavor << " (unknown)\n";
9997 outs() << " count " << count << "\n";
9998 outs() << " state (unknown)\n";
9999 begin += count * sizeof(uint32_t);
10000 }
10001 }
10002 } else {
10003 while (begin < end) {
10004 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
10005 memcpy((char *)&flavor, begin, sizeof(uint32_t));
10006 begin += sizeof(uint32_t);
10007 } else {
10008 flavor = 0;
10009 begin = end;
10010 }
10011 if (isLittleEndian != sys::IsLittleEndianHost)
10012 sys::swapByteOrder(flavor);
10013 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
10014 memcpy((char *)&count, begin, sizeof(uint32_t));
10015 begin += sizeof(uint32_t);
10016 } else {
10017 count = 0;
10018 begin = end;
10019 }
10020 if (isLittleEndian != sys::IsLittleEndianHost)
10021 sys::swapByteOrder(count);
10022 outs() << " flavor " << flavor << "\n";
10023 outs() << " count " << count << "\n";
10024 outs() << " state (Unknown cputype/cpusubtype)\n";
10025 begin += count * sizeof(uint32_t);
10026 }
10027 }
10028 }
10029
PrintDylibCommand(MachO::dylib_command dl,const char * Ptr)10030 static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) {
10031 if (dl.cmd == MachO::LC_ID_DYLIB)
10032 outs() << " cmd LC_ID_DYLIB\n";
10033 else if (dl.cmd == MachO::LC_LOAD_DYLIB)
10034 outs() << " cmd LC_LOAD_DYLIB\n";
10035 else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB)
10036 outs() << " cmd LC_LOAD_WEAK_DYLIB\n";
10037 else if (dl.cmd == MachO::LC_REEXPORT_DYLIB)
10038 outs() << " cmd LC_REEXPORT_DYLIB\n";
10039 else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB)
10040 outs() << " cmd LC_LAZY_LOAD_DYLIB\n";
10041 else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
10042 outs() << " cmd LC_LOAD_UPWARD_DYLIB\n";
10043 else
10044 outs() << " cmd " << dl.cmd << " (unknown)\n";
10045 outs() << " cmdsize " << dl.cmdsize;
10046 if (dl.cmdsize < sizeof(struct MachO::dylib_command))
10047 outs() << " Incorrect size\n";
10048 else
10049 outs() << "\n";
10050 if (dl.dylib.name < dl.cmdsize) {
10051 const char *P = (const char *)(Ptr) + dl.dylib.name;
10052 outs() << " name " << P << " (offset " << dl.dylib.name << ")\n";
10053 } else {
10054 outs() << " name ?(bad offset " << dl.dylib.name << ")\n";
10055 }
10056 outs() << " time stamp " << dl.dylib.timestamp << " ";
10057 time_t t = dl.dylib.timestamp;
10058 outs() << ctime(&t);
10059 outs() << " current version ";
10060 if (dl.dylib.current_version == 0xffffffff)
10061 outs() << "n/a\n";
10062 else
10063 outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "."
10064 << ((dl.dylib.current_version >> 8) & 0xff) << "."
10065 << (dl.dylib.current_version & 0xff) << "\n";
10066 outs() << "compatibility version ";
10067 if (dl.dylib.compatibility_version == 0xffffffff)
10068 outs() << "n/a\n";
10069 else
10070 outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
10071 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
10072 << (dl.dylib.compatibility_version & 0xff) << "\n";
10073 }
10074
PrintLinkEditDataCommand(MachO::linkedit_data_command ld,uint32_t object_size)10075 static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld,
10076 uint32_t object_size) {
10077 if (ld.cmd == MachO::LC_CODE_SIGNATURE)
10078 outs() << " cmd LC_CODE_SIGNATURE\n";
10079 else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO)
10080 outs() << " cmd LC_SEGMENT_SPLIT_INFO\n";
10081 else if (ld.cmd == MachO::LC_FUNCTION_STARTS)
10082 outs() << " cmd LC_FUNCTION_STARTS\n";
10083 else if (ld.cmd == MachO::LC_DATA_IN_CODE)
10084 outs() << " cmd LC_DATA_IN_CODE\n";
10085 else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS)
10086 outs() << " cmd LC_DYLIB_CODE_SIGN_DRS\n";
10087 else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT)
10088 outs() << " cmd LC_LINKER_OPTIMIZATION_HINT\n";
10089 else if (ld.cmd == MachO::LC_DYLD_EXPORTS_TRIE)
10090 outs() << " cmd LC_DYLD_EXPORTS_TRIE\n";
10091 else if (ld.cmd == MachO::LC_DYLD_CHAINED_FIXUPS)
10092 outs() << " cmd LC_DYLD_CHAINED_FIXUPS\n";
10093 else
10094 outs() << " cmd " << ld.cmd << " (?)\n";
10095 outs() << " cmdsize " << ld.cmdsize;
10096 if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command))
10097 outs() << " Incorrect size\n";
10098 else
10099 outs() << "\n";
10100 outs() << " dataoff " << ld.dataoff;
10101 if (ld.dataoff > object_size)
10102 outs() << " (past end of file)\n";
10103 else
10104 outs() << "\n";
10105 outs() << " datasize " << ld.datasize;
10106 uint64_t big_size = ld.dataoff;
10107 big_size += ld.datasize;
10108 if (big_size > object_size)
10109 outs() << " (past end of file)\n";
10110 else
10111 outs() << "\n";
10112 }
10113
PrintLoadCommands(const MachOObjectFile * Obj,uint32_t filetype,uint32_t cputype,bool verbose)10114 static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t filetype,
10115 uint32_t cputype, bool verbose) {
10116 StringRef Buf = Obj->getData();
10117 unsigned Index = 0;
10118 for (const auto &Command : Obj->load_commands()) {
10119 outs() << "Load command " << Index++ << "\n";
10120 if (Command.C.cmd == MachO::LC_SEGMENT) {
10121 MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command);
10122 const char *sg_segname = SLC.segname;
10123 PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr,
10124 SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot,
10125 SLC.initprot, SLC.nsects, SLC.flags, Buf.size(),
10126 verbose);
10127 for (unsigned j = 0; j < SLC.nsects; j++) {
10128 MachO::section S = Obj->getSection(Command, j);
10129 PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align,
10130 S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2,
10131 SLC.cmd, sg_segname, filetype, Buf.size(), verbose);
10132 }
10133 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
10134 MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command);
10135 const char *sg_segname = SLC_64.segname;
10136 PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname,
10137 SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff,
10138 SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot,
10139 SLC_64.nsects, SLC_64.flags, Buf.size(), verbose);
10140 for (unsigned j = 0; j < SLC_64.nsects; j++) {
10141 MachO::section_64 S_64 = Obj->getSection64(Command, j);
10142 PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size,
10143 S_64.offset, S_64.align, S_64.reloff, S_64.nreloc,
10144 S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd,
10145 sg_segname, filetype, Buf.size(), verbose);
10146 }
10147 } else if (Command.C.cmd == MachO::LC_SYMTAB) {
10148 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
10149 PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size());
10150 } else if (Command.C.cmd == MachO::LC_DYSYMTAB) {
10151 MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand();
10152 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
10153 PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(),
10154 Obj->is64Bit());
10155 } else if (Command.C.cmd == MachO::LC_DYLD_INFO ||
10156 Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) {
10157 MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command);
10158 PrintDyldInfoLoadCommand(DyldInfo, Buf.size());
10159 } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER ||
10160 Command.C.cmd == MachO::LC_ID_DYLINKER ||
10161 Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) {
10162 MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command);
10163 PrintDyldLoadCommand(Dyld, Command.Ptr);
10164 } else if (Command.C.cmd == MachO::LC_UUID) {
10165 MachO::uuid_command Uuid = Obj->getUuidCommand(Command);
10166 PrintUuidLoadCommand(Uuid);
10167 } else if (Command.C.cmd == MachO::LC_RPATH) {
10168 MachO::rpath_command Rpath = Obj->getRpathCommand(Command);
10169 PrintRpathLoadCommand(Rpath, Command.Ptr);
10170 } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX ||
10171 Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS ||
10172 Command.C.cmd == MachO::LC_VERSION_MIN_TVOS ||
10173 Command.C.cmd == MachO::LC_VERSION_MIN_WATCHOS) {
10174 MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command);
10175 PrintVersionMinLoadCommand(Vd);
10176 } else if (Command.C.cmd == MachO::LC_NOTE) {
10177 MachO::note_command Nt = Obj->getNoteLoadCommand(Command);
10178 PrintNoteLoadCommand(Nt);
10179 } else if (Command.C.cmd == MachO::LC_BUILD_VERSION) {
10180 MachO::build_version_command Bv =
10181 Obj->getBuildVersionLoadCommand(Command);
10182 PrintBuildVersionLoadCommand(Obj, Bv, verbose);
10183 } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) {
10184 MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command);
10185 PrintSourceVersionCommand(Sd);
10186 } else if (Command.C.cmd == MachO::LC_MAIN) {
10187 MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command);
10188 PrintEntryPointCommand(Ep);
10189 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO) {
10190 MachO::encryption_info_command Ei =
10191 Obj->getEncryptionInfoCommand(Command);
10192 PrintEncryptionInfoCommand(Ei, Buf.size());
10193 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO_64) {
10194 MachO::encryption_info_command_64 Ei =
10195 Obj->getEncryptionInfoCommand64(Command);
10196 PrintEncryptionInfoCommand64(Ei, Buf.size());
10197 } else if (Command.C.cmd == MachO::LC_LINKER_OPTION) {
10198 MachO::linker_option_command Lo =
10199 Obj->getLinkerOptionLoadCommand(Command);
10200 PrintLinkerOptionCommand(Lo, Command.Ptr);
10201 } else if (Command.C.cmd == MachO::LC_SUB_FRAMEWORK) {
10202 MachO::sub_framework_command Sf = Obj->getSubFrameworkCommand(Command);
10203 PrintSubFrameworkCommand(Sf, Command.Ptr);
10204 } else if (Command.C.cmd == MachO::LC_SUB_UMBRELLA) {
10205 MachO::sub_umbrella_command Sf = Obj->getSubUmbrellaCommand(Command);
10206 PrintSubUmbrellaCommand(Sf, Command.Ptr);
10207 } else if (Command.C.cmd == MachO::LC_SUB_LIBRARY) {
10208 MachO::sub_library_command Sl = Obj->getSubLibraryCommand(Command);
10209 PrintSubLibraryCommand(Sl, Command.Ptr);
10210 } else if (Command.C.cmd == MachO::LC_SUB_CLIENT) {
10211 MachO::sub_client_command Sc = Obj->getSubClientCommand(Command);
10212 PrintSubClientCommand(Sc, Command.Ptr);
10213 } else if (Command.C.cmd == MachO::LC_ROUTINES) {
10214 MachO::routines_command Rc = Obj->getRoutinesCommand(Command);
10215 PrintRoutinesCommand(Rc);
10216 } else if (Command.C.cmd == MachO::LC_ROUTINES_64) {
10217 MachO::routines_command_64 Rc = Obj->getRoutinesCommand64(Command);
10218 PrintRoutinesCommand64(Rc);
10219 } else if (Command.C.cmd == MachO::LC_THREAD ||
10220 Command.C.cmd == MachO::LC_UNIXTHREAD) {
10221 MachO::thread_command Tc = Obj->getThreadCommand(Command);
10222 PrintThreadCommand(Tc, Command.Ptr, Obj->isLittleEndian(), cputype);
10223 } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB ||
10224 Command.C.cmd == MachO::LC_ID_DYLIB ||
10225 Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
10226 Command.C.cmd == MachO::LC_REEXPORT_DYLIB ||
10227 Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
10228 Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
10229 MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command);
10230 PrintDylibCommand(Dl, Command.Ptr);
10231 } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE ||
10232 Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO ||
10233 Command.C.cmd == MachO::LC_FUNCTION_STARTS ||
10234 Command.C.cmd == MachO::LC_DATA_IN_CODE ||
10235 Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS ||
10236 Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT ||
10237 Command.C.cmd == MachO::LC_DYLD_EXPORTS_TRIE ||
10238 Command.C.cmd == MachO::LC_DYLD_CHAINED_FIXUPS) {
10239 MachO::linkedit_data_command Ld =
10240 Obj->getLinkeditDataLoadCommand(Command);
10241 PrintLinkEditDataCommand(Ld, Buf.size());
10242 } else {
10243 outs() << " cmd ?(" << format("0x%08" PRIx32, Command.C.cmd)
10244 << ")\n";
10245 outs() << " cmdsize " << Command.C.cmdsize << "\n";
10246 // TODO: get and print the raw bytes of the load command.
10247 }
10248 // TODO: print all the other kinds of load commands.
10249 }
10250 }
10251
PrintMachHeader(const MachOObjectFile * Obj,bool verbose)10252 static void PrintMachHeader(const MachOObjectFile *Obj, bool verbose) {
10253 if (Obj->is64Bit()) {
10254 MachO::mach_header_64 H_64;
10255 H_64 = Obj->getHeader64();
10256 PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype,
10257 H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose);
10258 } else {
10259 MachO::mach_header H;
10260 H = Obj->getHeader();
10261 PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds,
10262 H.sizeofcmds, H.flags, verbose);
10263 }
10264 }
10265
printMachOFileHeader(const object::ObjectFile * Obj)10266 void objdump::printMachOFileHeader(const object::ObjectFile *Obj) {
10267 const MachOObjectFile *file = cast<const MachOObjectFile>(Obj);
10268 PrintMachHeader(file, Verbose);
10269 }
10270
printMachOLoadCommands(const object::ObjectFile * Obj)10271 void objdump::printMachOLoadCommands(const object::ObjectFile *Obj) {
10272 const MachOObjectFile *file = cast<const MachOObjectFile>(Obj);
10273 uint32_t filetype = 0;
10274 uint32_t cputype = 0;
10275 if (file->is64Bit()) {
10276 MachO::mach_header_64 H_64;
10277 H_64 = file->getHeader64();
10278 filetype = H_64.filetype;
10279 cputype = H_64.cputype;
10280 } else {
10281 MachO::mach_header H;
10282 H = file->getHeader();
10283 filetype = H.filetype;
10284 cputype = H.cputype;
10285 }
10286 PrintLoadCommands(file, filetype, cputype, Verbose);
10287 }
10288
10289 //===----------------------------------------------------------------------===//
10290 // export trie dumping
10291 //===----------------------------------------------------------------------===//
10292
printMachOExportsTrie(const object::MachOObjectFile * Obj)10293 static void printMachOExportsTrie(const object::MachOObjectFile *Obj) {
10294 uint64_t BaseSegmentAddress = 0;
10295 for (const auto &Command : Obj->load_commands()) {
10296 if (Command.C.cmd == MachO::LC_SEGMENT) {
10297 MachO::segment_command Seg = Obj->getSegmentLoadCommand(Command);
10298 if (Seg.fileoff == 0 && Seg.filesize != 0) {
10299 BaseSegmentAddress = Seg.vmaddr;
10300 break;
10301 }
10302 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
10303 MachO::segment_command_64 Seg = Obj->getSegment64LoadCommand(Command);
10304 if (Seg.fileoff == 0 && Seg.filesize != 0) {
10305 BaseSegmentAddress = Seg.vmaddr;
10306 break;
10307 }
10308 }
10309 }
10310 Error Err = Error::success();
10311 for (const object::ExportEntry &Entry : Obj->exports(Err)) {
10312 uint64_t Flags = Entry.flags();
10313 bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT);
10314 bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION);
10315 bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
10316 MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL);
10317 bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
10318 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE);
10319 bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER);
10320 if (ReExport)
10321 outs() << "[re-export] ";
10322 else
10323 outs() << format("0x%08llX ",
10324 Entry.address() + BaseSegmentAddress);
10325 outs() << Entry.name();
10326 if (WeakDef || ThreadLocal || Resolver || Abs) {
10327 ListSeparator LS;
10328 outs() << " [";
10329 if (WeakDef)
10330 outs() << LS << "weak_def";
10331 if (ThreadLocal)
10332 outs() << LS << "per-thread";
10333 if (Abs)
10334 outs() << LS << "absolute";
10335 if (Resolver)
10336 outs() << LS << format("resolver=0x%08llX", Entry.other());
10337 outs() << "]";
10338 }
10339 if (ReExport) {
10340 StringRef DylibName = "unknown";
10341 int Ordinal = Entry.other() - 1;
10342 Obj->getLibraryShortNameByIndex(Ordinal, DylibName);
10343 if (Entry.otherName().empty())
10344 outs() << " (from " << DylibName << ")";
10345 else
10346 outs() << " (" << Entry.otherName() << " from " << DylibName << ")";
10347 }
10348 outs() << "\n";
10349 }
10350 if (Err)
10351 reportError(std::move(Err), Obj->getFileName());
10352 }
10353
10354 //===----------------------------------------------------------------------===//
10355 // rebase table dumping
10356 //===----------------------------------------------------------------------===//
10357
printMachORebaseTable(object::MachOObjectFile * Obj)10358 static void printMachORebaseTable(object::MachOObjectFile *Obj) {
10359 outs() << "segment section address type\n";
10360 Error Err = Error::success();
10361 for (const object::MachORebaseEntry &Entry : Obj->rebaseTable(Err)) {
10362 StringRef SegmentName = Entry.segmentName();
10363 StringRef SectionName = Entry.sectionName();
10364 uint64_t Address = Entry.address();
10365
10366 // Table lines look like: __DATA __nl_symbol_ptr 0x0000F00C pointer
10367 outs() << format("%-8s %-18s 0x%08" PRIX64 " %s\n",
10368 SegmentName.str().c_str(), SectionName.str().c_str(),
10369 Address, Entry.typeName().str().c_str());
10370 }
10371 if (Err)
10372 reportError(std::move(Err), Obj->getFileName());
10373 }
10374
ordinalName(const object::MachOObjectFile * Obj,int Ordinal)10375 static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) {
10376 StringRef DylibName;
10377 switch (Ordinal) {
10378 case MachO::BIND_SPECIAL_DYLIB_SELF:
10379 return "this-image";
10380 case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE:
10381 return "main-executable";
10382 case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP:
10383 return "flat-namespace";
10384 default:
10385 if (Ordinal > 0) {
10386 std::error_code EC =
10387 Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName);
10388 if (EC)
10389 return "<<bad library ordinal>>";
10390 return DylibName;
10391 }
10392 }
10393 return "<<unknown special ordinal>>";
10394 }
10395
10396 //===----------------------------------------------------------------------===//
10397 // bind table dumping
10398 //===----------------------------------------------------------------------===//
10399
printMachOBindTable(object::MachOObjectFile * Obj)10400 static void printMachOBindTable(object::MachOObjectFile *Obj) {
10401 // Build table of sections so names can used in final output.
10402 outs() << "segment section address type "
10403 "addend dylib symbol\n";
10404 Error Err = Error::success();
10405 for (const object::MachOBindEntry &Entry : Obj->bindTable(Err)) {
10406 StringRef SegmentName = Entry.segmentName();
10407 StringRef SectionName = Entry.sectionName();
10408 uint64_t Address = Entry.address();
10409
10410 // Table lines look like:
10411 // __DATA __got 0x00012010 pointer 0 libSystem ___stack_chk_guard
10412 StringRef Attr;
10413 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT)
10414 Attr = " (weak_import)";
10415 outs() << left_justify(SegmentName, 8) << " "
10416 << left_justify(SectionName, 18) << " "
10417 << format_hex(Address, 10, true) << " "
10418 << left_justify(Entry.typeName(), 8) << " "
10419 << format_decimal(Entry.addend(), 8) << " "
10420 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
10421 << Entry.symbolName() << Attr << "\n";
10422 }
10423 if (Err)
10424 reportError(std::move(Err), Obj->getFileName());
10425 }
10426
10427 //===----------------------------------------------------------------------===//
10428 // lazy bind table dumping
10429 //===----------------------------------------------------------------------===//
10430
printMachOLazyBindTable(object::MachOObjectFile * Obj)10431 static void printMachOLazyBindTable(object::MachOObjectFile *Obj) {
10432 outs() << "segment section address "
10433 "dylib symbol\n";
10434 Error Err = Error::success();
10435 for (const object::MachOBindEntry &Entry : Obj->lazyBindTable(Err)) {
10436 StringRef SegmentName = Entry.segmentName();
10437 StringRef SectionName = Entry.sectionName();
10438 uint64_t Address = Entry.address();
10439
10440 // Table lines look like:
10441 // __DATA __got 0x00012010 libSystem ___stack_chk_guard
10442 outs() << left_justify(SegmentName, 8) << " "
10443 << left_justify(SectionName, 18) << " "
10444 << format_hex(Address, 10, true) << " "
10445 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
10446 << Entry.symbolName() << "\n";
10447 }
10448 if (Err)
10449 reportError(std::move(Err), Obj->getFileName());
10450 }
10451
10452 //===----------------------------------------------------------------------===//
10453 // weak bind table dumping
10454 //===----------------------------------------------------------------------===//
10455
printMachOWeakBindTable(object::MachOObjectFile * Obj)10456 static void printMachOWeakBindTable(object::MachOObjectFile *Obj) {
10457 outs() << "segment section address "
10458 "type addend symbol\n";
10459 Error Err = Error::success();
10460 for (const object::MachOBindEntry &Entry : Obj->weakBindTable(Err)) {
10461 // Strong symbols don't have a location to update.
10462 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) {
10463 outs() << " strong "
10464 << Entry.symbolName() << "\n";
10465 continue;
10466 }
10467 StringRef SegmentName = Entry.segmentName();
10468 StringRef SectionName = Entry.sectionName();
10469 uint64_t Address = Entry.address();
10470
10471 // Table lines look like:
10472 // __DATA __data 0x00001000 pointer 0 _foo
10473 outs() << left_justify(SegmentName, 8) << " "
10474 << left_justify(SectionName, 18) << " "
10475 << format_hex(Address, 10, true) << " "
10476 << left_justify(Entry.typeName(), 8) << " "
10477 << format_decimal(Entry.addend(), 8) << " " << Entry.symbolName()
10478 << "\n";
10479 }
10480 if (Err)
10481 reportError(std::move(Err), Obj->getFileName());
10482 }
10483
10484 // get_dyld_bind_info_symbolname() is used for disassembly and passed an
10485 // address, ReferenceValue, in the Mach-O file and looks in the dyld bind
10486 // information for that address. If the address is found its binding symbol
10487 // name is returned. If not nullptr is returned.
get_dyld_bind_info_symbolname(uint64_t ReferenceValue,struct DisassembleInfo * info)10488 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
10489 struct DisassembleInfo *info) {
10490 if (info->bindtable == nullptr) {
10491 info->bindtable = std::make_unique<SymbolAddressMap>();
10492 Error Err = Error::success();
10493 for (const object::MachOBindEntry &Entry : info->O->bindTable(Err)) {
10494 uint64_t Address = Entry.address();
10495 StringRef name = Entry.symbolName();
10496 if (!name.empty())
10497 (*info->bindtable)[Address] = name;
10498 }
10499 if (Err)
10500 reportError(std::move(Err), info->O->getFileName());
10501 }
10502 auto name = info->bindtable->lookup(ReferenceValue);
10503 return !name.empty() ? name.data() : nullptr;
10504 }
10505
printLazyBindTable(ObjectFile * o)10506 void objdump::printLazyBindTable(ObjectFile *o) {
10507 outs() << "\nLazy bind table:\n";
10508 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10509 printMachOLazyBindTable(MachO);
10510 else
10511 WithColor::error()
10512 << "This operation is only currently supported "
10513 "for Mach-O executable files.\n";
10514 }
10515
printWeakBindTable(ObjectFile * o)10516 void objdump::printWeakBindTable(ObjectFile *o) {
10517 outs() << "\nWeak bind table:\n";
10518 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10519 printMachOWeakBindTable(MachO);
10520 else
10521 WithColor::error()
10522 << "This operation is only currently supported "
10523 "for Mach-O executable files.\n";
10524 }
10525
printExportsTrie(const ObjectFile * o)10526 void objdump::printExportsTrie(const ObjectFile *o) {
10527 outs() << "\nExports trie:\n";
10528 if (const MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10529 printMachOExportsTrie(MachO);
10530 else
10531 WithColor::error()
10532 << "This operation is only currently supported "
10533 "for Mach-O executable files.\n";
10534 }
10535
printRebaseTable(ObjectFile * o)10536 void objdump::printRebaseTable(ObjectFile *o) {
10537 outs() << "\nRebase table:\n";
10538 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10539 printMachORebaseTable(MachO);
10540 else
10541 WithColor::error()
10542 << "This operation is only currently supported "
10543 "for Mach-O executable files.\n";
10544 }
10545
printBindTable(ObjectFile * o)10546 void objdump::printBindTable(ObjectFile *o) {
10547 outs() << "\nBind table:\n";
10548 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10549 printMachOBindTable(MachO);
10550 else
10551 WithColor::error()
10552 << "This operation is only currently supported "
10553 "for Mach-O executable files.\n";
10554 }
10555