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 "llvm-objdump.h"
16 #include "llvm-c/Disassembler.h"
17 #include "llvm/ADT/STLExtras.h"
18 #include "llvm/ADT/StringExtras.h"
19 #include "llvm/ADT/Triple.h"
20 #include "llvm/BinaryFormat/MachO.h"
21 #include "llvm/Config/config.h"
22 #include "llvm/DebugInfo/DIContext.h"
23 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
24 #include "llvm/Demangle/Demangle.h"
25 #include "llvm/MC/MCAsmInfo.h"
26 #include "llvm/MC/MCContext.h"
27 #include "llvm/MC/MCDisassembler/MCDisassembler.h"
28 #include "llvm/MC/MCInst.h"
29 #include "llvm/MC/MCInstPrinter.h"
30 #include "llvm/MC/MCInstrDesc.h"
31 #include "llvm/MC/MCInstrInfo.h"
32 #include "llvm/MC/MCRegisterInfo.h"
33 #include "llvm/MC/MCSubtargetInfo.h"
34 #include "llvm/MC/MCTargetOptions.h"
35 #include "llvm/Object/MachO.h"
36 #include "llvm/Object/MachOUniversal.h"
37 #include "llvm/Support/Casting.h"
38 #include "llvm/Support/CommandLine.h"
39 #include "llvm/Support/Debug.h"
40 #include "llvm/Support/Endian.h"
41 #include "llvm/Support/Format.h"
42 #include "llvm/Support/FormattedStream.h"
43 #include "llvm/Support/GraphWriter.h"
44 #include "llvm/Support/LEB128.h"
45 #include "llvm/Support/MemoryBuffer.h"
46 #include "llvm/Support/TargetRegistry.h"
47 #include "llvm/Support/TargetSelect.h"
48 #include "llvm/Support/ToolOutputFile.h"
49 #include "llvm/Support/WithColor.h"
50 #include "llvm/Support/raw_ostream.h"
51 #include <algorithm>
52 #include <cstring>
53 #include <system_error>
54 
55 #ifdef HAVE_LIBXAR
56 extern "C" {
57 #include <xar/xar.h>
58 }
59 #endif
60 
61 using namespace llvm;
62 using namespace llvm::object;
63 using namespace llvm::objdump;
64 
65 cl::OptionCategory objdump::MachOCat("llvm-objdump MachO Specific Options");
66 
67 cl::opt<bool> objdump::FirstPrivateHeader(
68     "private-header",
69     cl::desc("Display only the first format specific file header"),
70     cl::cat(MachOCat));
71 
72 cl::opt<bool> objdump::ExportsTrie("exports-trie",
73                                    cl::desc("Display mach-o exported symbols"),
74                                    cl::cat(MachOCat));
75 
76 cl::opt<bool> objdump::Rebase("rebase",
77                               cl::desc("Display mach-o rebasing info"),
78                               cl::cat(MachOCat));
79 
80 cl::opt<bool> objdump::Bind("bind", cl::desc("Display mach-o binding info"),
81                             cl::cat(MachOCat));
82 
83 cl::opt<bool> objdump::LazyBind("lazy-bind",
84                                 cl::desc("Display mach-o lazy binding info"),
85                                 cl::cat(MachOCat));
86 
87 cl::opt<bool> objdump::WeakBind("weak-bind",
88                                 cl::desc("Display mach-o weak binding info"),
89                                 cl::cat(MachOCat));
90 
91 static cl::opt<bool>
92     UseDbg("g", cl::Grouping,
93            cl::desc("Print line information from debug info if available"),
94            cl::cat(MachOCat));
95 
96 static cl::opt<std::string> DSYMFile("dsym",
97                                      cl::desc("Use .dSYM file for debug info"),
98                                      cl::cat(MachOCat));
99 
100 static cl::opt<bool> FullLeadingAddr("full-leading-addr",
101                                      cl::desc("Print full leading address"),
102                                      cl::cat(MachOCat));
103 
104 static cl::opt<bool> NoLeadingHeaders("no-leading-headers",
105                                       cl::desc("Print no leading headers"),
106                                       cl::cat(MachOCat));
107 
108 cl::opt<bool> objdump::UniversalHeaders(
109     "universal-headers",
110     cl::desc("Print Mach-O universal headers (requires --macho)"),
111     cl::cat(MachOCat));
112 
113 static cl::opt<bool> ArchiveMemberOffsets(
114     "archive-member-offsets",
115     cl::desc("Print the offset to each archive member for Mach-O archives "
116              "(requires --macho and --archive-headers)"),
117     cl::cat(MachOCat));
118 
119 cl::opt<bool> objdump::IndirectSymbols(
120     "indirect-symbols",
121     cl::desc(
122         "Print indirect symbol table for Mach-O objects (requires --macho)"),
123     cl::cat(MachOCat));
124 
125 cl::opt<bool> objdump::DataInCode(
126     "data-in-code",
127     cl::desc(
128         "Print the data in code table for Mach-O objects (requires --macho)"),
129     cl::cat(MachOCat));
130 
131 cl::opt<bool>
132     objdump::LinkOptHints("link-opt-hints",
133                           cl::desc("Print the linker optimization hints for "
134                                    "Mach-O objects (requires --macho)"),
135                           cl::cat(MachOCat));
136 
137 cl::opt<bool>
138     objdump::InfoPlist("info-plist",
139                        cl::desc("Print the info plist section as strings for "
140                                 "Mach-O objects (requires --macho)"),
141                        cl::cat(MachOCat));
142 
143 cl::opt<bool>
144     objdump::DylibsUsed("dylibs-used",
145                         cl::desc("Print the shared libraries used for linked "
146                                  "Mach-O files (requires --macho)"),
147                         cl::cat(MachOCat));
148 
149 cl::opt<bool> objdump::DylibId("dylib-id",
150                                cl::desc("Print the shared library's id for the "
151                                         "dylib Mach-O file (requires --macho)"),
152                                cl::cat(MachOCat));
153 
154 static cl::opt<bool>
155     NonVerbose("non-verbose",
156                cl::desc("Print the info for Mach-O objects in non-verbose or "
157                         "numeric form (requires --macho)"),
158                cl::cat(MachOCat));
159 
160 cl::opt<bool>
161     objdump::ObjcMetaData("objc-meta-data",
162                           cl::desc("Print the Objective-C runtime meta data "
163                                    "for Mach-O files (requires --macho)"),
164                           cl::cat(MachOCat));
165 
166 static cl::opt<std::string> DisSymName(
167     "dis-symname",
168     cl::desc("disassemble just this symbol's instructions (requires --macho)"),
169     cl::cat(MachOCat));
170 
171 static cl::opt<bool> NoSymbolicOperands(
172     "no-symbolic-operands",
173     cl::desc("do not symbolic operands when disassembling (requires --macho)"),
174     cl::cat(MachOCat));
175 
176 static cl::list<std::string>
177     ArchFlags("arch", cl::desc("architecture(s) from a Mach-O file to dump"),
178               cl::ZeroOrMore, cl::cat(MachOCat));
179 
180 static bool ArchAll = false;
181 
182 static std::string ThumbTripleName;
183 
GetTarget(const MachOObjectFile * MachOObj,const char ** McpuDefault,const Target ** ThumbTarget)184 static const Target *GetTarget(const MachOObjectFile *MachOObj,
185                                const char **McpuDefault,
186                                const Target **ThumbTarget) {
187   // Figure out the target triple.
188   Triple TT(TripleName);
189   if (TripleName.empty()) {
190     TT = MachOObj->getArchTriple(McpuDefault);
191     TripleName = TT.str();
192   }
193 
194   if (TT.getArch() == Triple::arm) {
195     // We've inferred a 32-bit ARM target from the object file. All MachO CPUs
196     // that support ARM are also capable of Thumb mode.
197     Triple ThumbTriple = TT;
198     std::string ThumbName = (Twine("thumb") + TT.getArchName().substr(3)).str();
199     ThumbTriple.setArchName(ThumbName);
200     ThumbTripleName = ThumbTriple.str();
201   }
202 
203   // Get the target specific parser.
204   std::string Error;
205   const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
206   if (TheTarget && ThumbTripleName.empty())
207     return TheTarget;
208 
209   *ThumbTarget = TargetRegistry::lookupTarget(ThumbTripleName, Error);
210   if (*ThumbTarget)
211     return TheTarget;
212 
213   WithColor::error(errs(), "llvm-objdump") << "unable to get target for '";
214   if (!TheTarget)
215     errs() << TripleName;
216   else
217     errs() << ThumbTripleName;
218   errs() << "', see --version and --triple.\n";
219   return nullptr;
220 }
221 
222 namespace {
223 struct SymbolSorter {
operator ()__anonabe85be80111::SymbolSorter224   bool operator()(const SymbolRef &A, const SymbolRef &B) {
225     Expected<SymbolRef::Type> ATypeOrErr = A.getType();
226     if (!ATypeOrErr)
227       reportError(ATypeOrErr.takeError(), A.getObject()->getFileName());
228     SymbolRef::Type AType = *ATypeOrErr;
229     Expected<SymbolRef::Type> BTypeOrErr = B.getType();
230     if (!BTypeOrErr)
231       reportError(BTypeOrErr.takeError(), B.getObject()->getFileName());
232     SymbolRef::Type BType = *BTypeOrErr;
233     uint64_t AAddr =
234         (AType != SymbolRef::ST_Function) ? 0 : cantFail(A.getValue());
235     uint64_t BAddr =
236         (BType != SymbolRef::ST_Function) ? 0 : cantFail(B.getValue());
237     return AAddr < BAddr;
238   }
239 };
240 } // namespace
241 
242 // Types for the storted data in code table that is built before disassembly
243 // and the predicate function to sort them.
244 typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
245 typedef std::vector<DiceTableEntry> DiceTable;
246 typedef DiceTable::iterator dice_table_iterator;
247 
248 #ifdef HAVE_LIBXAR
249 namespace {
250 struct ScopedXarFile {
251   xar_t xar;
ScopedXarFile__anonabe85be80211::ScopedXarFile252   ScopedXarFile(const char *filename, int32_t flags)
253       : xar(xar_open(filename, flags)) {}
~ScopedXarFile__anonabe85be80211::ScopedXarFile254   ~ScopedXarFile() {
255     if (xar)
256       xar_close(xar);
257   }
258   ScopedXarFile(const ScopedXarFile &) = delete;
259   ScopedXarFile &operator=(const ScopedXarFile &) = delete;
operator xar_t__anonabe85be80211::ScopedXarFile260   operator xar_t() { return xar; }
261 };
262 
263 struct ScopedXarIter {
264   xar_iter_t iter;
ScopedXarIter__anonabe85be80211::ScopedXarIter265   ScopedXarIter() : iter(xar_iter_new()) {}
~ScopedXarIter__anonabe85be80211::ScopedXarIter266   ~ScopedXarIter() {
267     if (iter)
268       xar_iter_free(iter);
269   }
270   ScopedXarIter(const ScopedXarIter &) = delete;
271   ScopedXarIter &operator=(const ScopedXarIter &) = delete;
operator xar_iter_t__anonabe85be80211::ScopedXarIter272   operator xar_iter_t() { return iter; }
273 };
274 } // namespace
275 #endif // defined(HAVE_LIBXAR)
276 
277 // This is used to search for a data in code table entry for the PC being
278 // disassembled.  The j parameter has the PC in j.first.  A single data in code
279 // table entry can cover many bytes for each of its Kind's.  So if the offset,
280 // aka the i.first value, of the data in code table entry plus its Length
281 // covers the PC being searched for this will return true.  If not it will
282 // return false.
compareDiceTableEntries(const DiceTableEntry & i,const DiceTableEntry & j)283 static bool compareDiceTableEntries(const DiceTableEntry &i,
284                                     const DiceTableEntry &j) {
285   uint16_t Length;
286   i.second.getLength(Length);
287 
288   return j.first >= i.first && j.first < i.first + Length;
289 }
290 
DumpDataInCode(const uint8_t * bytes,uint64_t Length,unsigned short Kind)291 static uint64_t DumpDataInCode(const uint8_t *bytes, uint64_t Length,
292                                unsigned short Kind) {
293   uint32_t Value, Size = 1;
294 
295   switch (Kind) {
296   default:
297   case MachO::DICE_KIND_DATA:
298     if (Length >= 4) {
299       if (!NoShowRawInsn)
300         dumpBytes(makeArrayRef(bytes, 4), outs());
301       Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
302       outs() << "\t.long " << Value;
303       Size = 4;
304     } else if (Length >= 2) {
305       if (!NoShowRawInsn)
306         dumpBytes(makeArrayRef(bytes, 2), outs());
307       Value = bytes[1] << 8 | bytes[0];
308       outs() << "\t.short " << Value;
309       Size = 2;
310     } else {
311       if (!NoShowRawInsn)
312         dumpBytes(makeArrayRef(bytes, 2), outs());
313       Value = bytes[0];
314       outs() << "\t.byte " << Value;
315       Size = 1;
316     }
317     if (Kind == MachO::DICE_KIND_DATA)
318       outs() << "\t@ KIND_DATA\n";
319     else
320       outs() << "\t@ data in code kind = " << Kind << "\n";
321     break;
322   case MachO::DICE_KIND_JUMP_TABLE8:
323     if (!NoShowRawInsn)
324       dumpBytes(makeArrayRef(bytes, 1), outs());
325     Value = bytes[0];
326     outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n";
327     Size = 1;
328     break;
329   case MachO::DICE_KIND_JUMP_TABLE16:
330     if (!NoShowRawInsn)
331       dumpBytes(makeArrayRef(bytes, 2), outs());
332     Value = bytes[1] << 8 | bytes[0];
333     outs() << "\t.short " << format("%5u", Value & 0xffff)
334            << "\t@ KIND_JUMP_TABLE16\n";
335     Size = 2;
336     break;
337   case MachO::DICE_KIND_JUMP_TABLE32:
338   case MachO::DICE_KIND_ABS_JUMP_TABLE32:
339     if (!NoShowRawInsn)
340       dumpBytes(makeArrayRef(bytes, 4), outs());
341     Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
342     outs() << "\t.long " << Value;
343     if (Kind == MachO::DICE_KIND_JUMP_TABLE32)
344       outs() << "\t@ KIND_JUMP_TABLE32\n";
345     else
346       outs() << "\t@ KIND_ABS_JUMP_TABLE32\n";
347     Size = 4;
348     break;
349   }
350   return Size;
351 }
352 
getSectionsAndSymbols(MachOObjectFile * MachOObj,std::vector<SectionRef> & Sections,std::vector<SymbolRef> & Symbols,SmallVectorImpl<uint64_t> & FoundFns,uint64_t & BaseSegmentAddress)353 static void getSectionsAndSymbols(MachOObjectFile *MachOObj,
354                                   std::vector<SectionRef> &Sections,
355                                   std::vector<SymbolRef> &Symbols,
356                                   SmallVectorImpl<uint64_t> &FoundFns,
357                                   uint64_t &BaseSegmentAddress) {
358   const StringRef FileName = MachOObj->getFileName();
359   for (const SymbolRef &Symbol : MachOObj->symbols()) {
360     StringRef SymName = unwrapOrError(Symbol.getName(), FileName);
361     if (!SymName.startswith("ltmp"))
362       Symbols.push_back(Symbol);
363   }
364 
365   for (const SectionRef &Section : MachOObj->sections())
366     Sections.push_back(Section);
367 
368   bool BaseSegmentAddressSet = false;
369   for (const auto &Command : MachOObj->load_commands()) {
370     if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) {
371       // We found a function starts segment, parse the addresses for later
372       // consumption.
373       MachO::linkedit_data_command LLC =
374           MachOObj->getLinkeditDataLoadCommand(Command);
375 
376       MachOObj->ReadULEB128s(LLC.dataoff, FoundFns);
377     } else if (Command.C.cmd == MachO::LC_SEGMENT) {
378       MachO::segment_command SLC = MachOObj->getSegmentLoadCommand(Command);
379       StringRef SegName = SLC.segname;
380       if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
381         BaseSegmentAddressSet = true;
382         BaseSegmentAddress = SLC.vmaddr;
383       }
384     } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
385       MachO::segment_command_64 SLC = MachOObj->getSegment64LoadCommand(Command);
386       StringRef SegName = SLC.segname;
387       if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
388         BaseSegmentAddressSet = true;
389         BaseSegmentAddress = SLC.vmaddr;
390       }
391     }
392   }
393 }
394 
DumpAndSkipDataInCode(uint64_t PC,const uint8_t * bytes,DiceTable & Dices,uint64_t & InstSize)395 static bool DumpAndSkipDataInCode(uint64_t PC, const uint8_t *bytes,
396                                  DiceTable &Dices, uint64_t &InstSize) {
397   // Check the data in code table here to see if this is data not an
398   // instruction to be disassembled.
399   DiceTable Dice;
400   Dice.push_back(std::make_pair(PC, DiceRef()));
401   dice_table_iterator DTI =
402       std::search(Dices.begin(), Dices.end(), Dice.begin(), Dice.end(),
403                   compareDiceTableEntries);
404   if (DTI != Dices.end()) {
405     uint16_t Length;
406     DTI->second.getLength(Length);
407     uint16_t Kind;
408     DTI->second.getKind(Kind);
409     InstSize = DumpDataInCode(bytes, Length, Kind);
410     if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) &&
411         (PC == (DTI->first + Length - 1)) && (Length & 1))
412       InstSize++;
413     return true;
414   }
415   return false;
416 }
417 
printRelocationTargetName(const MachOObjectFile * O,const MachO::any_relocation_info & RE,raw_string_ostream & Fmt)418 static void printRelocationTargetName(const MachOObjectFile *O,
419                                       const MachO::any_relocation_info &RE,
420                                       raw_string_ostream &Fmt) {
421   // Target of a scattered relocation is an address.  In the interest of
422   // generating pretty output, scan through the symbol table looking for a
423   // symbol that aligns with that address.  If we find one, print it.
424   // Otherwise, we just print the hex address of the target.
425   const StringRef FileName = O->getFileName();
426   if (O->isRelocationScattered(RE)) {
427     uint32_t Val = O->getPlainRelocationSymbolNum(RE);
428 
429     for (const SymbolRef &Symbol : O->symbols()) {
430       uint64_t Addr = unwrapOrError(Symbol.getAddress(), FileName);
431       if (Addr != Val)
432         continue;
433       Fmt << unwrapOrError(Symbol.getName(), FileName);
434       return;
435     }
436 
437     // If we couldn't find a symbol that this relocation refers to, try
438     // to find a section beginning instead.
439     for (const SectionRef &Section : ToolSectionFilter(*O)) {
440       uint64_t Addr = Section.getAddress();
441       if (Addr != Val)
442         continue;
443       StringRef NameOrErr = unwrapOrError(Section.getName(), O->getFileName());
444       Fmt << NameOrErr;
445       return;
446     }
447 
448     Fmt << format("0x%x", Val);
449     return;
450   }
451 
452   StringRef S;
453   bool isExtern = O->getPlainRelocationExternal(RE);
454   uint64_t Val = O->getPlainRelocationSymbolNum(RE);
455 
456   if (O->getAnyRelocationType(RE) == MachO::ARM64_RELOC_ADDEND &&
457       (O->getArch() == Triple::aarch64 || O->getArch() == Triple::aarch64_be)) {
458     Fmt << format("0x%0" PRIx64, Val);
459     return;
460   }
461 
462   if (isExtern) {
463     symbol_iterator SI = O->symbol_begin();
464     advance(SI, Val);
465     S = unwrapOrError(SI->getName(), FileName);
466   } else {
467     section_iterator SI = O->section_begin();
468     // Adjust for the fact that sections are 1-indexed.
469     if (Val == 0) {
470       Fmt << "0 (?,?)";
471       return;
472     }
473     uint32_t I = Val - 1;
474     while (I != 0 && SI != O->section_end()) {
475       --I;
476       advance(SI, 1);
477     }
478     if (SI == O->section_end()) {
479       Fmt << Val << " (?,?)";
480     } else {
481       if (Expected<StringRef> NameOrErr = SI->getName())
482         S = *NameOrErr;
483       else
484         consumeError(NameOrErr.takeError());
485     }
486   }
487 
488   Fmt << S;
489 }
490 
getMachORelocationValueString(const MachOObjectFile * Obj,const RelocationRef & RelRef,SmallVectorImpl<char> & Result)491 Error objdump::getMachORelocationValueString(const MachOObjectFile *Obj,
492                                              const RelocationRef &RelRef,
493                                              SmallVectorImpl<char> &Result) {
494   DataRefImpl Rel = RelRef.getRawDataRefImpl();
495   MachO::any_relocation_info RE = Obj->getRelocation(Rel);
496 
497   unsigned Arch = Obj->getArch();
498 
499   std::string FmtBuf;
500   raw_string_ostream Fmt(FmtBuf);
501   unsigned Type = Obj->getAnyRelocationType(RE);
502   bool IsPCRel = Obj->getAnyRelocationPCRel(RE);
503 
504   // Determine any addends that should be displayed with the relocation.
505   // These require decoding the relocation type, which is triple-specific.
506 
507   // X86_64 has entirely custom relocation types.
508   if (Arch == Triple::x86_64) {
509     switch (Type) {
510     case MachO::X86_64_RELOC_GOT_LOAD:
511     case MachO::X86_64_RELOC_GOT: {
512       printRelocationTargetName(Obj, RE, Fmt);
513       Fmt << "@GOT";
514       if (IsPCRel)
515         Fmt << "PCREL";
516       break;
517     }
518     case MachO::X86_64_RELOC_SUBTRACTOR: {
519       DataRefImpl RelNext = Rel;
520       Obj->moveRelocationNext(RelNext);
521       MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
522 
523       // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type
524       // X86_64_RELOC_UNSIGNED.
525       // NOTE: Scattered relocations don't exist on x86_64.
526       unsigned RType = Obj->getAnyRelocationType(RENext);
527       if (RType != MachO::X86_64_RELOC_UNSIGNED)
528         reportError(Obj->getFileName(), "Expected X86_64_RELOC_UNSIGNED after "
529                                         "X86_64_RELOC_SUBTRACTOR.");
530 
531       // The X86_64_RELOC_UNSIGNED contains the minuend symbol;
532       // X86_64_RELOC_SUBTRACTOR contains the subtrahend.
533       printRelocationTargetName(Obj, RENext, Fmt);
534       Fmt << "-";
535       printRelocationTargetName(Obj, RE, Fmt);
536       break;
537     }
538     case MachO::X86_64_RELOC_TLV:
539       printRelocationTargetName(Obj, RE, Fmt);
540       Fmt << "@TLV";
541       if (IsPCRel)
542         Fmt << "P";
543       break;
544     case MachO::X86_64_RELOC_SIGNED_1:
545       printRelocationTargetName(Obj, RE, Fmt);
546       Fmt << "-1";
547       break;
548     case MachO::X86_64_RELOC_SIGNED_2:
549       printRelocationTargetName(Obj, RE, Fmt);
550       Fmt << "-2";
551       break;
552     case MachO::X86_64_RELOC_SIGNED_4:
553       printRelocationTargetName(Obj, RE, Fmt);
554       Fmt << "-4";
555       break;
556     default:
557       printRelocationTargetName(Obj, RE, Fmt);
558       break;
559     }
560     // X86 and ARM share some relocation types in common.
561   } else if (Arch == Triple::x86 || Arch == Triple::arm ||
562              Arch == Triple::ppc) {
563     // Generic relocation types...
564     switch (Type) {
565     case MachO::GENERIC_RELOC_PAIR: // prints no info
566       return Error::success();
567     case MachO::GENERIC_RELOC_SECTDIFF: {
568       DataRefImpl RelNext = Rel;
569       Obj->moveRelocationNext(RelNext);
570       MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
571 
572       // X86 sect diff's must be followed by a relocation of type
573       // GENERIC_RELOC_PAIR.
574       unsigned RType = Obj->getAnyRelocationType(RENext);
575 
576       if (RType != MachO::GENERIC_RELOC_PAIR)
577         reportError(Obj->getFileName(), "Expected GENERIC_RELOC_PAIR after "
578                                         "GENERIC_RELOC_SECTDIFF.");
579 
580       printRelocationTargetName(Obj, RE, Fmt);
581       Fmt << "-";
582       printRelocationTargetName(Obj, RENext, Fmt);
583       break;
584     }
585     }
586 
587     if (Arch == Triple::x86 || Arch == Triple::ppc) {
588       switch (Type) {
589       case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: {
590         DataRefImpl RelNext = Rel;
591         Obj->moveRelocationNext(RelNext);
592         MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
593 
594         // X86 sect diff's must be followed by a relocation of type
595         // GENERIC_RELOC_PAIR.
596         unsigned RType = Obj->getAnyRelocationType(RENext);
597         if (RType != MachO::GENERIC_RELOC_PAIR)
598           reportError(Obj->getFileName(), "Expected GENERIC_RELOC_PAIR after "
599                                           "GENERIC_RELOC_LOCAL_SECTDIFF.");
600 
601         printRelocationTargetName(Obj, RE, Fmt);
602         Fmt << "-";
603         printRelocationTargetName(Obj, RENext, Fmt);
604         break;
605       }
606       case MachO::GENERIC_RELOC_TLV: {
607         printRelocationTargetName(Obj, RE, Fmt);
608         Fmt << "@TLV";
609         if (IsPCRel)
610           Fmt << "P";
611         break;
612       }
613       default:
614         printRelocationTargetName(Obj, RE, Fmt);
615       }
616     } else { // ARM-specific relocations
617       switch (Type) {
618       case MachO::ARM_RELOC_HALF:
619       case MachO::ARM_RELOC_HALF_SECTDIFF: {
620         // Half relocations steal a bit from the length field to encode
621         // whether this is an upper16 or a lower16 relocation.
622         bool isUpper = (Obj->getAnyRelocationLength(RE) & 0x1) == 1;
623 
624         if (isUpper)
625           Fmt << ":upper16:(";
626         else
627           Fmt << ":lower16:(";
628         printRelocationTargetName(Obj, RE, Fmt);
629 
630         DataRefImpl RelNext = Rel;
631         Obj->moveRelocationNext(RelNext);
632         MachO::any_relocation_info RENext = Obj->getRelocation(RelNext);
633 
634         // ARM half relocs must be followed by a relocation of type
635         // ARM_RELOC_PAIR.
636         unsigned RType = Obj->getAnyRelocationType(RENext);
637         if (RType != MachO::ARM_RELOC_PAIR)
638           reportError(Obj->getFileName(), "Expected ARM_RELOC_PAIR after "
639                                           "ARM_RELOC_HALF");
640 
641         // NOTE: The half of the target virtual address is stashed in the
642         // address field of the secondary relocation, but we can't reverse
643         // engineer the constant offset from it without decoding the movw/movt
644         // instruction to find the other half in its immediate field.
645 
646         // ARM_RELOC_HALF_SECTDIFF encodes the second section in the
647         // symbol/section pointer of the follow-on relocation.
648         if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) {
649           Fmt << "-";
650           printRelocationTargetName(Obj, RENext, Fmt);
651         }
652 
653         Fmt << ")";
654         break;
655       }
656       default: {
657         printRelocationTargetName(Obj, RE, Fmt);
658       }
659       }
660     }
661   } else
662     printRelocationTargetName(Obj, RE, Fmt);
663 
664   Fmt.flush();
665   Result.append(FmtBuf.begin(), FmtBuf.end());
666   return Error::success();
667 }
668 
PrintIndirectSymbolTable(MachOObjectFile * O,bool verbose,uint32_t n,uint32_t count,uint32_t stride,uint64_t addr)669 static void PrintIndirectSymbolTable(MachOObjectFile *O, bool verbose,
670                                      uint32_t n, uint32_t count,
671                                      uint32_t stride, uint64_t addr) {
672   MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
673   uint32_t nindirectsyms = Dysymtab.nindirectsyms;
674   if (n > nindirectsyms)
675     outs() << " (entries start past the end of the indirect symbol "
676               "table) (reserved1 field greater than the table size)";
677   else if (n + count > nindirectsyms)
678     outs() << " (entries extends past the end of the indirect symbol "
679               "table)";
680   outs() << "\n";
681   uint32_t cputype = O->getHeader().cputype;
682   if (cputype & MachO::CPU_ARCH_ABI64)
683     outs() << "address            index";
684   else
685     outs() << "address    index";
686   if (verbose)
687     outs() << " name\n";
688   else
689     outs() << "\n";
690   for (uint32_t j = 0; j < count && n + j < nindirectsyms; j++) {
691     if (cputype & MachO::CPU_ARCH_ABI64)
692       outs() << format("0x%016" PRIx64, addr + j * stride) << " ";
693     else
694       outs() << format("0x%08" PRIx32, (uint32_t)addr + j * stride) << " ";
695     MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
696     uint32_t indirect_symbol = O->getIndirectSymbolTableEntry(Dysymtab, n + j);
697     if (indirect_symbol == MachO::INDIRECT_SYMBOL_LOCAL) {
698       outs() << "LOCAL\n";
699       continue;
700     }
701     if (indirect_symbol ==
702         (MachO::INDIRECT_SYMBOL_LOCAL | MachO::INDIRECT_SYMBOL_ABS)) {
703       outs() << "LOCAL ABSOLUTE\n";
704       continue;
705     }
706     if (indirect_symbol == MachO::INDIRECT_SYMBOL_ABS) {
707       outs() << "ABSOLUTE\n";
708       continue;
709     }
710     outs() << format("%5u ", indirect_symbol);
711     if (verbose) {
712       MachO::symtab_command Symtab = O->getSymtabLoadCommand();
713       if (indirect_symbol < Symtab.nsyms) {
714         symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol);
715         SymbolRef Symbol = *Sym;
716         outs() << unwrapOrError(Symbol.getName(), O->getFileName());
717       } else {
718         outs() << "?";
719       }
720     }
721     outs() << "\n";
722   }
723 }
724 
PrintIndirectSymbols(MachOObjectFile * O,bool verbose)725 static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) {
726   for (const auto &Load : O->load_commands()) {
727     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
728       MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load);
729       for (unsigned J = 0; J < Seg.nsects; ++J) {
730         MachO::section_64 Sec = O->getSection64(Load, J);
731         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
732         if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
733             section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
734             section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
735             section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
736             section_type == MachO::S_SYMBOL_STUBS) {
737           uint32_t stride;
738           if (section_type == MachO::S_SYMBOL_STUBS)
739             stride = Sec.reserved2;
740           else
741             stride = 8;
742           if (stride == 0) {
743             outs() << "Can't print indirect symbols for (" << Sec.segname << ","
744                    << Sec.sectname << ") "
745                    << "(size of stubs in reserved2 field is zero)\n";
746             continue;
747           }
748           uint32_t count = Sec.size / stride;
749           outs() << "Indirect symbols for (" << Sec.segname << ","
750                  << Sec.sectname << ") " << count << " entries";
751           uint32_t n = Sec.reserved1;
752           PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
753         }
754       }
755     } else if (Load.C.cmd == MachO::LC_SEGMENT) {
756       MachO::segment_command Seg = O->getSegmentLoadCommand(Load);
757       for (unsigned J = 0; J < Seg.nsects; ++J) {
758         MachO::section Sec = O->getSection(Load, J);
759         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
760         if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
761             section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
762             section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
763             section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
764             section_type == MachO::S_SYMBOL_STUBS) {
765           uint32_t stride;
766           if (section_type == MachO::S_SYMBOL_STUBS)
767             stride = Sec.reserved2;
768           else
769             stride = 4;
770           if (stride == 0) {
771             outs() << "Can't print indirect symbols for (" << Sec.segname << ","
772                    << Sec.sectname << ") "
773                    << "(size of stubs in reserved2 field is zero)\n";
774             continue;
775           }
776           uint32_t count = Sec.size / stride;
777           outs() << "Indirect symbols for (" << Sec.segname << ","
778                  << Sec.sectname << ") " << count << " entries";
779           uint32_t n = Sec.reserved1;
780           PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
781         }
782       }
783     }
784   }
785 }
786 
PrintRType(const uint64_t cputype,const unsigned r_type)787 static void PrintRType(const uint64_t cputype, const unsigned r_type) {
788   static char const *generic_r_types[] = {
789     "VANILLA ", "PAIR    ", "SECTDIF ", "PBLAPTR ", "LOCSDIF ", "TLV     ",
790     "  6 (?) ", "  7 (?) ", "  8 (?) ", "  9 (?) ", " 10 (?) ", " 11 (?) ",
791     " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
792   };
793   static char const *x86_64_r_types[] = {
794     "UNSIGND ", "SIGNED  ", "BRANCH  ", "GOT_LD  ", "GOT     ", "SUB     ",
795     "SIGNED1 ", "SIGNED2 ", "SIGNED4 ", "TLV     ", " 10 (?) ", " 11 (?) ",
796     " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
797   };
798   static char const *arm_r_types[] = {
799     "VANILLA ", "PAIR    ", "SECTDIFF", "LOCSDIF ", "PBLAPTR ",
800     "BR24    ", "T_BR22  ", "T_BR32  ", "HALF    ", "HALFDIF ",
801     " 10 (?) ", " 11 (?) ", " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
802   };
803   static char const *arm64_r_types[] = {
804     "UNSIGND ", "SUB     ", "BR26    ", "PAGE21  ", "PAGOF12 ",
805     "GOTLDP  ", "GOTLDPOF", "PTRTGOT ", "TLVLDP  ", "TLVLDPOF",
806     "ADDEND  ", " 11 (?) ", " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) "
807   };
808 
809   if (r_type > 0xf){
810     outs() << format("%-7u", r_type) << " ";
811     return;
812   }
813   switch (cputype) {
814     case MachO::CPU_TYPE_I386:
815       outs() << generic_r_types[r_type];
816       break;
817     case MachO::CPU_TYPE_X86_64:
818       outs() << x86_64_r_types[r_type];
819       break;
820     case MachO::CPU_TYPE_ARM:
821       outs() << arm_r_types[r_type];
822       break;
823     case MachO::CPU_TYPE_ARM64:
824     case MachO::CPU_TYPE_ARM64_32:
825       outs() << arm64_r_types[r_type];
826       break;
827     default:
828       outs() << format("%-7u ", r_type);
829   }
830 }
831 
PrintRLength(const uint64_t cputype,const unsigned r_type,const unsigned r_length,const bool previous_arm_half)832 static void PrintRLength(const uint64_t cputype, const unsigned r_type,
833                          const unsigned r_length, const bool previous_arm_half){
834   if (cputype == MachO::CPU_TYPE_ARM &&
835       (r_type == MachO::ARM_RELOC_HALF ||
836        r_type == MachO::ARM_RELOC_HALF_SECTDIFF || previous_arm_half == true)) {
837     if ((r_length & 0x1) == 0)
838       outs() << "lo/";
839     else
840       outs() << "hi/";
841     if ((r_length & 0x1) == 0)
842       outs() << "arm ";
843     else
844       outs() << "thm ";
845   } else {
846     switch (r_length) {
847       case 0:
848         outs() << "byte   ";
849         break;
850       case 1:
851         outs() << "word   ";
852         break;
853       case 2:
854         outs() << "long   ";
855         break;
856       case 3:
857         if (cputype == MachO::CPU_TYPE_X86_64)
858           outs() << "quad   ";
859         else
860           outs() << format("?(%2d)  ", r_length);
861         break;
862       default:
863         outs() << format("?(%2d)  ", r_length);
864     }
865   }
866 }
867 
PrintRelocationEntries(const MachOObjectFile * O,const relocation_iterator Begin,const relocation_iterator End,const uint64_t cputype,const bool verbose)868 static void PrintRelocationEntries(const MachOObjectFile *O,
869                                    const relocation_iterator Begin,
870                                    const relocation_iterator End,
871                                    const uint64_t cputype,
872                                    const bool verbose) {
873   const MachO::symtab_command Symtab = O->getSymtabLoadCommand();
874   bool previous_arm_half = false;
875   bool previous_sectdiff = false;
876   uint32_t sectdiff_r_type = 0;
877 
878   for (relocation_iterator Reloc = Begin; Reloc != End; ++Reloc) {
879     const DataRefImpl Rel = Reloc->getRawDataRefImpl();
880     const MachO::any_relocation_info RE = O->getRelocation(Rel);
881     const unsigned r_type = O->getAnyRelocationType(RE);
882     const bool r_scattered = O->isRelocationScattered(RE);
883     const unsigned r_pcrel = O->getAnyRelocationPCRel(RE);
884     const unsigned r_length = O->getAnyRelocationLength(RE);
885     const unsigned r_address = O->getAnyRelocationAddress(RE);
886     const bool r_extern = (r_scattered ? false :
887                            O->getPlainRelocationExternal(RE));
888     const uint32_t r_value = (r_scattered ?
889                               O->getScatteredRelocationValue(RE) : 0);
890     const unsigned r_symbolnum = (r_scattered ? 0 :
891                                   O->getPlainRelocationSymbolNum(RE));
892 
893     if (r_scattered && cputype != MachO::CPU_TYPE_X86_64) {
894       if (verbose) {
895         // scattered: address
896         if ((cputype == MachO::CPU_TYPE_I386 &&
897              r_type == MachO::GENERIC_RELOC_PAIR) ||
898             (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR))
899           outs() << "         ";
900         else
901           outs() << format("%08x ", (unsigned int)r_address);
902 
903         // scattered: pcrel
904         if (r_pcrel)
905           outs() << "True  ";
906         else
907           outs() << "False ";
908 
909         // scattered: length
910         PrintRLength(cputype, r_type, r_length, previous_arm_half);
911 
912         // scattered: extern & type
913         outs() << "n/a    ";
914         PrintRType(cputype, r_type);
915 
916         // scattered: scattered & value
917         outs() << format("True      0x%08x", (unsigned int)r_value);
918         if (previous_sectdiff == false) {
919           if ((cputype == MachO::CPU_TYPE_ARM &&
920                r_type == MachO::ARM_RELOC_PAIR))
921             outs() << format(" half = 0x%04x ", (unsigned int)r_address);
922         } else if (cputype == MachO::CPU_TYPE_ARM &&
923                    sectdiff_r_type == MachO::ARM_RELOC_HALF_SECTDIFF)
924           outs() << format(" other_half = 0x%04x ", (unsigned int)r_address);
925         if ((cputype == MachO::CPU_TYPE_I386 &&
926              (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
927               r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) ||
928             (cputype == MachO::CPU_TYPE_ARM &&
929              (sectdiff_r_type == MachO::ARM_RELOC_SECTDIFF ||
930               sectdiff_r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
931               sectdiff_r_type == MachO::ARM_RELOC_HALF_SECTDIFF))) {
932           previous_sectdiff = true;
933           sectdiff_r_type = r_type;
934         } else {
935           previous_sectdiff = false;
936           sectdiff_r_type = 0;
937         }
938         if (cputype == MachO::CPU_TYPE_ARM &&
939             (r_type == MachO::ARM_RELOC_HALF ||
940              r_type == MachO::ARM_RELOC_HALF_SECTDIFF))
941           previous_arm_half = true;
942         else
943           previous_arm_half = false;
944         outs() << "\n";
945       }
946       else {
947         // scattered: address pcrel length extern type scattered value
948         outs() << format("%08x %1d     %-2d     n/a    %-7d 1         0x%08x\n",
949                          (unsigned int)r_address, r_pcrel, r_length, r_type,
950                          (unsigned int)r_value);
951       }
952     }
953     else {
954       if (verbose) {
955         // plain: address
956         if (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR)
957           outs() << "         ";
958         else
959           outs() << format("%08x ", (unsigned int)r_address);
960 
961         // plain: pcrel
962         if (r_pcrel)
963           outs() << "True  ";
964         else
965           outs() << "False ";
966 
967         // plain: length
968         PrintRLength(cputype, r_type, r_length, previous_arm_half);
969 
970         if (r_extern) {
971           // plain: extern & type & scattered
972           outs() << "True   ";
973           PrintRType(cputype, r_type);
974           outs() << "False     ";
975 
976           // plain: symbolnum/value
977           if (r_symbolnum > Symtab.nsyms)
978             outs() << format("?(%d)\n", r_symbolnum);
979           else {
980             SymbolRef Symbol = *O->getSymbolByIndex(r_symbolnum);
981             Expected<StringRef> SymNameNext = Symbol.getName();
982             const char *name = NULL;
983             if (SymNameNext)
984               name = SymNameNext->data();
985             if (name == NULL)
986               outs() << format("?(%d)\n", r_symbolnum);
987             else
988               outs() << name << "\n";
989           }
990         }
991         else {
992           // plain: extern & type & scattered
993           outs() << "False  ";
994           PrintRType(cputype, r_type);
995           outs() << "False     ";
996 
997           // plain: symbolnum/value
998           if (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR)
999             outs() << format("other_half = 0x%04x\n", (unsigned int)r_address);
1000           else if ((cputype == MachO::CPU_TYPE_ARM64 ||
1001                     cputype == MachO::CPU_TYPE_ARM64_32) &&
1002                    r_type == MachO::ARM64_RELOC_ADDEND)
1003             outs() << format("addend = 0x%06x\n", (unsigned int)r_symbolnum);
1004           else {
1005             outs() << format("%d ", r_symbolnum);
1006             if (r_symbolnum == MachO::R_ABS)
1007               outs() << "R_ABS\n";
1008             else {
1009               // in this case, r_symbolnum is actually a 1-based section number
1010               uint32_t nsects = O->section_end()->getRawDataRefImpl().d.a;
1011               if (r_symbolnum > 0 && r_symbolnum <= nsects) {
1012                 object::DataRefImpl DRI;
1013                 DRI.d.a = r_symbolnum-1;
1014                 StringRef SegName = O->getSectionFinalSegmentName(DRI);
1015                 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI))
1016                   outs() << "(" << SegName << "," << *NameOrErr << ")\n";
1017                 else
1018                   outs() << "(?,?)\n";
1019               }
1020               else {
1021                 outs() << "(?,?)\n";
1022               }
1023             }
1024           }
1025         }
1026         if (cputype == MachO::CPU_TYPE_ARM &&
1027             (r_type == MachO::ARM_RELOC_HALF ||
1028              r_type == MachO::ARM_RELOC_HALF_SECTDIFF))
1029           previous_arm_half = true;
1030         else
1031           previous_arm_half = false;
1032       }
1033       else {
1034         // plain: address pcrel length extern type scattered symbolnum/section
1035         outs() << format("%08x %1d     %-2d     %1d      %-7d 0         %d\n",
1036                          (unsigned int)r_address, r_pcrel, r_length, r_extern,
1037                          r_type, r_symbolnum);
1038       }
1039     }
1040   }
1041 }
1042 
PrintRelocations(const MachOObjectFile * O,const bool verbose)1043 static void PrintRelocations(const MachOObjectFile *O, const bool verbose) {
1044   const uint64_t cputype = O->getHeader().cputype;
1045   const MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
1046   if (Dysymtab.nextrel != 0) {
1047     outs() << "External relocation information " << Dysymtab.nextrel
1048            << " entries";
1049     outs() << "\naddress  pcrel length extern type    scattered "
1050               "symbolnum/value\n";
1051     PrintRelocationEntries(O, O->extrel_begin(), O->extrel_end(), cputype,
1052                            verbose);
1053   }
1054   if (Dysymtab.nlocrel != 0) {
1055     outs() << format("Local relocation information %u entries",
1056                      Dysymtab.nlocrel);
1057     outs() << "\naddress  pcrel length extern type    scattered "
1058               "symbolnum/value\n";
1059     PrintRelocationEntries(O, O->locrel_begin(), O->locrel_end(), cputype,
1060                            verbose);
1061   }
1062   for (const auto &Load : O->load_commands()) {
1063     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
1064       const MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load);
1065       for (unsigned J = 0; J < Seg.nsects; ++J) {
1066         const MachO::section_64 Sec = O->getSection64(Load, J);
1067         if (Sec.nreloc != 0) {
1068           DataRefImpl DRI;
1069           DRI.d.a = J;
1070           const StringRef SegName = O->getSectionFinalSegmentName(DRI);
1071           if (Expected<StringRef> NameOrErr = O->getSectionName(DRI))
1072             outs() << "Relocation information (" << SegName << "," << *NameOrErr
1073                    << format(") %u entries", Sec.nreloc);
1074           else
1075             outs() << "Relocation information (" << SegName << ",?) "
1076                    << format("%u entries", Sec.nreloc);
1077           outs() << "\naddress  pcrel length extern type    scattered "
1078                     "symbolnum/value\n";
1079           PrintRelocationEntries(O, O->section_rel_begin(DRI),
1080                                  O->section_rel_end(DRI), cputype, verbose);
1081         }
1082       }
1083     } else if (Load.C.cmd == MachO::LC_SEGMENT) {
1084       const MachO::segment_command Seg = O->getSegmentLoadCommand(Load);
1085       for (unsigned J = 0; J < Seg.nsects; ++J) {
1086         const MachO::section Sec = O->getSection(Load, J);
1087         if (Sec.nreloc != 0) {
1088           DataRefImpl DRI;
1089           DRI.d.a = J;
1090           const StringRef SegName = O->getSectionFinalSegmentName(DRI);
1091           if (Expected<StringRef> NameOrErr = O->getSectionName(DRI))
1092             outs() << "Relocation information (" << SegName << "," << *NameOrErr
1093                    << format(") %u entries", Sec.nreloc);
1094           else
1095             outs() << "Relocation information (" << SegName << ",?) "
1096                    << format("%u entries", Sec.nreloc);
1097           outs() << "\naddress  pcrel length extern type    scattered "
1098                     "symbolnum/value\n";
1099           PrintRelocationEntries(O, O->section_rel_begin(DRI),
1100                                  O->section_rel_end(DRI), cputype, verbose);
1101         }
1102       }
1103     }
1104   }
1105 }
1106 
PrintDataInCodeTable(MachOObjectFile * O,bool verbose)1107 static void PrintDataInCodeTable(MachOObjectFile *O, bool verbose) {
1108   MachO::linkedit_data_command DIC = O->getDataInCodeLoadCommand();
1109   uint32_t nentries = DIC.datasize / sizeof(struct MachO::data_in_code_entry);
1110   outs() << "Data in code table (" << nentries << " entries)\n";
1111   outs() << "offset     length kind\n";
1112   for (dice_iterator DI = O->begin_dices(), DE = O->end_dices(); DI != DE;
1113        ++DI) {
1114     uint32_t Offset;
1115     DI->getOffset(Offset);
1116     outs() << format("0x%08" PRIx32, Offset) << " ";
1117     uint16_t Length;
1118     DI->getLength(Length);
1119     outs() << format("%6u", Length) << " ";
1120     uint16_t Kind;
1121     DI->getKind(Kind);
1122     if (verbose) {
1123       switch (Kind) {
1124       case MachO::DICE_KIND_DATA:
1125         outs() << "DATA";
1126         break;
1127       case MachO::DICE_KIND_JUMP_TABLE8:
1128         outs() << "JUMP_TABLE8";
1129         break;
1130       case MachO::DICE_KIND_JUMP_TABLE16:
1131         outs() << "JUMP_TABLE16";
1132         break;
1133       case MachO::DICE_KIND_JUMP_TABLE32:
1134         outs() << "JUMP_TABLE32";
1135         break;
1136       case MachO::DICE_KIND_ABS_JUMP_TABLE32:
1137         outs() << "ABS_JUMP_TABLE32";
1138         break;
1139       default:
1140         outs() << format("0x%04" PRIx32, Kind);
1141         break;
1142       }
1143     } else
1144       outs() << format("0x%04" PRIx32, Kind);
1145     outs() << "\n";
1146   }
1147 }
1148 
PrintLinkOptHints(MachOObjectFile * O)1149 static void PrintLinkOptHints(MachOObjectFile *O) {
1150   MachO::linkedit_data_command LohLC = O->getLinkOptHintsLoadCommand();
1151   const char *loh = O->getData().substr(LohLC.dataoff, 1).data();
1152   uint32_t nloh = LohLC.datasize;
1153   outs() << "Linker optimiztion hints (" << nloh << " total bytes)\n";
1154   for (uint32_t i = 0; i < nloh;) {
1155     unsigned n;
1156     uint64_t identifier = decodeULEB128((const uint8_t *)(loh + i), &n);
1157     i += n;
1158     outs() << "    identifier " << identifier << " ";
1159     if (i >= nloh)
1160       return;
1161     switch (identifier) {
1162     case 1:
1163       outs() << "AdrpAdrp\n";
1164       break;
1165     case 2:
1166       outs() << "AdrpLdr\n";
1167       break;
1168     case 3:
1169       outs() << "AdrpAddLdr\n";
1170       break;
1171     case 4:
1172       outs() << "AdrpLdrGotLdr\n";
1173       break;
1174     case 5:
1175       outs() << "AdrpAddStr\n";
1176       break;
1177     case 6:
1178       outs() << "AdrpLdrGotStr\n";
1179       break;
1180     case 7:
1181       outs() << "AdrpAdd\n";
1182       break;
1183     case 8:
1184       outs() << "AdrpLdrGot\n";
1185       break;
1186     default:
1187       outs() << "Unknown identifier value\n";
1188       break;
1189     }
1190     uint64_t narguments = decodeULEB128((const uint8_t *)(loh + i), &n);
1191     i += n;
1192     outs() << "    narguments " << narguments << "\n";
1193     if (i >= nloh)
1194       return;
1195 
1196     for (uint32_t j = 0; j < narguments; j++) {
1197       uint64_t value = decodeULEB128((const uint8_t *)(loh + i), &n);
1198       i += n;
1199       outs() << "\tvalue " << format("0x%" PRIx64, value) << "\n";
1200       if (i >= nloh)
1201         return;
1202     }
1203   }
1204 }
1205 
PrintDylibs(MachOObjectFile * O,bool JustId)1206 static void PrintDylibs(MachOObjectFile *O, bool JustId) {
1207   unsigned Index = 0;
1208   for (const auto &Load : O->load_commands()) {
1209     if ((JustId && Load.C.cmd == MachO::LC_ID_DYLIB) ||
1210         (!JustId && (Load.C.cmd == MachO::LC_ID_DYLIB ||
1211                      Load.C.cmd == MachO::LC_LOAD_DYLIB ||
1212                      Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
1213                      Load.C.cmd == MachO::LC_REEXPORT_DYLIB ||
1214                      Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
1215                      Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB))) {
1216       MachO::dylib_command dl = O->getDylibIDLoadCommand(Load);
1217       if (dl.dylib.name < dl.cmdsize) {
1218         const char *p = (const char *)(Load.Ptr) + dl.dylib.name;
1219         if (JustId)
1220           outs() << p << "\n";
1221         else {
1222           outs() << "\t" << p;
1223           outs() << " (compatibility version "
1224                  << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
1225                  << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
1226                  << (dl.dylib.compatibility_version & 0xff) << ",";
1227           outs() << " current version "
1228                  << ((dl.dylib.current_version >> 16) & 0xffff) << "."
1229                  << ((dl.dylib.current_version >> 8) & 0xff) << "."
1230                  << (dl.dylib.current_version & 0xff);
1231           if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB)
1232             outs() << ", weak";
1233           if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB)
1234             outs() << ", reexport";
1235           if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
1236             outs() << ", upward";
1237           if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB)
1238             outs() << ", lazy";
1239           outs() << ")\n";
1240         }
1241       } else {
1242         outs() << "\tBad offset (" << dl.dylib.name << ") for name of ";
1243         if (Load.C.cmd == MachO::LC_ID_DYLIB)
1244           outs() << "LC_ID_DYLIB ";
1245         else if (Load.C.cmd == MachO::LC_LOAD_DYLIB)
1246           outs() << "LC_LOAD_DYLIB ";
1247         else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB)
1248           outs() << "LC_LOAD_WEAK_DYLIB ";
1249         else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB)
1250           outs() << "LC_LAZY_LOAD_DYLIB ";
1251         else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB)
1252           outs() << "LC_REEXPORT_DYLIB ";
1253         else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
1254           outs() << "LC_LOAD_UPWARD_DYLIB ";
1255         else
1256           outs() << "LC_??? ";
1257         outs() << "command " << Index++ << "\n";
1258       }
1259     }
1260   }
1261 }
1262 
1263 typedef DenseMap<uint64_t, StringRef> SymbolAddressMap;
1264 
CreateSymbolAddressMap(MachOObjectFile * O,SymbolAddressMap * AddrMap)1265 static void CreateSymbolAddressMap(MachOObjectFile *O,
1266                                    SymbolAddressMap *AddrMap) {
1267   // Create a map of symbol addresses to symbol names.
1268   const StringRef FileName = O->getFileName();
1269   for (const SymbolRef &Symbol : O->symbols()) {
1270     SymbolRef::Type ST = unwrapOrError(Symbol.getType(), FileName);
1271     if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
1272         ST == SymbolRef::ST_Other) {
1273       uint64_t Address = cantFail(Symbol.getValue());
1274       StringRef SymName = unwrapOrError(Symbol.getName(), FileName);
1275       if (!SymName.startswith(".objc"))
1276         (*AddrMap)[Address] = SymName;
1277     }
1278   }
1279 }
1280 
1281 // GuessSymbolName is passed the address of what might be a symbol and a
1282 // pointer to the SymbolAddressMap.  It returns the name of a symbol
1283 // with that address or nullptr if no symbol is found with that address.
GuessSymbolName(uint64_t value,SymbolAddressMap * AddrMap)1284 static const char *GuessSymbolName(uint64_t value, SymbolAddressMap *AddrMap) {
1285   const char *SymbolName = nullptr;
1286   // A DenseMap can't lookup up some values.
1287   if (value != 0xffffffffffffffffULL && value != 0xfffffffffffffffeULL) {
1288     StringRef name = AddrMap->lookup(value);
1289     if (!name.empty())
1290       SymbolName = name.data();
1291   }
1292   return SymbolName;
1293 }
1294 
DumpCstringChar(const char c)1295 static void DumpCstringChar(const char c) {
1296   char p[2];
1297   p[0] = c;
1298   p[1] = '\0';
1299   outs().write_escaped(p);
1300 }
1301 
DumpCstringSection(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1302 static void DumpCstringSection(MachOObjectFile *O, const char *sect,
1303                                uint32_t sect_size, uint64_t sect_addr,
1304                                bool print_addresses) {
1305   for (uint32_t i = 0; i < sect_size; i++) {
1306     if (print_addresses) {
1307       if (O->is64Bit())
1308         outs() << format("%016" PRIx64, sect_addr + i) << "  ";
1309       else
1310         outs() << format("%08" PRIx64, sect_addr + i) << "  ";
1311     }
1312     for (; i < sect_size && sect[i] != '\0'; i++)
1313       DumpCstringChar(sect[i]);
1314     if (i < sect_size && sect[i] == '\0')
1315       outs() << "\n";
1316   }
1317 }
1318 
DumpLiteral4(uint32_t l,float f)1319 static void DumpLiteral4(uint32_t l, float f) {
1320   outs() << format("0x%08" PRIx32, l);
1321   if ((l & 0x7f800000) != 0x7f800000)
1322     outs() << format(" (%.16e)\n", f);
1323   else {
1324     if (l == 0x7f800000)
1325       outs() << " (+Infinity)\n";
1326     else if (l == 0xff800000)
1327       outs() << " (-Infinity)\n";
1328     else if ((l & 0x00400000) == 0x00400000)
1329       outs() << " (non-signaling Not-a-Number)\n";
1330     else
1331       outs() << " (signaling Not-a-Number)\n";
1332   }
1333 }
1334 
DumpLiteral4Section(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1335 static void DumpLiteral4Section(MachOObjectFile *O, const char *sect,
1336                                 uint32_t sect_size, uint64_t sect_addr,
1337                                 bool print_addresses) {
1338   for (uint32_t i = 0; i < sect_size; i += sizeof(float)) {
1339     if (print_addresses) {
1340       if (O->is64Bit())
1341         outs() << format("%016" PRIx64, sect_addr + i) << "  ";
1342       else
1343         outs() << format("%08" PRIx64, sect_addr + i) << "  ";
1344     }
1345     float f;
1346     memcpy(&f, sect + i, sizeof(float));
1347     if (O->isLittleEndian() != sys::IsLittleEndianHost)
1348       sys::swapByteOrder(f);
1349     uint32_t l;
1350     memcpy(&l, sect + i, sizeof(uint32_t));
1351     if (O->isLittleEndian() != sys::IsLittleEndianHost)
1352       sys::swapByteOrder(l);
1353     DumpLiteral4(l, f);
1354   }
1355 }
1356 
DumpLiteral8(MachOObjectFile * O,uint32_t l0,uint32_t l1,double d)1357 static void DumpLiteral8(MachOObjectFile *O, uint32_t l0, uint32_t l1,
1358                          double d) {
1359   outs() << format("0x%08" PRIx32, l0) << " " << format("0x%08" PRIx32, l1);
1360   uint32_t Hi, Lo;
1361   Hi = (O->isLittleEndian()) ? l1 : l0;
1362   Lo = (O->isLittleEndian()) ? l0 : l1;
1363 
1364   // Hi is the high word, so this is equivalent to if(isfinite(d))
1365   if ((Hi & 0x7ff00000) != 0x7ff00000)
1366     outs() << format(" (%.16e)\n", d);
1367   else {
1368     if (Hi == 0x7ff00000 && Lo == 0)
1369       outs() << " (+Infinity)\n";
1370     else if (Hi == 0xfff00000 && Lo == 0)
1371       outs() << " (-Infinity)\n";
1372     else if ((Hi & 0x00080000) == 0x00080000)
1373       outs() << " (non-signaling Not-a-Number)\n";
1374     else
1375       outs() << " (signaling Not-a-Number)\n";
1376   }
1377 }
1378 
DumpLiteral8Section(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1379 static void DumpLiteral8Section(MachOObjectFile *O, const char *sect,
1380                                 uint32_t sect_size, uint64_t sect_addr,
1381                                 bool print_addresses) {
1382   for (uint32_t i = 0; i < sect_size; i += sizeof(double)) {
1383     if (print_addresses) {
1384       if (O->is64Bit())
1385         outs() << format("%016" PRIx64, sect_addr + i) << "  ";
1386       else
1387         outs() << format("%08" PRIx64, sect_addr + i) << "  ";
1388     }
1389     double d;
1390     memcpy(&d, sect + i, sizeof(double));
1391     if (O->isLittleEndian() != sys::IsLittleEndianHost)
1392       sys::swapByteOrder(d);
1393     uint32_t l0, l1;
1394     memcpy(&l0, sect + i, sizeof(uint32_t));
1395     memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
1396     if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1397       sys::swapByteOrder(l0);
1398       sys::swapByteOrder(l1);
1399     }
1400     DumpLiteral8(O, l0, l1, d);
1401   }
1402 }
1403 
DumpLiteral16(uint32_t l0,uint32_t l1,uint32_t l2,uint32_t l3)1404 static void DumpLiteral16(uint32_t l0, uint32_t l1, uint32_t l2, uint32_t l3) {
1405   outs() << format("0x%08" PRIx32, l0) << " ";
1406   outs() << format("0x%08" PRIx32, l1) << " ";
1407   outs() << format("0x%08" PRIx32, l2) << " ";
1408   outs() << format("0x%08" PRIx32, l3) << "\n";
1409 }
1410 
DumpLiteral16Section(MachOObjectFile * O,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1411 static void DumpLiteral16Section(MachOObjectFile *O, const char *sect,
1412                                  uint32_t sect_size, uint64_t sect_addr,
1413                                  bool print_addresses) {
1414   for (uint32_t i = 0; i < sect_size; i += 16) {
1415     if (print_addresses) {
1416       if (O->is64Bit())
1417         outs() << format("%016" PRIx64, sect_addr + i) << "  ";
1418       else
1419         outs() << format("%08" PRIx64, sect_addr + i) << "  ";
1420     }
1421     uint32_t l0, l1, l2, l3;
1422     memcpy(&l0, sect + i, sizeof(uint32_t));
1423     memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
1424     memcpy(&l2, sect + i + 2 * sizeof(uint32_t), sizeof(uint32_t));
1425     memcpy(&l3, sect + i + 3 * sizeof(uint32_t), sizeof(uint32_t));
1426     if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1427       sys::swapByteOrder(l0);
1428       sys::swapByteOrder(l1);
1429       sys::swapByteOrder(l2);
1430       sys::swapByteOrder(l3);
1431     }
1432     DumpLiteral16(l0, l1, l2, l3);
1433   }
1434 }
1435 
DumpLiteralPointerSection(MachOObjectFile * O,const SectionRef & Section,const char * sect,uint32_t sect_size,uint64_t sect_addr,bool print_addresses)1436 static void DumpLiteralPointerSection(MachOObjectFile *O,
1437                                       const SectionRef &Section,
1438                                       const char *sect, uint32_t sect_size,
1439                                       uint64_t sect_addr,
1440                                       bool print_addresses) {
1441   // Collect the literal sections in this Mach-O file.
1442   std::vector<SectionRef> LiteralSections;
1443   for (const SectionRef &Section : O->sections()) {
1444     DataRefImpl Ref = Section.getRawDataRefImpl();
1445     uint32_t section_type;
1446     if (O->is64Bit()) {
1447       const MachO::section_64 Sec = O->getSection64(Ref);
1448       section_type = Sec.flags & MachO::SECTION_TYPE;
1449     } else {
1450       const MachO::section Sec = O->getSection(Ref);
1451       section_type = Sec.flags & MachO::SECTION_TYPE;
1452     }
1453     if (section_type == MachO::S_CSTRING_LITERALS ||
1454         section_type == MachO::S_4BYTE_LITERALS ||
1455         section_type == MachO::S_8BYTE_LITERALS ||
1456         section_type == MachO::S_16BYTE_LITERALS)
1457       LiteralSections.push_back(Section);
1458   }
1459 
1460   // Set the size of the literal pointer.
1461   uint32_t lp_size = O->is64Bit() ? 8 : 4;
1462 
1463   // Collect the external relocation symbols for the literal pointers.
1464   std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
1465   for (const RelocationRef &Reloc : Section.relocations()) {
1466     DataRefImpl Rel;
1467     MachO::any_relocation_info RE;
1468     bool isExtern = false;
1469     Rel = Reloc.getRawDataRefImpl();
1470     RE = O->getRelocation(Rel);
1471     isExtern = O->getPlainRelocationExternal(RE);
1472     if (isExtern) {
1473       uint64_t RelocOffset = Reloc.getOffset();
1474       symbol_iterator RelocSym = Reloc.getSymbol();
1475       Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
1476     }
1477   }
1478   array_pod_sort(Relocs.begin(), Relocs.end());
1479 
1480   // Dump each literal pointer.
1481   for (uint32_t i = 0; i < sect_size; i += lp_size) {
1482     if (print_addresses) {
1483       if (O->is64Bit())
1484         outs() << format("%016" PRIx64, sect_addr + i) << "  ";
1485       else
1486         outs() << format("%08" PRIx64, sect_addr + i) << "  ";
1487     }
1488     uint64_t lp;
1489     if (O->is64Bit()) {
1490       memcpy(&lp, sect + i, sizeof(uint64_t));
1491       if (O->isLittleEndian() != sys::IsLittleEndianHost)
1492         sys::swapByteOrder(lp);
1493     } else {
1494       uint32_t li;
1495       memcpy(&li, sect + i, sizeof(uint32_t));
1496       if (O->isLittleEndian() != sys::IsLittleEndianHost)
1497         sys::swapByteOrder(li);
1498       lp = li;
1499     }
1500 
1501     // First look for an external relocation entry for this literal pointer.
1502     auto Reloc = find_if(Relocs, [&](const std::pair<uint64_t, SymbolRef> &P) {
1503       return P.first == i;
1504     });
1505     if (Reloc != Relocs.end()) {
1506       symbol_iterator RelocSym = Reloc->second;
1507       StringRef SymName = unwrapOrError(RelocSym->getName(), O->getFileName());
1508       outs() << "external relocation entry for symbol:" << SymName << "\n";
1509       continue;
1510     }
1511 
1512     // For local references see what the section the literal pointer points to.
1513     auto Sect = find_if(LiteralSections, [&](const SectionRef &R) {
1514       return lp >= R.getAddress() && lp < R.getAddress() + R.getSize();
1515     });
1516     if (Sect == LiteralSections.end()) {
1517       outs() << format("0x%" PRIx64, lp) << " (not in a literal section)\n";
1518       continue;
1519     }
1520 
1521     uint64_t SectAddress = Sect->getAddress();
1522     uint64_t SectSize = Sect->getSize();
1523 
1524     StringRef SectName;
1525     Expected<StringRef> SectNameOrErr = Sect->getName();
1526     if (SectNameOrErr)
1527       SectName = *SectNameOrErr;
1528     else
1529       consumeError(SectNameOrErr.takeError());
1530 
1531     DataRefImpl Ref = Sect->getRawDataRefImpl();
1532     StringRef SegmentName = O->getSectionFinalSegmentName(Ref);
1533     outs() << SegmentName << ":" << SectName << ":";
1534 
1535     uint32_t section_type;
1536     if (O->is64Bit()) {
1537       const MachO::section_64 Sec = O->getSection64(Ref);
1538       section_type = Sec.flags & MachO::SECTION_TYPE;
1539     } else {
1540       const MachO::section Sec = O->getSection(Ref);
1541       section_type = Sec.flags & MachO::SECTION_TYPE;
1542     }
1543 
1544     StringRef BytesStr = unwrapOrError(Sect->getContents(), O->getFileName());
1545 
1546     const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
1547 
1548     switch (section_type) {
1549     case MachO::S_CSTRING_LITERALS:
1550       for (uint64_t i = lp - SectAddress; i < SectSize && Contents[i] != '\0';
1551            i++) {
1552         DumpCstringChar(Contents[i]);
1553       }
1554       outs() << "\n";
1555       break;
1556     case MachO::S_4BYTE_LITERALS:
1557       float f;
1558       memcpy(&f, Contents + (lp - SectAddress), sizeof(float));
1559       uint32_t l;
1560       memcpy(&l, Contents + (lp - SectAddress), sizeof(uint32_t));
1561       if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1562         sys::swapByteOrder(f);
1563         sys::swapByteOrder(l);
1564       }
1565       DumpLiteral4(l, f);
1566       break;
1567     case MachO::S_8BYTE_LITERALS: {
1568       double d;
1569       memcpy(&d, Contents + (lp - SectAddress), sizeof(double));
1570       uint32_t l0, l1;
1571       memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
1572       memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
1573              sizeof(uint32_t));
1574       if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1575         sys::swapByteOrder(f);
1576         sys::swapByteOrder(l0);
1577         sys::swapByteOrder(l1);
1578       }
1579       DumpLiteral8(O, l0, l1, d);
1580       break;
1581     }
1582     case MachO::S_16BYTE_LITERALS: {
1583       uint32_t l0, l1, l2, l3;
1584       memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
1585       memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
1586              sizeof(uint32_t));
1587       memcpy(&l2, Contents + (lp - SectAddress) + 2 * sizeof(uint32_t),
1588              sizeof(uint32_t));
1589       memcpy(&l3, Contents + (lp - SectAddress) + 3 * sizeof(uint32_t),
1590              sizeof(uint32_t));
1591       if (O->isLittleEndian() != sys::IsLittleEndianHost) {
1592         sys::swapByteOrder(l0);
1593         sys::swapByteOrder(l1);
1594         sys::swapByteOrder(l2);
1595         sys::swapByteOrder(l3);
1596       }
1597       DumpLiteral16(l0, l1, l2, l3);
1598       break;
1599     }
1600     }
1601   }
1602 }
1603 
DumpInitTermPointerSection(MachOObjectFile * O,const SectionRef & Section,const char * sect,uint32_t sect_size,uint64_t sect_addr,SymbolAddressMap * AddrMap,bool verbose)1604 static void DumpInitTermPointerSection(MachOObjectFile *O,
1605                                        const SectionRef &Section,
1606                                        const char *sect,
1607                                        uint32_t sect_size, uint64_t sect_addr,
1608                                        SymbolAddressMap *AddrMap,
1609                                        bool verbose) {
1610   uint32_t stride;
1611   stride = (O->is64Bit()) ? sizeof(uint64_t) : sizeof(uint32_t);
1612 
1613   // Collect the external relocation symbols for the pointers.
1614   std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
1615   for (const RelocationRef &Reloc : Section.relocations()) {
1616     DataRefImpl Rel;
1617     MachO::any_relocation_info RE;
1618     bool isExtern = false;
1619     Rel = Reloc.getRawDataRefImpl();
1620     RE = O->getRelocation(Rel);
1621     isExtern = O->getPlainRelocationExternal(RE);
1622     if (isExtern) {
1623       uint64_t RelocOffset = Reloc.getOffset();
1624       symbol_iterator RelocSym = Reloc.getSymbol();
1625       Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
1626     }
1627   }
1628   array_pod_sort(Relocs.begin(), Relocs.end());
1629 
1630   for (uint32_t i = 0; i < sect_size; i += stride) {
1631     const char *SymbolName = nullptr;
1632     uint64_t p;
1633     if (O->is64Bit()) {
1634       outs() << format("0x%016" PRIx64, sect_addr + i * stride) << " ";
1635       uint64_t pointer_value;
1636       memcpy(&pointer_value, sect + i, stride);
1637       if (O->isLittleEndian() != sys::IsLittleEndianHost)
1638         sys::swapByteOrder(pointer_value);
1639       outs() << format("0x%016" PRIx64, pointer_value);
1640       p = pointer_value;
1641     } else {
1642       outs() << format("0x%08" PRIx64, sect_addr + i * stride) << " ";
1643       uint32_t pointer_value;
1644       memcpy(&pointer_value, sect + i, stride);
1645       if (O->isLittleEndian() != sys::IsLittleEndianHost)
1646         sys::swapByteOrder(pointer_value);
1647       outs() << format("0x%08" PRIx32, pointer_value);
1648       p = pointer_value;
1649     }
1650     if (verbose) {
1651       // First look for an external relocation entry for this pointer.
1652       auto Reloc = find_if(Relocs, [&](const std::pair<uint64_t, SymbolRef> &P) {
1653         return P.first == i;
1654       });
1655       if (Reloc != Relocs.end()) {
1656         symbol_iterator RelocSym = Reloc->second;
1657         outs() << " " << unwrapOrError(RelocSym->getName(), O->getFileName());
1658       } else {
1659         SymbolName = GuessSymbolName(p, AddrMap);
1660         if (SymbolName)
1661           outs() << " " << SymbolName;
1662       }
1663     }
1664     outs() << "\n";
1665   }
1666 }
1667 
DumpRawSectionContents(MachOObjectFile * O,const char * sect,uint32_t size,uint64_t addr)1668 static void DumpRawSectionContents(MachOObjectFile *O, const char *sect,
1669                                    uint32_t size, uint64_t addr) {
1670   uint32_t cputype = O->getHeader().cputype;
1671   if (cputype == MachO::CPU_TYPE_I386 || cputype == MachO::CPU_TYPE_X86_64) {
1672     uint32_t j;
1673     for (uint32_t i = 0; i < size; i += j, addr += j) {
1674       if (O->is64Bit())
1675         outs() << format("%016" PRIx64, addr) << "\t";
1676       else
1677         outs() << format("%08" PRIx64, addr) << "\t";
1678       for (j = 0; j < 16 && i + j < size; j++) {
1679         uint8_t byte_word = *(sect + i + j);
1680         outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
1681       }
1682       outs() << "\n";
1683     }
1684   } else {
1685     uint32_t j;
1686     for (uint32_t i = 0; i < size; i += j, addr += j) {
1687       if (O->is64Bit())
1688         outs() << format("%016" PRIx64, addr) << "\t";
1689       else
1690         outs() << format("%08" PRIx64, addr) << "\t";
1691       for (j = 0; j < 4 * sizeof(int32_t) && i + j < size;
1692            j += sizeof(int32_t)) {
1693         if (i + j + sizeof(int32_t) <= size) {
1694           uint32_t long_word;
1695           memcpy(&long_word, sect + i + j, sizeof(int32_t));
1696           if (O->isLittleEndian() != sys::IsLittleEndianHost)
1697             sys::swapByteOrder(long_word);
1698           outs() << format("%08" PRIx32, long_word) << " ";
1699         } else {
1700           for (uint32_t k = 0; i + j + k < size; k++) {
1701             uint8_t byte_word = *(sect + i + j + k);
1702             outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
1703           }
1704         }
1705       }
1706       outs() << "\n";
1707     }
1708   }
1709 }
1710 
1711 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
1712                              StringRef DisSegName, StringRef DisSectName);
1713 static void DumpProtocolSection(MachOObjectFile *O, const char *sect,
1714                                 uint32_t size, uint32_t addr);
1715 #ifdef HAVE_LIBXAR
1716 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect,
1717                                 uint32_t size, bool verbose,
1718                                 bool PrintXarHeader, bool PrintXarFileHeaders,
1719                                 std::string XarMemberName);
1720 #endif // defined(HAVE_LIBXAR)
1721 
DumpSectionContents(StringRef Filename,MachOObjectFile * O,bool verbose)1722 static void DumpSectionContents(StringRef Filename, MachOObjectFile *O,
1723                                 bool verbose) {
1724   SymbolAddressMap AddrMap;
1725   if (verbose)
1726     CreateSymbolAddressMap(O, &AddrMap);
1727 
1728   for (unsigned i = 0; i < FilterSections.size(); ++i) {
1729     StringRef DumpSection = FilterSections[i];
1730     std::pair<StringRef, StringRef> DumpSegSectName;
1731     DumpSegSectName = DumpSection.split(',');
1732     StringRef DumpSegName, DumpSectName;
1733     if (!DumpSegSectName.second.empty()) {
1734       DumpSegName = DumpSegSectName.first;
1735       DumpSectName = DumpSegSectName.second;
1736     } else {
1737       DumpSegName = "";
1738       DumpSectName = DumpSegSectName.first;
1739     }
1740     for (const SectionRef &Section : O->sections()) {
1741       StringRef SectName;
1742       Expected<StringRef> SecNameOrErr = Section.getName();
1743       if (SecNameOrErr)
1744         SectName = *SecNameOrErr;
1745       else
1746         consumeError(SecNameOrErr.takeError());
1747 
1748       if (!DumpSection.empty())
1749         FoundSectionSet.insert(DumpSection);
1750 
1751       DataRefImpl Ref = Section.getRawDataRefImpl();
1752       StringRef SegName = O->getSectionFinalSegmentName(Ref);
1753       if ((DumpSegName.empty() || SegName == DumpSegName) &&
1754           (SectName == DumpSectName)) {
1755 
1756         uint32_t section_flags;
1757         if (O->is64Bit()) {
1758           const MachO::section_64 Sec = O->getSection64(Ref);
1759           section_flags = Sec.flags;
1760 
1761         } else {
1762           const MachO::section Sec = O->getSection(Ref);
1763           section_flags = Sec.flags;
1764         }
1765         uint32_t section_type = section_flags & MachO::SECTION_TYPE;
1766 
1767         StringRef BytesStr =
1768             unwrapOrError(Section.getContents(), O->getFileName());
1769         const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1770         uint32_t sect_size = BytesStr.size();
1771         uint64_t sect_addr = Section.getAddress();
1772 
1773         if (!NoLeadingHeaders)
1774           outs() << "Contents of (" << SegName << "," << SectName
1775                  << ") section\n";
1776 
1777         if (verbose) {
1778           if ((section_flags & MachO::S_ATTR_PURE_INSTRUCTIONS) ||
1779               (section_flags & MachO::S_ATTR_SOME_INSTRUCTIONS)) {
1780             DisassembleMachO(Filename, O, SegName, SectName);
1781             continue;
1782           }
1783           if (SegName == "__TEXT" && SectName == "__info_plist") {
1784             outs() << sect;
1785             continue;
1786           }
1787           if (SegName == "__OBJC" && SectName == "__protocol") {
1788             DumpProtocolSection(O, sect, sect_size, sect_addr);
1789             continue;
1790           }
1791 #ifdef HAVE_LIBXAR
1792           if (SegName == "__LLVM" && SectName == "__bundle") {
1793             DumpBitcodeSection(O, sect, sect_size, verbose, !NoSymbolicOperands,
1794                                ArchiveHeaders, "");
1795             continue;
1796           }
1797 #endif // defined(HAVE_LIBXAR)
1798           switch (section_type) {
1799           case MachO::S_REGULAR:
1800             DumpRawSectionContents(O, sect, sect_size, sect_addr);
1801             break;
1802           case MachO::S_ZEROFILL:
1803             outs() << "zerofill section and has no contents in the file\n";
1804             break;
1805           case MachO::S_CSTRING_LITERALS:
1806             DumpCstringSection(O, sect, sect_size, sect_addr, !NoLeadingAddr);
1807             break;
1808           case MachO::S_4BYTE_LITERALS:
1809             DumpLiteral4Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
1810             break;
1811           case MachO::S_8BYTE_LITERALS:
1812             DumpLiteral8Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
1813             break;
1814           case MachO::S_16BYTE_LITERALS:
1815             DumpLiteral16Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
1816             break;
1817           case MachO::S_LITERAL_POINTERS:
1818             DumpLiteralPointerSection(O, Section, sect, sect_size, sect_addr,
1819                                       !NoLeadingAddr);
1820             break;
1821           case MachO::S_MOD_INIT_FUNC_POINTERS:
1822           case MachO::S_MOD_TERM_FUNC_POINTERS:
1823             DumpInitTermPointerSection(O, Section, sect, sect_size, sect_addr,
1824                                        &AddrMap, verbose);
1825             break;
1826           default:
1827             outs() << "Unknown section type ("
1828                    << format("0x%08" PRIx32, section_type) << ")\n";
1829             DumpRawSectionContents(O, sect, sect_size, sect_addr);
1830             break;
1831           }
1832         } else {
1833           if (section_type == MachO::S_ZEROFILL)
1834             outs() << "zerofill section and has no contents in the file\n";
1835           else
1836             DumpRawSectionContents(O, sect, sect_size, sect_addr);
1837         }
1838       }
1839     }
1840   }
1841 }
1842 
DumpInfoPlistSectionContents(StringRef Filename,MachOObjectFile * O)1843 static void DumpInfoPlistSectionContents(StringRef Filename,
1844                                          MachOObjectFile *O) {
1845   for (const SectionRef &Section : O->sections()) {
1846     StringRef SectName;
1847     Expected<StringRef> SecNameOrErr = Section.getName();
1848     if (SecNameOrErr)
1849       SectName = *SecNameOrErr;
1850     else
1851       consumeError(SecNameOrErr.takeError());
1852 
1853     DataRefImpl Ref = Section.getRawDataRefImpl();
1854     StringRef SegName = O->getSectionFinalSegmentName(Ref);
1855     if (SegName == "__TEXT" && SectName == "__info_plist") {
1856       if (!NoLeadingHeaders)
1857         outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
1858       StringRef BytesStr =
1859           unwrapOrError(Section.getContents(), O->getFileName());
1860       const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1861       outs() << format("%.*s", BytesStr.size(), sect) << "\n";
1862       return;
1863     }
1864   }
1865 }
1866 
1867 // checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file
1868 // and if it is and there is a list of architecture flags is specified then
1869 // check to make sure this Mach-O file is one of those architectures or all
1870 // architectures were specified.  If not then an error is generated and this
1871 // routine returns false.  Else it returns true.
checkMachOAndArchFlags(ObjectFile * O,StringRef Filename)1872 static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) {
1873   auto *MachO = dyn_cast<MachOObjectFile>(O);
1874 
1875   if (!MachO || ArchAll || ArchFlags.empty())
1876     return true;
1877 
1878   MachO::mach_header H;
1879   MachO::mach_header_64 H_64;
1880   Triple T;
1881   const char *McpuDefault, *ArchFlag;
1882   if (MachO->is64Bit()) {
1883     H_64 = MachO->MachOObjectFile::getHeader64();
1884     T = MachOObjectFile::getArchTriple(H_64.cputype, H_64.cpusubtype,
1885                                        &McpuDefault, &ArchFlag);
1886   } else {
1887     H = MachO->MachOObjectFile::getHeader();
1888     T = MachOObjectFile::getArchTriple(H.cputype, H.cpusubtype,
1889                                        &McpuDefault, &ArchFlag);
1890   }
1891   const std::string ArchFlagName(ArchFlag);
1892   if (!llvm::is_contained(ArchFlags, ArchFlagName)) {
1893     WithColor::error(errs(), "llvm-objdump")
1894         << Filename << ": no architecture specified.\n";
1895     return false;
1896   }
1897   return true;
1898 }
1899 
1900 static void printObjcMetaData(MachOObjectFile *O, bool verbose);
1901 
1902 // ProcessMachO() is passed a single opened Mach-O file, which may be an
1903 // archive member and or in a slice of a universal file.  It prints the
1904 // the file name and header info and then processes it according to the
1905 // command line options.
ProcessMachO(StringRef Name,MachOObjectFile * MachOOF,StringRef ArchiveMemberName=StringRef (),StringRef ArchitectureName=StringRef ())1906 static void ProcessMachO(StringRef Name, MachOObjectFile *MachOOF,
1907                          StringRef ArchiveMemberName = StringRef(),
1908                          StringRef ArchitectureName = StringRef()) {
1909   // If we are doing some processing here on the Mach-O file print the header
1910   // info.  And don't print it otherwise like in the case of printing the
1911   // UniversalHeaders or ArchiveHeaders.
1912   if (Disassemble || Relocations || PrivateHeaders || ExportsTrie || Rebase ||
1913       Bind || SymbolTable || LazyBind || WeakBind || IndirectSymbols ||
1914       DataInCode || LinkOptHints || DylibsUsed || DylibId || ObjcMetaData ||
1915       (!FilterSections.empty())) {
1916     if (!NoLeadingHeaders) {
1917       outs() << Name;
1918       if (!ArchiveMemberName.empty())
1919         outs() << '(' << ArchiveMemberName << ')';
1920       if (!ArchitectureName.empty())
1921         outs() << " (architecture " << ArchitectureName << ")";
1922       outs() << ":\n";
1923     }
1924   }
1925   // To use the report_error() form with an ArchiveName and FileName set
1926   // these up based on what is passed for Name and ArchiveMemberName.
1927   StringRef ArchiveName;
1928   StringRef FileName;
1929   if (!ArchiveMemberName.empty()) {
1930     ArchiveName = Name;
1931     FileName = ArchiveMemberName;
1932   } else {
1933     ArchiveName = StringRef();
1934     FileName = Name;
1935   }
1936 
1937   // If we need the symbol table to do the operation then check it here to
1938   // produce a good error message as to where the Mach-O file comes from in
1939   // the error message.
1940   if (Disassemble || IndirectSymbols || !FilterSections.empty() || UnwindInfo)
1941     if (Error Err = MachOOF->checkSymbolTable())
1942       reportError(std::move(Err), FileName, ArchiveName, ArchitectureName);
1943 
1944   if (DisassembleAll) {
1945     for (const SectionRef &Section : MachOOF->sections()) {
1946       StringRef SectName;
1947       if (Expected<StringRef> NameOrErr = Section.getName())
1948         SectName = *NameOrErr;
1949       else
1950         consumeError(NameOrErr.takeError());
1951 
1952       if (SectName.equals("__text")) {
1953         DataRefImpl Ref = Section.getRawDataRefImpl();
1954         StringRef SegName = MachOOF->getSectionFinalSegmentName(Ref);
1955         DisassembleMachO(FileName, MachOOF, SegName, SectName);
1956       }
1957     }
1958   }
1959   else if (Disassemble) {
1960     if (MachOOF->getHeader().filetype == MachO::MH_KEXT_BUNDLE &&
1961         MachOOF->getHeader().cputype == MachO::CPU_TYPE_ARM64)
1962       DisassembleMachO(FileName, MachOOF, "__TEXT_EXEC", "__text");
1963     else
1964       DisassembleMachO(FileName, MachOOF, "__TEXT", "__text");
1965   }
1966   if (IndirectSymbols)
1967     PrintIndirectSymbols(MachOOF, !NonVerbose);
1968   if (DataInCode)
1969     PrintDataInCodeTable(MachOOF, !NonVerbose);
1970   if (LinkOptHints)
1971     PrintLinkOptHints(MachOOF);
1972   if (Relocations)
1973     PrintRelocations(MachOOF, !NonVerbose);
1974   if (SectionHeaders)
1975     printSectionHeaders(MachOOF);
1976   if (SectionContents)
1977     printSectionContents(MachOOF);
1978   if (!FilterSections.empty())
1979     DumpSectionContents(FileName, MachOOF, !NonVerbose);
1980   if (InfoPlist)
1981     DumpInfoPlistSectionContents(FileName, MachOOF);
1982   if (DylibsUsed)
1983     PrintDylibs(MachOOF, false);
1984   if (DylibId)
1985     PrintDylibs(MachOOF, true);
1986   if (SymbolTable)
1987     printSymbolTable(MachOOF, ArchiveName, ArchitectureName);
1988   if (UnwindInfo)
1989     printMachOUnwindInfo(MachOOF);
1990   if (PrivateHeaders) {
1991     printMachOFileHeader(MachOOF);
1992     printMachOLoadCommands(MachOOF);
1993   }
1994   if (FirstPrivateHeader)
1995     printMachOFileHeader(MachOOF);
1996   if (ObjcMetaData)
1997     printObjcMetaData(MachOOF, !NonVerbose);
1998   if (ExportsTrie)
1999     printExportsTrie(MachOOF);
2000   if (Rebase)
2001     printRebaseTable(MachOOF);
2002   if (Bind)
2003     printBindTable(MachOOF);
2004   if (LazyBind)
2005     printLazyBindTable(MachOOF);
2006   if (WeakBind)
2007     printWeakBindTable(MachOOF);
2008 
2009   if (DwarfDumpType != DIDT_Null) {
2010     std::unique_ptr<DIContext> DICtx = DWARFContext::create(*MachOOF);
2011     // Dump the complete DWARF structure.
2012     DIDumpOptions DumpOpts;
2013     DumpOpts.DumpType = DwarfDumpType;
2014     DICtx->dump(outs(), DumpOpts);
2015   }
2016 }
2017 
2018 // printUnknownCPUType() helps print_fat_headers for unknown CPU's.
printUnknownCPUType(uint32_t cputype,uint32_t cpusubtype)2019 static void printUnknownCPUType(uint32_t cputype, uint32_t cpusubtype) {
2020   outs() << "    cputype (" << cputype << ")\n";
2021   outs() << "    cpusubtype (" << cpusubtype << ")\n";
2022 }
2023 
2024 // printCPUType() helps print_fat_headers by printing the cputype and
2025 // pusubtype (symbolically for the one's it knows about).
printCPUType(uint32_t cputype,uint32_t cpusubtype)2026 static void printCPUType(uint32_t cputype, uint32_t cpusubtype) {
2027   switch (cputype) {
2028   case MachO::CPU_TYPE_I386:
2029     switch (cpusubtype) {
2030     case MachO::CPU_SUBTYPE_I386_ALL:
2031       outs() << "    cputype CPU_TYPE_I386\n";
2032       outs() << "    cpusubtype CPU_SUBTYPE_I386_ALL\n";
2033       break;
2034     default:
2035       printUnknownCPUType(cputype, cpusubtype);
2036       break;
2037     }
2038     break;
2039   case MachO::CPU_TYPE_X86_64:
2040     switch (cpusubtype) {
2041     case MachO::CPU_SUBTYPE_X86_64_ALL:
2042       outs() << "    cputype CPU_TYPE_X86_64\n";
2043       outs() << "    cpusubtype CPU_SUBTYPE_X86_64_ALL\n";
2044       break;
2045     case MachO::CPU_SUBTYPE_X86_64_H:
2046       outs() << "    cputype CPU_TYPE_X86_64\n";
2047       outs() << "    cpusubtype CPU_SUBTYPE_X86_64_H\n";
2048       break;
2049     default:
2050       printUnknownCPUType(cputype, cpusubtype);
2051       break;
2052     }
2053     break;
2054   case MachO::CPU_TYPE_ARM:
2055     switch (cpusubtype) {
2056     case MachO::CPU_SUBTYPE_ARM_ALL:
2057       outs() << "    cputype CPU_TYPE_ARM\n";
2058       outs() << "    cpusubtype CPU_SUBTYPE_ARM_ALL\n";
2059       break;
2060     case MachO::CPU_SUBTYPE_ARM_V4T:
2061       outs() << "    cputype CPU_TYPE_ARM\n";
2062       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V4T\n";
2063       break;
2064     case MachO::CPU_SUBTYPE_ARM_V5TEJ:
2065       outs() << "    cputype CPU_TYPE_ARM\n";
2066       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V5TEJ\n";
2067       break;
2068     case MachO::CPU_SUBTYPE_ARM_XSCALE:
2069       outs() << "    cputype CPU_TYPE_ARM\n";
2070       outs() << "    cpusubtype CPU_SUBTYPE_ARM_XSCALE\n";
2071       break;
2072     case MachO::CPU_SUBTYPE_ARM_V6:
2073       outs() << "    cputype CPU_TYPE_ARM\n";
2074       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V6\n";
2075       break;
2076     case MachO::CPU_SUBTYPE_ARM_V6M:
2077       outs() << "    cputype CPU_TYPE_ARM\n";
2078       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V6M\n";
2079       break;
2080     case MachO::CPU_SUBTYPE_ARM_V7:
2081       outs() << "    cputype CPU_TYPE_ARM\n";
2082       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7\n";
2083       break;
2084     case MachO::CPU_SUBTYPE_ARM_V7EM:
2085       outs() << "    cputype CPU_TYPE_ARM\n";
2086       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7EM\n";
2087       break;
2088     case MachO::CPU_SUBTYPE_ARM_V7K:
2089       outs() << "    cputype CPU_TYPE_ARM\n";
2090       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7K\n";
2091       break;
2092     case MachO::CPU_SUBTYPE_ARM_V7M:
2093       outs() << "    cputype CPU_TYPE_ARM\n";
2094       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7M\n";
2095       break;
2096     case MachO::CPU_SUBTYPE_ARM_V7S:
2097       outs() << "    cputype CPU_TYPE_ARM\n";
2098       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7S\n";
2099       break;
2100     default:
2101       printUnknownCPUType(cputype, cpusubtype);
2102       break;
2103     }
2104     break;
2105   case MachO::CPU_TYPE_ARM64:
2106     switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2107     case MachO::CPU_SUBTYPE_ARM64_ALL:
2108       outs() << "    cputype CPU_TYPE_ARM64\n";
2109       outs() << "    cpusubtype CPU_SUBTYPE_ARM64_ALL\n";
2110       break;
2111     case MachO::CPU_SUBTYPE_ARM64_V8:
2112       outs() << "    cputype CPU_TYPE_ARM64\n";
2113       outs() << "    cpusubtype CPU_SUBTYPE_ARM64_V8\n";
2114       break;
2115     case MachO::CPU_SUBTYPE_ARM64E:
2116       outs() << "    cputype CPU_TYPE_ARM64\n";
2117       outs() << "    cpusubtype CPU_SUBTYPE_ARM64E\n";
2118       break;
2119     default:
2120       printUnknownCPUType(cputype, cpusubtype);
2121       break;
2122     }
2123     break;
2124   case MachO::CPU_TYPE_ARM64_32:
2125     switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2126     case MachO::CPU_SUBTYPE_ARM64_32_V8:
2127       outs() << "    cputype CPU_TYPE_ARM64_32\n";
2128       outs() << "    cpusubtype CPU_SUBTYPE_ARM64_32_V8\n";
2129       break;
2130     default:
2131       printUnknownCPUType(cputype, cpusubtype);
2132       break;
2133     }
2134     break;
2135   default:
2136     printUnknownCPUType(cputype, cpusubtype);
2137     break;
2138   }
2139 }
2140 
printMachOUniversalHeaders(const object::MachOUniversalBinary * UB,bool verbose)2141 static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB,
2142                                        bool verbose) {
2143   outs() << "Fat headers\n";
2144   if (verbose) {
2145     if (UB->getMagic() == MachO::FAT_MAGIC)
2146       outs() << "fat_magic FAT_MAGIC\n";
2147     else // UB->getMagic() == MachO::FAT_MAGIC_64
2148       outs() << "fat_magic FAT_MAGIC_64\n";
2149   } else
2150     outs() << "fat_magic " << format("0x%" PRIx32, MachO::FAT_MAGIC) << "\n";
2151 
2152   uint32_t nfat_arch = UB->getNumberOfObjects();
2153   StringRef Buf = UB->getData();
2154   uint64_t size = Buf.size();
2155   uint64_t big_size = sizeof(struct MachO::fat_header) +
2156                       nfat_arch * sizeof(struct MachO::fat_arch);
2157   outs() << "nfat_arch " << UB->getNumberOfObjects();
2158   if (nfat_arch == 0)
2159     outs() << " (malformed, contains zero architecture types)\n";
2160   else if (big_size > size)
2161     outs() << " (malformed, architectures past end of file)\n";
2162   else
2163     outs() << "\n";
2164 
2165   for (uint32_t i = 0; i < nfat_arch; ++i) {
2166     MachOUniversalBinary::ObjectForArch OFA(UB, i);
2167     uint32_t cputype = OFA.getCPUType();
2168     uint32_t cpusubtype = OFA.getCPUSubType();
2169     outs() << "architecture ";
2170     for (uint32_t j = 0; i != 0 && j <= i - 1; j++) {
2171       MachOUniversalBinary::ObjectForArch other_OFA(UB, j);
2172       uint32_t other_cputype = other_OFA.getCPUType();
2173       uint32_t other_cpusubtype = other_OFA.getCPUSubType();
2174       if (cputype != 0 && cpusubtype != 0 && cputype == other_cputype &&
2175           (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) ==
2176               (other_cpusubtype & ~MachO::CPU_SUBTYPE_MASK)) {
2177         outs() << "(illegal duplicate architecture) ";
2178         break;
2179       }
2180     }
2181     if (verbose) {
2182       outs() << OFA.getArchFlagName() << "\n";
2183       printCPUType(cputype, cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2184     } else {
2185       outs() << i << "\n";
2186       outs() << "    cputype " << cputype << "\n";
2187       outs() << "    cpusubtype " << (cpusubtype & ~MachO::CPU_SUBTYPE_MASK)
2188              << "\n";
2189     }
2190     if (verbose &&
2191         (cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64)
2192       outs() << "    capabilities CPU_SUBTYPE_LIB64\n";
2193     else
2194       outs() << "    capabilities "
2195              << format("0x%" PRIx32,
2196                        (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24) << "\n";
2197     outs() << "    offset " << OFA.getOffset();
2198     if (OFA.getOffset() > size)
2199       outs() << " (past end of file)";
2200     if (OFA.getOffset() % (1ull << OFA.getAlign()) != 0)
2201       outs() << " (not aligned on it's alignment (2^" << OFA.getAlign() << ")";
2202     outs() << "\n";
2203     outs() << "    size " << OFA.getSize();
2204     big_size = OFA.getOffset() + OFA.getSize();
2205     if (big_size > size)
2206       outs() << " (past end of file)";
2207     outs() << "\n";
2208     outs() << "    align 2^" << OFA.getAlign() << " (" << (1 << OFA.getAlign())
2209            << ")\n";
2210   }
2211 }
2212 
printArchiveChild(StringRef Filename,const Archive::Child & C,size_t ChildIndex,bool verbose,bool print_offset,StringRef ArchitectureName=StringRef ())2213 static void printArchiveChild(StringRef Filename, const Archive::Child &C,
2214                               size_t ChildIndex, bool verbose,
2215                               bool print_offset,
2216                               StringRef ArchitectureName = StringRef()) {
2217   if (print_offset)
2218     outs() << C.getChildOffset() << "\t";
2219   sys::fs::perms Mode =
2220       unwrapOrError(C.getAccessMode(), getFileNameForError(C, ChildIndex),
2221                     Filename, ArchitectureName);
2222   if (verbose) {
2223     // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG.
2224     // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG.
2225     outs() << "-";
2226     outs() << ((Mode & sys::fs::owner_read) ? "r" : "-");
2227     outs() << ((Mode & sys::fs::owner_write) ? "w" : "-");
2228     outs() << ((Mode & sys::fs::owner_exe) ? "x" : "-");
2229     outs() << ((Mode & sys::fs::group_read) ? "r" : "-");
2230     outs() << ((Mode & sys::fs::group_write) ? "w" : "-");
2231     outs() << ((Mode & sys::fs::group_exe) ? "x" : "-");
2232     outs() << ((Mode & sys::fs::others_read) ? "r" : "-");
2233     outs() << ((Mode & sys::fs::others_write) ? "w" : "-");
2234     outs() << ((Mode & sys::fs::others_exe) ? "x" : "-");
2235   } else {
2236     outs() << format("0%o ", Mode);
2237   }
2238 
2239   outs() << format("%3d/%-3d %5" PRId64 " ",
2240                    unwrapOrError(C.getUID(), getFileNameForError(C, ChildIndex),
2241                                  Filename, ArchitectureName),
2242                    unwrapOrError(C.getGID(), getFileNameForError(C, ChildIndex),
2243                                  Filename, ArchitectureName),
2244                    unwrapOrError(C.getRawSize(),
2245                                  getFileNameForError(C, ChildIndex), Filename,
2246                                  ArchitectureName));
2247 
2248   StringRef RawLastModified = C.getRawLastModified();
2249   if (verbose) {
2250     unsigned Seconds;
2251     if (RawLastModified.getAsInteger(10, Seconds))
2252       outs() << "(date: \"" << RawLastModified
2253              << "\" contains non-decimal chars) ";
2254     else {
2255       // Since cime(3) returns a 26 character string of the form:
2256       // "Sun Sep 16 01:03:52 1973\n\0"
2257       // just print 24 characters.
2258       time_t t = Seconds;
2259       outs() << format("%.24s ", ctime(&t));
2260     }
2261   } else {
2262     outs() << RawLastModified << " ";
2263   }
2264 
2265   if (verbose) {
2266     Expected<StringRef> NameOrErr = C.getName();
2267     if (!NameOrErr) {
2268       consumeError(NameOrErr.takeError());
2269       outs() << unwrapOrError(C.getRawName(),
2270                               getFileNameForError(C, ChildIndex), Filename,
2271                               ArchitectureName)
2272              << "\n";
2273     } else {
2274       StringRef Name = NameOrErr.get();
2275       outs() << Name << "\n";
2276     }
2277   } else {
2278     outs() << unwrapOrError(C.getRawName(), getFileNameForError(C, ChildIndex),
2279                             Filename, ArchitectureName)
2280            << "\n";
2281   }
2282 }
2283 
printArchiveHeaders(StringRef Filename,Archive * A,bool verbose,bool print_offset,StringRef ArchitectureName=StringRef ())2284 static void printArchiveHeaders(StringRef Filename, Archive *A, bool verbose,
2285                                 bool print_offset,
2286                                 StringRef ArchitectureName = StringRef()) {
2287   Error Err = Error::success();
2288   size_t I = 0;
2289   for (const auto &C : A->children(Err, false))
2290     printArchiveChild(Filename, C, I++, verbose, print_offset,
2291                       ArchitectureName);
2292 
2293   if (Err)
2294     reportError(std::move(Err), Filename, "", ArchitectureName);
2295 }
2296 
ValidateArchFlags()2297 static bool ValidateArchFlags() {
2298   // Check for -arch all and verifiy the -arch flags are valid.
2299   for (unsigned i = 0; i < ArchFlags.size(); ++i) {
2300     if (ArchFlags[i] == "all") {
2301       ArchAll = true;
2302     } else {
2303       if (!MachOObjectFile::isValidArch(ArchFlags[i])) {
2304         WithColor::error(errs(), "llvm-objdump")
2305             << "unknown architecture named '" + ArchFlags[i] +
2306                    "'for the -arch option\n";
2307         return false;
2308       }
2309     }
2310   }
2311   return true;
2312 }
2313 
2314 // ParseInputMachO() parses the named Mach-O file in Filename and handles the
2315 // -arch flags selecting just those slices as specified by them and also parses
2316 // archive files.  Then for each individual Mach-O file ProcessMachO() is
2317 // called to process the file based on the command line options.
parseInputMachO(StringRef Filename)2318 void objdump::parseInputMachO(StringRef Filename) {
2319   if (!ValidateArchFlags())
2320     return;
2321 
2322   // Attempt to open the binary.
2323   Expected<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename);
2324   if (!BinaryOrErr) {
2325     if (Error E = isNotObjectErrorInvalidFileType(BinaryOrErr.takeError()))
2326       reportError(std::move(E), Filename);
2327     else
2328       outs() << Filename << ": is not an object file\n";
2329     return;
2330   }
2331   Binary &Bin = *BinaryOrErr.get().getBinary();
2332 
2333   if (Archive *A = dyn_cast<Archive>(&Bin)) {
2334     outs() << "Archive : " << Filename << "\n";
2335     if (ArchiveHeaders)
2336       printArchiveHeaders(Filename, A, !NonVerbose, ArchiveMemberOffsets);
2337 
2338     Error Err = Error::success();
2339     unsigned I = -1;
2340     for (auto &C : A->children(Err)) {
2341       ++I;
2342       Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2343       if (!ChildOrErr) {
2344         if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2345           reportError(std::move(E), getFileNameForError(C, I), Filename);
2346         continue;
2347       }
2348       if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
2349         if (!checkMachOAndArchFlags(O, Filename))
2350           return;
2351         ProcessMachO(Filename, O, O->getFileName());
2352       }
2353     }
2354     if (Err)
2355       reportError(std::move(Err), Filename);
2356     return;
2357   }
2358   if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) {
2359     parseInputMachO(UB);
2360     return;
2361   }
2362   if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) {
2363     if (!checkMachOAndArchFlags(O, Filename))
2364       return;
2365     if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O))
2366       ProcessMachO(Filename, MachOOF);
2367     else
2368       WithColor::error(errs(), "llvm-objdump")
2369           << Filename << "': "
2370           << "object is not a Mach-O file type.\n";
2371     return;
2372   }
2373   llvm_unreachable("Input object can't be invalid at this point");
2374 }
2375 
parseInputMachO(MachOUniversalBinary * UB)2376 void objdump::parseInputMachO(MachOUniversalBinary *UB) {
2377   if (!ValidateArchFlags())
2378     return;
2379 
2380   auto Filename = UB->getFileName();
2381 
2382   if (UniversalHeaders)
2383     printMachOUniversalHeaders(UB, !NonVerbose);
2384 
2385   // If we have a list of architecture flags specified dump only those.
2386   if (!ArchAll && !ArchFlags.empty()) {
2387     // Look for a slice in the universal binary that matches each ArchFlag.
2388     bool ArchFound;
2389     for (unsigned i = 0; i < ArchFlags.size(); ++i) {
2390       ArchFound = false;
2391       for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2392                                                   E = UB->end_objects();
2393             I != E; ++I) {
2394         if (ArchFlags[i] == I->getArchFlagName()) {
2395           ArchFound = true;
2396           Expected<std::unique_ptr<ObjectFile>> ObjOrErr =
2397               I->getAsObjectFile();
2398           std::string ArchitectureName;
2399           if (ArchFlags.size() > 1)
2400             ArchitectureName = I->getArchFlagName();
2401           if (ObjOrErr) {
2402             ObjectFile &O = *ObjOrErr.get();
2403             if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
2404               ProcessMachO(Filename, MachOOF, "", ArchitectureName);
2405           } else if (Error E = isNotObjectErrorInvalidFileType(
2406                          ObjOrErr.takeError())) {
2407             reportError(std::move(E), "", Filename, ArchitectureName);
2408             continue;
2409           } else if (Expected<std::unique_ptr<Archive>> AOrErr =
2410                          I->getAsArchive()) {
2411             std::unique_ptr<Archive> &A = *AOrErr;
2412             outs() << "Archive : " << Filename;
2413             if (!ArchitectureName.empty())
2414               outs() << " (architecture " << ArchitectureName << ")";
2415             outs() << "\n";
2416             if (ArchiveHeaders)
2417               printArchiveHeaders(Filename, A.get(), !NonVerbose,
2418                                   ArchiveMemberOffsets, ArchitectureName);
2419             Error Err = Error::success();
2420             unsigned I = -1;
2421             for (auto &C : A->children(Err)) {
2422               ++I;
2423               Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2424               if (!ChildOrErr) {
2425                 if (Error E =
2426                         isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2427                   reportError(std::move(E), getFileNameForError(C, I), Filename,
2428                               ArchitectureName);
2429                 continue;
2430               }
2431               if (MachOObjectFile *O =
2432                       dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
2433                 ProcessMachO(Filename, O, O->getFileName(), ArchitectureName);
2434             }
2435             if (Err)
2436               reportError(std::move(Err), Filename);
2437           } else {
2438             consumeError(AOrErr.takeError());
2439             reportError(Filename,
2440                         "Mach-O universal file for architecture " +
2441                             StringRef(I->getArchFlagName()) +
2442                             " is not a Mach-O file or an archive file");
2443           }
2444         }
2445       }
2446       if (!ArchFound) {
2447         WithColor::error(errs(), "llvm-objdump")
2448             << "file: " + Filename + " does not contain "
2449             << "architecture: " + ArchFlags[i] + "\n";
2450         return;
2451       }
2452     }
2453     return;
2454   }
2455   // No architecture flags were specified so if this contains a slice that
2456   // matches the host architecture dump only that.
2457   if (!ArchAll) {
2458     for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2459                                                 E = UB->end_objects();
2460           I != E; ++I) {
2461       if (MachOObjectFile::getHostArch().getArchName() ==
2462           I->getArchFlagName()) {
2463         Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
2464         std::string ArchiveName;
2465         ArchiveName.clear();
2466         if (ObjOrErr) {
2467           ObjectFile &O = *ObjOrErr.get();
2468           if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
2469             ProcessMachO(Filename, MachOOF);
2470         } else if (Error E =
2471                        isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) {
2472           reportError(std::move(E), Filename);
2473         } else if (Expected<std::unique_ptr<Archive>> AOrErr =
2474                        I->getAsArchive()) {
2475           std::unique_ptr<Archive> &A = *AOrErr;
2476           outs() << "Archive : " << Filename << "\n";
2477           if (ArchiveHeaders)
2478             printArchiveHeaders(Filename, A.get(), !NonVerbose,
2479                                 ArchiveMemberOffsets);
2480           Error Err = Error::success();
2481           unsigned I = -1;
2482           for (auto &C : A->children(Err)) {
2483             ++I;
2484             Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2485             if (!ChildOrErr) {
2486               if (Error E =
2487                       isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2488                 reportError(std::move(E), getFileNameForError(C, I), Filename);
2489               continue;
2490             }
2491             if (MachOObjectFile *O =
2492                     dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
2493               ProcessMachO(Filename, O, O->getFileName());
2494           }
2495           if (Err)
2496             reportError(std::move(Err), Filename);
2497         } else {
2498           consumeError(AOrErr.takeError());
2499           reportError(Filename, "Mach-O universal file for architecture " +
2500                                     StringRef(I->getArchFlagName()) +
2501                                     " is not a Mach-O file or an archive file");
2502         }
2503         return;
2504       }
2505     }
2506   }
2507   // Either all architectures have been specified or none have been specified
2508   // and this does not contain the host architecture so dump all the slices.
2509   bool moreThanOneArch = UB->getNumberOfObjects() > 1;
2510   for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2511                                               E = UB->end_objects();
2512         I != E; ++I) {
2513     Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
2514     std::string ArchitectureName;
2515     if (moreThanOneArch)
2516       ArchitectureName = I->getArchFlagName();
2517     if (ObjOrErr) {
2518       ObjectFile &Obj = *ObjOrErr.get();
2519       if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj))
2520         ProcessMachO(Filename, MachOOF, "", ArchitectureName);
2521     } else if (Error E =
2522                    isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) {
2523       reportError(std::move(E), Filename, "", ArchitectureName);
2524     } else if (Expected<std::unique_ptr<Archive>> AOrErr = I->getAsArchive()) {
2525       std::unique_ptr<Archive> &A = *AOrErr;
2526       outs() << "Archive : " << Filename;
2527       if (!ArchitectureName.empty())
2528         outs() << " (architecture " << ArchitectureName << ")";
2529       outs() << "\n";
2530       if (ArchiveHeaders)
2531         printArchiveHeaders(Filename, A.get(), !NonVerbose,
2532                             ArchiveMemberOffsets, ArchitectureName);
2533       Error Err = Error::success();
2534       unsigned I = -1;
2535       for (auto &C : A->children(Err)) {
2536         ++I;
2537         Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2538         if (!ChildOrErr) {
2539           if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2540             reportError(std::move(E), getFileNameForError(C, I), Filename,
2541                         ArchitectureName);
2542           continue;
2543         }
2544         if (MachOObjectFile *O =
2545                 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
2546           if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O))
2547             ProcessMachO(Filename, MachOOF, MachOOF->getFileName(),
2548                           ArchitectureName);
2549         }
2550       }
2551       if (Err)
2552         reportError(std::move(Err), Filename);
2553     } else {
2554       consumeError(AOrErr.takeError());
2555       reportError(Filename, "Mach-O universal file for architecture " +
2556                                 StringRef(I->getArchFlagName()) +
2557                                 " is not a Mach-O file or an archive file");
2558     }
2559   }
2560 }
2561 
2562 namespace {
2563 // The block of info used by the Symbolizer call backs.
2564 struct DisassembleInfo {
DisassembleInfo__anonabe85be80611::DisassembleInfo2565   DisassembleInfo(MachOObjectFile *O, SymbolAddressMap *AddrMap,
2566                   std::vector<SectionRef> *Sections, bool verbose)
2567     : verbose(verbose), O(O), AddrMap(AddrMap), Sections(Sections) {}
2568   bool verbose;
2569   MachOObjectFile *O;
2570   SectionRef S;
2571   SymbolAddressMap *AddrMap;
2572   std::vector<SectionRef> *Sections;
2573   const char *class_name = nullptr;
2574   const char *selector_name = nullptr;
2575   std::unique_ptr<char[]> method = nullptr;
2576   char *demangled_name = nullptr;
2577   uint64_t adrp_addr = 0;
2578   uint32_t adrp_inst = 0;
2579   std::unique_ptr<SymbolAddressMap> bindtable;
2580   uint32_t depth = 0;
2581 };
2582 } // namespace
2583 
2584 // SymbolizerGetOpInfo() is the operand information call back function.
2585 // This is called to get the symbolic information for operand(s) of an
2586 // instruction when it is being done.  This routine does this from
2587 // the relocation information, symbol table, etc. That block of information
2588 // is a pointer to the struct DisassembleInfo that was passed when the
2589 // disassembler context was created and passed to back to here when
2590 // called back by the disassembler for instruction operands that could have
2591 // relocation information. The address of the instruction containing operand is
2592 // at the Pc parameter.  The immediate value the operand has is passed in
2593 // op_info->Value and is at Offset past the start of the instruction and has a
2594 // byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the
2595 // LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol
2596 // names and addends of the symbolic expression to add for the operand.  The
2597 // value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic
2598 // information is returned then this function returns 1 else it returns 0.
SymbolizerGetOpInfo(void * DisInfo,uint64_t Pc,uint64_t Offset,uint64_t Size,int TagType,void * TagBuf)2599 static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset,
2600                                uint64_t Size, int TagType, void *TagBuf) {
2601   struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
2602   struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf;
2603   uint64_t value = op_info->Value;
2604 
2605   // Make sure all fields returned are zero if we don't set them.
2606   memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1));
2607   op_info->Value = value;
2608 
2609   // If the TagType is not the value 1 which it code knows about or if no
2610   // verbose symbolic information is wanted then just return 0, indicating no
2611   // information is being returned.
2612   if (TagType != 1 || !info->verbose)
2613     return 0;
2614 
2615   unsigned int Arch = info->O->getArch();
2616   if (Arch == Triple::x86) {
2617     if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
2618       return 0;
2619     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2620       // TODO:
2621       // Search the external relocation entries of a fully linked image
2622       // (if any) for an entry that matches this segment offset.
2623       // uint32_t seg_offset = (Pc + Offset);
2624       return 0;
2625     }
2626     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2627     // for an entry for this section offset.
2628     uint32_t sect_addr = info->S.getAddress();
2629     uint32_t sect_offset = (Pc + Offset) - sect_addr;
2630     bool reloc_found = false;
2631     DataRefImpl Rel;
2632     MachO::any_relocation_info RE;
2633     bool isExtern = false;
2634     SymbolRef Symbol;
2635     bool r_scattered = false;
2636     uint32_t r_value, pair_r_value, r_type;
2637     for (const RelocationRef &Reloc : info->S.relocations()) {
2638       uint64_t RelocOffset = Reloc.getOffset();
2639       if (RelocOffset == sect_offset) {
2640         Rel = Reloc.getRawDataRefImpl();
2641         RE = info->O->getRelocation(Rel);
2642         r_type = info->O->getAnyRelocationType(RE);
2643         r_scattered = info->O->isRelocationScattered(RE);
2644         if (r_scattered) {
2645           r_value = info->O->getScatteredRelocationValue(RE);
2646           if (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
2647               r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) {
2648             DataRefImpl RelNext = Rel;
2649             info->O->moveRelocationNext(RelNext);
2650             MachO::any_relocation_info RENext;
2651             RENext = info->O->getRelocation(RelNext);
2652             if (info->O->isRelocationScattered(RENext))
2653               pair_r_value = info->O->getScatteredRelocationValue(RENext);
2654             else
2655               return 0;
2656           }
2657         } else {
2658           isExtern = info->O->getPlainRelocationExternal(RE);
2659           if (isExtern) {
2660             symbol_iterator RelocSym = Reloc.getSymbol();
2661             Symbol = *RelocSym;
2662           }
2663         }
2664         reloc_found = true;
2665         break;
2666       }
2667     }
2668     if (reloc_found && isExtern) {
2669       op_info->AddSymbol.Present = 1;
2670       op_info->AddSymbol.Name =
2671           unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2672       // For i386 extern relocation entries the value in the instruction is
2673       // the offset from the symbol, and value is already set in op_info->Value.
2674       return 1;
2675     }
2676     if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
2677                         r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) {
2678       const char *add = GuessSymbolName(r_value, info->AddrMap);
2679       const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
2680       uint32_t offset = value - (r_value - pair_r_value);
2681       op_info->AddSymbol.Present = 1;
2682       if (add != nullptr)
2683         op_info->AddSymbol.Name = add;
2684       else
2685         op_info->AddSymbol.Value = r_value;
2686       op_info->SubtractSymbol.Present = 1;
2687       if (sub != nullptr)
2688         op_info->SubtractSymbol.Name = sub;
2689       else
2690         op_info->SubtractSymbol.Value = pair_r_value;
2691       op_info->Value = offset;
2692       return 1;
2693     }
2694     return 0;
2695   }
2696   if (Arch == Triple::x86_64) {
2697     if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
2698       return 0;
2699     // For non MH_OBJECT types, like MH_KEXT_BUNDLE, Search the external
2700     // relocation entries of a linked image (if any) for an entry that matches
2701     // this segment offset.
2702     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2703       uint64_t seg_offset = Pc + Offset;
2704       bool reloc_found = false;
2705       DataRefImpl Rel;
2706       MachO::any_relocation_info RE;
2707       bool isExtern = false;
2708       SymbolRef Symbol;
2709       for (const RelocationRef &Reloc : info->O->external_relocations()) {
2710         uint64_t RelocOffset = Reloc.getOffset();
2711         if (RelocOffset == seg_offset) {
2712           Rel = Reloc.getRawDataRefImpl();
2713           RE = info->O->getRelocation(Rel);
2714           // external relocation entries should always be external.
2715           isExtern = info->O->getPlainRelocationExternal(RE);
2716           if (isExtern) {
2717             symbol_iterator RelocSym = Reloc.getSymbol();
2718             Symbol = *RelocSym;
2719           }
2720           reloc_found = true;
2721           break;
2722         }
2723       }
2724       if (reloc_found && isExtern) {
2725         // The Value passed in will be adjusted by the Pc if the instruction
2726         // adds the Pc.  But for x86_64 external relocation entries the Value
2727         // is the offset from the external symbol.
2728         if (info->O->getAnyRelocationPCRel(RE))
2729           op_info->Value -= Pc + Offset + Size;
2730         const char *name =
2731             unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2732         op_info->AddSymbol.Present = 1;
2733         op_info->AddSymbol.Name = name;
2734         return 1;
2735       }
2736       return 0;
2737     }
2738     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2739     // for an entry for this section offset.
2740     uint64_t sect_addr = info->S.getAddress();
2741     uint64_t sect_offset = (Pc + Offset) - sect_addr;
2742     bool reloc_found = false;
2743     DataRefImpl Rel;
2744     MachO::any_relocation_info RE;
2745     bool isExtern = false;
2746     SymbolRef Symbol;
2747     for (const RelocationRef &Reloc : info->S.relocations()) {
2748       uint64_t RelocOffset = Reloc.getOffset();
2749       if (RelocOffset == sect_offset) {
2750         Rel = Reloc.getRawDataRefImpl();
2751         RE = info->O->getRelocation(Rel);
2752         // NOTE: Scattered relocations don't exist on x86_64.
2753         isExtern = info->O->getPlainRelocationExternal(RE);
2754         if (isExtern) {
2755           symbol_iterator RelocSym = Reloc.getSymbol();
2756           Symbol = *RelocSym;
2757         }
2758         reloc_found = true;
2759         break;
2760       }
2761     }
2762     if (reloc_found && isExtern) {
2763       // The Value passed in will be adjusted by the Pc if the instruction
2764       // adds the Pc.  But for x86_64 external relocation entries the Value
2765       // is the offset from the external symbol.
2766       if (info->O->getAnyRelocationPCRel(RE))
2767         op_info->Value -= Pc + Offset + Size;
2768       const char *name =
2769           unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2770       unsigned Type = info->O->getAnyRelocationType(RE);
2771       if (Type == MachO::X86_64_RELOC_SUBTRACTOR) {
2772         DataRefImpl RelNext = Rel;
2773         info->O->moveRelocationNext(RelNext);
2774         MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2775         unsigned TypeNext = info->O->getAnyRelocationType(RENext);
2776         bool isExternNext = info->O->getPlainRelocationExternal(RENext);
2777         unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext);
2778         if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) {
2779           op_info->SubtractSymbol.Present = 1;
2780           op_info->SubtractSymbol.Name = name;
2781           symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum);
2782           Symbol = *RelocSymNext;
2783           name = unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2784         }
2785       }
2786       // TODO: add the VariantKinds to op_info->VariantKind for relocation types
2787       // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT.
2788       op_info->AddSymbol.Present = 1;
2789       op_info->AddSymbol.Name = name;
2790       return 1;
2791     }
2792     return 0;
2793   }
2794   if (Arch == Triple::arm) {
2795     if (Offset != 0 || (Size != 4 && Size != 2))
2796       return 0;
2797     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2798       // TODO:
2799       // Search the external relocation entries of a fully linked image
2800       // (if any) for an entry that matches this segment offset.
2801       // uint32_t seg_offset = (Pc + Offset);
2802       return 0;
2803     }
2804     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2805     // for an entry for this section offset.
2806     uint32_t sect_addr = info->S.getAddress();
2807     uint32_t sect_offset = (Pc + Offset) - sect_addr;
2808     DataRefImpl Rel;
2809     MachO::any_relocation_info RE;
2810     bool isExtern = false;
2811     SymbolRef Symbol;
2812     bool r_scattered = false;
2813     uint32_t r_value, pair_r_value, r_type, r_length, other_half;
2814     auto Reloc =
2815         find_if(info->S.relocations(), [&](const RelocationRef &Reloc) {
2816           uint64_t RelocOffset = Reloc.getOffset();
2817           return RelocOffset == sect_offset;
2818         });
2819 
2820     if (Reloc == info->S.relocations().end())
2821       return 0;
2822 
2823     Rel = Reloc->getRawDataRefImpl();
2824     RE = info->O->getRelocation(Rel);
2825     r_length = info->O->getAnyRelocationLength(RE);
2826     r_scattered = info->O->isRelocationScattered(RE);
2827     if (r_scattered) {
2828       r_value = info->O->getScatteredRelocationValue(RE);
2829       r_type = info->O->getScatteredRelocationType(RE);
2830     } else {
2831       r_type = info->O->getAnyRelocationType(RE);
2832       isExtern = info->O->getPlainRelocationExternal(RE);
2833       if (isExtern) {
2834         symbol_iterator RelocSym = Reloc->getSymbol();
2835         Symbol = *RelocSym;
2836       }
2837     }
2838     if (r_type == MachO::ARM_RELOC_HALF ||
2839         r_type == MachO::ARM_RELOC_SECTDIFF ||
2840         r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
2841         r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2842       DataRefImpl RelNext = Rel;
2843       info->O->moveRelocationNext(RelNext);
2844       MachO::any_relocation_info RENext;
2845       RENext = info->O->getRelocation(RelNext);
2846       other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff;
2847       if (info->O->isRelocationScattered(RENext))
2848         pair_r_value = info->O->getScatteredRelocationValue(RENext);
2849     }
2850 
2851     if (isExtern) {
2852       const char *name =
2853           unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2854       op_info->AddSymbol.Present = 1;
2855       op_info->AddSymbol.Name = name;
2856       switch (r_type) {
2857       case MachO::ARM_RELOC_HALF:
2858         if ((r_length & 0x1) == 1) {
2859           op_info->Value = value << 16 | other_half;
2860           op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2861         } else {
2862           op_info->Value = other_half << 16 | value;
2863           op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2864         }
2865         break;
2866       default:
2867         break;
2868       }
2869       return 1;
2870     }
2871     // If we have a branch that is not an external relocation entry then
2872     // return 0 so the code in tryAddingSymbolicOperand() can use the
2873     // SymbolLookUp call back with the branch target address to look up the
2874     // symbol and possibility add an annotation for a symbol stub.
2875     if (isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 ||
2876                           r_type == MachO::ARM_THUMB_RELOC_BR22))
2877       return 0;
2878 
2879     uint32_t offset = 0;
2880     if (r_type == MachO::ARM_RELOC_HALF ||
2881         r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2882       if ((r_length & 0x1) == 1)
2883         value = value << 16 | other_half;
2884       else
2885         value = other_half << 16 | value;
2886     }
2887     if (r_scattered && (r_type != MachO::ARM_RELOC_HALF &&
2888                         r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) {
2889       offset = value - r_value;
2890       value = r_value;
2891     }
2892 
2893     if (r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2894       if ((r_length & 0x1) == 1)
2895         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2896       else
2897         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2898       const char *add = GuessSymbolName(r_value, info->AddrMap);
2899       const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
2900       int32_t offset = value - (r_value - pair_r_value);
2901       op_info->AddSymbol.Present = 1;
2902       if (add != nullptr)
2903         op_info->AddSymbol.Name = add;
2904       else
2905         op_info->AddSymbol.Value = r_value;
2906       op_info->SubtractSymbol.Present = 1;
2907       if (sub != nullptr)
2908         op_info->SubtractSymbol.Name = sub;
2909       else
2910         op_info->SubtractSymbol.Value = pair_r_value;
2911       op_info->Value = offset;
2912       return 1;
2913     }
2914 
2915     op_info->AddSymbol.Present = 1;
2916     op_info->Value = offset;
2917     if (r_type == MachO::ARM_RELOC_HALF) {
2918       if ((r_length & 0x1) == 1)
2919         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2920       else
2921         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2922     }
2923     const char *add = GuessSymbolName(value, info->AddrMap);
2924     if (add != nullptr) {
2925       op_info->AddSymbol.Name = add;
2926       return 1;
2927     }
2928     op_info->AddSymbol.Value = value;
2929     return 1;
2930   }
2931   if (Arch == Triple::aarch64) {
2932     if (Offset != 0 || Size != 4)
2933       return 0;
2934     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2935       // TODO:
2936       // Search the external relocation entries of a fully linked image
2937       // (if any) for an entry that matches this segment offset.
2938       // uint64_t seg_offset = (Pc + Offset);
2939       return 0;
2940     }
2941     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2942     // for an entry for this section offset.
2943     uint64_t sect_addr = info->S.getAddress();
2944     uint64_t sect_offset = (Pc + Offset) - sect_addr;
2945     auto Reloc =
2946         find_if(info->S.relocations(), [&](const RelocationRef &Reloc) {
2947           uint64_t RelocOffset = Reloc.getOffset();
2948           return RelocOffset == sect_offset;
2949         });
2950 
2951     if (Reloc == info->S.relocations().end())
2952       return 0;
2953 
2954     DataRefImpl Rel = Reloc->getRawDataRefImpl();
2955     MachO::any_relocation_info RE = info->O->getRelocation(Rel);
2956     uint32_t r_type = info->O->getAnyRelocationType(RE);
2957     if (r_type == MachO::ARM64_RELOC_ADDEND) {
2958       DataRefImpl RelNext = Rel;
2959       info->O->moveRelocationNext(RelNext);
2960       MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2961       if (value == 0) {
2962         value = info->O->getPlainRelocationSymbolNum(RENext);
2963         op_info->Value = value;
2964       }
2965     }
2966     // NOTE: Scattered relocations don't exist on arm64.
2967     if (!info->O->getPlainRelocationExternal(RE))
2968       return 0;
2969     const char *name =
2970         unwrapOrError(Reloc->getSymbol()->getName(), info->O->getFileName())
2971             .data();
2972     op_info->AddSymbol.Present = 1;
2973     op_info->AddSymbol.Name = name;
2974 
2975     switch (r_type) {
2976     case MachO::ARM64_RELOC_PAGE21:
2977       /* @page */
2978       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE;
2979       break;
2980     case MachO::ARM64_RELOC_PAGEOFF12:
2981       /* @pageoff */
2982       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF;
2983       break;
2984     case MachO::ARM64_RELOC_GOT_LOAD_PAGE21:
2985       /* @gotpage */
2986       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE;
2987       break;
2988     case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12:
2989       /* @gotpageoff */
2990       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF;
2991       break;
2992     case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21:
2993       /* @tvlppage is not implemented in llvm-mc */
2994       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP;
2995       break;
2996     case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12:
2997       /* @tvlppageoff is not implemented in llvm-mc */
2998       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF;
2999       break;
3000     default:
3001     case MachO::ARM64_RELOC_BRANCH26:
3002       op_info->VariantKind = LLVMDisassembler_VariantKind_None;
3003       break;
3004     }
3005     return 1;
3006   }
3007   return 0;
3008 }
3009 
3010 // GuessCstringPointer is passed the address of what might be a pointer to a
3011 // literal string in a cstring section.  If that address is in a cstring section
3012 // it returns a pointer to that string.  Else it returns nullptr.
GuessCstringPointer(uint64_t ReferenceValue,struct DisassembleInfo * info)3013 static const char *GuessCstringPointer(uint64_t ReferenceValue,
3014                                        struct DisassembleInfo *info) {
3015   for (const auto &Load : info->O->load_commands()) {
3016     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3017       MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3018       for (unsigned J = 0; J < Seg.nsects; ++J) {
3019         MachO::section_64 Sec = info->O->getSection64(Load, J);
3020         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3021         if (section_type == MachO::S_CSTRING_LITERALS &&
3022             ReferenceValue >= Sec.addr &&
3023             ReferenceValue < Sec.addr + Sec.size) {
3024           uint64_t sect_offset = ReferenceValue - Sec.addr;
3025           uint64_t object_offset = Sec.offset + sect_offset;
3026           StringRef MachOContents = info->O->getData();
3027           uint64_t object_size = MachOContents.size();
3028           const char *object_addr = (const char *)MachOContents.data();
3029           if (object_offset < object_size) {
3030             const char *name = object_addr + object_offset;
3031             return name;
3032           } else {
3033             return nullptr;
3034           }
3035         }
3036       }
3037     } else if (Load.C.cmd == MachO::LC_SEGMENT) {
3038       MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
3039       for (unsigned J = 0; J < Seg.nsects; ++J) {
3040         MachO::section Sec = info->O->getSection(Load, J);
3041         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3042         if (section_type == MachO::S_CSTRING_LITERALS &&
3043             ReferenceValue >= Sec.addr &&
3044             ReferenceValue < Sec.addr + Sec.size) {
3045           uint64_t sect_offset = ReferenceValue - Sec.addr;
3046           uint64_t object_offset = Sec.offset + sect_offset;
3047           StringRef MachOContents = info->O->getData();
3048           uint64_t object_size = MachOContents.size();
3049           const char *object_addr = (const char *)MachOContents.data();
3050           if (object_offset < object_size) {
3051             const char *name = object_addr + object_offset;
3052             return name;
3053           } else {
3054             return nullptr;
3055           }
3056         }
3057       }
3058     }
3059   }
3060   return nullptr;
3061 }
3062 
3063 // GuessIndirectSymbol returns the name of the indirect symbol for the
3064 // ReferenceValue passed in or nullptr.  This is used when ReferenceValue maybe
3065 // an address of a symbol stub or a lazy or non-lazy pointer to associate the
3066 // symbol name being referenced by the stub or pointer.
GuessIndirectSymbol(uint64_t ReferenceValue,struct DisassembleInfo * info)3067 static const char *GuessIndirectSymbol(uint64_t ReferenceValue,
3068                                        struct DisassembleInfo *info) {
3069   MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand();
3070   MachO::symtab_command Symtab = info->O->getSymtabLoadCommand();
3071   for (const auto &Load : info->O->load_commands()) {
3072     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3073       MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3074       for (unsigned J = 0; J < Seg.nsects; ++J) {
3075         MachO::section_64 Sec = info->O->getSection64(Load, J);
3076         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3077         if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3078              section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3079              section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3080              section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
3081              section_type == MachO::S_SYMBOL_STUBS) &&
3082             ReferenceValue >= Sec.addr &&
3083             ReferenceValue < Sec.addr + Sec.size) {
3084           uint32_t stride;
3085           if (section_type == MachO::S_SYMBOL_STUBS)
3086             stride = Sec.reserved2;
3087           else
3088             stride = 8;
3089           if (stride == 0)
3090             return nullptr;
3091           uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
3092           if (index < Dysymtab.nindirectsyms) {
3093             uint32_t indirect_symbol =
3094                 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
3095             if (indirect_symbol < Symtab.nsyms) {
3096               symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
3097               return unwrapOrError(Sym->getName(), info->O->getFileName())
3098                   .data();
3099             }
3100           }
3101         }
3102       }
3103     } else if (Load.C.cmd == MachO::LC_SEGMENT) {
3104       MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
3105       for (unsigned J = 0; J < Seg.nsects; ++J) {
3106         MachO::section Sec = info->O->getSection(Load, J);
3107         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3108         if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3109              section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3110              section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3111              section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
3112              section_type == MachO::S_SYMBOL_STUBS) &&
3113             ReferenceValue >= Sec.addr &&
3114             ReferenceValue < Sec.addr + Sec.size) {
3115           uint32_t stride;
3116           if (section_type == MachO::S_SYMBOL_STUBS)
3117             stride = Sec.reserved2;
3118           else
3119             stride = 4;
3120           if (stride == 0)
3121             return nullptr;
3122           uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
3123           if (index < Dysymtab.nindirectsyms) {
3124             uint32_t indirect_symbol =
3125                 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
3126             if (indirect_symbol < Symtab.nsyms) {
3127               symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
3128               return unwrapOrError(Sym->getName(), info->O->getFileName())
3129                   .data();
3130             }
3131           }
3132         }
3133       }
3134     }
3135   }
3136   return nullptr;
3137 }
3138 
3139 // method_reference() is called passing it the ReferenceName that might be
3140 // a reference it to an Objective-C method call.  If so then it allocates and
3141 // assembles a method call string with the values last seen and saved in
3142 // the DisassembleInfo's class_name and selector_name fields.  This is saved
3143 // into the method field of the info and any previous string is free'ed.
3144 // Then the class_name field in the info is set to nullptr.  The method call
3145 // string is set into ReferenceName and ReferenceType is set to
3146 // LLVMDisassembler_ReferenceType_Out_Objc_Message.  If this not a method call
3147 // then both ReferenceType and ReferenceName are left unchanged.
method_reference(struct DisassembleInfo * info,uint64_t * ReferenceType,const char ** ReferenceName)3148 static void method_reference(struct DisassembleInfo *info,
3149                              uint64_t *ReferenceType,
3150                              const char **ReferenceName) {
3151   unsigned int Arch = info->O->getArch();
3152   if (*ReferenceName != nullptr) {
3153     if (strcmp(*ReferenceName, "_objc_msgSend") == 0) {
3154       if (info->selector_name != nullptr) {
3155         if (info->class_name != nullptr) {
3156           info->method = std::make_unique<char[]>(
3157               5 + strlen(info->class_name) + strlen(info->selector_name));
3158           char *method = info->method.get();
3159           if (method != nullptr) {
3160             strcpy(method, "+[");
3161             strcat(method, info->class_name);
3162             strcat(method, " ");
3163             strcat(method, info->selector_name);
3164             strcat(method, "]");
3165             *ReferenceName = method;
3166             *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3167           }
3168         } else {
3169           info->method =
3170               std::make_unique<char[]>(9 + strlen(info->selector_name));
3171           char *method = info->method.get();
3172           if (method != nullptr) {
3173             if (Arch == Triple::x86_64)
3174               strcpy(method, "-[%rdi ");
3175             else if (Arch == Triple::aarch64)
3176               strcpy(method, "-[x0 ");
3177             else
3178               strcpy(method, "-[r? ");
3179             strcat(method, info->selector_name);
3180             strcat(method, "]");
3181             *ReferenceName = method;
3182             *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3183           }
3184         }
3185         info->class_name = nullptr;
3186       }
3187     } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) {
3188       if (info->selector_name != nullptr) {
3189         info->method =
3190             std::make_unique<char[]>(17 + strlen(info->selector_name));
3191         char *method = info->method.get();
3192         if (method != nullptr) {
3193           if (Arch == Triple::x86_64)
3194             strcpy(method, "-[[%rdi super] ");
3195           else if (Arch == Triple::aarch64)
3196             strcpy(method, "-[[x0 super] ");
3197           else
3198             strcpy(method, "-[[r? super] ");
3199           strcat(method, info->selector_name);
3200           strcat(method, "]");
3201           *ReferenceName = method;
3202           *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3203         }
3204         info->class_name = nullptr;
3205       }
3206     }
3207   }
3208 }
3209 
3210 // GuessPointerPointer() is passed the address of what might be a pointer to
3211 // a reference to an Objective-C class, selector, message ref or cfstring.
3212 // If so the value of the pointer is returned and one of the booleans are set
3213 // 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)3214 static uint64_t GuessPointerPointer(uint64_t ReferenceValue,
3215                                     struct DisassembleInfo *info,
3216                                     bool &classref, bool &selref, bool &msgref,
3217                                     bool &cfstring) {
3218   classref = false;
3219   selref = false;
3220   msgref = false;
3221   cfstring = false;
3222   for (const auto &Load : info->O->load_commands()) {
3223     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3224       MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3225       for (unsigned J = 0; J < Seg.nsects; ++J) {
3226         MachO::section_64 Sec = info->O->getSection64(Load, J);
3227         if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 ||
3228              strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
3229              strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 ||
3230              strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 ||
3231              strncmp(Sec.sectname, "__cfstring", 16) == 0) &&
3232             ReferenceValue >= Sec.addr &&
3233             ReferenceValue < Sec.addr + Sec.size) {
3234           uint64_t sect_offset = ReferenceValue - Sec.addr;
3235           uint64_t object_offset = Sec.offset + sect_offset;
3236           StringRef MachOContents = info->O->getData();
3237           uint64_t object_size = MachOContents.size();
3238           const char *object_addr = (const char *)MachOContents.data();
3239           if (object_offset < object_size) {
3240             uint64_t pointer_value;
3241             memcpy(&pointer_value, object_addr + object_offset,
3242                    sizeof(uint64_t));
3243             if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
3244               sys::swapByteOrder(pointer_value);
3245             if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0)
3246               selref = true;
3247             else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
3248                      strncmp(Sec.sectname, "__objc_superrefs", 16) == 0)
3249               classref = true;
3250             else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 &&
3251                      ReferenceValue + 8 < Sec.addr + Sec.size) {
3252               msgref = true;
3253               memcpy(&pointer_value, object_addr + object_offset + 8,
3254                      sizeof(uint64_t));
3255               if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
3256                 sys::swapByteOrder(pointer_value);
3257             } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0)
3258               cfstring = true;
3259             return pointer_value;
3260           } else {
3261             return 0;
3262           }
3263         }
3264       }
3265     }
3266     // TODO: Look for LC_SEGMENT for 32-bit Mach-O files.
3267   }
3268   return 0;
3269 }
3270 
3271 // get_pointer_64 returns a pointer to the bytes in the object file at the
3272 // Address from a section in the Mach-O file.  And indirectly returns the
3273 // offset into the section, number of bytes left in the section past the offset
3274 // and which section is was being referenced.  If the Address is not in a
3275 // section nullptr is returned.
get_pointer_64(uint64_t Address,uint32_t & offset,uint32_t & left,SectionRef & S,DisassembleInfo * info,bool objc_only=false)3276 static const char *get_pointer_64(uint64_t Address, uint32_t &offset,
3277                                   uint32_t &left, SectionRef &S,
3278                                   DisassembleInfo *info,
3279                                   bool objc_only = false) {
3280   offset = 0;
3281   left = 0;
3282   S = SectionRef();
3283   for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) {
3284     uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress();
3285     uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize();
3286     if (SectSize == 0)
3287       continue;
3288     if (objc_only) {
3289       StringRef SectName;
3290       Expected<StringRef> SecNameOrErr =
3291           ((*(info->Sections))[SectIdx]).getName();
3292       if (SecNameOrErr)
3293         SectName = *SecNameOrErr;
3294       else
3295         consumeError(SecNameOrErr.takeError());
3296 
3297       DataRefImpl Ref = ((*(info->Sections))[SectIdx]).getRawDataRefImpl();
3298       StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
3299       if (SegName != "__OBJC" && SectName != "__cstring")
3300         continue;
3301     }
3302     if (Address >= SectAddress && Address < SectAddress + SectSize) {
3303       S = (*(info->Sections))[SectIdx];
3304       offset = Address - SectAddress;
3305       left = SectSize - offset;
3306       StringRef SectContents = unwrapOrError(
3307           ((*(info->Sections))[SectIdx]).getContents(), info->O->getFileName());
3308       return SectContents.data() + offset;
3309     }
3310   }
3311   return nullptr;
3312 }
3313 
get_pointer_32(uint32_t Address,uint32_t & offset,uint32_t & left,SectionRef & S,DisassembleInfo * info,bool objc_only=false)3314 static const char *get_pointer_32(uint32_t Address, uint32_t &offset,
3315                                   uint32_t &left, SectionRef &S,
3316                                   DisassembleInfo *info,
3317                                   bool objc_only = false) {
3318   return get_pointer_64(Address, offset, left, S, info, objc_only);
3319 }
3320 
3321 // get_symbol_64() returns the name of a symbol (or nullptr) and the address of
3322 // the symbol indirectly through n_value. Based on the relocation information
3323 // for the specified section offset in the specified section reference.
3324 // If no relocation information is found and a non-zero ReferenceValue for the
3325 // 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)3326 static const char *get_symbol_64(uint32_t sect_offset, SectionRef S,
3327                                  DisassembleInfo *info, uint64_t &n_value,
3328                                  uint64_t ReferenceValue = 0) {
3329   n_value = 0;
3330   if (!info->verbose)
3331     return nullptr;
3332 
3333   // See if there is an external relocation entry at the sect_offset.
3334   bool reloc_found = false;
3335   DataRefImpl Rel;
3336   MachO::any_relocation_info RE;
3337   bool isExtern = false;
3338   SymbolRef Symbol;
3339   for (const RelocationRef &Reloc : S.relocations()) {
3340     uint64_t RelocOffset = Reloc.getOffset();
3341     if (RelocOffset == sect_offset) {
3342       Rel = Reloc.getRawDataRefImpl();
3343       RE = info->O->getRelocation(Rel);
3344       if (info->O->isRelocationScattered(RE))
3345         continue;
3346       isExtern = info->O->getPlainRelocationExternal(RE);
3347       if (isExtern) {
3348         symbol_iterator RelocSym = Reloc.getSymbol();
3349         Symbol = *RelocSym;
3350       }
3351       reloc_found = true;
3352       break;
3353     }
3354   }
3355   // If there is an external relocation entry for a symbol in this section
3356   // at this section_offset then use that symbol's value for the n_value
3357   // and return its name.
3358   const char *SymbolName = nullptr;
3359   if (reloc_found && isExtern) {
3360     n_value = cantFail(Symbol.getValue());
3361     StringRef Name = unwrapOrError(Symbol.getName(), info->O->getFileName());
3362     if (!Name.empty()) {
3363       SymbolName = Name.data();
3364       return SymbolName;
3365     }
3366   }
3367 
3368   // TODO: For fully linked images, look through the external relocation
3369   // entries off the dynamic symtab command. For these the r_offset is from the
3370   // start of the first writeable segment in the Mach-O file.  So the offset
3371   // to this section from that segment is passed to this routine by the caller,
3372   // as the database_offset. Which is the difference of the section's starting
3373   // address and the first writable segment.
3374   //
3375   // NOTE: need add passing the database_offset to this routine.
3376 
3377   // We did not find an external relocation entry so look up the ReferenceValue
3378   // as an address of a symbol and if found return that symbol's name.
3379   SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
3380 
3381   return SymbolName;
3382 }
3383 
get_symbol_32(uint32_t sect_offset,SectionRef S,DisassembleInfo * info,uint32_t ReferenceValue)3384 static const char *get_symbol_32(uint32_t sect_offset, SectionRef S,
3385                                  DisassembleInfo *info,
3386                                  uint32_t ReferenceValue) {
3387   uint64_t n_value64;
3388   return get_symbol_64(sect_offset, S, info, n_value64, ReferenceValue);
3389 }
3390 
3391 namespace {
3392 
3393 // These are structs in the Objective-C meta data and read to produce the
3394 // comments for disassembly.  While these are part of the ABI they are no
3395 // public defintions.  So the are here not in include/llvm/BinaryFormat/MachO.h
3396 // .
3397 
3398 // The cfstring object in a 64-bit Mach-O file.
3399 struct cfstring64_t {
3400   uint64_t isa;        // class64_t * (64-bit pointer)
3401   uint64_t flags;      // flag bits
3402   uint64_t characters; // char * (64-bit pointer)
3403   uint64_t length;     // number of non-NULL characters in above
3404 };
3405 
3406 // The class object in a 64-bit Mach-O file.
3407 struct class64_t {
3408   uint64_t isa;        // class64_t * (64-bit pointer)
3409   uint64_t superclass; // class64_t * (64-bit pointer)
3410   uint64_t cache;      // Cache (64-bit pointer)
3411   uint64_t vtable;     // IMP * (64-bit pointer)
3412   uint64_t data;       // class_ro64_t * (64-bit pointer)
3413 };
3414 
3415 struct class32_t {
3416   uint32_t isa;        /* class32_t * (32-bit pointer) */
3417   uint32_t superclass; /* class32_t * (32-bit pointer) */
3418   uint32_t cache;      /* Cache (32-bit pointer) */
3419   uint32_t vtable;     /* IMP * (32-bit pointer) */
3420   uint32_t data;       /* class_ro32_t * (32-bit pointer) */
3421 };
3422 
3423 struct class_ro64_t {
3424   uint32_t flags;
3425   uint32_t instanceStart;
3426   uint32_t instanceSize;
3427   uint32_t reserved;
3428   uint64_t ivarLayout;     // const uint8_t * (64-bit pointer)
3429   uint64_t name;           // const char * (64-bit pointer)
3430   uint64_t baseMethods;    // const method_list_t * (64-bit pointer)
3431   uint64_t baseProtocols;  // const protocol_list_t * (64-bit pointer)
3432   uint64_t ivars;          // const ivar_list_t * (64-bit pointer)
3433   uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer)
3434   uint64_t baseProperties; // const struct objc_property_list (64-bit pointer)
3435 };
3436 
3437 struct class_ro32_t {
3438   uint32_t flags;
3439   uint32_t instanceStart;
3440   uint32_t instanceSize;
3441   uint32_t ivarLayout;     /* const uint8_t * (32-bit pointer) */
3442   uint32_t name;           /* const char * (32-bit pointer) */
3443   uint32_t baseMethods;    /* const method_list_t * (32-bit pointer) */
3444   uint32_t baseProtocols;  /* const protocol_list_t * (32-bit pointer) */
3445   uint32_t ivars;          /* const ivar_list_t * (32-bit pointer) */
3446   uint32_t weakIvarLayout; /* const uint8_t * (32-bit pointer) */
3447   uint32_t baseProperties; /* const struct objc_property_list *
3448                                                    (32-bit pointer) */
3449 };
3450 
3451 /* Values for class_ro{64,32}_t->flags */
3452 #define RO_META (1 << 0)
3453 #define RO_ROOT (1 << 1)
3454 #define RO_HAS_CXX_STRUCTORS (1 << 2)
3455 
3456 struct method_list64_t {
3457   uint32_t entsize;
3458   uint32_t count;
3459   /* struct method64_t first;  These structures follow inline */
3460 };
3461 
3462 struct method_list32_t {
3463   uint32_t entsize;
3464   uint32_t count;
3465   /* struct method32_t first;  These structures follow inline */
3466 };
3467 
3468 struct method64_t {
3469   uint64_t name;  /* SEL (64-bit pointer) */
3470   uint64_t types; /* const char * (64-bit pointer) */
3471   uint64_t imp;   /* IMP (64-bit pointer) */
3472 };
3473 
3474 struct method32_t {
3475   uint32_t name;  /* SEL (32-bit pointer) */
3476   uint32_t types; /* const char * (32-bit pointer) */
3477   uint32_t imp;   /* IMP (32-bit pointer) */
3478 };
3479 
3480 struct protocol_list64_t {
3481   uint64_t count; /* uintptr_t (a 64-bit value) */
3482   /* struct protocol64_t * list[0];  These pointers follow inline */
3483 };
3484 
3485 struct protocol_list32_t {
3486   uint32_t count; /* uintptr_t (a 32-bit value) */
3487   /* struct protocol32_t * list[0];  These pointers follow inline */
3488 };
3489 
3490 struct protocol64_t {
3491   uint64_t isa;                     /* id * (64-bit pointer) */
3492   uint64_t name;                    /* const char * (64-bit pointer) */
3493   uint64_t protocols;               /* struct protocol_list64_t *
3494                                                     (64-bit pointer) */
3495   uint64_t instanceMethods;         /* method_list_t * (64-bit pointer) */
3496   uint64_t classMethods;            /* method_list_t * (64-bit pointer) */
3497   uint64_t optionalInstanceMethods; /* method_list_t * (64-bit pointer) */
3498   uint64_t optionalClassMethods;    /* method_list_t * (64-bit pointer) */
3499   uint64_t instanceProperties;      /* struct objc_property_list *
3500                                                        (64-bit pointer) */
3501 };
3502 
3503 struct protocol32_t {
3504   uint32_t isa;                     /* id * (32-bit pointer) */
3505   uint32_t name;                    /* const char * (32-bit pointer) */
3506   uint32_t protocols;               /* struct protocol_list_t *
3507                                                     (32-bit pointer) */
3508   uint32_t instanceMethods;         /* method_list_t * (32-bit pointer) */
3509   uint32_t classMethods;            /* method_list_t * (32-bit pointer) */
3510   uint32_t optionalInstanceMethods; /* method_list_t * (32-bit pointer) */
3511   uint32_t optionalClassMethods;    /* method_list_t * (32-bit pointer) */
3512   uint32_t instanceProperties;      /* struct objc_property_list *
3513                                                        (32-bit pointer) */
3514 };
3515 
3516 struct ivar_list64_t {
3517   uint32_t entsize;
3518   uint32_t count;
3519   /* struct ivar64_t first;  These structures follow inline */
3520 };
3521 
3522 struct ivar_list32_t {
3523   uint32_t entsize;
3524   uint32_t count;
3525   /* struct ivar32_t first;  These structures follow inline */
3526 };
3527 
3528 struct ivar64_t {
3529   uint64_t offset; /* uintptr_t * (64-bit pointer) */
3530   uint64_t name;   /* const char * (64-bit pointer) */
3531   uint64_t type;   /* const char * (64-bit pointer) */
3532   uint32_t alignment;
3533   uint32_t size;
3534 };
3535 
3536 struct ivar32_t {
3537   uint32_t offset; /* uintptr_t * (32-bit pointer) */
3538   uint32_t name;   /* const char * (32-bit pointer) */
3539   uint32_t type;   /* const char * (32-bit pointer) */
3540   uint32_t alignment;
3541   uint32_t size;
3542 };
3543 
3544 struct objc_property_list64 {
3545   uint32_t entsize;
3546   uint32_t count;
3547   /* struct objc_property64 first;  These structures follow inline */
3548 };
3549 
3550 struct objc_property_list32 {
3551   uint32_t entsize;
3552   uint32_t count;
3553   /* struct objc_property32 first;  These structures follow inline */
3554 };
3555 
3556 struct objc_property64 {
3557   uint64_t name;       /* const char * (64-bit pointer) */
3558   uint64_t attributes; /* const char * (64-bit pointer) */
3559 };
3560 
3561 struct objc_property32 {
3562   uint32_t name;       /* const char * (32-bit pointer) */
3563   uint32_t attributes; /* const char * (32-bit pointer) */
3564 };
3565 
3566 struct category64_t {
3567   uint64_t name;               /* const char * (64-bit pointer) */
3568   uint64_t cls;                /* struct class_t * (64-bit pointer) */
3569   uint64_t instanceMethods;    /* struct method_list_t * (64-bit pointer) */
3570   uint64_t classMethods;       /* struct method_list_t * (64-bit pointer) */
3571   uint64_t protocols;          /* struct protocol_list_t * (64-bit pointer) */
3572   uint64_t instanceProperties; /* struct objc_property_list *
3573                                   (64-bit pointer) */
3574 };
3575 
3576 struct category32_t {
3577   uint32_t name;               /* const char * (32-bit pointer) */
3578   uint32_t cls;                /* struct class_t * (32-bit pointer) */
3579   uint32_t instanceMethods;    /* struct method_list_t * (32-bit pointer) */
3580   uint32_t classMethods;       /* struct method_list_t * (32-bit pointer) */
3581   uint32_t protocols;          /* struct protocol_list_t * (32-bit pointer) */
3582   uint32_t instanceProperties; /* struct objc_property_list *
3583                                   (32-bit pointer) */
3584 };
3585 
3586 struct objc_image_info64 {
3587   uint32_t version;
3588   uint32_t flags;
3589 };
3590 struct objc_image_info32 {
3591   uint32_t version;
3592   uint32_t flags;
3593 };
3594 struct imageInfo_t {
3595   uint32_t version;
3596   uint32_t flags;
3597 };
3598 /* masks for objc_image_info.flags */
3599 #define OBJC_IMAGE_IS_REPLACEMENT (1 << 0)
3600 #define OBJC_IMAGE_SUPPORTS_GC (1 << 1)
3601 #define OBJC_IMAGE_IS_SIMULATED (1 << 5)
3602 #define OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES (1 << 6)
3603 
3604 struct message_ref64 {
3605   uint64_t imp; /* IMP (64-bit pointer) */
3606   uint64_t sel; /* SEL (64-bit pointer) */
3607 };
3608 
3609 struct message_ref32 {
3610   uint32_t imp; /* IMP (32-bit pointer) */
3611   uint32_t sel; /* SEL (32-bit pointer) */
3612 };
3613 
3614 // Objective-C 1 (32-bit only) meta data structs.
3615 
3616 struct objc_module_t {
3617   uint32_t version;
3618   uint32_t size;
3619   uint32_t name;   /* char * (32-bit pointer) */
3620   uint32_t symtab; /* struct objc_symtab * (32-bit pointer) */
3621 };
3622 
3623 struct objc_symtab_t {
3624   uint32_t sel_ref_cnt;
3625   uint32_t refs; /* SEL * (32-bit pointer) */
3626   uint16_t cls_def_cnt;
3627   uint16_t cat_def_cnt;
3628   // uint32_t defs[1];        /* void * (32-bit pointer) variable size */
3629 };
3630 
3631 struct objc_class_t {
3632   uint32_t isa;         /* struct objc_class * (32-bit pointer) */
3633   uint32_t super_class; /* struct objc_class * (32-bit pointer) */
3634   uint32_t name;        /* const char * (32-bit pointer) */
3635   int32_t version;
3636   int32_t info;
3637   int32_t instance_size;
3638   uint32_t ivars;       /* struct objc_ivar_list * (32-bit pointer) */
3639   uint32_t methodLists; /* struct objc_method_list ** (32-bit pointer) */
3640   uint32_t cache;       /* struct objc_cache * (32-bit pointer) */
3641   uint32_t protocols;   /* struct objc_protocol_list * (32-bit pointer) */
3642 };
3643 
3644 #define CLS_GETINFO(cls, infomask) ((cls)->info & (infomask))
3645 // class is not a metaclass
3646 #define CLS_CLASS 0x1
3647 // class is a metaclass
3648 #define CLS_META 0x2
3649 
3650 struct objc_category_t {
3651   uint32_t category_name;    /* char * (32-bit pointer) */
3652   uint32_t class_name;       /* char * (32-bit pointer) */
3653   uint32_t instance_methods; /* struct objc_method_list * (32-bit pointer) */
3654   uint32_t class_methods;    /* struct objc_method_list * (32-bit pointer) */
3655   uint32_t protocols;        /* struct objc_protocol_list * (32-bit ptr) */
3656 };
3657 
3658 struct objc_ivar_t {
3659   uint32_t ivar_name; /* char * (32-bit pointer) */
3660   uint32_t ivar_type; /* char * (32-bit pointer) */
3661   int32_t ivar_offset;
3662 };
3663 
3664 struct objc_ivar_list_t {
3665   int32_t ivar_count;
3666   // struct objc_ivar_t ivar_list[1];          /* variable length structure */
3667 };
3668 
3669 struct objc_method_list_t {
3670   uint32_t obsolete; /* struct objc_method_list * (32-bit pointer) */
3671   int32_t method_count;
3672   // struct objc_method_t method_list[1];      /* variable length structure */
3673 };
3674 
3675 struct objc_method_t {
3676   uint32_t method_name;  /* SEL, aka struct objc_selector * (32-bit pointer) */
3677   uint32_t method_types; /* char * (32-bit pointer) */
3678   uint32_t method_imp;   /* IMP, aka function pointer, (*IMP)(id, SEL, ...)
3679                             (32-bit pointer) */
3680 };
3681 
3682 struct objc_protocol_list_t {
3683   uint32_t next; /* struct objc_protocol_list * (32-bit pointer) */
3684   int32_t count;
3685   // uint32_t list[1];   /* Protocol *, aka struct objc_protocol_t *
3686   //                        (32-bit pointer) */
3687 };
3688 
3689 struct objc_protocol_t {
3690   uint32_t isa;              /* struct objc_class * (32-bit pointer) */
3691   uint32_t protocol_name;    /* char * (32-bit pointer) */
3692   uint32_t protocol_list;    /* struct objc_protocol_list * (32-bit pointer) */
3693   uint32_t instance_methods; /* struct objc_method_description_list *
3694                                 (32-bit pointer) */
3695   uint32_t class_methods;    /* struct objc_method_description_list *
3696                                 (32-bit pointer) */
3697 };
3698 
3699 struct objc_method_description_list_t {
3700   int32_t count;
3701   // struct objc_method_description_t list[1];
3702 };
3703 
3704 struct objc_method_description_t {
3705   uint32_t name;  /* SEL, aka struct objc_selector * (32-bit pointer) */
3706   uint32_t types; /* char * (32-bit pointer) */
3707 };
3708 
swapStruct(struct cfstring64_t & cfs)3709 inline void swapStruct(struct cfstring64_t &cfs) {
3710   sys::swapByteOrder(cfs.isa);
3711   sys::swapByteOrder(cfs.flags);
3712   sys::swapByteOrder(cfs.characters);
3713   sys::swapByteOrder(cfs.length);
3714 }
3715 
swapStruct(struct class64_t & c)3716 inline void swapStruct(struct class64_t &c) {
3717   sys::swapByteOrder(c.isa);
3718   sys::swapByteOrder(c.superclass);
3719   sys::swapByteOrder(c.cache);
3720   sys::swapByteOrder(c.vtable);
3721   sys::swapByteOrder(c.data);
3722 }
3723 
swapStruct(struct class32_t & c)3724 inline void swapStruct(struct class32_t &c) {
3725   sys::swapByteOrder(c.isa);
3726   sys::swapByteOrder(c.superclass);
3727   sys::swapByteOrder(c.cache);
3728   sys::swapByteOrder(c.vtable);
3729   sys::swapByteOrder(c.data);
3730 }
3731 
swapStruct(struct class_ro64_t & cro)3732 inline void swapStruct(struct class_ro64_t &cro) {
3733   sys::swapByteOrder(cro.flags);
3734   sys::swapByteOrder(cro.instanceStart);
3735   sys::swapByteOrder(cro.instanceSize);
3736   sys::swapByteOrder(cro.reserved);
3737   sys::swapByteOrder(cro.ivarLayout);
3738   sys::swapByteOrder(cro.name);
3739   sys::swapByteOrder(cro.baseMethods);
3740   sys::swapByteOrder(cro.baseProtocols);
3741   sys::swapByteOrder(cro.ivars);
3742   sys::swapByteOrder(cro.weakIvarLayout);
3743   sys::swapByteOrder(cro.baseProperties);
3744 }
3745 
swapStruct(struct class_ro32_t & cro)3746 inline void swapStruct(struct class_ro32_t &cro) {
3747   sys::swapByteOrder(cro.flags);
3748   sys::swapByteOrder(cro.instanceStart);
3749   sys::swapByteOrder(cro.instanceSize);
3750   sys::swapByteOrder(cro.ivarLayout);
3751   sys::swapByteOrder(cro.name);
3752   sys::swapByteOrder(cro.baseMethods);
3753   sys::swapByteOrder(cro.baseProtocols);
3754   sys::swapByteOrder(cro.ivars);
3755   sys::swapByteOrder(cro.weakIvarLayout);
3756   sys::swapByteOrder(cro.baseProperties);
3757 }
3758 
swapStruct(struct method_list64_t & ml)3759 inline void swapStruct(struct method_list64_t &ml) {
3760   sys::swapByteOrder(ml.entsize);
3761   sys::swapByteOrder(ml.count);
3762 }
3763 
swapStruct(struct method_list32_t & ml)3764 inline void swapStruct(struct method_list32_t &ml) {
3765   sys::swapByteOrder(ml.entsize);
3766   sys::swapByteOrder(ml.count);
3767 }
3768 
swapStruct(struct method64_t & m)3769 inline void swapStruct(struct method64_t &m) {
3770   sys::swapByteOrder(m.name);
3771   sys::swapByteOrder(m.types);
3772   sys::swapByteOrder(m.imp);
3773 }
3774 
swapStruct(struct method32_t & m)3775 inline void swapStruct(struct method32_t &m) {
3776   sys::swapByteOrder(m.name);
3777   sys::swapByteOrder(m.types);
3778   sys::swapByteOrder(m.imp);
3779 }
3780 
swapStruct(struct protocol_list64_t & pl)3781 inline void swapStruct(struct protocol_list64_t &pl) {
3782   sys::swapByteOrder(pl.count);
3783 }
3784 
swapStruct(struct protocol_list32_t & pl)3785 inline void swapStruct(struct protocol_list32_t &pl) {
3786   sys::swapByteOrder(pl.count);
3787 }
3788 
swapStruct(struct protocol64_t & p)3789 inline void swapStruct(struct protocol64_t &p) {
3790   sys::swapByteOrder(p.isa);
3791   sys::swapByteOrder(p.name);
3792   sys::swapByteOrder(p.protocols);
3793   sys::swapByteOrder(p.instanceMethods);
3794   sys::swapByteOrder(p.classMethods);
3795   sys::swapByteOrder(p.optionalInstanceMethods);
3796   sys::swapByteOrder(p.optionalClassMethods);
3797   sys::swapByteOrder(p.instanceProperties);
3798 }
3799 
swapStruct(struct protocol32_t & p)3800 inline void swapStruct(struct protocol32_t &p) {
3801   sys::swapByteOrder(p.isa);
3802   sys::swapByteOrder(p.name);
3803   sys::swapByteOrder(p.protocols);
3804   sys::swapByteOrder(p.instanceMethods);
3805   sys::swapByteOrder(p.classMethods);
3806   sys::swapByteOrder(p.optionalInstanceMethods);
3807   sys::swapByteOrder(p.optionalClassMethods);
3808   sys::swapByteOrder(p.instanceProperties);
3809 }
3810 
swapStruct(struct ivar_list64_t & il)3811 inline void swapStruct(struct ivar_list64_t &il) {
3812   sys::swapByteOrder(il.entsize);
3813   sys::swapByteOrder(il.count);
3814 }
3815 
swapStruct(struct ivar_list32_t & il)3816 inline void swapStruct(struct ivar_list32_t &il) {
3817   sys::swapByteOrder(il.entsize);
3818   sys::swapByteOrder(il.count);
3819 }
3820 
swapStruct(struct ivar64_t & i)3821 inline void swapStruct(struct ivar64_t &i) {
3822   sys::swapByteOrder(i.offset);
3823   sys::swapByteOrder(i.name);
3824   sys::swapByteOrder(i.type);
3825   sys::swapByteOrder(i.alignment);
3826   sys::swapByteOrder(i.size);
3827 }
3828 
swapStruct(struct ivar32_t & i)3829 inline void swapStruct(struct ivar32_t &i) {
3830   sys::swapByteOrder(i.offset);
3831   sys::swapByteOrder(i.name);
3832   sys::swapByteOrder(i.type);
3833   sys::swapByteOrder(i.alignment);
3834   sys::swapByteOrder(i.size);
3835 }
3836 
swapStruct(struct objc_property_list64 & pl)3837 inline void swapStruct(struct objc_property_list64 &pl) {
3838   sys::swapByteOrder(pl.entsize);
3839   sys::swapByteOrder(pl.count);
3840 }
3841 
swapStruct(struct objc_property_list32 & pl)3842 inline void swapStruct(struct objc_property_list32 &pl) {
3843   sys::swapByteOrder(pl.entsize);
3844   sys::swapByteOrder(pl.count);
3845 }
3846 
swapStruct(struct objc_property64 & op)3847 inline void swapStruct(struct objc_property64 &op) {
3848   sys::swapByteOrder(op.name);
3849   sys::swapByteOrder(op.attributes);
3850 }
3851 
swapStruct(struct objc_property32 & op)3852 inline void swapStruct(struct objc_property32 &op) {
3853   sys::swapByteOrder(op.name);
3854   sys::swapByteOrder(op.attributes);
3855 }
3856 
swapStruct(struct category64_t & c)3857 inline void swapStruct(struct category64_t &c) {
3858   sys::swapByteOrder(c.name);
3859   sys::swapByteOrder(c.cls);
3860   sys::swapByteOrder(c.instanceMethods);
3861   sys::swapByteOrder(c.classMethods);
3862   sys::swapByteOrder(c.protocols);
3863   sys::swapByteOrder(c.instanceProperties);
3864 }
3865 
swapStruct(struct category32_t & c)3866 inline void swapStruct(struct category32_t &c) {
3867   sys::swapByteOrder(c.name);
3868   sys::swapByteOrder(c.cls);
3869   sys::swapByteOrder(c.instanceMethods);
3870   sys::swapByteOrder(c.classMethods);
3871   sys::swapByteOrder(c.protocols);
3872   sys::swapByteOrder(c.instanceProperties);
3873 }
3874 
swapStruct(struct objc_image_info64 & o)3875 inline void swapStruct(struct objc_image_info64 &o) {
3876   sys::swapByteOrder(o.version);
3877   sys::swapByteOrder(o.flags);
3878 }
3879 
swapStruct(struct objc_image_info32 & o)3880 inline void swapStruct(struct objc_image_info32 &o) {
3881   sys::swapByteOrder(o.version);
3882   sys::swapByteOrder(o.flags);
3883 }
3884 
swapStruct(struct imageInfo_t & o)3885 inline void swapStruct(struct imageInfo_t &o) {
3886   sys::swapByteOrder(o.version);
3887   sys::swapByteOrder(o.flags);
3888 }
3889 
swapStruct(struct message_ref64 & mr)3890 inline void swapStruct(struct message_ref64 &mr) {
3891   sys::swapByteOrder(mr.imp);
3892   sys::swapByteOrder(mr.sel);
3893 }
3894 
swapStruct(struct message_ref32 & mr)3895 inline void swapStruct(struct message_ref32 &mr) {
3896   sys::swapByteOrder(mr.imp);
3897   sys::swapByteOrder(mr.sel);
3898 }
3899 
swapStruct(struct objc_module_t & module)3900 inline void swapStruct(struct objc_module_t &module) {
3901   sys::swapByteOrder(module.version);
3902   sys::swapByteOrder(module.size);
3903   sys::swapByteOrder(module.name);
3904   sys::swapByteOrder(module.symtab);
3905 }
3906 
swapStruct(struct objc_symtab_t & symtab)3907 inline void swapStruct(struct objc_symtab_t &symtab) {
3908   sys::swapByteOrder(symtab.sel_ref_cnt);
3909   sys::swapByteOrder(symtab.refs);
3910   sys::swapByteOrder(symtab.cls_def_cnt);
3911   sys::swapByteOrder(symtab.cat_def_cnt);
3912 }
3913 
swapStruct(struct objc_class_t & objc_class)3914 inline void swapStruct(struct objc_class_t &objc_class) {
3915   sys::swapByteOrder(objc_class.isa);
3916   sys::swapByteOrder(objc_class.super_class);
3917   sys::swapByteOrder(objc_class.name);
3918   sys::swapByteOrder(objc_class.version);
3919   sys::swapByteOrder(objc_class.info);
3920   sys::swapByteOrder(objc_class.instance_size);
3921   sys::swapByteOrder(objc_class.ivars);
3922   sys::swapByteOrder(objc_class.methodLists);
3923   sys::swapByteOrder(objc_class.cache);
3924   sys::swapByteOrder(objc_class.protocols);
3925 }
3926 
swapStruct(struct objc_category_t & objc_category)3927 inline void swapStruct(struct objc_category_t &objc_category) {
3928   sys::swapByteOrder(objc_category.category_name);
3929   sys::swapByteOrder(objc_category.class_name);
3930   sys::swapByteOrder(objc_category.instance_methods);
3931   sys::swapByteOrder(objc_category.class_methods);
3932   sys::swapByteOrder(objc_category.protocols);
3933 }
3934 
swapStruct(struct objc_ivar_list_t & objc_ivar_list)3935 inline void swapStruct(struct objc_ivar_list_t &objc_ivar_list) {
3936   sys::swapByteOrder(objc_ivar_list.ivar_count);
3937 }
3938 
swapStruct(struct objc_ivar_t & objc_ivar)3939 inline void swapStruct(struct objc_ivar_t &objc_ivar) {
3940   sys::swapByteOrder(objc_ivar.ivar_name);
3941   sys::swapByteOrder(objc_ivar.ivar_type);
3942   sys::swapByteOrder(objc_ivar.ivar_offset);
3943 }
3944 
swapStruct(struct objc_method_list_t & method_list)3945 inline void swapStruct(struct objc_method_list_t &method_list) {
3946   sys::swapByteOrder(method_list.obsolete);
3947   sys::swapByteOrder(method_list.method_count);
3948 }
3949 
swapStruct(struct objc_method_t & method)3950 inline void swapStruct(struct objc_method_t &method) {
3951   sys::swapByteOrder(method.method_name);
3952   sys::swapByteOrder(method.method_types);
3953   sys::swapByteOrder(method.method_imp);
3954 }
3955 
swapStruct(struct objc_protocol_list_t & protocol_list)3956 inline void swapStruct(struct objc_protocol_list_t &protocol_list) {
3957   sys::swapByteOrder(protocol_list.next);
3958   sys::swapByteOrder(protocol_list.count);
3959 }
3960 
swapStruct(struct objc_protocol_t & protocol)3961 inline void swapStruct(struct objc_protocol_t &protocol) {
3962   sys::swapByteOrder(protocol.isa);
3963   sys::swapByteOrder(protocol.protocol_name);
3964   sys::swapByteOrder(protocol.protocol_list);
3965   sys::swapByteOrder(protocol.instance_methods);
3966   sys::swapByteOrder(protocol.class_methods);
3967 }
3968 
swapStruct(struct objc_method_description_list_t & mdl)3969 inline void swapStruct(struct objc_method_description_list_t &mdl) {
3970   sys::swapByteOrder(mdl.count);
3971 }
3972 
swapStruct(struct objc_method_description_t & md)3973 inline void swapStruct(struct objc_method_description_t &md) {
3974   sys::swapByteOrder(md.name);
3975   sys::swapByteOrder(md.types);
3976 }
3977 
3978 } // namespace
3979 
3980 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
3981                                                  struct DisassembleInfo *info);
3982 
3983 // get_objc2_64bit_class_name() is used for disassembly and is passed a pointer
3984 // to an Objective-C class and returns the class name.  It is also passed the
3985 // address of the pointer, so when the pointer is zero as it can be in an .o
3986 // file, that is used to look for an external relocation entry with a symbol
3987 // name.
get_objc2_64bit_class_name(uint64_t pointer_value,uint64_t ReferenceValue,struct DisassembleInfo * info)3988 static const char *get_objc2_64bit_class_name(uint64_t pointer_value,
3989                                               uint64_t ReferenceValue,
3990                                               struct DisassembleInfo *info) {
3991   const char *r;
3992   uint32_t offset, left;
3993   SectionRef S;
3994 
3995   // The pointer_value can be 0 in an object file and have a relocation
3996   // entry for the class symbol at the ReferenceValue (the address of the
3997   // pointer).
3998   if (pointer_value == 0) {
3999     r = get_pointer_64(ReferenceValue, offset, left, S, info);
4000     if (r == nullptr || left < sizeof(uint64_t))
4001       return nullptr;
4002     uint64_t n_value;
4003     const char *symbol_name = get_symbol_64(offset, S, info, n_value);
4004     if (symbol_name == nullptr)
4005       return nullptr;
4006     const char *class_name = strrchr(symbol_name, '$');
4007     if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0')
4008       return class_name + 2;
4009     else
4010       return nullptr;
4011   }
4012 
4013   // The case were the pointer_value is non-zero and points to a class defined
4014   // in this Mach-O file.
4015   r = get_pointer_64(pointer_value, offset, left, S, info);
4016   if (r == nullptr || left < sizeof(struct class64_t))
4017     return nullptr;
4018   struct class64_t c;
4019   memcpy(&c, r, sizeof(struct class64_t));
4020   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4021     swapStruct(c);
4022   if (c.data == 0)
4023     return nullptr;
4024   r = get_pointer_64(c.data, offset, left, S, info);
4025   if (r == nullptr || left < sizeof(struct class_ro64_t))
4026     return nullptr;
4027   struct class_ro64_t cro;
4028   memcpy(&cro, r, sizeof(struct class_ro64_t));
4029   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4030     swapStruct(cro);
4031   if (cro.name == 0)
4032     return nullptr;
4033   const char *name = get_pointer_64(cro.name, offset, left, S, info);
4034   return name;
4035 }
4036 
4037 // get_objc2_64bit_cfstring_name is used for disassembly and is passed a
4038 // pointer to a cfstring and returns its name or nullptr.
get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,struct DisassembleInfo * info)4039 static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,
4040                                                  struct DisassembleInfo *info) {
4041   const char *r, *name;
4042   uint32_t offset, left;
4043   SectionRef S;
4044   struct cfstring64_t cfs;
4045   uint64_t cfs_characters;
4046 
4047   r = get_pointer_64(ReferenceValue, offset, left, S, info);
4048   if (r == nullptr || left < sizeof(struct cfstring64_t))
4049     return nullptr;
4050   memcpy(&cfs, r, sizeof(struct cfstring64_t));
4051   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4052     swapStruct(cfs);
4053   if (cfs.characters == 0) {
4054     uint64_t n_value;
4055     const char *symbol_name = get_symbol_64(
4056         offset + offsetof(struct cfstring64_t, characters), S, info, n_value);
4057     if (symbol_name == nullptr)
4058       return nullptr;
4059     cfs_characters = n_value;
4060   } else
4061     cfs_characters = cfs.characters;
4062   name = get_pointer_64(cfs_characters, offset, left, S, info);
4063 
4064   return name;
4065 }
4066 
4067 // get_objc2_64bit_selref() is used for disassembly and is passed a the address
4068 // of a pointer to an Objective-C selector reference when the pointer value is
4069 // zero as in a .o file and is likely to have a external relocation entry with
4070 // who's symbol's n_value is the real pointer to the selector name.  If that is
4071 // the case the real pointer to the selector name is returned else 0 is
4072 // returned
get_objc2_64bit_selref(uint64_t ReferenceValue,struct DisassembleInfo * info)4073 static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue,
4074                                        struct DisassembleInfo *info) {
4075   uint32_t offset, left;
4076   SectionRef S;
4077 
4078   const char *r = get_pointer_64(ReferenceValue, offset, left, S, info);
4079   if (r == nullptr || left < sizeof(uint64_t))
4080     return 0;
4081   uint64_t n_value;
4082   const char *symbol_name = get_symbol_64(offset, S, info, n_value);
4083   if (symbol_name == nullptr)
4084     return 0;
4085   return n_value;
4086 }
4087 
get_section(MachOObjectFile * O,const char * segname,const char * sectname)4088 static const SectionRef get_section(MachOObjectFile *O, const char *segname,
4089                                     const char *sectname) {
4090   for (const SectionRef &Section : O->sections()) {
4091     StringRef SectName;
4092     Expected<StringRef> SecNameOrErr = Section.getName();
4093     if (SecNameOrErr)
4094       SectName = *SecNameOrErr;
4095     else
4096       consumeError(SecNameOrErr.takeError());
4097 
4098     DataRefImpl Ref = Section.getRawDataRefImpl();
4099     StringRef SegName = O->getSectionFinalSegmentName(Ref);
4100     if (SegName == segname && SectName == sectname)
4101       return Section;
4102   }
4103   return SectionRef();
4104 }
4105 
4106 static void
walk_pointer_list_64(const char * listname,const SectionRef S,MachOObjectFile * O,struct DisassembleInfo * info,void (* func)(uint64_t,struct DisassembleInfo * info))4107 walk_pointer_list_64(const char *listname, const SectionRef S,
4108                      MachOObjectFile *O, struct DisassembleInfo *info,
4109                      void (*func)(uint64_t, struct DisassembleInfo *info)) {
4110   if (S == SectionRef())
4111     return;
4112 
4113   StringRef SectName;
4114   Expected<StringRef> SecNameOrErr = S.getName();
4115   if (SecNameOrErr)
4116     SectName = *SecNameOrErr;
4117   else
4118     consumeError(SecNameOrErr.takeError());
4119 
4120   DataRefImpl Ref = S.getRawDataRefImpl();
4121   StringRef SegName = O->getSectionFinalSegmentName(Ref);
4122   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
4123 
4124   StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName());
4125   const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
4126 
4127   for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint64_t)) {
4128     uint32_t left = S.getSize() - i;
4129     uint32_t size = left < sizeof(uint64_t) ? left : sizeof(uint64_t);
4130     uint64_t p = 0;
4131     memcpy(&p, Contents + i, size);
4132     if (i + sizeof(uint64_t) > S.getSize())
4133       outs() << listname << " list pointer extends past end of (" << SegName
4134              << "," << SectName << ") section\n";
4135     outs() << format("%016" PRIx64, S.getAddress() + i) << " ";
4136 
4137     if (O->isLittleEndian() != sys::IsLittleEndianHost)
4138       sys::swapByteOrder(p);
4139 
4140     uint64_t n_value = 0;
4141     const char *name = get_symbol_64(i, S, info, n_value, p);
4142     if (name == nullptr)
4143       name = get_dyld_bind_info_symbolname(S.getAddress() + i, info);
4144 
4145     if (n_value != 0) {
4146       outs() << format("0x%" PRIx64, n_value);
4147       if (p != 0)
4148         outs() << " + " << format("0x%" PRIx64, p);
4149     } else
4150       outs() << format("0x%" PRIx64, p);
4151     if (name != nullptr)
4152       outs() << " " << name;
4153     outs() << "\n";
4154 
4155     p += n_value;
4156     if (func)
4157       func(p, info);
4158   }
4159 }
4160 
4161 static void
walk_pointer_list_32(const char * listname,const SectionRef S,MachOObjectFile * O,struct DisassembleInfo * info,void (* func)(uint32_t,struct DisassembleInfo * info))4162 walk_pointer_list_32(const char *listname, const SectionRef S,
4163                      MachOObjectFile *O, struct DisassembleInfo *info,
4164                      void (*func)(uint32_t, struct DisassembleInfo *info)) {
4165   if (S == SectionRef())
4166     return;
4167 
4168   StringRef SectName = unwrapOrError(S.getName(), O->getFileName());
4169   DataRefImpl Ref = S.getRawDataRefImpl();
4170   StringRef SegName = O->getSectionFinalSegmentName(Ref);
4171   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
4172 
4173   StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName());
4174   const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
4175 
4176   for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint32_t)) {
4177     uint32_t left = S.getSize() - i;
4178     uint32_t size = left < sizeof(uint32_t) ? left : sizeof(uint32_t);
4179     uint32_t p = 0;
4180     memcpy(&p, Contents + i, size);
4181     if (i + sizeof(uint32_t) > S.getSize())
4182       outs() << listname << " list pointer extends past end of (" << SegName
4183              << "," << SectName << ") section\n";
4184     uint32_t Address = S.getAddress() + i;
4185     outs() << format("%08" PRIx32, Address) << " ";
4186 
4187     if (O->isLittleEndian() != sys::IsLittleEndianHost)
4188       sys::swapByteOrder(p);
4189     outs() << format("0x%" PRIx32, p);
4190 
4191     const char *name = get_symbol_32(i, S, info, p);
4192     if (name != nullptr)
4193       outs() << " " << name;
4194     outs() << "\n";
4195 
4196     if (func)
4197       func(p, info);
4198   }
4199 }
4200 
print_layout_map(const char * layout_map,uint32_t left)4201 static void print_layout_map(const char *layout_map, uint32_t left) {
4202   if (layout_map == nullptr)
4203     return;
4204   outs() << "                layout map: ";
4205   do {
4206     outs() << format("0x%02" PRIx32, (*layout_map) & 0xff) << " ";
4207     left--;
4208     layout_map++;
4209   } while (*layout_map != '\0' && left != 0);
4210   outs() << "\n";
4211 }
4212 
print_layout_map64(uint64_t p,struct DisassembleInfo * info)4213 static void print_layout_map64(uint64_t p, struct DisassembleInfo *info) {
4214   uint32_t offset, left;
4215   SectionRef S;
4216   const char *layout_map;
4217 
4218   if (p == 0)
4219     return;
4220   layout_map = get_pointer_64(p, offset, left, S, info);
4221   print_layout_map(layout_map, left);
4222 }
4223 
print_layout_map32(uint32_t p,struct DisassembleInfo * info)4224 static void print_layout_map32(uint32_t p, struct DisassembleInfo *info) {
4225   uint32_t offset, left;
4226   SectionRef S;
4227   const char *layout_map;
4228 
4229   if (p == 0)
4230     return;
4231   layout_map = get_pointer_32(p, offset, left, S, info);
4232   print_layout_map(layout_map, left);
4233 }
4234 
print_method_list64_t(uint64_t p,struct DisassembleInfo * info,const char * indent)4235 static void print_method_list64_t(uint64_t p, struct DisassembleInfo *info,
4236                                   const char *indent) {
4237   struct method_list64_t ml;
4238   struct method64_t m;
4239   const char *r;
4240   uint32_t offset, xoffset, left, i;
4241   SectionRef S, xS;
4242   const char *name, *sym_name;
4243   uint64_t n_value;
4244 
4245   r = get_pointer_64(p, offset, left, S, info);
4246   if (r == nullptr)
4247     return;
4248   memset(&ml, '\0', sizeof(struct method_list64_t));
4249   if (left < sizeof(struct method_list64_t)) {
4250     memcpy(&ml, r, left);
4251     outs() << "   (method_list_t entends past the end of the section)\n";
4252   } else
4253     memcpy(&ml, r, sizeof(struct method_list64_t));
4254   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4255     swapStruct(ml);
4256   outs() << indent << "\t\t   entsize " << ml.entsize << "\n";
4257   outs() << indent << "\t\t     count " << ml.count << "\n";
4258 
4259   p += sizeof(struct method_list64_t);
4260   offset += sizeof(struct method_list64_t);
4261   for (i = 0; i < ml.count; i++) {
4262     r = get_pointer_64(p, offset, left, S, info);
4263     if (r == nullptr)
4264       return;
4265     memset(&m, '\0', sizeof(struct method64_t));
4266     if (left < sizeof(struct method64_t)) {
4267       memcpy(&m, r, left);
4268       outs() << indent << "   (method_t extends past the end of the section)\n";
4269     } else
4270       memcpy(&m, r, sizeof(struct method64_t));
4271     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4272       swapStruct(m);
4273 
4274     outs() << indent << "\t\t      name ";
4275     sym_name = get_symbol_64(offset + offsetof(struct method64_t, name), S,
4276                              info, n_value, m.name);
4277     if (n_value != 0) {
4278       if (info->verbose && sym_name != nullptr)
4279         outs() << sym_name;
4280       else
4281         outs() << format("0x%" PRIx64, n_value);
4282       if (m.name != 0)
4283         outs() << " + " << format("0x%" PRIx64, m.name);
4284     } else
4285       outs() << format("0x%" PRIx64, m.name);
4286     name = get_pointer_64(m.name + n_value, xoffset, left, xS, info);
4287     if (name != nullptr)
4288       outs() << format(" %.*s", left, name);
4289     outs() << "\n";
4290 
4291     outs() << indent << "\t\t     types ";
4292     sym_name = get_symbol_64(offset + offsetof(struct method64_t, types), S,
4293                              info, n_value, m.types);
4294     if (n_value != 0) {
4295       if (info->verbose && sym_name != nullptr)
4296         outs() << sym_name;
4297       else
4298         outs() << format("0x%" PRIx64, n_value);
4299       if (m.types != 0)
4300         outs() << " + " << format("0x%" PRIx64, m.types);
4301     } else
4302       outs() << format("0x%" PRIx64, m.types);
4303     name = get_pointer_64(m.types + n_value, xoffset, left, xS, info);
4304     if (name != nullptr)
4305       outs() << format(" %.*s", left, name);
4306     outs() << "\n";
4307 
4308     outs() << indent << "\t\t       imp ";
4309     name = get_symbol_64(offset + offsetof(struct method64_t, imp), S, info,
4310                          n_value, m.imp);
4311     if (info->verbose && name == nullptr) {
4312       if (n_value != 0) {
4313         outs() << format("0x%" PRIx64, n_value) << " ";
4314         if (m.imp != 0)
4315           outs() << "+ " << format("0x%" PRIx64, m.imp) << " ";
4316       } else
4317         outs() << format("0x%" PRIx64, m.imp) << " ";
4318     }
4319     if (name != nullptr)
4320       outs() << name;
4321     outs() << "\n";
4322 
4323     p += sizeof(struct method64_t);
4324     offset += sizeof(struct method64_t);
4325   }
4326 }
4327 
print_method_list32_t(uint64_t p,struct DisassembleInfo * info,const char * indent)4328 static void print_method_list32_t(uint64_t p, struct DisassembleInfo *info,
4329                                   const char *indent) {
4330   struct method_list32_t ml;
4331   struct method32_t m;
4332   const char *r, *name;
4333   uint32_t offset, xoffset, left, i;
4334   SectionRef S, xS;
4335 
4336   r = get_pointer_32(p, offset, left, S, info);
4337   if (r == nullptr)
4338     return;
4339   memset(&ml, '\0', sizeof(struct method_list32_t));
4340   if (left < sizeof(struct method_list32_t)) {
4341     memcpy(&ml, r, left);
4342     outs() << "   (method_list_t entends past the end of the section)\n";
4343   } else
4344     memcpy(&ml, r, sizeof(struct method_list32_t));
4345   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4346     swapStruct(ml);
4347   outs() << indent << "\t\t   entsize " << ml.entsize << "\n";
4348   outs() << indent << "\t\t     count " << ml.count << "\n";
4349 
4350   p += sizeof(struct method_list32_t);
4351   offset += sizeof(struct method_list32_t);
4352   for (i = 0; i < ml.count; i++) {
4353     r = get_pointer_32(p, offset, left, S, info);
4354     if (r == nullptr)
4355       return;
4356     memset(&m, '\0', sizeof(struct method32_t));
4357     if (left < sizeof(struct method32_t)) {
4358       memcpy(&ml, r, left);
4359       outs() << indent << "   (method_t entends past the end of the section)\n";
4360     } else
4361       memcpy(&m, r, sizeof(struct method32_t));
4362     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4363       swapStruct(m);
4364 
4365     outs() << indent << "\t\t      name " << format("0x%" PRIx32, m.name);
4366     name = get_pointer_32(m.name, xoffset, left, xS, info);
4367     if (name != nullptr)
4368       outs() << format(" %.*s", left, name);
4369     outs() << "\n";
4370 
4371     outs() << indent << "\t\t     types " << format("0x%" PRIx32, m.types);
4372     name = get_pointer_32(m.types, xoffset, left, xS, info);
4373     if (name != nullptr)
4374       outs() << format(" %.*s", left, name);
4375     outs() << "\n";
4376 
4377     outs() << indent << "\t\t       imp " << format("0x%" PRIx32, m.imp);
4378     name = get_symbol_32(offset + offsetof(struct method32_t, imp), S, info,
4379                          m.imp);
4380     if (name != nullptr)
4381       outs() << " " << name;
4382     outs() << "\n";
4383 
4384     p += sizeof(struct method32_t);
4385     offset += sizeof(struct method32_t);
4386   }
4387 }
4388 
print_method_list(uint32_t p,struct DisassembleInfo * info)4389 static bool print_method_list(uint32_t p, struct DisassembleInfo *info) {
4390   uint32_t offset, left, xleft;
4391   SectionRef S;
4392   struct objc_method_list_t method_list;
4393   struct objc_method_t method;
4394   const char *r, *methods, *name, *SymbolName;
4395   int32_t i;
4396 
4397   r = get_pointer_32(p, offset, left, S, info, true);
4398   if (r == nullptr)
4399     return true;
4400 
4401   outs() << "\n";
4402   if (left > sizeof(struct objc_method_list_t)) {
4403     memcpy(&method_list, r, sizeof(struct objc_method_list_t));
4404   } else {
4405     outs() << "\t\t objc_method_list extends past end of the section\n";
4406     memset(&method_list, '\0', sizeof(struct objc_method_list_t));
4407     memcpy(&method_list, r, left);
4408   }
4409   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4410     swapStruct(method_list);
4411 
4412   outs() << "\t\t         obsolete "
4413          << format("0x%08" PRIx32, method_list.obsolete) << "\n";
4414   outs() << "\t\t     method_count " << method_list.method_count << "\n";
4415 
4416   methods = r + sizeof(struct objc_method_list_t);
4417   for (i = 0; i < method_list.method_count; i++) {
4418     if ((i + 1) * sizeof(struct objc_method_t) > left) {
4419       outs() << "\t\t remaining method's extend past the of the section\n";
4420       break;
4421     }
4422     memcpy(&method, methods + i * sizeof(struct objc_method_t),
4423            sizeof(struct objc_method_t));
4424     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4425       swapStruct(method);
4426 
4427     outs() << "\t\t      method_name "
4428            << format("0x%08" PRIx32, method.method_name);
4429     if (info->verbose) {
4430       name = get_pointer_32(method.method_name, offset, xleft, S, info, true);
4431       if (name != nullptr)
4432         outs() << format(" %.*s", xleft, name);
4433       else
4434         outs() << " (not in an __OBJC section)";
4435     }
4436     outs() << "\n";
4437 
4438     outs() << "\t\t     method_types "
4439            << format("0x%08" PRIx32, method.method_types);
4440     if (info->verbose) {
4441       name = get_pointer_32(method.method_types, offset, xleft, S, info, true);
4442       if (name != nullptr)
4443         outs() << format(" %.*s", xleft, name);
4444       else
4445         outs() << " (not in an __OBJC section)";
4446     }
4447     outs() << "\n";
4448 
4449     outs() << "\t\t       method_imp "
4450            << format("0x%08" PRIx32, method.method_imp) << " ";
4451     if (info->verbose) {
4452       SymbolName = GuessSymbolName(method.method_imp, info->AddrMap);
4453       if (SymbolName != nullptr)
4454         outs() << SymbolName;
4455     }
4456     outs() << "\n";
4457   }
4458   return false;
4459 }
4460 
print_protocol_list64_t(uint64_t p,struct DisassembleInfo * info)4461 static void print_protocol_list64_t(uint64_t p, struct DisassembleInfo *info) {
4462   struct protocol_list64_t pl;
4463   uint64_t q, n_value;
4464   struct protocol64_t pc;
4465   const char *r;
4466   uint32_t offset, xoffset, left, i;
4467   SectionRef S, xS;
4468   const char *name, *sym_name;
4469 
4470   r = get_pointer_64(p, offset, left, S, info);
4471   if (r == nullptr)
4472     return;
4473   memset(&pl, '\0', sizeof(struct protocol_list64_t));
4474   if (left < sizeof(struct protocol_list64_t)) {
4475     memcpy(&pl, r, left);
4476     outs() << "   (protocol_list_t entends past the end of the section)\n";
4477   } else
4478     memcpy(&pl, r, sizeof(struct protocol_list64_t));
4479   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4480     swapStruct(pl);
4481   outs() << "                      count " << pl.count << "\n";
4482 
4483   p += sizeof(struct protocol_list64_t);
4484   offset += sizeof(struct protocol_list64_t);
4485   for (i = 0; i < pl.count; i++) {
4486     r = get_pointer_64(p, offset, left, S, info);
4487     if (r == nullptr)
4488       return;
4489     q = 0;
4490     if (left < sizeof(uint64_t)) {
4491       memcpy(&q, r, left);
4492       outs() << "   (protocol_t * entends past the end of the section)\n";
4493     } else
4494       memcpy(&q, r, sizeof(uint64_t));
4495     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4496       sys::swapByteOrder(q);
4497 
4498     outs() << "\t\t      list[" << i << "] ";
4499     sym_name = get_symbol_64(offset, S, info, n_value, q);
4500     if (n_value != 0) {
4501       if (info->verbose && sym_name != nullptr)
4502         outs() << sym_name;
4503       else
4504         outs() << format("0x%" PRIx64, n_value);
4505       if (q != 0)
4506         outs() << " + " << format("0x%" PRIx64, q);
4507     } else
4508       outs() << format("0x%" PRIx64, q);
4509     outs() << " (struct protocol_t *)\n";
4510 
4511     r = get_pointer_64(q + n_value, offset, left, S, info);
4512     if (r == nullptr)
4513       return;
4514     memset(&pc, '\0', sizeof(struct protocol64_t));
4515     if (left < sizeof(struct protocol64_t)) {
4516       memcpy(&pc, r, left);
4517       outs() << "   (protocol_t entends past the end of the section)\n";
4518     } else
4519       memcpy(&pc, r, sizeof(struct protocol64_t));
4520     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4521       swapStruct(pc);
4522 
4523     outs() << "\t\t\t      isa " << format("0x%" PRIx64, pc.isa) << "\n";
4524 
4525     outs() << "\t\t\t     name ";
4526     sym_name = get_symbol_64(offset + offsetof(struct protocol64_t, name), S,
4527                              info, n_value, pc.name);
4528     if (n_value != 0) {
4529       if (info->verbose && sym_name != nullptr)
4530         outs() << sym_name;
4531       else
4532         outs() << format("0x%" PRIx64, n_value);
4533       if (pc.name != 0)
4534         outs() << " + " << format("0x%" PRIx64, pc.name);
4535     } else
4536       outs() << format("0x%" PRIx64, pc.name);
4537     name = get_pointer_64(pc.name + n_value, xoffset, left, xS, info);
4538     if (name != nullptr)
4539       outs() << format(" %.*s", left, name);
4540     outs() << "\n";
4541 
4542     outs() << "\t\t\tprotocols " << format("0x%" PRIx64, pc.protocols) << "\n";
4543 
4544     outs() << "\t\t  instanceMethods ";
4545     sym_name =
4546         get_symbol_64(offset + offsetof(struct protocol64_t, instanceMethods),
4547                       S, info, n_value, pc.instanceMethods);
4548     if (n_value != 0) {
4549       if (info->verbose && sym_name != nullptr)
4550         outs() << sym_name;
4551       else
4552         outs() << format("0x%" PRIx64, n_value);
4553       if (pc.instanceMethods != 0)
4554         outs() << " + " << format("0x%" PRIx64, pc.instanceMethods);
4555     } else
4556       outs() << format("0x%" PRIx64, pc.instanceMethods);
4557     outs() << " (struct method_list_t *)\n";
4558     if (pc.instanceMethods + n_value != 0)
4559       print_method_list64_t(pc.instanceMethods + n_value, info, "\t");
4560 
4561     outs() << "\t\t     classMethods ";
4562     sym_name =
4563         get_symbol_64(offset + offsetof(struct protocol64_t, classMethods), S,
4564                       info, n_value, pc.classMethods);
4565     if (n_value != 0) {
4566       if (info->verbose && sym_name != nullptr)
4567         outs() << sym_name;
4568       else
4569         outs() << format("0x%" PRIx64, n_value);
4570       if (pc.classMethods != 0)
4571         outs() << " + " << format("0x%" PRIx64, pc.classMethods);
4572     } else
4573       outs() << format("0x%" PRIx64, pc.classMethods);
4574     outs() << " (struct method_list_t *)\n";
4575     if (pc.classMethods + n_value != 0)
4576       print_method_list64_t(pc.classMethods + n_value, info, "\t");
4577 
4578     outs() << "\t  optionalInstanceMethods "
4579            << format("0x%" PRIx64, pc.optionalInstanceMethods) << "\n";
4580     outs() << "\t     optionalClassMethods "
4581            << format("0x%" PRIx64, pc.optionalClassMethods) << "\n";
4582     outs() << "\t       instanceProperties "
4583            << format("0x%" PRIx64, pc.instanceProperties) << "\n";
4584 
4585     p += sizeof(uint64_t);
4586     offset += sizeof(uint64_t);
4587   }
4588 }
4589 
print_protocol_list32_t(uint32_t p,struct DisassembleInfo * info)4590 static void print_protocol_list32_t(uint32_t p, struct DisassembleInfo *info) {
4591   struct protocol_list32_t pl;
4592   uint32_t q;
4593   struct protocol32_t pc;
4594   const char *r;
4595   uint32_t offset, xoffset, left, i;
4596   SectionRef S, xS;
4597   const char *name;
4598 
4599   r = get_pointer_32(p, offset, left, S, info);
4600   if (r == nullptr)
4601     return;
4602   memset(&pl, '\0', sizeof(struct protocol_list32_t));
4603   if (left < sizeof(struct protocol_list32_t)) {
4604     memcpy(&pl, r, left);
4605     outs() << "   (protocol_list_t entends past the end of the section)\n";
4606   } else
4607     memcpy(&pl, r, sizeof(struct protocol_list32_t));
4608   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4609     swapStruct(pl);
4610   outs() << "                      count " << pl.count << "\n";
4611 
4612   p += sizeof(struct protocol_list32_t);
4613   offset += sizeof(struct protocol_list32_t);
4614   for (i = 0; i < pl.count; i++) {
4615     r = get_pointer_32(p, offset, left, S, info);
4616     if (r == nullptr)
4617       return;
4618     q = 0;
4619     if (left < sizeof(uint32_t)) {
4620       memcpy(&q, r, left);
4621       outs() << "   (protocol_t * entends past the end of the section)\n";
4622     } else
4623       memcpy(&q, r, sizeof(uint32_t));
4624     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4625       sys::swapByteOrder(q);
4626     outs() << "\t\t      list[" << i << "] " << format("0x%" PRIx32, q)
4627            << " (struct protocol_t *)\n";
4628     r = get_pointer_32(q, offset, left, S, info);
4629     if (r == nullptr)
4630       return;
4631     memset(&pc, '\0', sizeof(struct protocol32_t));
4632     if (left < sizeof(struct protocol32_t)) {
4633       memcpy(&pc, r, left);
4634       outs() << "   (protocol_t entends past the end of the section)\n";
4635     } else
4636       memcpy(&pc, r, sizeof(struct protocol32_t));
4637     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4638       swapStruct(pc);
4639     outs() << "\t\t\t      isa " << format("0x%" PRIx32, pc.isa) << "\n";
4640     outs() << "\t\t\t     name " << format("0x%" PRIx32, pc.name);
4641     name = get_pointer_32(pc.name, xoffset, left, xS, info);
4642     if (name != nullptr)
4643       outs() << format(" %.*s", left, name);
4644     outs() << "\n";
4645     outs() << "\t\t\tprotocols " << format("0x%" PRIx32, pc.protocols) << "\n";
4646     outs() << "\t\t  instanceMethods "
4647            << format("0x%" PRIx32, pc.instanceMethods)
4648            << " (struct method_list_t *)\n";
4649     if (pc.instanceMethods != 0)
4650       print_method_list32_t(pc.instanceMethods, info, "\t");
4651     outs() << "\t\t     classMethods " << format("0x%" PRIx32, pc.classMethods)
4652            << " (struct method_list_t *)\n";
4653     if (pc.classMethods != 0)
4654       print_method_list32_t(pc.classMethods, info, "\t");
4655     outs() << "\t  optionalInstanceMethods "
4656            << format("0x%" PRIx32, pc.optionalInstanceMethods) << "\n";
4657     outs() << "\t     optionalClassMethods "
4658            << format("0x%" PRIx32, pc.optionalClassMethods) << "\n";
4659     outs() << "\t       instanceProperties "
4660            << format("0x%" PRIx32, pc.instanceProperties) << "\n";
4661     p += sizeof(uint32_t);
4662     offset += sizeof(uint32_t);
4663   }
4664 }
4665 
print_indent(uint32_t indent)4666 static void print_indent(uint32_t indent) {
4667   for (uint32_t i = 0; i < indent;) {
4668     if (indent - i >= 8) {
4669       outs() << "\t";
4670       i += 8;
4671     } else {
4672       for (uint32_t j = i; j < indent; j++)
4673         outs() << " ";
4674       return;
4675     }
4676   }
4677 }
4678 
print_method_description_list(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4679 static bool print_method_description_list(uint32_t p, uint32_t indent,
4680                                           struct DisassembleInfo *info) {
4681   uint32_t offset, left, xleft;
4682   SectionRef S;
4683   struct objc_method_description_list_t mdl;
4684   struct objc_method_description_t md;
4685   const char *r, *list, *name;
4686   int32_t i;
4687 
4688   r = get_pointer_32(p, offset, left, S, info, true);
4689   if (r == nullptr)
4690     return true;
4691 
4692   outs() << "\n";
4693   if (left > sizeof(struct objc_method_description_list_t)) {
4694     memcpy(&mdl, r, sizeof(struct objc_method_description_list_t));
4695   } else {
4696     print_indent(indent);
4697     outs() << " objc_method_description_list extends past end of the section\n";
4698     memset(&mdl, '\0', sizeof(struct objc_method_description_list_t));
4699     memcpy(&mdl, r, left);
4700   }
4701   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4702     swapStruct(mdl);
4703 
4704   print_indent(indent);
4705   outs() << "        count " << mdl.count << "\n";
4706 
4707   list = r + sizeof(struct objc_method_description_list_t);
4708   for (i = 0; i < mdl.count; i++) {
4709     if ((i + 1) * sizeof(struct objc_method_description_t) > left) {
4710       print_indent(indent);
4711       outs() << " remaining list entries extend past the of the section\n";
4712       break;
4713     }
4714     print_indent(indent);
4715     outs() << "        list[" << i << "]\n";
4716     memcpy(&md, list + i * sizeof(struct objc_method_description_t),
4717            sizeof(struct objc_method_description_t));
4718     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4719       swapStruct(md);
4720 
4721     print_indent(indent);
4722     outs() << "             name " << format("0x%08" PRIx32, md.name);
4723     if (info->verbose) {
4724       name = get_pointer_32(md.name, offset, xleft, S, info, true);
4725       if (name != nullptr)
4726         outs() << format(" %.*s", xleft, name);
4727       else
4728         outs() << " (not in an __OBJC section)";
4729     }
4730     outs() << "\n";
4731 
4732     print_indent(indent);
4733     outs() << "            types " << format("0x%08" PRIx32, md.types);
4734     if (info->verbose) {
4735       name = get_pointer_32(md.types, offset, xleft, S, info, true);
4736       if (name != nullptr)
4737         outs() << format(" %.*s", xleft, name);
4738       else
4739         outs() << " (not in an __OBJC section)";
4740     }
4741     outs() << "\n";
4742   }
4743   return false;
4744 }
4745 
4746 static bool print_protocol_list(uint32_t p, uint32_t indent,
4747                                 struct DisassembleInfo *info);
4748 
print_protocol(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4749 static bool print_protocol(uint32_t p, uint32_t indent,
4750                            struct DisassembleInfo *info) {
4751   uint32_t offset, left;
4752   SectionRef S;
4753   struct objc_protocol_t protocol;
4754   const char *r, *name;
4755 
4756   r = get_pointer_32(p, offset, left, S, info, true);
4757   if (r == nullptr)
4758     return true;
4759 
4760   outs() << "\n";
4761   if (left >= sizeof(struct objc_protocol_t)) {
4762     memcpy(&protocol, r, sizeof(struct objc_protocol_t));
4763   } else {
4764     print_indent(indent);
4765     outs() << "            Protocol extends past end of the section\n";
4766     memset(&protocol, '\0', sizeof(struct objc_protocol_t));
4767     memcpy(&protocol, r, left);
4768   }
4769   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4770     swapStruct(protocol);
4771 
4772   print_indent(indent);
4773   outs() << "              isa " << format("0x%08" PRIx32, protocol.isa)
4774          << "\n";
4775 
4776   print_indent(indent);
4777   outs() << "    protocol_name "
4778          << format("0x%08" PRIx32, protocol.protocol_name);
4779   if (info->verbose) {
4780     name = get_pointer_32(protocol.protocol_name, offset, left, S, info, true);
4781     if (name != nullptr)
4782       outs() << format(" %.*s", left, name);
4783     else
4784       outs() << " (not in an __OBJC section)";
4785   }
4786   outs() << "\n";
4787 
4788   print_indent(indent);
4789   outs() << "    protocol_list "
4790          << format("0x%08" PRIx32, protocol.protocol_list);
4791   if (print_protocol_list(protocol.protocol_list, indent + 4, info))
4792     outs() << " (not in an __OBJC section)\n";
4793 
4794   print_indent(indent);
4795   outs() << " instance_methods "
4796          << format("0x%08" PRIx32, protocol.instance_methods);
4797   if (print_method_description_list(protocol.instance_methods, indent, info))
4798     outs() << " (not in an __OBJC section)\n";
4799 
4800   print_indent(indent);
4801   outs() << "    class_methods "
4802          << format("0x%08" PRIx32, protocol.class_methods);
4803   if (print_method_description_list(protocol.class_methods, indent, info))
4804     outs() << " (not in an __OBJC section)\n";
4805 
4806   return false;
4807 }
4808 
print_protocol_list(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4809 static bool print_protocol_list(uint32_t p, uint32_t indent,
4810                                 struct DisassembleInfo *info) {
4811   uint32_t offset, left, l;
4812   SectionRef S;
4813   struct objc_protocol_list_t protocol_list;
4814   const char *r, *list;
4815   int32_t i;
4816 
4817   r = get_pointer_32(p, offset, left, S, info, true);
4818   if (r == nullptr)
4819     return true;
4820 
4821   outs() << "\n";
4822   if (left > sizeof(struct objc_protocol_list_t)) {
4823     memcpy(&protocol_list, r, sizeof(struct objc_protocol_list_t));
4824   } else {
4825     outs() << "\t\t objc_protocol_list_t extends past end of the section\n";
4826     memset(&protocol_list, '\0', sizeof(struct objc_protocol_list_t));
4827     memcpy(&protocol_list, r, left);
4828   }
4829   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4830     swapStruct(protocol_list);
4831 
4832   print_indent(indent);
4833   outs() << "         next " << format("0x%08" PRIx32, protocol_list.next)
4834          << "\n";
4835   print_indent(indent);
4836   outs() << "        count " << protocol_list.count << "\n";
4837 
4838   list = r + sizeof(struct objc_protocol_list_t);
4839   for (i = 0; i < protocol_list.count; i++) {
4840     if ((i + 1) * sizeof(uint32_t) > left) {
4841       outs() << "\t\t remaining list entries extend past the of the section\n";
4842       break;
4843     }
4844     memcpy(&l, list + i * sizeof(uint32_t), sizeof(uint32_t));
4845     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4846       sys::swapByteOrder(l);
4847 
4848     print_indent(indent);
4849     outs() << "      list[" << i << "] " << format("0x%08" PRIx32, l);
4850     if (print_protocol(l, indent, info))
4851       outs() << "(not in an __OBJC section)\n";
4852   }
4853   return false;
4854 }
4855 
print_ivar_list64_t(uint64_t p,struct DisassembleInfo * info)4856 static void print_ivar_list64_t(uint64_t p, struct DisassembleInfo *info) {
4857   struct ivar_list64_t il;
4858   struct ivar64_t i;
4859   const char *r;
4860   uint32_t offset, xoffset, left, j;
4861   SectionRef S, xS;
4862   const char *name, *sym_name, *ivar_offset_p;
4863   uint64_t ivar_offset, n_value;
4864 
4865   r = get_pointer_64(p, offset, left, S, info);
4866   if (r == nullptr)
4867     return;
4868   memset(&il, '\0', sizeof(struct ivar_list64_t));
4869   if (left < sizeof(struct ivar_list64_t)) {
4870     memcpy(&il, r, left);
4871     outs() << "   (ivar_list_t entends past the end of the section)\n";
4872   } else
4873     memcpy(&il, r, sizeof(struct ivar_list64_t));
4874   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4875     swapStruct(il);
4876   outs() << "                    entsize " << il.entsize << "\n";
4877   outs() << "                      count " << il.count << "\n";
4878 
4879   p += sizeof(struct ivar_list64_t);
4880   offset += sizeof(struct ivar_list64_t);
4881   for (j = 0; j < il.count; j++) {
4882     r = get_pointer_64(p, offset, left, S, info);
4883     if (r == nullptr)
4884       return;
4885     memset(&i, '\0', sizeof(struct ivar64_t));
4886     if (left < sizeof(struct ivar64_t)) {
4887       memcpy(&i, r, left);
4888       outs() << "   (ivar_t entends past the end of the section)\n";
4889     } else
4890       memcpy(&i, r, sizeof(struct ivar64_t));
4891     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4892       swapStruct(i);
4893 
4894     outs() << "\t\t\t   offset ";
4895     sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, offset), S,
4896                              info, n_value, i.offset);
4897     if (n_value != 0) {
4898       if (info->verbose && sym_name != nullptr)
4899         outs() << sym_name;
4900       else
4901         outs() << format("0x%" PRIx64, n_value);
4902       if (i.offset != 0)
4903         outs() << " + " << format("0x%" PRIx64, i.offset);
4904     } else
4905       outs() << format("0x%" PRIx64, i.offset);
4906     ivar_offset_p = get_pointer_64(i.offset + n_value, xoffset, left, xS, info);
4907     if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) {
4908       memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset));
4909       if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4910         sys::swapByteOrder(ivar_offset);
4911       outs() << " " << ivar_offset << "\n";
4912     } else
4913       outs() << "\n";
4914 
4915     outs() << "\t\t\t     name ";
4916     sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, name), S, info,
4917                              n_value, i.name);
4918     if (n_value != 0) {
4919       if (info->verbose && sym_name != nullptr)
4920         outs() << sym_name;
4921       else
4922         outs() << format("0x%" PRIx64, n_value);
4923       if (i.name != 0)
4924         outs() << " + " << format("0x%" PRIx64, i.name);
4925     } else
4926       outs() << format("0x%" PRIx64, i.name);
4927     name = get_pointer_64(i.name + n_value, xoffset, left, xS, info);
4928     if (name != nullptr)
4929       outs() << format(" %.*s", left, name);
4930     outs() << "\n";
4931 
4932     outs() << "\t\t\t     type ";
4933     sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, type), S, info,
4934                              n_value, i.name);
4935     name = get_pointer_64(i.type + n_value, xoffset, left, xS, info);
4936     if (n_value != 0) {
4937       if (info->verbose && sym_name != nullptr)
4938         outs() << sym_name;
4939       else
4940         outs() << format("0x%" PRIx64, n_value);
4941       if (i.type != 0)
4942         outs() << " + " << format("0x%" PRIx64, i.type);
4943     } else
4944       outs() << format("0x%" PRIx64, i.type);
4945     if (name != nullptr)
4946       outs() << format(" %.*s", left, name);
4947     outs() << "\n";
4948 
4949     outs() << "\t\t\talignment " << i.alignment << "\n";
4950     outs() << "\t\t\t     size " << i.size << "\n";
4951 
4952     p += sizeof(struct ivar64_t);
4953     offset += sizeof(struct ivar64_t);
4954   }
4955 }
4956 
print_ivar_list32_t(uint32_t p,struct DisassembleInfo * info)4957 static void print_ivar_list32_t(uint32_t p, struct DisassembleInfo *info) {
4958   struct ivar_list32_t il;
4959   struct ivar32_t i;
4960   const char *r;
4961   uint32_t offset, xoffset, left, j;
4962   SectionRef S, xS;
4963   const char *name, *ivar_offset_p;
4964   uint32_t ivar_offset;
4965 
4966   r = get_pointer_32(p, offset, left, S, info);
4967   if (r == nullptr)
4968     return;
4969   memset(&il, '\0', sizeof(struct ivar_list32_t));
4970   if (left < sizeof(struct ivar_list32_t)) {
4971     memcpy(&il, r, left);
4972     outs() << "   (ivar_list_t entends past the end of the section)\n";
4973   } else
4974     memcpy(&il, r, sizeof(struct ivar_list32_t));
4975   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4976     swapStruct(il);
4977   outs() << "                    entsize " << il.entsize << "\n";
4978   outs() << "                      count " << il.count << "\n";
4979 
4980   p += sizeof(struct ivar_list32_t);
4981   offset += sizeof(struct ivar_list32_t);
4982   for (j = 0; j < il.count; j++) {
4983     r = get_pointer_32(p, offset, left, S, info);
4984     if (r == nullptr)
4985       return;
4986     memset(&i, '\0', sizeof(struct ivar32_t));
4987     if (left < sizeof(struct ivar32_t)) {
4988       memcpy(&i, r, left);
4989       outs() << "   (ivar_t entends past the end of the section)\n";
4990     } else
4991       memcpy(&i, r, sizeof(struct ivar32_t));
4992     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4993       swapStruct(i);
4994 
4995     outs() << "\t\t\t   offset " << format("0x%" PRIx32, i.offset);
4996     ivar_offset_p = get_pointer_32(i.offset, xoffset, left, xS, info);
4997     if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) {
4998       memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset));
4999       if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5000         sys::swapByteOrder(ivar_offset);
5001       outs() << " " << ivar_offset << "\n";
5002     } else
5003       outs() << "\n";
5004 
5005     outs() << "\t\t\t     name " << format("0x%" PRIx32, i.name);
5006     name = get_pointer_32(i.name, xoffset, left, xS, info);
5007     if (name != nullptr)
5008       outs() << format(" %.*s", left, name);
5009     outs() << "\n";
5010 
5011     outs() << "\t\t\t     type " << format("0x%" PRIx32, i.type);
5012     name = get_pointer_32(i.type, xoffset, left, xS, info);
5013     if (name != nullptr)
5014       outs() << format(" %.*s", left, name);
5015     outs() << "\n";
5016 
5017     outs() << "\t\t\talignment " << i.alignment << "\n";
5018     outs() << "\t\t\t     size " << i.size << "\n";
5019 
5020     p += sizeof(struct ivar32_t);
5021     offset += sizeof(struct ivar32_t);
5022   }
5023 }
5024 
print_objc_property_list64(uint64_t p,struct DisassembleInfo * info)5025 static void print_objc_property_list64(uint64_t p,
5026                                        struct DisassembleInfo *info) {
5027   struct objc_property_list64 opl;
5028   struct objc_property64 op;
5029   const char *r;
5030   uint32_t offset, xoffset, left, j;
5031   SectionRef S, xS;
5032   const char *name, *sym_name;
5033   uint64_t n_value;
5034 
5035   r = get_pointer_64(p, offset, left, S, info);
5036   if (r == nullptr)
5037     return;
5038   memset(&opl, '\0', sizeof(struct objc_property_list64));
5039   if (left < sizeof(struct objc_property_list64)) {
5040     memcpy(&opl, r, left);
5041     outs() << "   (objc_property_list entends past the end of the section)\n";
5042   } else
5043     memcpy(&opl, r, sizeof(struct objc_property_list64));
5044   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5045     swapStruct(opl);
5046   outs() << "                    entsize " << opl.entsize << "\n";
5047   outs() << "                      count " << opl.count << "\n";
5048 
5049   p += sizeof(struct objc_property_list64);
5050   offset += sizeof(struct objc_property_list64);
5051   for (j = 0; j < opl.count; j++) {
5052     r = get_pointer_64(p, offset, left, S, info);
5053     if (r == nullptr)
5054       return;
5055     memset(&op, '\0', sizeof(struct objc_property64));
5056     if (left < sizeof(struct objc_property64)) {
5057       memcpy(&op, r, left);
5058       outs() << "   (objc_property entends past the end of the section)\n";
5059     } else
5060       memcpy(&op, r, sizeof(struct objc_property64));
5061     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5062       swapStruct(op);
5063 
5064     outs() << "\t\t\t     name ";
5065     sym_name = get_symbol_64(offset + offsetof(struct objc_property64, name), S,
5066                              info, n_value, op.name);
5067     if (n_value != 0) {
5068       if (info->verbose && sym_name != nullptr)
5069         outs() << sym_name;
5070       else
5071         outs() << format("0x%" PRIx64, n_value);
5072       if (op.name != 0)
5073         outs() << " + " << format("0x%" PRIx64, op.name);
5074     } else
5075       outs() << format("0x%" PRIx64, op.name);
5076     name = get_pointer_64(op.name + n_value, xoffset, left, xS, info);
5077     if (name != nullptr)
5078       outs() << format(" %.*s", left, name);
5079     outs() << "\n";
5080 
5081     outs() << "\t\t\tattributes ";
5082     sym_name =
5083         get_symbol_64(offset + offsetof(struct objc_property64, attributes), S,
5084                       info, n_value, op.attributes);
5085     if (n_value != 0) {
5086       if (info->verbose && sym_name != nullptr)
5087         outs() << sym_name;
5088       else
5089         outs() << format("0x%" PRIx64, n_value);
5090       if (op.attributes != 0)
5091         outs() << " + " << format("0x%" PRIx64, op.attributes);
5092     } else
5093       outs() << format("0x%" PRIx64, op.attributes);
5094     name = get_pointer_64(op.attributes + n_value, xoffset, left, xS, info);
5095     if (name != nullptr)
5096       outs() << format(" %.*s", left, name);
5097     outs() << "\n";
5098 
5099     p += sizeof(struct objc_property64);
5100     offset += sizeof(struct objc_property64);
5101   }
5102 }
5103 
print_objc_property_list32(uint32_t p,struct DisassembleInfo * info)5104 static void print_objc_property_list32(uint32_t p,
5105                                        struct DisassembleInfo *info) {
5106   struct objc_property_list32 opl;
5107   struct objc_property32 op;
5108   const char *r;
5109   uint32_t offset, xoffset, left, j;
5110   SectionRef S, xS;
5111   const char *name;
5112 
5113   r = get_pointer_32(p, offset, left, S, info);
5114   if (r == nullptr)
5115     return;
5116   memset(&opl, '\0', sizeof(struct objc_property_list32));
5117   if (left < sizeof(struct objc_property_list32)) {
5118     memcpy(&opl, r, left);
5119     outs() << "   (objc_property_list entends past the end of the section)\n";
5120   } else
5121     memcpy(&opl, r, sizeof(struct objc_property_list32));
5122   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5123     swapStruct(opl);
5124   outs() << "                    entsize " << opl.entsize << "\n";
5125   outs() << "                      count " << opl.count << "\n";
5126 
5127   p += sizeof(struct objc_property_list32);
5128   offset += sizeof(struct objc_property_list32);
5129   for (j = 0; j < opl.count; j++) {
5130     r = get_pointer_32(p, offset, left, S, info);
5131     if (r == nullptr)
5132       return;
5133     memset(&op, '\0', sizeof(struct objc_property32));
5134     if (left < sizeof(struct objc_property32)) {
5135       memcpy(&op, r, left);
5136       outs() << "   (objc_property entends past the end of the section)\n";
5137     } else
5138       memcpy(&op, r, sizeof(struct objc_property32));
5139     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5140       swapStruct(op);
5141 
5142     outs() << "\t\t\t     name " << format("0x%" PRIx32, op.name);
5143     name = get_pointer_32(op.name, xoffset, left, xS, info);
5144     if (name != nullptr)
5145       outs() << format(" %.*s", left, name);
5146     outs() << "\n";
5147 
5148     outs() << "\t\t\tattributes " << format("0x%" PRIx32, op.attributes);
5149     name = get_pointer_32(op.attributes, xoffset, left, xS, info);
5150     if (name != nullptr)
5151       outs() << format(" %.*s", left, name);
5152     outs() << "\n";
5153 
5154     p += sizeof(struct objc_property32);
5155     offset += sizeof(struct objc_property32);
5156   }
5157 }
5158 
print_class_ro64_t(uint64_t p,struct DisassembleInfo * info,bool & is_meta_class)5159 static bool print_class_ro64_t(uint64_t p, struct DisassembleInfo *info,
5160                                bool &is_meta_class) {
5161   struct class_ro64_t cro;
5162   const char *r;
5163   uint32_t offset, xoffset, left;
5164   SectionRef S, xS;
5165   const char *name, *sym_name;
5166   uint64_t n_value;
5167 
5168   r = get_pointer_64(p, offset, left, S, info);
5169   if (r == nullptr || left < sizeof(struct class_ro64_t))
5170     return false;
5171   memcpy(&cro, r, sizeof(struct class_ro64_t));
5172   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5173     swapStruct(cro);
5174   outs() << "                    flags " << format("0x%" PRIx32, cro.flags);
5175   if (cro.flags & RO_META)
5176     outs() << " RO_META";
5177   if (cro.flags & RO_ROOT)
5178     outs() << " RO_ROOT";
5179   if (cro.flags & RO_HAS_CXX_STRUCTORS)
5180     outs() << " RO_HAS_CXX_STRUCTORS";
5181   outs() << "\n";
5182   outs() << "            instanceStart " << cro.instanceStart << "\n";
5183   outs() << "             instanceSize " << cro.instanceSize << "\n";
5184   outs() << "                 reserved " << format("0x%" PRIx32, cro.reserved)
5185          << "\n";
5186   outs() << "               ivarLayout " << format("0x%" PRIx64, cro.ivarLayout)
5187          << "\n";
5188   print_layout_map64(cro.ivarLayout, info);
5189 
5190   outs() << "                     name ";
5191   sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, name), S,
5192                            info, n_value, cro.name);
5193   if (n_value != 0) {
5194     if (info->verbose && sym_name != nullptr)
5195       outs() << sym_name;
5196     else
5197       outs() << format("0x%" PRIx64, n_value);
5198     if (cro.name != 0)
5199       outs() << " + " << format("0x%" PRIx64, cro.name);
5200   } else
5201     outs() << format("0x%" PRIx64, cro.name);
5202   name = get_pointer_64(cro.name + n_value, xoffset, left, xS, info);
5203   if (name != nullptr)
5204     outs() << format(" %.*s", left, name);
5205   outs() << "\n";
5206 
5207   outs() << "              baseMethods ";
5208   sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, baseMethods),
5209                            S, info, n_value, cro.baseMethods);
5210   if (n_value != 0) {
5211     if (info->verbose && sym_name != nullptr)
5212       outs() << sym_name;
5213     else
5214       outs() << format("0x%" PRIx64, n_value);
5215     if (cro.baseMethods != 0)
5216       outs() << " + " << format("0x%" PRIx64, cro.baseMethods);
5217   } else
5218     outs() << format("0x%" PRIx64, cro.baseMethods);
5219   outs() << " (struct method_list_t *)\n";
5220   if (cro.baseMethods + n_value != 0)
5221     print_method_list64_t(cro.baseMethods + n_value, info, "");
5222 
5223   outs() << "            baseProtocols ";
5224   sym_name =
5225       get_symbol_64(offset + offsetof(struct class_ro64_t, baseProtocols), S,
5226                     info, n_value, cro.baseProtocols);
5227   if (n_value != 0) {
5228     if (info->verbose && sym_name != nullptr)
5229       outs() << sym_name;
5230     else
5231       outs() << format("0x%" PRIx64, n_value);
5232     if (cro.baseProtocols != 0)
5233       outs() << " + " << format("0x%" PRIx64, cro.baseProtocols);
5234   } else
5235     outs() << format("0x%" PRIx64, cro.baseProtocols);
5236   outs() << "\n";
5237   if (cro.baseProtocols + n_value != 0)
5238     print_protocol_list64_t(cro.baseProtocols + n_value, info);
5239 
5240   outs() << "                    ivars ";
5241   sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, ivars), S,
5242                            info, n_value, cro.ivars);
5243   if (n_value != 0) {
5244     if (info->verbose && sym_name != nullptr)
5245       outs() << sym_name;
5246     else
5247       outs() << format("0x%" PRIx64, n_value);
5248     if (cro.ivars != 0)
5249       outs() << " + " << format("0x%" PRIx64, cro.ivars);
5250   } else
5251     outs() << format("0x%" PRIx64, cro.ivars);
5252   outs() << "\n";
5253   if (cro.ivars + n_value != 0)
5254     print_ivar_list64_t(cro.ivars + n_value, info);
5255 
5256   outs() << "           weakIvarLayout ";
5257   sym_name =
5258       get_symbol_64(offset + offsetof(struct class_ro64_t, weakIvarLayout), S,
5259                     info, n_value, cro.weakIvarLayout);
5260   if (n_value != 0) {
5261     if (info->verbose && sym_name != nullptr)
5262       outs() << sym_name;
5263     else
5264       outs() << format("0x%" PRIx64, n_value);
5265     if (cro.weakIvarLayout != 0)
5266       outs() << " + " << format("0x%" PRIx64, cro.weakIvarLayout);
5267   } else
5268     outs() << format("0x%" PRIx64, cro.weakIvarLayout);
5269   outs() << "\n";
5270   print_layout_map64(cro.weakIvarLayout + n_value, info);
5271 
5272   outs() << "           baseProperties ";
5273   sym_name =
5274       get_symbol_64(offset + offsetof(struct class_ro64_t, baseProperties), S,
5275                     info, n_value, cro.baseProperties);
5276   if (n_value != 0) {
5277     if (info->verbose && sym_name != nullptr)
5278       outs() << sym_name;
5279     else
5280       outs() << format("0x%" PRIx64, n_value);
5281     if (cro.baseProperties != 0)
5282       outs() << " + " << format("0x%" PRIx64, cro.baseProperties);
5283   } else
5284     outs() << format("0x%" PRIx64, cro.baseProperties);
5285   outs() << "\n";
5286   if (cro.baseProperties + n_value != 0)
5287     print_objc_property_list64(cro.baseProperties + n_value, info);
5288 
5289   is_meta_class = (cro.flags & RO_META) != 0;
5290   return true;
5291 }
5292 
print_class_ro32_t(uint32_t p,struct DisassembleInfo * info,bool & is_meta_class)5293 static bool print_class_ro32_t(uint32_t p, struct DisassembleInfo *info,
5294                                bool &is_meta_class) {
5295   struct class_ro32_t cro;
5296   const char *r;
5297   uint32_t offset, xoffset, left;
5298   SectionRef S, xS;
5299   const char *name;
5300 
5301   r = get_pointer_32(p, offset, left, S, info);
5302   if (r == nullptr)
5303     return false;
5304   memset(&cro, '\0', sizeof(struct class_ro32_t));
5305   if (left < sizeof(struct class_ro32_t)) {
5306     memcpy(&cro, r, left);
5307     outs() << "   (class_ro_t entends past the end of the section)\n";
5308   } else
5309     memcpy(&cro, r, sizeof(struct class_ro32_t));
5310   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5311     swapStruct(cro);
5312   outs() << "                    flags " << format("0x%" PRIx32, cro.flags);
5313   if (cro.flags & RO_META)
5314     outs() << " RO_META";
5315   if (cro.flags & RO_ROOT)
5316     outs() << " RO_ROOT";
5317   if (cro.flags & RO_HAS_CXX_STRUCTORS)
5318     outs() << " RO_HAS_CXX_STRUCTORS";
5319   outs() << "\n";
5320   outs() << "            instanceStart " << cro.instanceStart << "\n";
5321   outs() << "             instanceSize " << cro.instanceSize << "\n";
5322   outs() << "               ivarLayout " << format("0x%" PRIx32, cro.ivarLayout)
5323          << "\n";
5324   print_layout_map32(cro.ivarLayout, info);
5325 
5326   outs() << "                     name " << format("0x%" PRIx32, cro.name);
5327   name = get_pointer_32(cro.name, xoffset, left, xS, info);
5328   if (name != nullptr)
5329     outs() << format(" %.*s", left, name);
5330   outs() << "\n";
5331 
5332   outs() << "              baseMethods "
5333          << format("0x%" PRIx32, cro.baseMethods)
5334          << " (struct method_list_t *)\n";
5335   if (cro.baseMethods != 0)
5336     print_method_list32_t(cro.baseMethods, info, "");
5337 
5338   outs() << "            baseProtocols "
5339          << format("0x%" PRIx32, cro.baseProtocols) << "\n";
5340   if (cro.baseProtocols != 0)
5341     print_protocol_list32_t(cro.baseProtocols, info);
5342   outs() << "                    ivars " << format("0x%" PRIx32, cro.ivars)
5343          << "\n";
5344   if (cro.ivars != 0)
5345     print_ivar_list32_t(cro.ivars, info);
5346   outs() << "           weakIvarLayout "
5347          << format("0x%" PRIx32, cro.weakIvarLayout) << "\n";
5348   print_layout_map32(cro.weakIvarLayout, info);
5349   outs() << "           baseProperties "
5350          << format("0x%" PRIx32, cro.baseProperties) << "\n";
5351   if (cro.baseProperties != 0)
5352     print_objc_property_list32(cro.baseProperties, info);
5353   is_meta_class = (cro.flags & RO_META) != 0;
5354   return true;
5355 }
5356 
print_class64_t(uint64_t p,struct DisassembleInfo * info)5357 static void print_class64_t(uint64_t p, struct DisassembleInfo *info) {
5358   struct class64_t c;
5359   const char *r;
5360   uint32_t offset, left;
5361   SectionRef S;
5362   const char *name;
5363   uint64_t isa_n_value, n_value;
5364 
5365   r = get_pointer_64(p, offset, left, S, info);
5366   if (r == nullptr || left < sizeof(struct class64_t))
5367     return;
5368   memcpy(&c, r, sizeof(struct class64_t));
5369   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5370     swapStruct(c);
5371 
5372   outs() << "           isa " << format("0x%" PRIx64, c.isa);
5373   name = get_symbol_64(offset + offsetof(struct class64_t, isa), S, info,
5374                        isa_n_value, c.isa);
5375   if (name != nullptr)
5376     outs() << " " << name;
5377   outs() << "\n";
5378 
5379   outs() << "    superclass " << format("0x%" PRIx64, c.superclass);
5380   name = get_symbol_64(offset + offsetof(struct class64_t, superclass), S, info,
5381                        n_value, c.superclass);
5382   if (name != nullptr)
5383     outs() << " " << name;
5384   else {
5385     name = get_dyld_bind_info_symbolname(S.getAddress() +
5386              offset + offsetof(struct class64_t, superclass), info);
5387     if (name != nullptr)
5388       outs() << " " << name;
5389   }
5390   outs() << "\n";
5391 
5392   outs() << "         cache " << format("0x%" PRIx64, c.cache);
5393   name = get_symbol_64(offset + offsetof(struct class64_t, cache), S, info,
5394                        n_value, c.cache);
5395   if (name != nullptr)
5396     outs() << " " << name;
5397   outs() << "\n";
5398 
5399   outs() << "        vtable " << format("0x%" PRIx64, c.vtable);
5400   name = get_symbol_64(offset + offsetof(struct class64_t, vtable), S, info,
5401                        n_value, c.vtable);
5402   if (name != nullptr)
5403     outs() << " " << name;
5404   outs() << "\n";
5405 
5406   name = get_symbol_64(offset + offsetof(struct class64_t, data), S, info,
5407                        n_value, c.data);
5408   outs() << "          data ";
5409   if (n_value != 0) {
5410     if (info->verbose && name != nullptr)
5411       outs() << name;
5412     else
5413       outs() << format("0x%" PRIx64, n_value);
5414     if (c.data != 0)
5415       outs() << " + " << format("0x%" PRIx64, c.data);
5416   } else
5417     outs() << format("0x%" PRIx64, c.data);
5418   outs() << " (struct class_ro_t *)";
5419 
5420   // This is a Swift class if some of the low bits of the pointer are set.
5421   if ((c.data + n_value) & 0x7)
5422     outs() << " Swift class";
5423   outs() << "\n";
5424   bool is_meta_class;
5425   if (!print_class_ro64_t((c.data + n_value) & ~0x7, info, is_meta_class))
5426     return;
5427 
5428   if (!is_meta_class &&
5429       c.isa + isa_n_value != p &&
5430       c.isa + isa_n_value != 0 &&
5431       info->depth < 100) {
5432       info->depth++;
5433       outs() << "Meta Class\n";
5434       print_class64_t(c.isa + isa_n_value, info);
5435   }
5436 }
5437 
print_class32_t(uint32_t p,struct DisassembleInfo * info)5438 static void print_class32_t(uint32_t p, struct DisassembleInfo *info) {
5439   struct class32_t c;
5440   const char *r;
5441   uint32_t offset, left;
5442   SectionRef S;
5443   const char *name;
5444 
5445   r = get_pointer_32(p, offset, left, S, info);
5446   if (r == nullptr)
5447     return;
5448   memset(&c, '\0', sizeof(struct class32_t));
5449   if (left < sizeof(struct class32_t)) {
5450     memcpy(&c, r, left);
5451     outs() << "   (class_t entends past the end of the section)\n";
5452   } else
5453     memcpy(&c, r, sizeof(struct class32_t));
5454   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5455     swapStruct(c);
5456 
5457   outs() << "           isa " << format("0x%" PRIx32, c.isa);
5458   name =
5459       get_symbol_32(offset + offsetof(struct class32_t, isa), S, info, c.isa);
5460   if (name != nullptr)
5461     outs() << " " << name;
5462   outs() << "\n";
5463 
5464   outs() << "    superclass " << format("0x%" PRIx32, c.superclass);
5465   name = get_symbol_32(offset + offsetof(struct class32_t, superclass), S, info,
5466                        c.superclass);
5467   if (name != nullptr)
5468     outs() << " " << name;
5469   outs() << "\n";
5470 
5471   outs() << "         cache " << format("0x%" PRIx32, c.cache);
5472   name = get_symbol_32(offset + offsetof(struct class32_t, cache), S, info,
5473                        c.cache);
5474   if (name != nullptr)
5475     outs() << " " << name;
5476   outs() << "\n";
5477 
5478   outs() << "        vtable " << format("0x%" PRIx32, c.vtable);
5479   name = get_symbol_32(offset + offsetof(struct class32_t, vtable), S, info,
5480                        c.vtable);
5481   if (name != nullptr)
5482     outs() << " " << name;
5483   outs() << "\n";
5484 
5485   name =
5486       get_symbol_32(offset + offsetof(struct class32_t, data), S, info, c.data);
5487   outs() << "          data " << format("0x%" PRIx32, c.data)
5488          << " (struct class_ro_t *)";
5489 
5490   // This is a Swift class if some of the low bits of the pointer are set.
5491   if (c.data & 0x3)
5492     outs() << " Swift class";
5493   outs() << "\n";
5494   bool is_meta_class;
5495   if (!print_class_ro32_t(c.data & ~0x3, info, is_meta_class))
5496     return;
5497 
5498   if (!is_meta_class) {
5499     outs() << "Meta Class\n";
5500     print_class32_t(c.isa, info);
5501   }
5502 }
5503 
print_objc_class_t(struct objc_class_t * objc_class,struct DisassembleInfo * info)5504 static void print_objc_class_t(struct objc_class_t *objc_class,
5505                                struct DisassembleInfo *info) {
5506   uint32_t offset, left, xleft;
5507   const char *name, *p, *ivar_list;
5508   SectionRef S;
5509   int32_t i;
5510   struct objc_ivar_list_t objc_ivar_list;
5511   struct objc_ivar_t ivar;
5512 
5513   outs() << "\t\t      isa " << format("0x%08" PRIx32, objc_class->isa);
5514   if (info->verbose && CLS_GETINFO(objc_class, CLS_META)) {
5515     name = get_pointer_32(objc_class->isa, offset, left, S, info, true);
5516     if (name != nullptr)
5517       outs() << format(" %.*s", left, name);
5518     else
5519       outs() << " (not in an __OBJC section)";
5520   }
5521   outs() << "\n";
5522 
5523   outs() << "\t      super_class "
5524          << format("0x%08" PRIx32, objc_class->super_class);
5525   if (info->verbose) {
5526     name = get_pointer_32(objc_class->super_class, offset, left, S, info, true);
5527     if (name != nullptr)
5528       outs() << format(" %.*s", left, name);
5529     else
5530       outs() << " (not in an __OBJC section)";
5531   }
5532   outs() << "\n";
5533 
5534   outs() << "\t\t     name " << format("0x%08" PRIx32, objc_class->name);
5535   if (info->verbose) {
5536     name = get_pointer_32(objc_class->name, offset, left, S, info, true);
5537     if (name != nullptr)
5538       outs() << format(" %.*s", left, name);
5539     else
5540       outs() << " (not in an __OBJC section)";
5541   }
5542   outs() << "\n";
5543 
5544   outs() << "\t\t  version " << format("0x%08" PRIx32, objc_class->version)
5545          << "\n";
5546 
5547   outs() << "\t\t     info " << format("0x%08" PRIx32, objc_class->info);
5548   if (info->verbose) {
5549     if (CLS_GETINFO(objc_class, CLS_CLASS))
5550       outs() << " CLS_CLASS";
5551     else if (CLS_GETINFO(objc_class, CLS_META))
5552       outs() << " CLS_META";
5553   }
5554   outs() << "\n";
5555 
5556   outs() << "\t    instance_size "
5557          << format("0x%08" PRIx32, objc_class->instance_size) << "\n";
5558 
5559   p = get_pointer_32(objc_class->ivars, offset, left, S, info, true);
5560   outs() << "\t\t    ivars " << format("0x%08" PRIx32, objc_class->ivars);
5561   if (p != nullptr) {
5562     if (left > sizeof(struct objc_ivar_list_t)) {
5563       outs() << "\n";
5564       memcpy(&objc_ivar_list, p, sizeof(struct objc_ivar_list_t));
5565     } else {
5566       outs() << " (entends past the end of the section)\n";
5567       memset(&objc_ivar_list, '\0', sizeof(struct objc_ivar_list_t));
5568       memcpy(&objc_ivar_list, p, left);
5569     }
5570     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5571       swapStruct(objc_ivar_list);
5572     outs() << "\t\t       ivar_count " << objc_ivar_list.ivar_count << "\n";
5573     ivar_list = p + sizeof(struct objc_ivar_list_t);
5574     for (i = 0; i < objc_ivar_list.ivar_count; i++) {
5575       if ((i + 1) * sizeof(struct objc_ivar_t) > left) {
5576         outs() << "\t\t remaining ivar's extend past the of the section\n";
5577         break;
5578       }
5579       memcpy(&ivar, ivar_list + i * sizeof(struct objc_ivar_t),
5580              sizeof(struct objc_ivar_t));
5581       if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5582         swapStruct(ivar);
5583 
5584       outs() << "\t\t\tivar_name " << format("0x%08" PRIx32, ivar.ivar_name);
5585       if (info->verbose) {
5586         name = get_pointer_32(ivar.ivar_name, offset, xleft, S, info, true);
5587         if (name != nullptr)
5588           outs() << format(" %.*s", xleft, name);
5589         else
5590           outs() << " (not in an __OBJC section)";
5591       }
5592       outs() << "\n";
5593 
5594       outs() << "\t\t\tivar_type " << format("0x%08" PRIx32, ivar.ivar_type);
5595       if (info->verbose) {
5596         name = get_pointer_32(ivar.ivar_type, offset, xleft, S, info, true);
5597         if (name != nullptr)
5598           outs() << format(" %.*s", xleft, name);
5599         else
5600           outs() << " (not in an __OBJC section)";
5601       }
5602       outs() << "\n";
5603 
5604       outs() << "\t\t      ivar_offset "
5605              << format("0x%08" PRIx32, ivar.ivar_offset) << "\n";
5606     }
5607   } else {
5608     outs() << " (not in an __OBJC section)\n";
5609   }
5610 
5611   outs() << "\t\t  methods " << format("0x%08" PRIx32, objc_class->methodLists);
5612   if (print_method_list(objc_class->methodLists, info))
5613     outs() << " (not in an __OBJC section)\n";
5614 
5615   outs() << "\t\t    cache " << format("0x%08" PRIx32, objc_class->cache)
5616          << "\n";
5617 
5618   outs() << "\t\tprotocols " << format("0x%08" PRIx32, objc_class->protocols);
5619   if (print_protocol_list(objc_class->protocols, 16, info))
5620     outs() << " (not in an __OBJC section)\n";
5621 }
5622 
print_objc_objc_category_t(struct objc_category_t * objc_category,struct DisassembleInfo * info)5623 static void print_objc_objc_category_t(struct objc_category_t *objc_category,
5624                                        struct DisassembleInfo *info) {
5625   uint32_t offset, left;
5626   const char *name;
5627   SectionRef S;
5628 
5629   outs() << "\t       category name "
5630          << format("0x%08" PRIx32, objc_category->category_name);
5631   if (info->verbose) {
5632     name = get_pointer_32(objc_category->category_name, offset, left, S, info,
5633                           true);
5634     if (name != nullptr)
5635       outs() << format(" %.*s", left, name);
5636     else
5637       outs() << " (not in an __OBJC section)";
5638   }
5639   outs() << "\n";
5640 
5641   outs() << "\t\t  class name "
5642          << format("0x%08" PRIx32, objc_category->class_name);
5643   if (info->verbose) {
5644     name =
5645         get_pointer_32(objc_category->class_name, offset, left, S, info, 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    instance methods "
5654          << format("0x%08" PRIx32, objc_category->instance_methods);
5655   if (print_method_list(objc_category->instance_methods, info))
5656     outs() << " (not in an __OBJC section)\n";
5657 
5658   outs() << "\t       class methods "
5659          << format("0x%08" PRIx32, objc_category->class_methods);
5660   if (print_method_list(objc_category->class_methods, info))
5661     outs() << " (not in an __OBJC section)\n";
5662 }
5663 
print_category64_t(uint64_t p,struct DisassembleInfo * info)5664 static void print_category64_t(uint64_t p, struct DisassembleInfo *info) {
5665   struct category64_t c;
5666   const char *r;
5667   uint32_t offset, xoffset, left;
5668   SectionRef S, xS;
5669   const char *name, *sym_name;
5670   uint64_t n_value;
5671 
5672   r = get_pointer_64(p, offset, left, S, info);
5673   if (r == nullptr)
5674     return;
5675   memset(&c, '\0', sizeof(struct category64_t));
5676   if (left < sizeof(struct category64_t)) {
5677     memcpy(&c, r, left);
5678     outs() << "   (category_t entends past the end of the section)\n";
5679   } else
5680     memcpy(&c, r, sizeof(struct category64_t));
5681   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5682     swapStruct(c);
5683 
5684   outs() << "              name ";
5685   sym_name = get_symbol_64(offset + offsetof(struct category64_t, name), S,
5686                            info, n_value, c.name);
5687   if (n_value != 0) {
5688     if (info->verbose && sym_name != nullptr)
5689       outs() << sym_name;
5690     else
5691       outs() << format("0x%" PRIx64, n_value);
5692     if (c.name != 0)
5693       outs() << " + " << format("0x%" PRIx64, c.name);
5694   } else
5695     outs() << format("0x%" PRIx64, c.name);
5696   name = get_pointer_64(c.name + n_value, xoffset, left, xS, info);
5697   if (name != nullptr)
5698     outs() << format(" %.*s", left, name);
5699   outs() << "\n";
5700 
5701   outs() << "               cls ";
5702   sym_name = get_symbol_64(offset + offsetof(struct category64_t, cls), S, info,
5703                            n_value, c.cls);
5704   if (n_value != 0) {
5705     if (info->verbose && sym_name != nullptr)
5706       outs() << sym_name;
5707     else
5708       outs() << format("0x%" PRIx64, n_value);
5709     if (c.cls != 0)
5710       outs() << " + " << format("0x%" PRIx64, c.cls);
5711   } else
5712     outs() << format("0x%" PRIx64, c.cls);
5713   outs() << "\n";
5714   if (c.cls + n_value != 0)
5715     print_class64_t(c.cls + n_value, info);
5716 
5717   outs() << "   instanceMethods ";
5718   sym_name =
5719       get_symbol_64(offset + offsetof(struct category64_t, instanceMethods), S,
5720                     info, n_value, c.instanceMethods);
5721   if (n_value != 0) {
5722     if (info->verbose && sym_name != nullptr)
5723       outs() << sym_name;
5724     else
5725       outs() << format("0x%" PRIx64, n_value);
5726     if (c.instanceMethods != 0)
5727       outs() << " + " << format("0x%" PRIx64, c.instanceMethods);
5728   } else
5729     outs() << format("0x%" PRIx64, c.instanceMethods);
5730   outs() << "\n";
5731   if (c.instanceMethods + n_value != 0)
5732     print_method_list64_t(c.instanceMethods + n_value, info, "");
5733 
5734   outs() << "      classMethods ";
5735   sym_name = get_symbol_64(offset + offsetof(struct category64_t, classMethods),
5736                            S, info, n_value, c.classMethods);
5737   if (n_value != 0) {
5738     if (info->verbose && sym_name != nullptr)
5739       outs() << sym_name;
5740     else
5741       outs() << format("0x%" PRIx64, n_value);
5742     if (c.classMethods != 0)
5743       outs() << " + " << format("0x%" PRIx64, c.classMethods);
5744   } else
5745     outs() << format("0x%" PRIx64, c.classMethods);
5746   outs() << "\n";
5747   if (c.classMethods + n_value != 0)
5748     print_method_list64_t(c.classMethods + n_value, info, "");
5749 
5750   outs() << "         protocols ";
5751   sym_name = get_symbol_64(offset + offsetof(struct category64_t, protocols), S,
5752                            info, n_value, c.protocols);
5753   if (n_value != 0) {
5754     if (info->verbose && sym_name != nullptr)
5755       outs() << sym_name;
5756     else
5757       outs() << format("0x%" PRIx64, n_value);
5758     if (c.protocols != 0)
5759       outs() << " + " << format("0x%" PRIx64, c.protocols);
5760   } else
5761     outs() << format("0x%" PRIx64, c.protocols);
5762   outs() << "\n";
5763   if (c.protocols + n_value != 0)
5764     print_protocol_list64_t(c.protocols + n_value, info);
5765 
5766   outs() << "instanceProperties ";
5767   sym_name =
5768       get_symbol_64(offset + offsetof(struct category64_t, instanceProperties),
5769                     S, info, n_value, c.instanceProperties);
5770   if (n_value != 0) {
5771     if (info->verbose && sym_name != nullptr)
5772       outs() << sym_name;
5773     else
5774       outs() << format("0x%" PRIx64, n_value);
5775     if (c.instanceProperties != 0)
5776       outs() << " + " << format("0x%" PRIx64, c.instanceProperties);
5777   } else
5778     outs() << format("0x%" PRIx64, c.instanceProperties);
5779   outs() << "\n";
5780   if (c.instanceProperties + n_value != 0)
5781     print_objc_property_list64(c.instanceProperties + n_value, info);
5782 }
5783 
print_category32_t(uint32_t p,struct DisassembleInfo * info)5784 static void print_category32_t(uint32_t p, struct DisassembleInfo *info) {
5785   struct category32_t c;
5786   const char *r;
5787   uint32_t offset, left;
5788   SectionRef S, xS;
5789   const char *name;
5790 
5791   r = get_pointer_32(p, offset, left, S, info);
5792   if (r == nullptr)
5793     return;
5794   memset(&c, '\0', sizeof(struct category32_t));
5795   if (left < sizeof(struct category32_t)) {
5796     memcpy(&c, r, left);
5797     outs() << "   (category_t entends past the end of the section)\n";
5798   } else
5799     memcpy(&c, r, sizeof(struct category32_t));
5800   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5801     swapStruct(c);
5802 
5803   outs() << "              name " << format("0x%" PRIx32, c.name);
5804   name = get_symbol_32(offset + offsetof(struct category32_t, name), S, info,
5805                        c.name);
5806   if (name)
5807     outs() << " " << name;
5808   outs() << "\n";
5809 
5810   outs() << "               cls " << format("0x%" PRIx32, c.cls) << "\n";
5811   if (c.cls != 0)
5812     print_class32_t(c.cls, info);
5813   outs() << "   instanceMethods " << format("0x%" PRIx32, c.instanceMethods)
5814          << "\n";
5815   if (c.instanceMethods != 0)
5816     print_method_list32_t(c.instanceMethods, info, "");
5817   outs() << "      classMethods " << format("0x%" PRIx32, c.classMethods)
5818          << "\n";
5819   if (c.classMethods != 0)
5820     print_method_list32_t(c.classMethods, info, "");
5821   outs() << "         protocols " << format("0x%" PRIx32, c.protocols) << "\n";
5822   if (c.protocols != 0)
5823     print_protocol_list32_t(c.protocols, info);
5824   outs() << "instanceProperties " << format("0x%" PRIx32, c.instanceProperties)
5825          << "\n";
5826   if (c.instanceProperties != 0)
5827     print_objc_property_list32(c.instanceProperties, info);
5828 }
5829 
print_message_refs64(SectionRef S,struct DisassembleInfo * info)5830 static void print_message_refs64(SectionRef S, struct DisassembleInfo *info) {
5831   uint32_t i, left, offset, xoffset;
5832   uint64_t p, n_value;
5833   struct message_ref64 mr;
5834   const char *name, *sym_name;
5835   const char *r;
5836   SectionRef xS;
5837 
5838   if (S == SectionRef())
5839     return;
5840 
5841   StringRef SectName;
5842   Expected<StringRef> SecNameOrErr = S.getName();
5843   if (SecNameOrErr)
5844     SectName = *SecNameOrErr;
5845   else
5846     consumeError(SecNameOrErr.takeError());
5847 
5848   DataRefImpl Ref = S.getRawDataRefImpl();
5849   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5850   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5851   offset = 0;
5852   for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) {
5853     p = S.getAddress() + i;
5854     r = get_pointer_64(p, offset, left, S, info);
5855     if (r == nullptr)
5856       return;
5857     memset(&mr, '\0', sizeof(struct message_ref64));
5858     if (left < sizeof(struct message_ref64)) {
5859       memcpy(&mr, r, left);
5860       outs() << "   (message_ref entends past the end of the section)\n";
5861     } else
5862       memcpy(&mr, r, sizeof(struct message_ref64));
5863     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5864       swapStruct(mr);
5865 
5866     outs() << "  imp ";
5867     name = get_symbol_64(offset + offsetof(struct message_ref64, imp), S, info,
5868                          n_value, mr.imp);
5869     if (n_value != 0) {
5870       outs() << format("0x%" PRIx64, n_value) << " ";
5871       if (mr.imp != 0)
5872         outs() << "+ " << format("0x%" PRIx64, mr.imp) << " ";
5873     } else
5874       outs() << format("0x%" PRIx64, mr.imp) << " ";
5875     if (name != nullptr)
5876       outs() << " " << name;
5877     outs() << "\n";
5878 
5879     outs() << "  sel ";
5880     sym_name = get_symbol_64(offset + offsetof(struct message_ref64, sel), S,
5881                              info, n_value, mr.sel);
5882     if (n_value != 0) {
5883       if (info->verbose && sym_name != nullptr)
5884         outs() << sym_name;
5885       else
5886         outs() << format("0x%" PRIx64, n_value);
5887       if (mr.sel != 0)
5888         outs() << " + " << format("0x%" PRIx64, mr.sel);
5889     } else
5890       outs() << format("0x%" PRIx64, mr.sel);
5891     name = get_pointer_64(mr.sel + n_value, xoffset, left, xS, info);
5892     if (name != nullptr)
5893       outs() << format(" %.*s", left, name);
5894     outs() << "\n";
5895 
5896     offset += sizeof(struct message_ref64);
5897   }
5898 }
5899 
print_message_refs32(SectionRef S,struct DisassembleInfo * info)5900 static void print_message_refs32(SectionRef S, struct DisassembleInfo *info) {
5901   uint32_t i, left, offset, xoffset, p;
5902   struct message_ref32 mr;
5903   const char *name, *r;
5904   SectionRef xS;
5905 
5906   if (S == SectionRef())
5907     return;
5908 
5909   StringRef SectName;
5910   Expected<StringRef> SecNameOrErr = S.getName();
5911   if (SecNameOrErr)
5912     SectName = *SecNameOrErr;
5913   else
5914     consumeError(SecNameOrErr.takeError());
5915 
5916   DataRefImpl Ref = S.getRawDataRefImpl();
5917   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5918   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5919   offset = 0;
5920   for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) {
5921     p = S.getAddress() + i;
5922     r = get_pointer_32(p, offset, left, S, info);
5923     if (r == nullptr)
5924       return;
5925     memset(&mr, '\0', sizeof(struct message_ref32));
5926     if (left < sizeof(struct message_ref32)) {
5927       memcpy(&mr, r, left);
5928       outs() << "   (message_ref entends past the end of the section)\n";
5929     } else
5930       memcpy(&mr, r, sizeof(struct message_ref32));
5931     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5932       swapStruct(mr);
5933 
5934     outs() << "  imp " << format("0x%" PRIx32, mr.imp);
5935     name = get_symbol_32(offset + offsetof(struct message_ref32, imp), S, info,
5936                          mr.imp);
5937     if (name != nullptr)
5938       outs() << " " << name;
5939     outs() << "\n";
5940 
5941     outs() << "  sel " << format("0x%" PRIx32, mr.sel);
5942     name = get_pointer_32(mr.sel, xoffset, left, xS, info);
5943     if (name != nullptr)
5944       outs() << " " << name;
5945     outs() << "\n";
5946 
5947     offset += sizeof(struct message_ref32);
5948   }
5949 }
5950 
print_image_info64(SectionRef S,struct DisassembleInfo * info)5951 static void print_image_info64(SectionRef S, struct DisassembleInfo *info) {
5952   uint32_t left, offset, swift_version;
5953   uint64_t p;
5954   struct objc_image_info64 o;
5955   const char *r;
5956 
5957   if (S == SectionRef())
5958     return;
5959 
5960   StringRef SectName;
5961   Expected<StringRef> SecNameOrErr = S.getName();
5962   if (SecNameOrErr)
5963     SectName = *SecNameOrErr;
5964   else
5965     consumeError(SecNameOrErr.takeError());
5966 
5967   DataRefImpl Ref = S.getRawDataRefImpl();
5968   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5969   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5970   p = S.getAddress();
5971   r = get_pointer_64(p, offset, left, S, info);
5972   if (r == nullptr)
5973     return;
5974   memset(&o, '\0', sizeof(struct objc_image_info64));
5975   if (left < sizeof(struct objc_image_info64)) {
5976     memcpy(&o, r, left);
5977     outs() << "   (objc_image_info entends past the end of the section)\n";
5978   } else
5979     memcpy(&o, r, sizeof(struct objc_image_info64));
5980   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5981     swapStruct(o);
5982   outs() << "  version " << o.version << "\n";
5983   outs() << "    flags " << format("0x%" PRIx32, o.flags);
5984   if (o.flags & OBJC_IMAGE_IS_REPLACEMENT)
5985     outs() << " OBJC_IMAGE_IS_REPLACEMENT";
5986   if (o.flags & OBJC_IMAGE_SUPPORTS_GC)
5987     outs() << " OBJC_IMAGE_SUPPORTS_GC";
5988   if (o.flags & OBJC_IMAGE_IS_SIMULATED)
5989     outs() << " OBJC_IMAGE_IS_SIMULATED";
5990   if (o.flags & OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES)
5991     outs() << " OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES";
5992   swift_version = (o.flags >> 8) & 0xff;
5993   if (swift_version != 0) {
5994     if (swift_version == 1)
5995       outs() << " Swift 1.0";
5996     else if (swift_version == 2)
5997       outs() << " Swift 1.1";
5998     else if(swift_version == 3)
5999       outs() << " Swift 2.0";
6000     else if(swift_version == 4)
6001       outs() << " Swift 3.0";
6002     else if(swift_version == 5)
6003       outs() << " Swift 4.0";
6004     else if(swift_version == 6)
6005       outs() << " Swift 4.1/Swift 4.2";
6006     else if(swift_version == 7)
6007       outs() << " Swift 5 or later";
6008     else
6009       outs() << " unknown future Swift version (" << swift_version << ")";
6010   }
6011   outs() << "\n";
6012 }
6013 
print_image_info32(SectionRef S,struct DisassembleInfo * info)6014 static void print_image_info32(SectionRef S, struct DisassembleInfo *info) {
6015   uint32_t left, offset, swift_version, p;
6016   struct objc_image_info32 o;
6017   const char *r;
6018 
6019   if (S == SectionRef())
6020     return;
6021 
6022   StringRef SectName;
6023   Expected<StringRef> SecNameOrErr = S.getName();
6024   if (SecNameOrErr)
6025     SectName = *SecNameOrErr;
6026   else
6027     consumeError(SecNameOrErr.takeError());
6028 
6029   DataRefImpl Ref = S.getRawDataRefImpl();
6030   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
6031   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
6032   p = S.getAddress();
6033   r = get_pointer_32(p, offset, left, S, info);
6034   if (r == nullptr)
6035     return;
6036   memset(&o, '\0', sizeof(struct objc_image_info32));
6037   if (left < sizeof(struct objc_image_info32)) {
6038     memcpy(&o, r, left);
6039     outs() << "   (objc_image_info entends past the end of the section)\n";
6040   } else
6041     memcpy(&o, r, sizeof(struct objc_image_info32));
6042   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
6043     swapStruct(o);
6044   outs() << "  version " << o.version << "\n";
6045   outs() << "    flags " << format("0x%" PRIx32, o.flags);
6046   if (o.flags & OBJC_IMAGE_IS_REPLACEMENT)
6047     outs() << " OBJC_IMAGE_IS_REPLACEMENT";
6048   if (o.flags & OBJC_IMAGE_SUPPORTS_GC)
6049     outs() << " OBJC_IMAGE_SUPPORTS_GC";
6050   swift_version = (o.flags >> 8) & 0xff;
6051   if (swift_version != 0) {
6052     if (swift_version == 1)
6053       outs() << " Swift 1.0";
6054     else if (swift_version == 2)
6055       outs() << " Swift 1.1";
6056     else if(swift_version == 3)
6057       outs() << " Swift 2.0";
6058     else if(swift_version == 4)
6059       outs() << " Swift 3.0";
6060     else if(swift_version == 5)
6061       outs() << " Swift 4.0";
6062     else if(swift_version == 6)
6063       outs() << " Swift 4.1/Swift 4.2";
6064     else if(swift_version == 7)
6065       outs() << " Swift 5 or later";
6066     else
6067       outs() << " unknown future Swift version (" << swift_version << ")";
6068   }
6069   outs() << "\n";
6070 }
6071 
print_image_info(SectionRef S,struct DisassembleInfo * info)6072 static void print_image_info(SectionRef S, struct DisassembleInfo *info) {
6073   uint32_t left, offset, p;
6074   struct imageInfo_t o;
6075   const char *r;
6076 
6077   StringRef SectName;
6078   Expected<StringRef> SecNameOrErr = S.getName();
6079   if (SecNameOrErr)
6080     SectName = *SecNameOrErr;
6081   else
6082     consumeError(SecNameOrErr.takeError());
6083 
6084   DataRefImpl Ref = S.getRawDataRefImpl();
6085   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
6086   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
6087   p = S.getAddress();
6088   r = get_pointer_32(p, offset, left, S, info);
6089   if (r == nullptr)
6090     return;
6091   memset(&o, '\0', sizeof(struct imageInfo_t));
6092   if (left < sizeof(struct imageInfo_t)) {
6093     memcpy(&o, r, left);
6094     outs() << " (imageInfo entends past the end of the section)\n";
6095   } else
6096     memcpy(&o, r, sizeof(struct imageInfo_t));
6097   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
6098     swapStruct(o);
6099   outs() << "  version " << o.version << "\n";
6100   outs() << "    flags " << format("0x%" PRIx32, o.flags);
6101   if (o.flags & 0x1)
6102     outs() << "  F&C";
6103   if (o.flags & 0x2)
6104     outs() << " GC";
6105   if (o.flags & 0x4)
6106     outs() << " GC-only";
6107   else
6108     outs() << " RR";
6109   outs() << "\n";
6110 }
6111 
printObjc2_64bit_MetaData(MachOObjectFile * O,bool verbose)6112 static void printObjc2_64bit_MetaData(MachOObjectFile *O, bool verbose) {
6113   SymbolAddressMap AddrMap;
6114   if (verbose)
6115     CreateSymbolAddressMap(O, &AddrMap);
6116 
6117   std::vector<SectionRef> Sections;
6118   for (const SectionRef &Section : O->sections())
6119     Sections.push_back(Section);
6120 
6121   struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6122 
6123   SectionRef CL = get_section(O, "__OBJC2", "__class_list");
6124   if (CL == SectionRef())
6125     CL = get_section(O, "__DATA", "__objc_classlist");
6126   if (CL == SectionRef())
6127     CL = get_section(O, "__DATA_CONST", "__objc_classlist");
6128   if (CL == SectionRef())
6129     CL = get_section(O, "__DATA_DIRTY", "__objc_classlist");
6130   info.S = CL;
6131   walk_pointer_list_64("class", CL, O, &info, print_class64_t);
6132 
6133   SectionRef CR = get_section(O, "__OBJC2", "__class_refs");
6134   if (CR == SectionRef())
6135     CR = get_section(O, "__DATA", "__objc_classrefs");
6136   if (CR == SectionRef())
6137     CR = get_section(O, "__DATA_CONST", "__objc_classrefs");
6138   if (CR == SectionRef())
6139     CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs");
6140   info.S = CR;
6141   walk_pointer_list_64("class refs", CR, O, &info, nullptr);
6142 
6143   SectionRef SR = get_section(O, "__OBJC2", "__super_refs");
6144   if (SR == SectionRef())
6145     SR = get_section(O, "__DATA", "__objc_superrefs");
6146   if (SR == SectionRef())
6147     SR = get_section(O, "__DATA_CONST", "__objc_superrefs");
6148   if (SR == SectionRef())
6149     SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs");
6150   info.S = SR;
6151   walk_pointer_list_64("super refs", SR, O, &info, nullptr);
6152 
6153   SectionRef CA = get_section(O, "__OBJC2", "__category_list");
6154   if (CA == SectionRef())
6155     CA = get_section(O, "__DATA", "__objc_catlist");
6156   if (CA == SectionRef())
6157     CA = get_section(O, "__DATA_CONST", "__objc_catlist");
6158   if (CA == SectionRef())
6159     CA = get_section(O, "__DATA_DIRTY", "__objc_catlist");
6160   info.S = CA;
6161   walk_pointer_list_64("category", CA, O, &info, print_category64_t);
6162 
6163   SectionRef PL = get_section(O, "__OBJC2", "__protocol_list");
6164   if (PL == SectionRef())
6165     PL = get_section(O, "__DATA", "__objc_protolist");
6166   if (PL == SectionRef())
6167     PL = get_section(O, "__DATA_CONST", "__objc_protolist");
6168   if (PL == SectionRef())
6169     PL = get_section(O, "__DATA_DIRTY", "__objc_protolist");
6170   info.S = PL;
6171   walk_pointer_list_64("protocol", PL, O, &info, nullptr);
6172 
6173   SectionRef MR = get_section(O, "__OBJC2", "__message_refs");
6174   if (MR == SectionRef())
6175     MR = get_section(O, "__DATA", "__objc_msgrefs");
6176   if (MR == SectionRef())
6177     MR = get_section(O, "__DATA_CONST", "__objc_msgrefs");
6178   if (MR == SectionRef())
6179     MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs");
6180   info.S = MR;
6181   print_message_refs64(MR, &info);
6182 
6183   SectionRef II = get_section(O, "__OBJC2", "__image_info");
6184   if (II == SectionRef())
6185     II = get_section(O, "__DATA", "__objc_imageinfo");
6186   if (II == SectionRef())
6187     II = get_section(O, "__DATA_CONST", "__objc_imageinfo");
6188   if (II == SectionRef())
6189     II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo");
6190   info.S = II;
6191   print_image_info64(II, &info);
6192 }
6193 
printObjc2_32bit_MetaData(MachOObjectFile * O,bool verbose)6194 static void printObjc2_32bit_MetaData(MachOObjectFile *O, bool verbose) {
6195   SymbolAddressMap AddrMap;
6196   if (verbose)
6197     CreateSymbolAddressMap(O, &AddrMap);
6198 
6199   std::vector<SectionRef> Sections;
6200   for (const SectionRef &Section : O->sections())
6201     Sections.push_back(Section);
6202 
6203   struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6204 
6205   SectionRef CL = get_section(O, "__OBJC2", "__class_list");
6206   if (CL == SectionRef())
6207     CL = get_section(O, "__DATA", "__objc_classlist");
6208   if (CL == SectionRef())
6209     CL = get_section(O, "__DATA_CONST", "__objc_classlist");
6210   if (CL == SectionRef())
6211     CL = get_section(O, "__DATA_DIRTY", "__objc_classlist");
6212   info.S = CL;
6213   walk_pointer_list_32("class", CL, O, &info, print_class32_t);
6214 
6215   SectionRef CR = get_section(O, "__OBJC2", "__class_refs");
6216   if (CR == SectionRef())
6217     CR = get_section(O, "__DATA", "__objc_classrefs");
6218   if (CR == SectionRef())
6219     CR = get_section(O, "__DATA_CONST", "__objc_classrefs");
6220   if (CR == SectionRef())
6221     CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs");
6222   info.S = CR;
6223   walk_pointer_list_32("class refs", CR, O, &info, nullptr);
6224 
6225   SectionRef SR = get_section(O, "__OBJC2", "__super_refs");
6226   if (SR == SectionRef())
6227     SR = get_section(O, "__DATA", "__objc_superrefs");
6228   if (SR == SectionRef())
6229     SR = get_section(O, "__DATA_CONST", "__objc_superrefs");
6230   if (SR == SectionRef())
6231     SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs");
6232   info.S = SR;
6233   walk_pointer_list_32("super refs", SR, O, &info, nullptr);
6234 
6235   SectionRef CA = get_section(O, "__OBJC2", "__category_list");
6236   if (CA == SectionRef())
6237     CA = get_section(O, "__DATA", "__objc_catlist");
6238   if (CA == SectionRef())
6239     CA = get_section(O, "__DATA_CONST", "__objc_catlist");
6240   if (CA == SectionRef())
6241     CA = get_section(O, "__DATA_DIRTY", "__objc_catlist");
6242   info.S = CA;
6243   walk_pointer_list_32("category", CA, O, &info, print_category32_t);
6244 
6245   SectionRef PL = get_section(O, "__OBJC2", "__protocol_list");
6246   if (PL == SectionRef())
6247     PL = get_section(O, "__DATA", "__objc_protolist");
6248   if (PL == SectionRef())
6249     PL = get_section(O, "__DATA_CONST", "__objc_protolist");
6250   if (PL == SectionRef())
6251     PL = get_section(O, "__DATA_DIRTY", "__objc_protolist");
6252   info.S = PL;
6253   walk_pointer_list_32("protocol", PL, O, &info, nullptr);
6254 
6255   SectionRef MR = get_section(O, "__OBJC2", "__message_refs");
6256   if (MR == SectionRef())
6257     MR = get_section(O, "__DATA", "__objc_msgrefs");
6258   if (MR == SectionRef())
6259     MR = get_section(O, "__DATA_CONST", "__objc_msgrefs");
6260   if (MR == SectionRef())
6261     MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs");
6262   info.S = MR;
6263   print_message_refs32(MR, &info);
6264 
6265   SectionRef II = get_section(O, "__OBJC2", "__image_info");
6266   if (II == SectionRef())
6267     II = get_section(O, "__DATA", "__objc_imageinfo");
6268   if (II == SectionRef())
6269     II = get_section(O, "__DATA_CONST", "__objc_imageinfo");
6270   if (II == SectionRef())
6271     II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo");
6272   info.S = II;
6273   print_image_info32(II, &info);
6274 }
6275 
printObjc1_32bit_MetaData(MachOObjectFile * O,bool verbose)6276 static bool printObjc1_32bit_MetaData(MachOObjectFile *O, bool verbose) {
6277   uint32_t i, j, p, offset, xoffset, left, defs_left, def;
6278   const char *r, *name, *defs;
6279   struct objc_module_t module;
6280   SectionRef S, xS;
6281   struct objc_symtab_t symtab;
6282   struct objc_class_t objc_class;
6283   struct objc_category_t objc_category;
6284 
6285   outs() << "Objective-C segment\n";
6286   S = get_section(O, "__OBJC", "__module_info");
6287   if (S == SectionRef())
6288     return false;
6289 
6290   SymbolAddressMap AddrMap;
6291   if (verbose)
6292     CreateSymbolAddressMap(O, &AddrMap);
6293 
6294   std::vector<SectionRef> Sections;
6295   for (const SectionRef &Section : O->sections())
6296     Sections.push_back(Section);
6297 
6298   struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6299 
6300   for (i = 0; i < S.getSize(); i += sizeof(struct objc_module_t)) {
6301     p = S.getAddress() + i;
6302     r = get_pointer_32(p, offset, left, S, &info, true);
6303     if (r == nullptr)
6304       return true;
6305     memset(&module, '\0', sizeof(struct objc_module_t));
6306     if (left < sizeof(struct objc_module_t)) {
6307       memcpy(&module, r, left);
6308       outs() << "   (module extends past end of __module_info section)\n";
6309     } else
6310       memcpy(&module, r, sizeof(struct objc_module_t));
6311     if (O->isLittleEndian() != sys::IsLittleEndianHost)
6312       swapStruct(module);
6313 
6314     outs() << "Module " << format("0x%" PRIx32, p) << "\n";
6315     outs() << "    version " << module.version << "\n";
6316     outs() << "       size " << module.size << "\n";
6317     outs() << "       name ";
6318     name = get_pointer_32(module.name, xoffset, left, xS, &info, true);
6319     if (name != nullptr)
6320       outs() << format("%.*s", left, name);
6321     else
6322       outs() << format("0x%08" PRIx32, module.name)
6323              << "(not in an __OBJC section)";
6324     outs() << "\n";
6325 
6326     r = get_pointer_32(module.symtab, xoffset, left, xS, &info, true);
6327     if (module.symtab == 0 || r == nullptr) {
6328       outs() << "     symtab " << format("0x%08" PRIx32, module.symtab)
6329              << " (not in an __OBJC section)\n";
6330       continue;
6331     }
6332     outs() << "     symtab " << format("0x%08" PRIx32, module.symtab) << "\n";
6333     memset(&symtab, '\0', sizeof(struct objc_symtab_t));
6334     defs_left = 0;
6335     defs = nullptr;
6336     if (left < sizeof(struct objc_symtab_t)) {
6337       memcpy(&symtab, r, left);
6338       outs() << "\tsymtab extends past end of an __OBJC section)\n";
6339     } else {
6340       memcpy(&symtab, r, sizeof(struct objc_symtab_t));
6341       if (left > sizeof(struct objc_symtab_t)) {
6342         defs_left = left - sizeof(struct objc_symtab_t);
6343         defs = r + sizeof(struct objc_symtab_t);
6344       }
6345     }
6346     if (O->isLittleEndian() != sys::IsLittleEndianHost)
6347       swapStruct(symtab);
6348 
6349     outs() << "\tsel_ref_cnt " << symtab.sel_ref_cnt << "\n";
6350     r = get_pointer_32(symtab.refs, xoffset, left, xS, &info, true);
6351     outs() << "\trefs " << format("0x%08" PRIx32, symtab.refs);
6352     if (r == nullptr)
6353       outs() << " (not in an __OBJC section)";
6354     outs() << "\n";
6355     outs() << "\tcls_def_cnt " << symtab.cls_def_cnt << "\n";
6356     outs() << "\tcat_def_cnt " << symtab.cat_def_cnt << "\n";
6357     if (symtab.cls_def_cnt > 0)
6358       outs() << "\tClass Definitions\n";
6359     for (j = 0; j < symtab.cls_def_cnt; j++) {
6360       if ((j + 1) * sizeof(uint32_t) > defs_left) {
6361         outs() << "\t(remaining class defs entries entends past the end of the "
6362                << "section)\n";
6363         break;
6364       }
6365       memcpy(&def, defs + j * sizeof(uint32_t), sizeof(uint32_t));
6366       if (O->isLittleEndian() != sys::IsLittleEndianHost)
6367         sys::swapByteOrder(def);
6368 
6369       r = get_pointer_32(def, xoffset, left, xS, &info, true);
6370       outs() << "\tdefs[" << j << "] " << format("0x%08" PRIx32, def);
6371       if (r != nullptr) {
6372         if (left > sizeof(struct objc_class_t)) {
6373           outs() << "\n";
6374           memcpy(&objc_class, r, sizeof(struct objc_class_t));
6375         } else {
6376           outs() << " (entends past the end of the section)\n";
6377           memset(&objc_class, '\0', sizeof(struct objc_class_t));
6378           memcpy(&objc_class, r, left);
6379         }
6380         if (O->isLittleEndian() != sys::IsLittleEndianHost)
6381           swapStruct(objc_class);
6382         print_objc_class_t(&objc_class, &info);
6383       } else {
6384         outs() << "(not in an __OBJC section)\n";
6385       }
6386 
6387       if (CLS_GETINFO(&objc_class, CLS_CLASS)) {
6388         outs() << "\tMeta Class";
6389         r = get_pointer_32(objc_class.isa, xoffset, left, xS, &info, true);
6390         if (r != nullptr) {
6391           if (left > sizeof(struct objc_class_t)) {
6392             outs() << "\n";
6393             memcpy(&objc_class, r, sizeof(struct objc_class_t));
6394           } else {
6395             outs() << " (entends past the end of the section)\n";
6396             memset(&objc_class, '\0', sizeof(struct objc_class_t));
6397             memcpy(&objc_class, r, left);
6398           }
6399           if (O->isLittleEndian() != sys::IsLittleEndianHost)
6400             swapStruct(objc_class);
6401           print_objc_class_t(&objc_class, &info);
6402         } else {
6403           outs() << "(not in an __OBJC section)\n";
6404         }
6405       }
6406     }
6407     if (symtab.cat_def_cnt > 0)
6408       outs() << "\tCategory Definitions\n";
6409     for (j = 0; j < symtab.cat_def_cnt; j++) {
6410       if ((j + symtab.cls_def_cnt + 1) * sizeof(uint32_t) > defs_left) {
6411         outs() << "\t(remaining category defs entries entends past the end of "
6412                << "the section)\n";
6413         break;
6414       }
6415       memcpy(&def, defs + (j + symtab.cls_def_cnt) * sizeof(uint32_t),
6416              sizeof(uint32_t));
6417       if (O->isLittleEndian() != sys::IsLittleEndianHost)
6418         sys::swapByteOrder(def);
6419 
6420       r = get_pointer_32(def, xoffset, left, xS, &info, true);
6421       outs() << "\tdefs[" << j + symtab.cls_def_cnt << "] "
6422              << format("0x%08" PRIx32, def);
6423       if (r != nullptr) {
6424         if (left > sizeof(struct objc_category_t)) {
6425           outs() << "\n";
6426           memcpy(&objc_category, r, sizeof(struct objc_category_t));
6427         } else {
6428           outs() << " (entends past the end of the section)\n";
6429           memset(&objc_category, '\0', sizeof(struct objc_category_t));
6430           memcpy(&objc_category, r, left);
6431         }
6432         if (O->isLittleEndian() != sys::IsLittleEndianHost)
6433           swapStruct(objc_category);
6434         print_objc_objc_category_t(&objc_category, &info);
6435       } else {
6436         outs() << "(not in an __OBJC section)\n";
6437       }
6438     }
6439   }
6440   const SectionRef II = get_section(O, "__OBJC", "__image_info");
6441   if (II != SectionRef())
6442     print_image_info(II, &info);
6443 
6444   return true;
6445 }
6446 
DumpProtocolSection(MachOObjectFile * O,const char * sect,uint32_t size,uint32_t addr)6447 static void DumpProtocolSection(MachOObjectFile *O, const char *sect,
6448                                 uint32_t size, uint32_t addr) {
6449   SymbolAddressMap AddrMap;
6450   CreateSymbolAddressMap(O, &AddrMap);
6451 
6452   std::vector<SectionRef> Sections;
6453   for (const SectionRef &Section : O->sections())
6454     Sections.push_back(Section);
6455 
6456   struct DisassembleInfo info(O, &AddrMap, &Sections, true);
6457 
6458   const char *p;
6459   struct objc_protocol_t protocol;
6460   uint32_t left, paddr;
6461   for (p = sect; p < sect + size; p += sizeof(struct objc_protocol_t)) {
6462     memset(&protocol, '\0', sizeof(struct objc_protocol_t));
6463     left = size - (p - sect);
6464     if (left < sizeof(struct objc_protocol_t)) {
6465       outs() << "Protocol extends past end of __protocol section\n";
6466       memcpy(&protocol, p, left);
6467     } else
6468       memcpy(&protocol, p, sizeof(struct objc_protocol_t));
6469     if (O->isLittleEndian() != sys::IsLittleEndianHost)
6470       swapStruct(protocol);
6471     paddr = addr + (p - sect);
6472     outs() << "Protocol " << format("0x%" PRIx32, paddr);
6473     if (print_protocol(paddr, 0, &info))
6474       outs() << "(not in an __OBJC section)\n";
6475   }
6476 }
6477 
6478 #ifdef HAVE_LIBXAR
swapStruct(struct xar_header & xar)6479 static inline void swapStruct(struct xar_header &xar) {
6480   sys::swapByteOrder(xar.magic);
6481   sys::swapByteOrder(xar.size);
6482   sys::swapByteOrder(xar.version);
6483   sys::swapByteOrder(xar.toc_length_compressed);
6484   sys::swapByteOrder(xar.toc_length_uncompressed);
6485   sys::swapByteOrder(xar.cksum_alg);
6486 }
6487 
PrintModeVerbose(uint32_t mode)6488 static void PrintModeVerbose(uint32_t mode) {
6489   switch(mode & S_IFMT){
6490   case S_IFDIR:
6491     outs() << "d";
6492     break;
6493   case S_IFCHR:
6494     outs() << "c";
6495     break;
6496   case S_IFBLK:
6497     outs() << "b";
6498     break;
6499   case S_IFREG:
6500     outs() << "-";
6501     break;
6502   case S_IFLNK:
6503     outs() << "l";
6504     break;
6505   case S_IFSOCK:
6506     outs() << "s";
6507     break;
6508   default:
6509     outs() << "?";
6510     break;
6511   }
6512 
6513   /* owner permissions */
6514   if(mode & S_IREAD)
6515     outs() << "r";
6516   else
6517     outs() << "-";
6518   if(mode & S_IWRITE)
6519     outs() << "w";
6520   else
6521     outs() << "-";
6522   if(mode & S_ISUID)
6523     outs() << "s";
6524   else if(mode & S_IEXEC)
6525     outs() << "x";
6526   else
6527     outs() << "-";
6528 
6529   /* group permissions */
6530   if(mode & (S_IREAD >> 3))
6531     outs() << "r";
6532   else
6533     outs() << "-";
6534   if(mode & (S_IWRITE >> 3))
6535     outs() << "w";
6536   else
6537     outs() << "-";
6538   if(mode & S_ISGID)
6539     outs() << "s";
6540   else if(mode & (S_IEXEC >> 3))
6541     outs() << "x";
6542   else
6543     outs() << "-";
6544 
6545   /* other permissions */
6546   if(mode & (S_IREAD >> 6))
6547     outs() << "r";
6548   else
6549     outs() << "-";
6550   if(mode & (S_IWRITE >> 6))
6551     outs() << "w";
6552   else
6553     outs() << "-";
6554   if(mode & S_ISVTX)
6555     outs() << "t";
6556   else if(mode & (S_IEXEC >> 6))
6557     outs() << "x";
6558   else
6559     outs() << "-";
6560 }
6561 
PrintXarFilesSummary(const char * XarFilename,xar_t xar)6562 static void PrintXarFilesSummary(const char *XarFilename, xar_t xar) {
6563   xar_file_t xf;
6564   const char *key, *type, *mode, *user, *group, *size, *mtime, *name, *m;
6565   char *endp;
6566   uint32_t mode_value;
6567 
6568   ScopedXarIter xi;
6569   if (!xi) {
6570     WithColor::error(errs(), "llvm-objdump")
6571         << "can't obtain an xar iterator for xar archive " << XarFilename
6572         << "\n";
6573     return;
6574   }
6575 
6576   // Go through the xar's files.
6577   for (xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)) {
6578     ScopedXarIter xp;
6579     if(!xp){
6580       WithColor::error(errs(), "llvm-objdump")
6581           << "can't obtain an xar iterator for xar archive " << XarFilename
6582           << "\n";
6583       return;
6584     }
6585     type = nullptr;
6586     mode = nullptr;
6587     user = nullptr;
6588     group = nullptr;
6589     size = nullptr;
6590     mtime = nullptr;
6591     name = nullptr;
6592     for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){
6593       const char *val = nullptr;
6594       xar_prop_get(xf, key, &val);
6595 #if 0 // Useful for debugging.
6596       outs() << "key: " << key << " value: " << val << "\n";
6597 #endif
6598       if(strcmp(key, "type") == 0)
6599         type = val;
6600       if(strcmp(key, "mode") == 0)
6601         mode = val;
6602       if(strcmp(key, "user") == 0)
6603         user = val;
6604       if(strcmp(key, "group") == 0)
6605         group = val;
6606       if(strcmp(key, "data/size") == 0)
6607         size = val;
6608       if(strcmp(key, "mtime") == 0)
6609         mtime = val;
6610       if(strcmp(key, "name") == 0)
6611         name = val;
6612     }
6613     if(mode != nullptr){
6614       mode_value = strtoul(mode, &endp, 8);
6615       if(*endp != '\0')
6616         outs() << "(mode: \"" << mode << "\" contains non-octal chars) ";
6617       if(strcmp(type, "file") == 0)
6618         mode_value |= S_IFREG;
6619       PrintModeVerbose(mode_value);
6620       outs() << " ";
6621     }
6622     if(user != nullptr)
6623       outs() << format("%10s/", user);
6624     if(group != nullptr)
6625       outs() << format("%-10s ", group);
6626     if(size != nullptr)
6627       outs() << format("%7s ", size);
6628     if(mtime != nullptr){
6629       for(m = mtime; *m != 'T' && *m != '\0'; m++)
6630         outs() << *m;
6631       if(*m == 'T')
6632         m++;
6633       outs() << " ";
6634       for( ; *m != 'Z' && *m != '\0'; m++)
6635         outs() << *m;
6636       outs() << " ";
6637     }
6638     if(name != nullptr)
6639       outs() << name;
6640     outs() << "\n";
6641   }
6642 }
6643 
DumpBitcodeSection(MachOObjectFile * O,const char * sect,uint32_t size,bool verbose,bool PrintXarHeader,bool PrintXarFileHeaders,std::string XarMemberName)6644 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect,
6645                                 uint32_t size, bool verbose,
6646                                 bool PrintXarHeader, bool PrintXarFileHeaders,
6647                                 std::string XarMemberName) {
6648   if(size < sizeof(struct xar_header)) {
6649     outs() << "size of (__LLVM,__bundle) section too small (smaller than size "
6650               "of struct xar_header)\n";
6651     return;
6652   }
6653   struct xar_header XarHeader;
6654   memcpy(&XarHeader, sect, sizeof(struct xar_header));
6655   if (sys::IsLittleEndianHost)
6656     swapStruct(XarHeader);
6657   if (PrintXarHeader) {
6658     if (!XarMemberName.empty())
6659       outs() << "In xar member " << XarMemberName << ": ";
6660     else
6661       outs() << "For (__LLVM,__bundle) section: ";
6662     outs() << "xar header\n";
6663     if (XarHeader.magic == XAR_HEADER_MAGIC)
6664       outs() << "                  magic XAR_HEADER_MAGIC\n";
6665     else
6666       outs() << "                  magic "
6667              << format_hex(XarHeader.magic, 10, true)
6668              << " (not XAR_HEADER_MAGIC)\n";
6669     outs() << "                   size " << XarHeader.size << "\n";
6670     outs() << "                version " << XarHeader.version << "\n";
6671     outs() << "  toc_length_compressed " << XarHeader.toc_length_compressed
6672            << "\n";
6673     outs() << "toc_length_uncompressed " << XarHeader.toc_length_uncompressed
6674            << "\n";
6675     outs() << "              cksum_alg ";
6676     switch (XarHeader.cksum_alg) {
6677       case XAR_CKSUM_NONE:
6678         outs() << "XAR_CKSUM_NONE\n";
6679         break;
6680       case XAR_CKSUM_SHA1:
6681         outs() << "XAR_CKSUM_SHA1\n";
6682         break;
6683       case XAR_CKSUM_MD5:
6684         outs() << "XAR_CKSUM_MD5\n";
6685         break;
6686 #ifdef XAR_CKSUM_SHA256
6687       case XAR_CKSUM_SHA256:
6688         outs() << "XAR_CKSUM_SHA256\n";
6689         break;
6690 #endif
6691 #ifdef XAR_CKSUM_SHA512
6692       case XAR_CKSUM_SHA512:
6693         outs() << "XAR_CKSUM_SHA512\n";
6694         break;
6695 #endif
6696       default:
6697         outs() << XarHeader.cksum_alg << "\n";
6698     }
6699   }
6700 
6701   SmallString<128> XarFilename;
6702   int FD;
6703   std::error_code XarEC =
6704       sys::fs::createTemporaryFile("llvm-objdump", "xar", FD, XarFilename);
6705   if (XarEC) {
6706     WithColor::error(errs(), "llvm-objdump") << XarEC.message() << "\n";
6707     return;
6708   }
6709   ToolOutputFile XarFile(XarFilename, FD);
6710   raw_fd_ostream &XarOut = XarFile.os();
6711   StringRef XarContents(sect, size);
6712   XarOut << XarContents;
6713   XarOut.close();
6714   if (XarOut.has_error())
6715     return;
6716 
6717   ScopedXarFile xar(XarFilename.c_str(), READ);
6718   if (!xar) {
6719     WithColor::error(errs(), "llvm-objdump")
6720         << "can't create temporary xar archive " << XarFilename << "\n";
6721     return;
6722   }
6723 
6724   SmallString<128> TocFilename;
6725   std::error_code TocEC =
6726       sys::fs::createTemporaryFile("llvm-objdump", "toc", TocFilename);
6727   if (TocEC) {
6728     WithColor::error(errs(), "llvm-objdump") << TocEC.message() << "\n";
6729     return;
6730   }
6731   xar_serialize(xar, TocFilename.c_str());
6732 
6733   if (PrintXarFileHeaders) {
6734     if (!XarMemberName.empty())
6735       outs() << "In xar member " << XarMemberName << ": ";
6736     else
6737       outs() << "For (__LLVM,__bundle) section: ";
6738     outs() << "xar archive files:\n";
6739     PrintXarFilesSummary(XarFilename.c_str(), xar);
6740   }
6741 
6742   ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr =
6743     MemoryBuffer::getFileOrSTDIN(TocFilename.c_str());
6744   if (std::error_code EC = FileOrErr.getError()) {
6745     WithColor::error(errs(), "llvm-objdump") << EC.message() << "\n";
6746     return;
6747   }
6748   std::unique_ptr<MemoryBuffer> &Buffer = FileOrErr.get();
6749 
6750   if (!XarMemberName.empty())
6751     outs() << "In xar member " << XarMemberName << ": ";
6752   else
6753     outs() << "For (__LLVM,__bundle) section: ";
6754   outs() << "xar table of contents:\n";
6755   outs() << Buffer->getBuffer() << "\n";
6756 
6757   // TODO: Go through the xar's files.
6758   ScopedXarIter xi;
6759   if(!xi){
6760     WithColor::error(errs(), "llvm-objdump")
6761         << "can't obtain an xar iterator for xar archive "
6762         << XarFilename.c_str() << "\n";
6763     return;
6764   }
6765   for(xar_file_t xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)){
6766     const char *key;
6767     const char *member_name, *member_type, *member_size_string;
6768     size_t member_size;
6769 
6770     ScopedXarIter xp;
6771     if(!xp){
6772       WithColor::error(errs(), "llvm-objdump")
6773           << "can't obtain an xar iterator for xar archive "
6774           << XarFilename.c_str() << "\n";
6775       return;
6776     }
6777     member_name = NULL;
6778     member_type = NULL;
6779     member_size_string = NULL;
6780     for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){
6781       const char *val = nullptr;
6782       xar_prop_get(xf, key, &val);
6783 #if 0 // Useful for debugging.
6784       outs() << "key: " << key << " value: " << val << "\n";
6785 #endif
6786       if (strcmp(key, "name") == 0)
6787         member_name = val;
6788       if (strcmp(key, "type") == 0)
6789         member_type = val;
6790       if (strcmp(key, "data/size") == 0)
6791         member_size_string = val;
6792     }
6793     /*
6794      * If we find a file with a name, date/size and type properties
6795      * and with the type being "file" see if that is a xar file.
6796      */
6797     if (member_name != NULL && member_type != NULL &&
6798         strcmp(member_type, "file") == 0 &&
6799         member_size_string != NULL){
6800       // Extract the file into a buffer.
6801       char *endptr;
6802       member_size = strtoul(member_size_string, &endptr, 10);
6803       if (*endptr == '\0' && member_size != 0) {
6804         char *buffer;
6805         if (xar_extract_tobuffersz(xar, xf, &buffer, &member_size) == 0) {
6806 #if 0 // Useful for debugging.
6807           outs() << "xar member: " << member_name << " extracted\n";
6808 #endif
6809           // Set the XarMemberName we want to see printed in the header.
6810           std::string OldXarMemberName;
6811           // If XarMemberName is already set this is nested. So
6812           // save the old name and create the nested name.
6813           if (!XarMemberName.empty()) {
6814             OldXarMemberName = XarMemberName;
6815             XarMemberName =
6816                 (Twine("[") + XarMemberName + "]" + member_name).str();
6817           } else {
6818             OldXarMemberName = "";
6819             XarMemberName = member_name;
6820           }
6821           // See if this is could be a xar file (nested).
6822           if (member_size >= sizeof(struct xar_header)) {
6823 #if 0 // Useful for debugging.
6824             outs() << "could be a xar file: " << member_name << "\n";
6825 #endif
6826             memcpy((char *)&XarHeader, buffer, sizeof(struct xar_header));
6827             if (sys::IsLittleEndianHost)
6828               swapStruct(XarHeader);
6829             if (XarHeader.magic == XAR_HEADER_MAGIC)
6830               DumpBitcodeSection(O, buffer, member_size, verbose,
6831                                  PrintXarHeader, PrintXarFileHeaders,
6832                                  XarMemberName);
6833           }
6834           XarMemberName = OldXarMemberName;
6835           delete buffer;
6836         }
6837       }
6838     }
6839   }
6840 }
6841 #endif // defined(HAVE_LIBXAR)
6842 
printObjcMetaData(MachOObjectFile * O,bool verbose)6843 static void printObjcMetaData(MachOObjectFile *O, bool verbose) {
6844   if (O->is64Bit())
6845     printObjc2_64bit_MetaData(O, verbose);
6846   else {
6847     MachO::mach_header H;
6848     H = O->getHeader();
6849     if (H.cputype == MachO::CPU_TYPE_ARM)
6850       printObjc2_32bit_MetaData(O, verbose);
6851     else {
6852       // This is the 32-bit non-arm cputype case.  Which is normally
6853       // the first Objective-C ABI.  But it may be the case of a
6854       // binary for the iOS simulator which is the second Objective-C
6855       // ABI.  In that case printObjc1_32bit_MetaData() will determine that
6856       // and return false.
6857       if (!printObjc1_32bit_MetaData(O, verbose))
6858         printObjc2_32bit_MetaData(O, verbose);
6859     }
6860   }
6861 }
6862 
6863 // GuessLiteralPointer returns a string which for the item in the Mach-O file
6864 // for the address passed in as ReferenceValue for printing as a comment with
6865 // the instruction and also returns the corresponding type of that item
6866 // indirectly through ReferenceType.
6867 //
6868 // If ReferenceValue is an address of literal cstring then a pointer to the
6869 // cstring is returned and ReferenceType is set to
6870 // LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr .
6871 //
6872 // If ReferenceValue is an address of an Objective-C CFString, Selector ref or
6873 // Class ref that name is returned and the ReferenceType is set accordingly.
6874 //
6875 // Lastly, literals which are Symbol address in a literal pool are looked for
6876 // and if found the symbol name is returned and ReferenceType is set to
6877 // LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr .
6878 //
6879 // If there is no item in the Mach-O file for the address passed in as
6880 // ReferenceValue nullptr is returned and ReferenceType is unchanged.
GuessLiteralPointer(uint64_t ReferenceValue,uint64_t ReferencePC,uint64_t * ReferenceType,struct DisassembleInfo * info)6881 static const char *GuessLiteralPointer(uint64_t ReferenceValue,
6882                                        uint64_t ReferencePC,
6883                                        uint64_t *ReferenceType,
6884                                        struct DisassembleInfo *info) {
6885   // First see if there is an external relocation entry at the ReferencePC.
6886   if (info->O->getHeader().filetype == MachO::MH_OBJECT) {
6887     uint64_t sect_addr = info->S.getAddress();
6888     uint64_t sect_offset = ReferencePC - sect_addr;
6889     bool reloc_found = false;
6890     DataRefImpl Rel;
6891     MachO::any_relocation_info RE;
6892     bool isExtern = false;
6893     SymbolRef Symbol;
6894     for (const RelocationRef &Reloc : info->S.relocations()) {
6895       uint64_t RelocOffset = Reloc.getOffset();
6896       if (RelocOffset == sect_offset) {
6897         Rel = Reloc.getRawDataRefImpl();
6898         RE = info->O->getRelocation(Rel);
6899         if (info->O->isRelocationScattered(RE))
6900           continue;
6901         isExtern = info->O->getPlainRelocationExternal(RE);
6902         if (isExtern) {
6903           symbol_iterator RelocSym = Reloc.getSymbol();
6904           Symbol = *RelocSym;
6905         }
6906         reloc_found = true;
6907         break;
6908       }
6909     }
6910     // If there is an external relocation entry for a symbol in a section
6911     // then used that symbol's value for the value of the reference.
6912     if (reloc_found && isExtern) {
6913       if (info->O->getAnyRelocationPCRel(RE)) {
6914         unsigned Type = info->O->getAnyRelocationType(RE);
6915         if (Type == MachO::X86_64_RELOC_SIGNED) {
6916           ReferenceValue = cantFail(Symbol.getValue());
6917         }
6918       }
6919     }
6920   }
6921 
6922   // Look for literals such as Objective-C CFStrings refs, Selector refs,
6923   // Message refs and Class refs.
6924   bool classref, selref, msgref, cfstring;
6925   uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref,
6926                                                selref, msgref, cfstring);
6927   if (classref && pointer_value == 0) {
6928     // Note the ReferenceValue is a pointer into the __objc_classrefs section.
6929     // And the pointer_value in that section is typically zero as it will be
6930     // set by dyld as part of the "bind information".
6931     const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info);
6932     if (name != nullptr) {
6933       *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
6934       const char *class_name = strrchr(name, '$');
6935       if (class_name != nullptr && class_name[1] == '_' &&
6936           class_name[2] != '\0') {
6937         info->class_name = class_name + 2;
6938         return name;
6939       }
6940     }
6941   }
6942 
6943   if (classref) {
6944     *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
6945     const char *name =
6946         get_objc2_64bit_class_name(pointer_value, ReferenceValue, info);
6947     if (name != nullptr)
6948       info->class_name = name;
6949     else
6950       name = "bad class ref";
6951     return name;
6952   }
6953 
6954   if (cfstring) {
6955     *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref;
6956     const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info);
6957     return name;
6958   }
6959 
6960   if (selref && pointer_value == 0)
6961     pointer_value = get_objc2_64bit_selref(ReferenceValue, info);
6962 
6963   if (pointer_value != 0)
6964     ReferenceValue = pointer_value;
6965 
6966   const char *name = GuessCstringPointer(ReferenceValue, info);
6967   if (name) {
6968     if (pointer_value != 0 && selref) {
6969       *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref;
6970       info->selector_name = name;
6971     } else if (pointer_value != 0 && msgref) {
6972       info->class_name = nullptr;
6973       *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref;
6974       info->selector_name = name;
6975     } else
6976       *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr;
6977     return name;
6978   }
6979 
6980   // Lastly look for an indirect symbol with this ReferenceValue which is in
6981   // a literal pool.  If found return that symbol name.
6982   name = GuessIndirectSymbol(ReferenceValue, info);
6983   if (name) {
6984     *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr;
6985     return name;
6986   }
6987 
6988   return nullptr;
6989 }
6990 
6991 // SymbolizerSymbolLookUp is the symbol lookup function passed when creating
6992 // the Symbolizer.  It looks up the ReferenceValue using the info passed via the
6993 // pointer to the struct DisassembleInfo that was passed when MCSymbolizer
6994 // is created and returns the symbol name that matches the ReferenceValue or
6995 // nullptr if none.  The ReferenceType is passed in for the IN type of
6996 // reference the instruction is making from the values in defined in the header
6997 // "llvm-c/Disassembler.h".  On return the ReferenceType can set to a specific
6998 // Out type and the ReferenceName will also be set which is added as a comment
6999 // to the disassembled instruction.
7000 //
7001 // If the symbol name is a C++ mangled name then the demangled name is
7002 // returned through ReferenceName and ReferenceType is set to
7003 // LLVMDisassembler_ReferenceType_DeMangled_Name .
7004 //
7005 // When this is called to get a symbol name for a branch target then the
7006 // ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then
7007 // SymbolValue will be looked for in the indirect symbol table to determine if
7008 // it is an address for a symbol stub.  If so then the symbol name for that
7009 // stub is returned indirectly through ReferenceName and then ReferenceType is
7010 // set to LLVMDisassembler_ReferenceType_Out_SymbolStub.
7011 //
7012 // When this is called with an value loaded via a PC relative load then
7013 // ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the
7014 // SymbolValue is checked to be an address of literal pointer, symbol pointer,
7015 // or an Objective-C meta data reference.  If so the output ReferenceType is
7016 // 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)7017 static const char *SymbolizerSymbolLookUp(void *DisInfo,
7018                                           uint64_t ReferenceValue,
7019                                           uint64_t *ReferenceType,
7020                                           uint64_t ReferencePC,
7021                                           const char **ReferenceName) {
7022   struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
7023   // If no verbose symbolic information is wanted then just return nullptr.
7024   if (!info->verbose) {
7025     *ReferenceName = nullptr;
7026     *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7027     return nullptr;
7028   }
7029 
7030   const char *SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
7031 
7032   if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) {
7033     *ReferenceName = GuessIndirectSymbol(ReferenceValue, info);
7034     if (*ReferenceName != nullptr) {
7035       method_reference(info, ReferenceType, ReferenceName);
7036       if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message)
7037         *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub;
7038     } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
7039       if (info->demangled_name != nullptr)
7040         free(info->demangled_name);
7041       int status;
7042       info->demangled_name =
7043           itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status);
7044       if (info->demangled_name != nullptr) {
7045         *ReferenceName = info->demangled_name;
7046         *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
7047       } else
7048         *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7049     } else
7050       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7051   } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) {
7052     *ReferenceName =
7053         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7054     if (*ReferenceName)
7055       method_reference(info, ReferenceType, ReferenceName);
7056     else
7057       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7058     // If this is arm64 and the reference is an adrp instruction save the
7059     // instruction, passed in ReferenceValue and the address of the instruction
7060     // for use later if we see and add immediate instruction.
7061   } else if (info->O->getArch() == Triple::aarch64 &&
7062              *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
7063     info->adrp_inst = ReferenceValue;
7064     info->adrp_addr = ReferencePC;
7065     SymbolName = nullptr;
7066     *ReferenceName = nullptr;
7067     *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7068     // If this is arm64 and reference is an add immediate instruction and we
7069     // have
7070     // seen an adrp instruction just before it and the adrp's Xd register
7071     // matches
7072     // this add's Xn register reconstruct the value being referenced and look to
7073     // see if it is a literal pointer.  Note the add immediate instruction is
7074     // passed in ReferenceValue.
7075   } else if (info->O->getArch() == Triple::aarch64 &&
7076              *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri &&
7077              ReferencePC - 4 == info->adrp_addr &&
7078              (info->adrp_inst & 0x9f000000) == 0x90000000 &&
7079              (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
7080     uint32_t addxri_inst;
7081     uint64_t adrp_imm, addxri_imm;
7082 
7083     adrp_imm =
7084         ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
7085     if (info->adrp_inst & 0x0200000)
7086       adrp_imm |= 0xfffffffffc000000LL;
7087 
7088     addxri_inst = ReferenceValue;
7089     addxri_imm = (addxri_inst >> 10) & 0xfff;
7090     if (((addxri_inst >> 22) & 0x3) == 1)
7091       addxri_imm <<= 12;
7092 
7093     ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
7094                      (adrp_imm << 12) + addxri_imm;
7095 
7096     *ReferenceName =
7097         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7098     if (*ReferenceName == nullptr)
7099       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7100     // If this is arm64 and the reference is a load register instruction and we
7101     // have seen an adrp instruction just before it and the adrp's Xd register
7102     // matches this add's Xn register reconstruct the value being referenced and
7103     // look to see if it is a literal pointer.  Note the load register
7104     // instruction is passed in ReferenceValue.
7105   } else if (info->O->getArch() == Triple::aarch64 &&
7106              *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui &&
7107              ReferencePC - 4 == info->adrp_addr &&
7108              (info->adrp_inst & 0x9f000000) == 0x90000000 &&
7109              (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
7110     uint32_t ldrxui_inst;
7111     uint64_t adrp_imm, ldrxui_imm;
7112 
7113     adrp_imm =
7114         ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
7115     if (info->adrp_inst & 0x0200000)
7116       adrp_imm |= 0xfffffffffc000000LL;
7117 
7118     ldrxui_inst = ReferenceValue;
7119     ldrxui_imm = (ldrxui_inst >> 10) & 0xfff;
7120 
7121     ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
7122                      (adrp_imm << 12) + (ldrxui_imm << 3);
7123 
7124     *ReferenceName =
7125         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7126     if (*ReferenceName == nullptr)
7127       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7128   }
7129   // If this arm64 and is an load register (PC-relative) instruction the
7130   // ReferenceValue is the PC plus the immediate value.
7131   else if (info->O->getArch() == Triple::aarch64 &&
7132            (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl ||
7133             *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) {
7134     *ReferenceName =
7135         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7136     if (*ReferenceName == nullptr)
7137       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7138   } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
7139     if (info->demangled_name != nullptr)
7140       free(info->demangled_name);
7141     int status;
7142     info->demangled_name =
7143         itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status);
7144     if (info->demangled_name != nullptr) {
7145       *ReferenceName = info->demangled_name;
7146       *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
7147     }
7148   }
7149   else {
7150     *ReferenceName = nullptr;
7151     *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7152   }
7153 
7154   return SymbolName;
7155 }
7156 
7157 /// Emits the comments that are stored in the CommentStream.
7158 /// 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)7159 static void emitComments(raw_svector_ostream &CommentStream,
7160                          SmallString<128> &CommentsToEmit,
7161                          formatted_raw_ostream &FormattedOS,
7162                          const MCAsmInfo &MAI) {
7163   // Flush the stream before taking its content.
7164   StringRef Comments = CommentsToEmit.str();
7165   // Get the default information for printing a comment.
7166   StringRef CommentBegin = MAI.getCommentString();
7167   unsigned CommentColumn = MAI.getCommentColumn();
7168   bool IsFirst = true;
7169   while (!Comments.empty()) {
7170     if (!IsFirst)
7171       FormattedOS << '\n';
7172     // Emit a line of comments.
7173     FormattedOS.PadToColumn(CommentColumn);
7174     size_t Position = Comments.find('\n');
7175     FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position);
7176     // Move after the newline character.
7177     Comments = Comments.substr(Position + 1);
7178     IsFirst = false;
7179   }
7180   FormattedOS.flush();
7181 
7182   // Tell the comment stream that the vector changed underneath it.
7183   CommentsToEmit.clear();
7184 }
7185 
DisassembleMachO(StringRef Filename,MachOObjectFile * MachOOF,StringRef DisSegName,StringRef DisSectName)7186 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
7187                              StringRef DisSegName, StringRef DisSectName) {
7188   const char *McpuDefault = nullptr;
7189   const Target *ThumbTarget = nullptr;
7190   const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget);
7191   if (!TheTarget) {
7192     // GetTarget prints out stuff.
7193     return;
7194   }
7195   std::string MachOMCPU;
7196   if (MCPU.empty() && McpuDefault)
7197     MachOMCPU = McpuDefault;
7198   else
7199     MachOMCPU = MCPU;
7200 
7201 #define CHECK_TARGET_INFO_CREATION(NAME)                                       \
7202   do {                                                                         \
7203     if (!NAME) {                                                               \
7204       WithColor::error(errs(), "llvm-objdump")                                 \
7205           << "couldn't initialize disassembler for target " << TripleName      \
7206           << '\n';                                                             \
7207       return;                                                                  \
7208     }                                                                          \
7209   } while (false)
7210 #define CHECK_THUMB_TARGET_INFO_CREATION(NAME)                                 \
7211   do {                                                                         \
7212     if (!NAME) {                                                               \
7213       WithColor::error(errs(), "llvm-objdump")                                 \
7214           << "couldn't initialize disassembler for target " << ThumbTripleName \
7215           << '\n';                                                             \
7216       return;                                                                  \
7217     }                                                                          \
7218   } while (false)
7219 
7220   std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
7221   CHECK_TARGET_INFO_CREATION(InstrInfo);
7222   std::unique_ptr<const MCInstrInfo> ThumbInstrInfo;
7223   if (ThumbTarget) {
7224     ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo());
7225     CHECK_THUMB_TARGET_INFO_CREATION(ThumbInstrInfo);
7226   }
7227 
7228   // Package up features to be passed to target/subtarget
7229   std::string FeaturesStr;
7230   if (!MAttrs.empty()) {
7231     SubtargetFeatures Features;
7232     for (unsigned i = 0; i != MAttrs.size(); ++i)
7233       Features.AddFeature(MAttrs[i]);
7234     FeaturesStr = Features.getString();
7235   }
7236 
7237   MCTargetOptions MCOptions;
7238   // Set up disassembler.
7239   std::unique_ptr<const MCRegisterInfo> MRI(
7240       TheTarget->createMCRegInfo(TripleName));
7241   CHECK_TARGET_INFO_CREATION(MRI);
7242   std::unique_ptr<const MCAsmInfo> AsmInfo(
7243       TheTarget->createMCAsmInfo(*MRI, TripleName, MCOptions));
7244   CHECK_TARGET_INFO_CREATION(AsmInfo);
7245   std::unique_ptr<const MCSubtargetInfo> STI(
7246       TheTarget->createMCSubtargetInfo(TripleName, MachOMCPU, FeaturesStr));
7247   CHECK_TARGET_INFO_CREATION(STI);
7248   MCContext Ctx(AsmInfo.get(), MRI.get(), nullptr);
7249   std::unique_ptr<MCDisassembler> DisAsm(
7250       TheTarget->createMCDisassembler(*STI, Ctx));
7251   CHECK_TARGET_INFO_CREATION(DisAsm);
7252   std::unique_ptr<MCSymbolizer> Symbolizer;
7253   struct DisassembleInfo SymbolizerInfo(nullptr, nullptr, nullptr, false);
7254   std::unique_ptr<MCRelocationInfo> RelInfo(
7255       TheTarget->createMCRelocationInfo(TripleName, Ctx));
7256   if (RelInfo) {
7257     Symbolizer.reset(TheTarget->createMCSymbolizer(
7258         TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
7259         &SymbolizerInfo, &Ctx, std::move(RelInfo)));
7260     DisAsm->setSymbolizer(std::move(Symbolizer));
7261   }
7262   int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
7263   std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
7264       Triple(TripleName), AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI));
7265   CHECK_TARGET_INFO_CREATION(IP);
7266   // Set the display preference for hex vs. decimal immediates.
7267   IP->setPrintImmHex(PrintImmHex);
7268   // Comment stream and backing vector.
7269   SmallString<128> CommentsToEmit;
7270   raw_svector_ostream CommentStream(CommentsToEmit);
7271   // FIXME: Setting the CommentStream in the InstPrinter is problematic in that
7272   // if it is done then arm64 comments for string literals don't get printed
7273   // and some constant get printed instead and not setting it causes intel
7274   // (32-bit and 64-bit) comments printed with different spacing before the
7275   // comment causing different diffs with the 'C' disassembler library API.
7276   // IP->setCommentStream(CommentStream);
7277 
7278   // Set up separate thumb disassembler if needed.
7279   std::unique_ptr<const MCRegisterInfo> ThumbMRI;
7280   std::unique_ptr<const MCAsmInfo> ThumbAsmInfo;
7281   std::unique_ptr<const MCSubtargetInfo> ThumbSTI;
7282   std::unique_ptr<MCDisassembler> ThumbDisAsm;
7283   std::unique_ptr<MCInstPrinter> ThumbIP;
7284   std::unique_ptr<MCContext> ThumbCtx;
7285   std::unique_ptr<MCSymbolizer> ThumbSymbolizer;
7286   struct DisassembleInfo ThumbSymbolizerInfo(nullptr, nullptr, nullptr, false);
7287   std::unique_ptr<MCRelocationInfo> ThumbRelInfo;
7288   if (ThumbTarget) {
7289     ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName));
7290     CHECK_THUMB_TARGET_INFO_CREATION(ThumbMRI);
7291     ThumbAsmInfo.reset(
7292         ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName, MCOptions));
7293     CHECK_THUMB_TARGET_INFO_CREATION(ThumbAsmInfo);
7294     ThumbSTI.reset(
7295         ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MachOMCPU,
7296                                            FeaturesStr));
7297     CHECK_THUMB_TARGET_INFO_CREATION(ThumbSTI);
7298     ThumbCtx.reset(new MCContext(ThumbAsmInfo.get(), ThumbMRI.get(), nullptr));
7299     ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx));
7300     CHECK_THUMB_TARGET_INFO_CREATION(ThumbDisAsm);
7301     MCContext *PtrThumbCtx = ThumbCtx.get();
7302     ThumbRelInfo.reset(
7303         ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx));
7304     if (ThumbRelInfo) {
7305       ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer(
7306           ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
7307           &ThumbSymbolizerInfo, PtrThumbCtx, std::move(ThumbRelInfo)));
7308       ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer));
7309     }
7310     int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect();
7311     ThumbIP.reset(ThumbTarget->createMCInstPrinter(
7312         Triple(ThumbTripleName), ThumbAsmPrinterVariant, *ThumbAsmInfo,
7313         *ThumbInstrInfo, *ThumbMRI));
7314     CHECK_THUMB_TARGET_INFO_CREATION(ThumbIP);
7315     // Set the display preference for hex vs. decimal immediates.
7316     ThumbIP->setPrintImmHex(PrintImmHex);
7317   }
7318 
7319 #undef CHECK_TARGET_INFO_CREATION
7320 #undef CHECK_THUMB_TARGET_INFO_CREATION
7321 
7322   MachO::mach_header Header = MachOOF->getHeader();
7323 
7324   // FIXME: Using the -cfg command line option, this code used to be able to
7325   // annotate relocations with the referenced symbol's name, and if this was
7326   // inside a __[cf]string section, the data it points to. This is now replaced
7327   // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
7328   std::vector<SectionRef> Sections;
7329   std::vector<SymbolRef> Symbols;
7330   SmallVector<uint64_t, 8> FoundFns;
7331   uint64_t BaseSegmentAddress = 0;
7332 
7333   getSectionsAndSymbols(MachOOF, Sections, Symbols, FoundFns,
7334                         BaseSegmentAddress);
7335 
7336   // Sort the symbols by address, just in case they didn't come in that way.
7337   llvm::sort(Symbols, SymbolSorter());
7338 
7339   // Build a data in code table that is sorted on by the address of each entry.
7340   uint64_t BaseAddress = 0;
7341   if (Header.filetype == MachO::MH_OBJECT)
7342     BaseAddress = Sections[0].getAddress();
7343   else
7344     BaseAddress = BaseSegmentAddress;
7345   DiceTable Dices;
7346   for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
7347        DI != DE; ++DI) {
7348     uint32_t Offset;
7349     DI->getOffset(Offset);
7350     Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
7351   }
7352   array_pod_sort(Dices.begin(), Dices.end());
7353 
7354   // Try to find debug info and set up the DIContext for it.
7355   std::unique_ptr<DIContext> diContext;
7356   std::unique_ptr<Binary> DSYMBinary;
7357   std::unique_ptr<MemoryBuffer> DSYMBuf;
7358   if (UseDbg) {
7359     ObjectFile *DbgObj = MachOOF;
7360 
7361     // A separate DSym file path was specified, parse it as a macho file,
7362     // get the sections and supply it to the section name parsing machinery.
7363     if (!DSYMFile.empty()) {
7364       std::string DSYMPath(DSYMFile);
7365 
7366       // If DSYMPath is a .dSYM directory, append the Mach-O file.
7367       if (llvm::sys::fs::is_directory(DSYMPath) &&
7368           llvm::sys::path::extension(DSYMPath) == ".dSYM") {
7369         SmallString<128> ShortName(llvm::sys::path::filename(DSYMPath));
7370         llvm::sys::path::replace_extension(ShortName, "");
7371         SmallString<1024> FullPath(DSYMPath);
7372         llvm::sys::path::append(FullPath, "Contents", "Resources", "DWARF",
7373                                 ShortName);
7374         DSYMPath = std::string(FullPath.str());
7375       }
7376 
7377       // Load the file.
7378       ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr =
7379           MemoryBuffer::getFileOrSTDIN(DSYMPath);
7380       if (std::error_code EC = BufOrErr.getError()) {
7381         reportError(errorCodeToError(EC), DSYMPath);
7382         return;
7383       }
7384 
7385       // We need to keep the file alive, because we're replacing DbgObj with it.
7386       DSYMBuf = std::move(BufOrErr.get());
7387 
7388       Expected<std::unique_ptr<Binary>> BinaryOrErr =
7389       createBinary(DSYMBuf.get()->getMemBufferRef());
7390       if (!BinaryOrErr) {
7391         reportError(BinaryOrErr.takeError(), DSYMPath);
7392         return;
7393       }
7394 
7395       // We need to keep the Binary alive with the buffer
7396       DSYMBinary = std::move(BinaryOrErr.get());
7397       if (ObjectFile *O = dyn_cast<ObjectFile>(DSYMBinary.get())) {
7398         // this is a Mach-O object file, use it
7399         if (MachOObjectFile *MachDSYM = dyn_cast<MachOObjectFile>(&*O)) {
7400           DbgObj = MachDSYM;
7401         }
7402         else {
7403           WithColor::error(errs(), "llvm-objdump")
7404             << DSYMPath << " is not a Mach-O file type.\n";
7405           return;
7406         }
7407       }
7408       else if (auto UB = dyn_cast<MachOUniversalBinary>(DSYMBinary.get())){
7409         // this is a Universal Binary, find a Mach-O for this architecture
7410         uint32_t CPUType, CPUSubType;
7411         const char *ArchFlag;
7412         if (MachOOF->is64Bit()) {
7413           const MachO::mach_header_64 H_64 = MachOOF->getHeader64();
7414           CPUType = H_64.cputype;
7415           CPUSubType = H_64.cpusubtype;
7416         } else {
7417           const MachO::mach_header H = MachOOF->getHeader();
7418           CPUType = H.cputype;
7419           CPUSubType = H.cpusubtype;
7420         }
7421         Triple T = MachOObjectFile::getArchTriple(CPUType, CPUSubType, nullptr,
7422                                                   &ArchFlag);
7423         Expected<std::unique_ptr<MachOObjectFile>> MachDSYM =
7424             UB->getMachOObjectForArch(ArchFlag);
7425         if (!MachDSYM) {
7426           reportError(MachDSYM.takeError(), DSYMPath);
7427           return;
7428         }
7429 
7430         // We need to keep the Binary alive with the buffer
7431         DbgObj = &*MachDSYM.get();
7432         DSYMBinary = std::move(*MachDSYM);
7433       }
7434       else {
7435         WithColor::error(errs(), "llvm-objdump")
7436           << DSYMPath << " is not a Mach-O or Universal file type.\n";
7437         return;
7438       }
7439     }
7440 
7441     // Setup the DIContext
7442     diContext = DWARFContext::create(*DbgObj);
7443   }
7444 
7445   if (FilterSections.empty())
7446     outs() << "(" << DisSegName << "," << DisSectName << ") section\n";
7447 
7448   for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
7449     Expected<StringRef> SecNameOrErr = Sections[SectIdx].getName();
7450     if (!SecNameOrErr) {
7451       consumeError(SecNameOrErr.takeError());
7452       continue;
7453     }
7454     if (*SecNameOrErr != DisSectName)
7455       continue;
7456 
7457     DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
7458 
7459     StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
7460     if (SegmentName != DisSegName)
7461       continue;
7462 
7463     StringRef BytesStr =
7464         unwrapOrError(Sections[SectIdx].getContents(), Filename);
7465     ArrayRef<uint8_t> Bytes = arrayRefFromStringRef(BytesStr);
7466     uint64_t SectAddress = Sections[SectIdx].getAddress();
7467 
7468     bool symbolTableWorked = false;
7469 
7470     // Create a map of symbol addresses to symbol names for use by
7471     // the SymbolizerSymbolLookUp() routine.
7472     SymbolAddressMap AddrMap;
7473     bool DisSymNameFound = false;
7474     for (const SymbolRef &Symbol : MachOOF->symbols()) {
7475       SymbolRef::Type ST =
7476           unwrapOrError(Symbol.getType(), MachOOF->getFileName());
7477       if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
7478           ST == SymbolRef::ST_Other) {
7479         uint64_t Address = cantFail(Symbol.getValue());
7480         StringRef SymName =
7481             unwrapOrError(Symbol.getName(), MachOOF->getFileName());
7482         AddrMap[Address] = SymName;
7483         if (!DisSymName.empty() && DisSymName == SymName)
7484           DisSymNameFound = true;
7485       }
7486     }
7487     if (!DisSymName.empty() && !DisSymNameFound) {
7488       outs() << "Can't find -dis-symname: " << DisSymName << "\n";
7489       return;
7490     }
7491     // Set up the block of info used by the Symbolizer call backs.
7492     SymbolizerInfo.verbose = !NoSymbolicOperands;
7493     SymbolizerInfo.O = MachOOF;
7494     SymbolizerInfo.S = Sections[SectIdx];
7495     SymbolizerInfo.AddrMap = &AddrMap;
7496     SymbolizerInfo.Sections = &Sections;
7497     // Same for the ThumbSymbolizer
7498     ThumbSymbolizerInfo.verbose = !NoSymbolicOperands;
7499     ThumbSymbolizerInfo.O = MachOOF;
7500     ThumbSymbolizerInfo.S = Sections[SectIdx];
7501     ThumbSymbolizerInfo.AddrMap = &AddrMap;
7502     ThumbSymbolizerInfo.Sections = &Sections;
7503 
7504     unsigned int Arch = MachOOF->getArch();
7505 
7506     // Skip all symbols if this is a stubs file.
7507     if (Bytes.empty())
7508       return;
7509 
7510     // If the section has symbols but no symbol at the start of the section
7511     // these are used to make sure the bytes before the first symbol are
7512     // disassembled.
7513     bool FirstSymbol = true;
7514     bool FirstSymbolAtSectionStart = true;
7515 
7516     // Disassemble symbol by symbol.
7517     for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
7518       StringRef SymName =
7519           unwrapOrError(Symbols[SymIdx].getName(), MachOOF->getFileName());
7520       SymbolRef::Type ST =
7521           unwrapOrError(Symbols[SymIdx].getType(), MachOOF->getFileName());
7522       if (ST != SymbolRef::ST_Function && ST != SymbolRef::ST_Data)
7523         continue;
7524 
7525       // Make sure the symbol is defined in this section.
7526       bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]);
7527       if (!containsSym) {
7528         if (!DisSymName.empty() && DisSymName == SymName) {
7529           outs() << "-dis-symname: " << DisSymName << " not in the section\n";
7530           return;
7531         }
7532         continue;
7533       }
7534       // The __mh_execute_header is special and we need to deal with that fact
7535       // this symbol is before the start of the (__TEXT,__text) section and at the
7536       // address of the start of the __TEXT segment.  This is because this symbol
7537       // is an N_SECT symbol in the (__TEXT,__text) but its address is before the
7538       // start of the section in a standard MH_EXECUTE filetype.
7539       if (!DisSymName.empty() && DisSymName == "__mh_execute_header") {
7540         outs() << "-dis-symname: __mh_execute_header not in any section\n";
7541         return;
7542       }
7543       // When this code is trying to disassemble a symbol at a time and in the
7544       // case there is only the __mh_execute_header symbol left as in a stripped
7545       // executable, we need to deal with this by ignoring this symbol so the
7546       // whole section is disassembled and this symbol is then not displayed.
7547       if (SymName == "__mh_execute_header" || SymName == "__mh_dylib_header" ||
7548           SymName == "__mh_bundle_header" || SymName == "__mh_object_header" ||
7549           SymName == "__mh_preload_header" || SymName == "__mh_dylinker_header")
7550         continue;
7551 
7552       // If we are only disassembling one symbol see if this is that symbol.
7553       if (!DisSymName.empty() && DisSymName != SymName)
7554         continue;
7555 
7556       // Start at the address of the symbol relative to the section's address.
7557       uint64_t SectSize = Sections[SectIdx].getSize();
7558       uint64_t Start = cantFail(Symbols[SymIdx].getValue());
7559       uint64_t SectionAddress = Sections[SectIdx].getAddress();
7560       Start -= SectionAddress;
7561 
7562       if (Start > SectSize) {
7563         outs() << "section data ends, " << SymName
7564                << " lies outside valid range\n";
7565         return;
7566       }
7567 
7568       // Stop disassembling either at the beginning of the next symbol or at
7569       // the end of the section.
7570       bool containsNextSym = false;
7571       uint64_t NextSym = 0;
7572       uint64_t NextSymIdx = SymIdx + 1;
7573       while (Symbols.size() > NextSymIdx) {
7574         SymbolRef::Type NextSymType = unwrapOrError(
7575             Symbols[NextSymIdx].getType(), MachOOF->getFileName());
7576         if (NextSymType == SymbolRef::ST_Function) {
7577           containsNextSym =
7578               Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]);
7579           NextSym = cantFail(Symbols[NextSymIdx].getValue());
7580           NextSym -= SectionAddress;
7581           break;
7582         }
7583         ++NextSymIdx;
7584       }
7585 
7586       uint64_t End = containsNextSym ? std::min(NextSym, SectSize) : SectSize;
7587       uint64_t Size;
7588 
7589       symbolTableWorked = true;
7590 
7591       DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl();
7592       uint32_t SymbolFlags = cantFail(MachOOF->getSymbolFlags(Symb));
7593       bool IsThumb = SymbolFlags & SymbolRef::SF_Thumb;
7594 
7595       // We only need the dedicated Thumb target if there's a real choice
7596       // (i.e. we're not targeting M-class) and the function is Thumb.
7597       bool UseThumbTarget = IsThumb && ThumbTarget;
7598 
7599       // If we are not specifying a symbol to start disassembly with and this
7600       // is the first symbol in the section but not at the start of the section
7601       // then move the disassembly index to the start of the section and
7602       // don't print the symbol name just yet.  This is so the bytes before the
7603       // first symbol are disassembled.
7604       uint64_t SymbolStart = Start;
7605       if (DisSymName.empty() && FirstSymbol && Start != 0) {
7606         FirstSymbolAtSectionStart = false;
7607         Start = 0;
7608       }
7609       else
7610         outs() << SymName << ":\n";
7611 
7612       DILineInfo lastLine;
7613       for (uint64_t Index = Start; Index < End; Index += Size) {
7614         MCInst Inst;
7615 
7616         // If this is the first symbol in the section and it was not at the
7617         // start of the section, see if we are at its Index now and if so print
7618         // the symbol name.
7619         if (FirstSymbol && !FirstSymbolAtSectionStart && Index == SymbolStart)
7620           outs() << SymName << ":\n";
7621 
7622         uint64_t PC = SectAddress + Index;
7623         if (!NoLeadingAddr) {
7624           if (FullLeadingAddr) {
7625             if (MachOOF->is64Bit())
7626               outs() << format("%016" PRIx64, PC);
7627             else
7628               outs() << format("%08" PRIx64, PC);
7629           } else {
7630             outs() << format("%8" PRIx64 ":", PC);
7631           }
7632         }
7633         if (!NoShowRawInsn || Arch == Triple::arm)
7634           outs() << "\t";
7635 
7636         if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, Size))
7637           continue;
7638 
7639         SmallVector<char, 64> AnnotationsBytes;
7640         raw_svector_ostream Annotations(AnnotationsBytes);
7641 
7642         bool gotInst;
7643         if (UseThumbTarget)
7644           gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index),
7645                                                 PC, Annotations);
7646         else
7647           gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC,
7648                                            Annotations);
7649         if (gotInst) {
7650           if (!NoShowRawInsn || Arch == Triple::arm) {
7651             dumpBytes(makeArrayRef(Bytes.data() + Index, Size), outs());
7652           }
7653           formatted_raw_ostream FormattedOS(outs());
7654           StringRef AnnotationsStr = Annotations.str();
7655           if (UseThumbTarget)
7656             ThumbIP->printInst(&Inst, PC, AnnotationsStr, *ThumbSTI,
7657                                FormattedOS);
7658           else
7659             IP->printInst(&Inst, PC, AnnotationsStr, *STI, FormattedOS);
7660           emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo);
7661 
7662           // Print debug info.
7663           if (diContext) {
7664             DILineInfo dli = diContext->getLineInfoForAddress({PC, SectIdx});
7665             // Print valid line info if it changed.
7666             if (dli != lastLine && dli.Line != 0)
7667               outs() << "\t## " << dli.FileName << ':' << dli.Line << ':'
7668                      << dli.Column;
7669             lastLine = dli;
7670           }
7671           outs() << "\n";
7672         } else {
7673           if (MachOOF->getArchTriple().isX86()) {
7674             outs() << format("\t.byte 0x%02x #bad opcode\n",
7675                              *(Bytes.data() + Index) & 0xff);
7676             Size = 1; // skip exactly one illegible byte and move on.
7677           } else if (Arch == Triple::aarch64 ||
7678                      (Arch == Triple::arm && !IsThumb)) {
7679             uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
7680                               (*(Bytes.data() + Index + 1) & 0xff) << 8 |
7681                               (*(Bytes.data() + Index + 2) & 0xff) << 16 |
7682                               (*(Bytes.data() + Index + 3) & 0xff) << 24;
7683             outs() << format("\t.long\t0x%08x\n", opcode);
7684             Size = 4;
7685           } else if (Arch == Triple::arm) {
7686             assert(IsThumb && "ARM mode should have been dealt with above");
7687             uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
7688                               (*(Bytes.data() + Index + 1) & 0xff) << 8;
7689             outs() << format("\t.short\t0x%04x\n", opcode);
7690             Size = 2;
7691           } else{
7692             WithColor::warning(errs(), "llvm-objdump")
7693                 << "invalid instruction encoding\n";
7694             if (Size == 0)
7695               Size = 1; // skip illegible bytes
7696           }
7697         }
7698       }
7699       // Now that we are done disassembled the first symbol set the bool that
7700       // were doing this to false.
7701       FirstSymbol = false;
7702     }
7703     if (!symbolTableWorked) {
7704       // Reading the symbol table didn't work, disassemble the whole section.
7705       uint64_t SectAddress = Sections[SectIdx].getAddress();
7706       uint64_t SectSize = Sections[SectIdx].getSize();
7707       uint64_t InstSize;
7708       for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
7709         MCInst Inst;
7710 
7711         uint64_t PC = SectAddress + Index;
7712 
7713         if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, InstSize))
7714           continue;
7715 
7716         SmallVector<char, 64> AnnotationsBytes;
7717         raw_svector_ostream Annotations(AnnotationsBytes);
7718         if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC,
7719                                    Annotations)) {
7720           if (!NoLeadingAddr) {
7721             if (FullLeadingAddr) {
7722               if (MachOOF->is64Bit())
7723                 outs() << format("%016" PRIx64, PC);
7724               else
7725                 outs() << format("%08" PRIx64, PC);
7726             } else {
7727               outs() << format("%8" PRIx64 ":", PC);
7728             }
7729           }
7730           if (!NoShowRawInsn || Arch == Triple::arm) {
7731             outs() << "\t";
7732             dumpBytes(makeArrayRef(Bytes.data() + Index, InstSize), outs());
7733           }
7734           StringRef AnnotationsStr = Annotations.str();
7735           IP->printInst(&Inst, PC, AnnotationsStr, *STI, outs());
7736           outs() << "\n";
7737         } else {
7738           if (MachOOF->getArchTriple().isX86()) {
7739             outs() << format("\t.byte 0x%02x #bad opcode\n",
7740                              *(Bytes.data() + Index) & 0xff);
7741             InstSize = 1; // skip exactly one illegible byte and move on.
7742           } else {
7743             WithColor::warning(errs(), "llvm-objdump")
7744                 << "invalid instruction encoding\n";
7745             if (InstSize == 0)
7746               InstSize = 1; // skip illegible bytes
7747           }
7748         }
7749       }
7750     }
7751     // The TripleName's need to be reset if we are called again for a different
7752     // architecture.
7753     TripleName = "";
7754     ThumbTripleName = "";
7755 
7756     if (SymbolizerInfo.demangled_name != nullptr)
7757       free(SymbolizerInfo.demangled_name);
7758     if (ThumbSymbolizerInfo.demangled_name != nullptr)
7759       free(ThumbSymbolizerInfo.demangled_name);
7760   }
7761 }
7762 
7763 //===----------------------------------------------------------------------===//
7764 // __compact_unwind section dumping
7765 //===----------------------------------------------------------------------===//
7766 
7767 namespace {
7768 
7769 template <typename T>
read(StringRef Contents,ptrdiff_t Offset)7770 static uint64_t read(StringRef Contents, ptrdiff_t Offset) {
7771   using llvm::support::little;
7772   using llvm::support::unaligned;
7773 
7774   if (Offset + sizeof(T) > Contents.size()) {
7775     outs() << "warning: attempt to read past end of buffer\n";
7776     return T();
7777   }
7778 
7779   uint64_t Val =
7780       support::endian::read<T, little, unaligned>(Contents.data() + Offset);
7781   return Val;
7782 }
7783 
7784 template <typename T>
readNext(StringRef Contents,ptrdiff_t & Offset)7785 static uint64_t readNext(StringRef Contents, ptrdiff_t &Offset) {
7786   T Val = read<T>(Contents, Offset);
7787   Offset += sizeof(T);
7788   return Val;
7789 }
7790 
7791 struct CompactUnwindEntry {
7792   uint32_t OffsetInSection;
7793 
7794   uint64_t FunctionAddr;
7795   uint32_t Length;
7796   uint32_t CompactEncoding;
7797   uint64_t PersonalityAddr;
7798   uint64_t LSDAAddr;
7799 
7800   RelocationRef FunctionReloc;
7801   RelocationRef PersonalityReloc;
7802   RelocationRef LSDAReloc;
7803 
CompactUnwindEntry__anonabe85be80a11::CompactUnwindEntry7804   CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64)
7805       : OffsetInSection(Offset) {
7806     if (Is64)
7807       read<uint64_t>(Contents, Offset);
7808     else
7809       read<uint32_t>(Contents, Offset);
7810   }
7811 
7812 private:
read__anonabe85be80a11::CompactUnwindEntry7813   template <typename UIntPtr> void read(StringRef Contents, ptrdiff_t Offset) {
7814     FunctionAddr = readNext<UIntPtr>(Contents, Offset);
7815     Length = readNext<uint32_t>(Contents, Offset);
7816     CompactEncoding = readNext<uint32_t>(Contents, Offset);
7817     PersonalityAddr = readNext<UIntPtr>(Contents, Offset);
7818     LSDAAddr = readNext<UIntPtr>(Contents, Offset);
7819   }
7820 };
7821 }
7822 
7823 /// Given a relocation from __compact_unwind, consisting of the RelocationRef
7824 /// and data being relocated, determine the best base Name and Addend to use for
7825 /// display purposes.
7826 ///
7827 /// 1. An Extern relocation will directly reference a symbol (and the data is
7828 ///    then already an addend), so use that.
7829 /// 2. Otherwise the data is an offset in the object file's layout; try to find
7830 //     a symbol before it in the same section, and use the offset from there.
7831 /// 3. Finally, if all that fails, fall back to an offset from the start of the
7832 ///    referenced section.
findUnwindRelocNameAddend(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const RelocationRef & Reloc,uint64_t Addr,StringRef & Name,uint64_t & Addend)7833 static void findUnwindRelocNameAddend(const MachOObjectFile *Obj,
7834                                       std::map<uint64_t, SymbolRef> &Symbols,
7835                                       const RelocationRef &Reloc, uint64_t Addr,
7836                                       StringRef &Name, uint64_t &Addend) {
7837   if (Reloc.getSymbol() != Obj->symbol_end()) {
7838     Name = unwrapOrError(Reloc.getSymbol()->getName(), Obj->getFileName());
7839     Addend = Addr;
7840     return;
7841   }
7842 
7843   auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl());
7844   SectionRef RelocSection = Obj->getAnyRelocationSection(RE);
7845 
7846   uint64_t SectionAddr = RelocSection.getAddress();
7847 
7848   auto Sym = Symbols.upper_bound(Addr);
7849   if (Sym == Symbols.begin()) {
7850     // The first symbol in the object is after this reference, the best we can
7851     // do is section-relative notation.
7852     if (Expected<StringRef> NameOrErr = RelocSection.getName())
7853       Name = *NameOrErr;
7854     else
7855       consumeError(NameOrErr.takeError());
7856 
7857     Addend = Addr - SectionAddr;
7858     return;
7859   }
7860 
7861   // Go back one so that SymbolAddress <= Addr.
7862   --Sym;
7863 
7864   section_iterator SymSection =
7865       unwrapOrError(Sym->second.getSection(), Obj->getFileName());
7866   if (RelocSection == *SymSection) {
7867     // There's a valid symbol in the same section before this reference.
7868     Name = unwrapOrError(Sym->second.getName(), Obj->getFileName());
7869     Addend = Addr - Sym->first;
7870     return;
7871   }
7872 
7873   // There is a symbol before this reference, but it's in a different
7874   // section. Probably not helpful to mention it, so use the section name.
7875   if (Expected<StringRef> NameOrErr = RelocSection.getName())
7876     Name = *NameOrErr;
7877   else
7878     consumeError(NameOrErr.takeError());
7879 
7880   Addend = Addr - SectionAddr;
7881 }
7882 
printUnwindRelocDest(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const RelocationRef & Reloc,uint64_t Addr)7883 static void printUnwindRelocDest(const MachOObjectFile *Obj,
7884                                  std::map<uint64_t, SymbolRef> &Symbols,
7885                                  const RelocationRef &Reloc, uint64_t Addr) {
7886   StringRef Name;
7887   uint64_t Addend;
7888 
7889   if (!Reloc.getObject())
7890     return;
7891 
7892   findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend);
7893 
7894   outs() << Name;
7895   if (Addend)
7896     outs() << " + " << format("0x%" PRIx64, Addend);
7897 }
7898 
7899 static void
printMachOCompactUnwindSection(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const SectionRef & CompactUnwind)7900 printMachOCompactUnwindSection(const MachOObjectFile *Obj,
7901                                std::map<uint64_t, SymbolRef> &Symbols,
7902                                const SectionRef &CompactUnwind) {
7903 
7904   if (!Obj->isLittleEndian()) {
7905     outs() << "Skipping big-endian __compact_unwind section\n";
7906     return;
7907   }
7908 
7909   bool Is64 = Obj->is64Bit();
7910   uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t);
7911   uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t);
7912 
7913   StringRef Contents =
7914       unwrapOrError(CompactUnwind.getContents(), Obj->getFileName());
7915   SmallVector<CompactUnwindEntry, 4> CompactUnwinds;
7916 
7917   // First populate the initial raw offsets, encodings and so on from the entry.
7918   for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) {
7919     CompactUnwindEntry Entry(Contents, Offset, Is64);
7920     CompactUnwinds.push_back(Entry);
7921   }
7922 
7923   // Next we need to look at the relocations to find out what objects are
7924   // actually being referred to.
7925   for (const RelocationRef &Reloc : CompactUnwind.relocations()) {
7926     uint64_t RelocAddress = Reloc.getOffset();
7927 
7928     uint32_t EntryIdx = RelocAddress / EntrySize;
7929     uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize;
7930     CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx];
7931 
7932     if (OffsetInEntry == 0)
7933       Entry.FunctionReloc = Reloc;
7934     else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t))
7935       Entry.PersonalityReloc = Reloc;
7936     else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t))
7937       Entry.LSDAReloc = Reloc;
7938     else {
7939       outs() << "Invalid relocation in __compact_unwind section\n";
7940       return;
7941     }
7942   }
7943 
7944   // Finally, we're ready to print the data we've gathered.
7945   outs() << "Contents of __compact_unwind section:\n";
7946   for (auto &Entry : CompactUnwinds) {
7947     outs() << "  Entry at offset "
7948            << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n";
7949 
7950     // 1. Start of the region this entry applies to.
7951     outs() << "    start:                " << format("0x%" PRIx64,
7952                                                      Entry.FunctionAddr) << ' ';
7953     printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr);
7954     outs() << '\n';
7955 
7956     // 2. Length of the region this entry applies to.
7957     outs() << "    length:               " << format("0x%" PRIx32, Entry.Length)
7958            << '\n';
7959     // 3. The 32-bit compact encoding.
7960     outs() << "    compact encoding:     "
7961            << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n';
7962 
7963     // 4. The personality function, if present.
7964     if (Entry.PersonalityReloc.getObject()) {
7965       outs() << "    personality function: "
7966              << format("0x%" PRIx64, Entry.PersonalityAddr) << ' ';
7967       printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc,
7968                            Entry.PersonalityAddr);
7969       outs() << '\n';
7970     }
7971 
7972     // 5. This entry's language-specific data area.
7973     if (Entry.LSDAReloc.getObject()) {
7974       outs() << "    LSDA:                 " << format("0x%" PRIx64,
7975                                                        Entry.LSDAAddr) << ' ';
7976       printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr);
7977       outs() << '\n';
7978     }
7979   }
7980 }
7981 
7982 //===----------------------------------------------------------------------===//
7983 // __unwind_info section dumping
7984 //===----------------------------------------------------------------------===//
7985 
printRegularSecondLevelUnwindPage(StringRef PageData)7986 static void printRegularSecondLevelUnwindPage(StringRef PageData) {
7987   ptrdiff_t Pos = 0;
7988   uint32_t Kind = readNext<uint32_t>(PageData, Pos);
7989   (void)Kind;
7990   assert(Kind == 2 && "kind for a regular 2nd level index should be 2");
7991 
7992   uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos);
7993   uint16_t NumEntries = readNext<uint16_t>(PageData, Pos);
7994 
7995   Pos = EntriesStart;
7996   for (unsigned i = 0; i < NumEntries; ++i) {
7997     uint32_t FunctionOffset = readNext<uint32_t>(PageData, Pos);
7998     uint32_t Encoding = readNext<uint32_t>(PageData, Pos);
7999 
8000     outs() << "      [" << i << "]: "
8001            << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8002            << ", "
8003            << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n';
8004   }
8005 }
8006 
printCompressedSecondLevelUnwindPage(StringRef PageData,uint32_t FunctionBase,const SmallVectorImpl<uint32_t> & CommonEncodings)8007 static void printCompressedSecondLevelUnwindPage(
8008     StringRef PageData, uint32_t FunctionBase,
8009     const SmallVectorImpl<uint32_t> &CommonEncodings) {
8010   ptrdiff_t Pos = 0;
8011   uint32_t Kind = readNext<uint32_t>(PageData, Pos);
8012   (void)Kind;
8013   assert(Kind == 3 && "kind for a compressed 2nd level index should be 3");
8014 
8015   uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos);
8016   uint16_t NumEntries = readNext<uint16_t>(PageData, Pos);
8017 
8018   uint16_t EncodingsStart = readNext<uint16_t>(PageData, Pos);
8019   readNext<uint16_t>(PageData, Pos);
8020   StringRef PageEncodings = PageData.substr(EncodingsStart, StringRef::npos);
8021 
8022   Pos = EntriesStart;
8023   for (unsigned i = 0; i < NumEntries; ++i) {
8024     uint32_t Entry = readNext<uint32_t>(PageData, Pos);
8025     uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff);
8026     uint32_t EncodingIdx = Entry >> 24;
8027 
8028     uint32_t Encoding;
8029     if (EncodingIdx < CommonEncodings.size())
8030       Encoding = CommonEncodings[EncodingIdx];
8031     else
8032       Encoding = read<uint32_t>(PageEncodings,
8033                                 sizeof(uint32_t) *
8034                                     (EncodingIdx - CommonEncodings.size()));
8035 
8036     outs() << "      [" << i << "]: "
8037            << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8038            << ", "
8039            << "encoding[" << EncodingIdx
8040            << "]=" << format("0x%08" PRIx32, Encoding) << '\n';
8041   }
8042 }
8043 
printMachOUnwindInfoSection(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const SectionRef & UnwindInfo)8044 static void printMachOUnwindInfoSection(const MachOObjectFile *Obj,
8045                                         std::map<uint64_t, SymbolRef> &Symbols,
8046                                         const SectionRef &UnwindInfo) {
8047 
8048   if (!Obj->isLittleEndian()) {
8049     outs() << "Skipping big-endian __unwind_info section\n";
8050     return;
8051   }
8052 
8053   outs() << "Contents of __unwind_info section:\n";
8054 
8055   StringRef Contents =
8056       unwrapOrError(UnwindInfo.getContents(), Obj->getFileName());
8057   ptrdiff_t Pos = 0;
8058 
8059   //===----------------------------------
8060   // Section header
8061   //===----------------------------------
8062 
8063   uint32_t Version = readNext<uint32_t>(Contents, Pos);
8064   outs() << "  Version:                                   "
8065          << format("0x%" PRIx32, Version) << '\n';
8066   if (Version != 1) {
8067     outs() << "    Skipping section with unknown version\n";
8068     return;
8069   }
8070 
8071   uint32_t CommonEncodingsStart = readNext<uint32_t>(Contents, Pos);
8072   outs() << "  Common encodings array section offset:     "
8073          << format("0x%" PRIx32, CommonEncodingsStart) << '\n';
8074   uint32_t NumCommonEncodings = readNext<uint32_t>(Contents, Pos);
8075   outs() << "  Number of common encodings in array:       "
8076          << format("0x%" PRIx32, NumCommonEncodings) << '\n';
8077 
8078   uint32_t PersonalitiesStart = readNext<uint32_t>(Contents, Pos);
8079   outs() << "  Personality function array section offset: "
8080          << format("0x%" PRIx32, PersonalitiesStart) << '\n';
8081   uint32_t NumPersonalities = readNext<uint32_t>(Contents, Pos);
8082   outs() << "  Number of personality functions in array:  "
8083          << format("0x%" PRIx32, NumPersonalities) << '\n';
8084 
8085   uint32_t IndicesStart = readNext<uint32_t>(Contents, Pos);
8086   outs() << "  Index array section offset:                "
8087          << format("0x%" PRIx32, IndicesStart) << '\n';
8088   uint32_t NumIndices = readNext<uint32_t>(Contents, Pos);
8089   outs() << "  Number of indices in array:                "
8090          << format("0x%" PRIx32, NumIndices) << '\n';
8091 
8092   //===----------------------------------
8093   // A shared list of common encodings
8094   //===----------------------------------
8095 
8096   // These occupy indices in the range [0, N] whenever an encoding is referenced
8097   // from a compressed 2nd level index table. In practice the linker only
8098   // creates ~128 of these, so that indices are available to embed encodings in
8099   // the 2nd level index.
8100 
8101   SmallVector<uint32_t, 64> CommonEncodings;
8102   outs() << "  Common encodings: (count = " << NumCommonEncodings << ")\n";
8103   Pos = CommonEncodingsStart;
8104   for (unsigned i = 0; i < NumCommonEncodings; ++i) {
8105     uint32_t Encoding = readNext<uint32_t>(Contents, Pos);
8106     CommonEncodings.push_back(Encoding);
8107 
8108     outs() << "    encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding)
8109            << '\n';
8110   }
8111 
8112   //===----------------------------------
8113   // Personality functions used in this executable
8114   //===----------------------------------
8115 
8116   // There should be only a handful of these (one per source language,
8117   // roughly). Particularly since they only get 2 bits in the compact encoding.
8118 
8119   outs() << "  Personality functions: (count = " << NumPersonalities << ")\n";
8120   Pos = PersonalitiesStart;
8121   for (unsigned i = 0; i < NumPersonalities; ++i) {
8122     uint32_t PersonalityFn = readNext<uint32_t>(Contents, Pos);
8123     outs() << "    personality[" << i + 1
8124            << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n';
8125   }
8126 
8127   //===----------------------------------
8128   // The level 1 index entries
8129   //===----------------------------------
8130 
8131   // These specify an approximate place to start searching for the more detailed
8132   // information, sorted by PC.
8133 
8134   struct IndexEntry {
8135     uint32_t FunctionOffset;
8136     uint32_t SecondLevelPageStart;
8137     uint32_t LSDAStart;
8138   };
8139 
8140   SmallVector<IndexEntry, 4> IndexEntries;
8141 
8142   outs() << "  Top level indices: (count = " << NumIndices << ")\n";
8143   Pos = IndicesStart;
8144   for (unsigned i = 0; i < NumIndices; ++i) {
8145     IndexEntry Entry;
8146 
8147     Entry.FunctionOffset = readNext<uint32_t>(Contents, Pos);
8148     Entry.SecondLevelPageStart = readNext<uint32_t>(Contents, Pos);
8149     Entry.LSDAStart = readNext<uint32_t>(Contents, Pos);
8150     IndexEntries.push_back(Entry);
8151 
8152     outs() << "    [" << i << "]: "
8153            << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset)
8154            << ", "
8155            << "2nd level page offset="
8156            << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", "
8157            << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n';
8158   }
8159 
8160   //===----------------------------------
8161   // Next come the LSDA tables
8162   //===----------------------------------
8163 
8164   // The LSDA layout is rather implicit: it's a contiguous array of entries from
8165   // the first top-level index's LSDAOffset to the last (sentinel).
8166 
8167   outs() << "  LSDA descriptors:\n";
8168   Pos = IndexEntries[0].LSDAStart;
8169   const uint32_t LSDASize = 2 * sizeof(uint32_t);
8170   int NumLSDAs =
8171       (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) / LSDASize;
8172 
8173   for (int i = 0; i < NumLSDAs; ++i) {
8174     uint32_t FunctionOffset = readNext<uint32_t>(Contents, Pos);
8175     uint32_t LSDAOffset = readNext<uint32_t>(Contents, Pos);
8176     outs() << "    [" << i << "]: "
8177            << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8178            << ", "
8179            << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n';
8180   }
8181 
8182   //===----------------------------------
8183   // Finally, the 2nd level indices
8184   //===----------------------------------
8185 
8186   // Generally these are 4K in size, and have 2 possible forms:
8187   //   + Regular stores up to 511 entries with disparate encodings
8188   //   + Compressed stores up to 1021 entries if few enough compact encoding
8189   //     values are used.
8190   outs() << "  Second level indices:\n";
8191   for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) {
8192     // The final sentinel top-level index has no associated 2nd level page
8193     if (IndexEntries[i].SecondLevelPageStart == 0)
8194       break;
8195 
8196     outs() << "    Second level index[" << i << "]: "
8197            << "offset in section="
8198            << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart)
8199            << ", "
8200            << "base function offset="
8201            << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n';
8202 
8203     Pos = IndexEntries[i].SecondLevelPageStart;
8204     if (Pos + sizeof(uint32_t) > Contents.size()) {
8205       outs() << "warning: invalid offset for second level page: " << Pos << '\n';
8206       continue;
8207     }
8208 
8209     uint32_t Kind =
8210         *reinterpret_cast<const support::ulittle32_t *>(Contents.data() + Pos);
8211     if (Kind == 2)
8212       printRegularSecondLevelUnwindPage(Contents.substr(Pos, 4096));
8213     else if (Kind == 3)
8214       printCompressedSecondLevelUnwindPage(Contents.substr(Pos, 4096),
8215                                            IndexEntries[i].FunctionOffset,
8216                                            CommonEncodings);
8217     else
8218       outs() << "    Skipping 2nd level page with unknown kind " << Kind
8219              << '\n';
8220   }
8221 }
8222 
printMachOUnwindInfo(const MachOObjectFile * Obj)8223 void objdump::printMachOUnwindInfo(const MachOObjectFile *Obj) {
8224   std::map<uint64_t, SymbolRef> Symbols;
8225   for (const SymbolRef &SymRef : Obj->symbols()) {
8226     // Discard any undefined or absolute symbols. They're not going to take part
8227     // in the convenience lookup for unwind info and just take up resources.
8228     auto SectOrErr = SymRef.getSection();
8229     if (!SectOrErr) {
8230       // TODO: Actually report errors helpfully.
8231       consumeError(SectOrErr.takeError());
8232       continue;
8233     }
8234     section_iterator Section = *SectOrErr;
8235     if (Section == Obj->section_end())
8236       continue;
8237 
8238     uint64_t Addr = cantFail(SymRef.getValue());
8239     Symbols.insert(std::make_pair(Addr, SymRef));
8240   }
8241 
8242   for (const SectionRef &Section : Obj->sections()) {
8243     StringRef SectName;
8244     if (Expected<StringRef> NameOrErr = Section.getName())
8245       SectName = *NameOrErr;
8246     else
8247       consumeError(NameOrErr.takeError());
8248 
8249     if (SectName == "__compact_unwind")
8250       printMachOCompactUnwindSection(Obj, Symbols, Section);
8251     else if (SectName == "__unwind_info")
8252       printMachOUnwindInfoSection(Obj, Symbols, Section);
8253   }
8254 }
8255 
PrintMachHeader(uint32_t magic,uint32_t cputype,uint32_t cpusubtype,uint32_t filetype,uint32_t ncmds,uint32_t sizeofcmds,uint32_t flags,bool verbose)8256 static void PrintMachHeader(uint32_t magic, uint32_t cputype,
8257                             uint32_t cpusubtype, uint32_t filetype,
8258                             uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags,
8259                             bool verbose) {
8260   outs() << "Mach header\n";
8261   outs() << "      magic cputype cpusubtype  caps    filetype ncmds "
8262             "sizeofcmds      flags\n";
8263   if (verbose) {
8264     if (magic == MachO::MH_MAGIC)
8265       outs() << "   MH_MAGIC";
8266     else if (magic == MachO::MH_MAGIC_64)
8267       outs() << "MH_MAGIC_64";
8268     else
8269       outs() << format(" 0x%08" PRIx32, magic);
8270     switch (cputype) {
8271     case MachO::CPU_TYPE_I386:
8272       outs() << "    I386";
8273       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8274       case MachO::CPU_SUBTYPE_I386_ALL:
8275         outs() << "        ALL";
8276         break;
8277       default:
8278         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8279         break;
8280       }
8281       break;
8282     case MachO::CPU_TYPE_X86_64:
8283       outs() << "  X86_64";
8284       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8285       case MachO::CPU_SUBTYPE_X86_64_ALL:
8286         outs() << "        ALL";
8287         break;
8288       case MachO::CPU_SUBTYPE_X86_64_H:
8289         outs() << "    Haswell";
8290         break;
8291       default:
8292         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8293         break;
8294       }
8295       break;
8296     case MachO::CPU_TYPE_ARM:
8297       outs() << "     ARM";
8298       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8299       case MachO::CPU_SUBTYPE_ARM_ALL:
8300         outs() << "        ALL";
8301         break;
8302       case MachO::CPU_SUBTYPE_ARM_V4T:
8303         outs() << "        V4T";
8304         break;
8305       case MachO::CPU_SUBTYPE_ARM_V5TEJ:
8306         outs() << "      V5TEJ";
8307         break;
8308       case MachO::CPU_SUBTYPE_ARM_XSCALE:
8309         outs() << "     XSCALE";
8310         break;
8311       case MachO::CPU_SUBTYPE_ARM_V6:
8312         outs() << "         V6";
8313         break;
8314       case MachO::CPU_SUBTYPE_ARM_V6M:
8315         outs() << "        V6M";
8316         break;
8317       case MachO::CPU_SUBTYPE_ARM_V7:
8318         outs() << "         V7";
8319         break;
8320       case MachO::CPU_SUBTYPE_ARM_V7EM:
8321         outs() << "       V7EM";
8322         break;
8323       case MachO::CPU_SUBTYPE_ARM_V7K:
8324         outs() << "        V7K";
8325         break;
8326       case MachO::CPU_SUBTYPE_ARM_V7M:
8327         outs() << "        V7M";
8328         break;
8329       case MachO::CPU_SUBTYPE_ARM_V7S:
8330         outs() << "        V7S";
8331         break;
8332       default:
8333         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8334         break;
8335       }
8336       break;
8337     case MachO::CPU_TYPE_ARM64:
8338       outs() << "   ARM64";
8339       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8340       case MachO::CPU_SUBTYPE_ARM64_ALL:
8341         outs() << "        ALL";
8342         break;
8343       case MachO::CPU_SUBTYPE_ARM64_V8:
8344         outs() << "         V8";
8345         break;
8346       case MachO::CPU_SUBTYPE_ARM64E:
8347         outs() << "          E";
8348         break;
8349       default:
8350         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8351         break;
8352       }
8353       break;
8354     case MachO::CPU_TYPE_ARM64_32:
8355       outs() << " ARM64_32";
8356       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8357       case MachO::CPU_SUBTYPE_ARM64_32_V8:
8358         outs() << "        V8";
8359         break;
8360       default:
8361         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8362         break;
8363       }
8364       break;
8365     case MachO::CPU_TYPE_POWERPC:
8366       outs() << "     PPC";
8367       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8368       case MachO::CPU_SUBTYPE_POWERPC_ALL:
8369         outs() << "        ALL";
8370         break;
8371       default:
8372         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8373         break;
8374       }
8375       break;
8376     case MachO::CPU_TYPE_POWERPC64:
8377       outs() << "   PPC64";
8378       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8379       case MachO::CPU_SUBTYPE_POWERPC_ALL:
8380         outs() << "        ALL";
8381         break;
8382       default:
8383         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8384         break;
8385       }
8386       break;
8387     default:
8388       outs() << format(" %7d", cputype);
8389       outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8390       break;
8391     }
8392     if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) {
8393       outs() << " LIB64";
8394     } else {
8395       outs() << format("  0x%02" PRIx32,
8396                        (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
8397     }
8398     switch (filetype) {
8399     case MachO::MH_OBJECT:
8400       outs() << "      OBJECT";
8401       break;
8402     case MachO::MH_EXECUTE:
8403       outs() << "     EXECUTE";
8404       break;
8405     case MachO::MH_FVMLIB:
8406       outs() << "      FVMLIB";
8407       break;
8408     case MachO::MH_CORE:
8409       outs() << "        CORE";
8410       break;
8411     case MachO::MH_PRELOAD:
8412       outs() << "     PRELOAD";
8413       break;
8414     case MachO::MH_DYLIB:
8415       outs() << "       DYLIB";
8416       break;
8417     case MachO::MH_DYLIB_STUB:
8418       outs() << "  DYLIB_STUB";
8419       break;
8420     case MachO::MH_DYLINKER:
8421       outs() << "    DYLINKER";
8422       break;
8423     case MachO::MH_BUNDLE:
8424       outs() << "      BUNDLE";
8425       break;
8426     case MachO::MH_DSYM:
8427       outs() << "        DSYM";
8428       break;
8429     case MachO::MH_KEXT_BUNDLE:
8430       outs() << "  KEXTBUNDLE";
8431       break;
8432     default:
8433       outs() << format("  %10u", filetype);
8434       break;
8435     }
8436     outs() << format(" %5u", ncmds);
8437     outs() << format(" %10u", sizeofcmds);
8438     uint32_t f = flags;
8439     if (f & MachO::MH_NOUNDEFS) {
8440       outs() << "   NOUNDEFS";
8441       f &= ~MachO::MH_NOUNDEFS;
8442     }
8443     if (f & MachO::MH_INCRLINK) {
8444       outs() << " INCRLINK";
8445       f &= ~MachO::MH_INCRLINK;
8446     }
8447     if (f & MachO::MH_DYLDLINK) {
8448       outs() << " DYLDLINK";
8449       f &= ~MachO::MH_DYLDLINK;
8450     }
8451     if (f & MachO::MH_BINDATLOAD) {
8452       outs() << " BINDATLOAD";
8453       f &= ~MachO::MH_BINDATLOAD;
8454     }
8455     if (f & MachO::MH_PREBOUND) {
8456       outs() << " PREBOUND";
8457       f &= ~MachO::MH_PREBOUND;
8458     }
8459     if (f & MachO::MH_SPLIT_SEGS) {
8460       outs() << " SPLIT_SEGS";
8461       f &= ~MachO::MH_SPLIT_SEGS;
8462     }
8463     if (f & MachO::MH_LAZY_INIT) {
8464       outs() << " LAZY_INIT";
8465       f &= ~MachO::MH_LAZY_INIT;
8466     }
8467     if (f & MachO::MH_TWOLEVEL) {
8468       outs() << " TWOLEVEL";
8469       f &= ~MachO::MH_TWOLEVEL;
8470     }
8471     if (f & MachO::MH_FORCE_FLAT) {
8472       outs() << " FORCE_FLAT";
8473       f &= ~MachO::MH_FORCE_FLAT;
8474     }
8475     if (f & MachO::MH_NOMULTIDEFS) {
8476       outs() << " NOMULTIDEFS";
8477       f &= ~MachO::MH_NOMULTIDEFS;
8478     }
8479     if (f & MachO::MH_NOFIXPREBINDING) {
8480       outs() << " NOFIXPREBINDING";
8481       f &= ~MachO::MH_NOFIXPREBINDING;
8482     }
8483     if (f & MachO::MH_PREBINDABLE) {
8484       outs() << " PREBINDABLE";
8485       f &= ~MachO::MH_PREBINDABLE;
8486     }
8487     if (f & MachO::MH_ALLMODSBOUND) {
8488       outs() << " ALLMODSBOUND";
8489       f &= ~MachO::MH_ALLMODSBOUND;
8490     }
8491     if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) {
8492       outs() << " SUBSECTIONS_VIA_SYMBOLS";
8493       f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
8494     }
8495     if (f & MachO::MH_CANONICAL) {
8496       outs() << " CANONICAL";
8497       f &= ~MachO::MH_CANONICAL;
8498     }
8499     if (f & MachO::MH_WEAK_DEFINES) {
8500       outs() << " WEAK_DEFINES";
8501       f &= ~MachO::MH_WEAK_DEFINES;
8502     }
8503     if (f & MachO::MH_BINDS_TO_WEAK) {
8504       outs() << " BINDS_TO_WEAK";
8505       f &= ~MachO::MH_BINDS_TO_WEAK;
8506     }
8507     if (f & MachO::MH_ALLOW_STACK_EXECUTION) {
8508       outs() << " ALLOW_STACK_EXECUTION";
8509       f &= ~MachO::MH_ALLOW_STACK_EXECUTION;
8510     }
8511     if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) {
8512       outs() << " DEAD_STRIPPABLE_DYLIB";
8513       f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB;
8514     }
8515     if (f & MachO::MH_PIE) {
8516       outs() << " PIE";
8517       f &= ~MachO::MH_PIE;
8518     }
8519     if (f & MachO::MH_NO_REEXPORTED_DYLIBS) {
8520       outs() << " NO_REEXPORTED_DYLIBS";
8521       f &= ~MachO::MH_NO_REEXPORTED_DYLIBS;
8522     }
8523     if (f & MachO::MH_HAS_TLV_DESCRIPTORS) {
8524       outs() << " MH_HAS_TLV_DESCRIPTORS";
8525       f &= ~MachO::MH_HAS_TLV_DESCRIPTORS;
8526     }
8527     if (f & MachO::MH_NO_HEAP_EXECUTION) {
8528       outs() << " MH_NO_HEAP_EXECUTION";
8529       f &= ~MachO::MH_NO_HEAP_EXECUTION;
8530     }
8531     if (f & MachO::MH_APP_EXTENSION_SAFE) {
8532       outs() << " APP_EXTENSION_SAFE";
8533       f &= ~MachO::MH_APP_EXTENSION_SAFE;
8534     }
8535     if (f & MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO) {
8536       outs() << " NLIST_OUTOFSYNC_WITH_DYLDINFO";
8537       f &= ~MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO;
8538     }
8539     if (f != 0 || flags == 0)
8540       outs() << format(" 0x%08" PRIx32, f);
8541   } else {
8542     outs() << format(" 0x%08" PRIx32, magic);
8543     outs() << format(" %7d", cputype);
8544     outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8545     outs() << format("  0x%02" PRIx32,
8546                      (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
8547     outs() << format("  %10u", filetype);
8548     outs() << format(" %5u", ncmds);
8549     outs() << format(" %10u", sizeofcmds);
8550     outs() << format(" 0x%08" PRIx32, flags);
8551   }
8552   outs() << "\n";
8553 }
8554 
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)8555 static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize,
8556                                 StringRef SegName, uint64_t vmaddr,
8557                                 uint64_t vmsize, uint64_t fileoff,
8558                                 uint64_t filesize, uint32_t maxprot,
8559                                 uint32_t initprot, uint32_t nsects,
8560                                 uint32_t flags, uint32_t object_size,
8561                                 bool verbose) {
8562   uint64_t expected_cmdsize;
8563   if (cmd == MachO::LC_SEGMENT) {
8564     outs() << "      cmd LC_SEGMENT\n";
8565     expected_cmdsize = nsects;
8566     expected_cmdsize *= sizeof(struct MachO::section);
8567     expected_cmdsize += sizeof(struct MachO::segment_command);
8568   } else {
8569     outs() << "      cmd LC_SEGMENT_64\n";
8570     expected_cmdsize = nsects;
8571     expected_cmdsize *= sizeof(struct MachO::section_64);
8572     expected_cmdsize += sizeof(struct MachO::segment_command_64);
8573   }
8574   outs() << "  cmdsize " << cmdsize;
8575   if (cmdsize != expected_cmdsize)
8576     outs() << " Inconsistent size\n";
8577   else
8578     outs() << "\n";
8579   outs() << "  segname " << SegName << "\n";
8580   if (cmd == MachO::LC_SEGMENT_64) {
8581     outs() << "   vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n";
8582     outs() << "   vmsize " << format("0x%016" PRIx64, vmsize) << "\n";
8583   } else {
8584     outs() << "   vmaddr " << format("0x%08" PRIx64, vmaddr) << "\n";
8585     outs() << "   vmsize " << format("0x%08" PRIx64, vmsize) << "\n";
8586   }
8587   outs() << "  fileoff " << fileoff;
8588   if (fileoff > object_size)
8589     outs() << " (past end of file)\n";
8590   else
8591     outs() << "\n";
8592   outs() << " filesize " << filesize;
8593   if (fileoff + filesize > object_size)
8594     outs() << " (past end of file)\n";
8595   else
8596     outs() << "\n";
8597   if (verbose) {
8598     if ((maxprot &
8599          ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
8600            MachO::VM_PROT_EXECUTE)) != 0)
8601       outs() << "  maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n";
8602     else {
8603       outs() << "  maxprot ";
8604       outs() << ((maxprot & MachO::VM_PROT_READ) ? "r" : "-");
8605       outs() << ((maxprot & MachO::VM_PROT_WRITE) ? "w" : "-");
8606       outs() << ((maxprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n");
8607     }
8608     if ((initprot &
8609          ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
8610            MachO::VM_PROT_EXECUTE)) != 0)
8611       outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n";
8612     else {
8613       outs() << " initprot ";
8614       outs() << ((initprot & MachO::VM_PROT_READ) ? "r" : "-");
8615       outs() << ((initprot & MachO::VM_PROT_WRITE) ? "w" : "-");
8616       outs() << ((initprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n");
8617     }
8618   } else {
8619     outs() << "  maxprot " << format("0x%08" PRIx32, maxprot) << "\n";
8620     outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n";
8621   }
8622   outs() << "   nsects " << nsects << "\n";
8623   if (verbose) {
8624     outs() << "    flags";
8625     if (flags == 0)
8626       outs() << " (none)\n";
8627     else {
8628       if (flags & MachO::SG_HIGHVM) {
8629         outs() << " HIGHVM";
8630         flags &= ~MachO::SG_HIGHVM;
8631       }
8632       if (flags & MachO::SG_FVMLIB) {
8633         outs() << " FVMLIB";
8634         flags &= ~MachO::SG_FVMLIB;
8635       }
8636       if (flags & MachO::SG_NORELOC) {
8637         outs() << " NORELOC";
8638         flags &= ~MachO::SG_NORELOC;
8639       }
8640       if (flags & MachO::SG_PROTECTED_VERSION_1) {
8641         outs() << " PROTECTED_VERSION_1";
8642         flags &= ~MachO::SG_PROTECTED_VERSION_1;
8643       }
8644       if (flags)
8645         outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n";
8646       else
8647         outs() << "\n";
8648     }
8649   } else {
8650     outs() << "    flags " << format("0x%" PRIx32, flags) << "\n";
8651   }
8652 }
8653 
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)8654 static void PrintSection(const char *sectname, const char *segname,
8655                          uint64_t addr, uint64_t size, uint32_t offset,
8656                          uint32_t align, uint32_t reloff, uint32_t nreloc,
8657                          uint32_t flags, uint32_t reserved1, uint32_t reserved2,
8658                          uint32_t cmd, const char *sg_segname,
8659                          uint32_t filetype, uint32_t object_size,
8660                          bool verbose) {
8661   outs() << "Section\n";
8662   outs() << "  sectname " << format("%.16s\n", sectname);
8663   outs() << "   segname " << format("%.16s", segname);
8664   if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0)
8665     outs() << " (does not match segment)\n";
8666   else
8667     outs() << "\n";
8668   if (cmd == MachO::LC_SEGMENT_64) {
8669     outs() << "      addr " << format("0x%016" PRIx64, addr) << "\n";
8670     outs() << "      size " << format("0x%016" PRIx64, size);
8671   } else {
8672     outs() << "      addr " << format("0x%08" PRIx64, addr) << "\n";
8673     outs() << "      size " << format("0x%08" PRIx64, size);
8674   }
8675   if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size)
8676     outs() << " (past end of file)\n";
8677   else
8678     outs() << "\n";
8679   outs() << "    offset " << offset;
8680   if (offset > object_size)
8681     outs() << " (past end of file)\n";
8682   else
8683     outs() << "\n";
8684   uint32_t align_shifted = 1 << align;
8685   outs() << "     align 2^" << align << " (" << align_shifted << ")\n";
8686   outs() << "    reloff " << reloff;
8687   if (reloff > object_size)
8688     outs() << " (past end of file)\n";
8689   else
8690     outs() << "\n";
8691   outs() << "    nreloc " << nreloc;
8692   if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size)
8693     outs() << " (past end of file)\n";
8694   else
8695     outs() << "\n";
8696   uint32_t section_type = flags & MachO::SECTION_TYPE;
8697   if (verbose) {
8698     outs() << "      type";
8699     if (section_type == MachO::S_REGULAR)
8700       outs() << " S_REGULAR\n";
8701     else if (section_type == MachO::S_ZEROFILL)
8702       outs() << " S_ZEROFILL\n";
8703     else if (section_type == MachO::S_CSTRING_LITERALS)
8704       outs() << " S_CSTRING_LITERALS\n";
8705     else if (section_type == MachO::S_4BYTE_LITERALS)
8706       outs() << " S_4BYTE_LITERALS\n";
8707     else if (section_type == MachO::S_8BYTE_LITERALS)
8708       outs() << " S_8BYTE_LITERALS\n";
8709     else if (section_type == MachO::S_16BYTE_LITERALS)
8710       outs() << " S_16BYTE_LITERALS\n";
8711     else if (section_type == MachO::S_LITERAL_POINTERS)
8712       outs() << " S_LITERAL_POINTERS\n";
8713     else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS)
8714       outs() << " S_NON_LAZY_SYMBOL_POINTERS\n";
8715     else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS)
8716       outs() << " S_LAZY_SYMBOL_POINTERS\n";
8717     else if (section_type == MachO::S_SYMBOL_STUBS)
8718       outs() << " S_SYMBOL_STUBS\n";
8719     else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS)
8720       outs() << " S_MOD_INIT_FUNC_POINTERS\n";
8721     else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS)
8722       outs() << " S_MOD_TERM_FUNC_POINTERS\n";
8723     else if (section_type == MachO::S_COALESCED)
8724       outs() << " S_COALESCED\n";
8725     else if (section_type == MachO::S_INTERPOSING)
8726       outs() << " S_INTERPOSING\n";
8727     else if (section_type == MachO::S_DTRACE_DOF)
8728       outs() << " S_DTRACE_DOF\n";
8729     else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS)
8730       outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n";
8731     else if (section_type == MachO::S_THREAD_LOCAL_REGULAR)
8732       outs() << " S_THREAD_LOCAL_REGULAR\n";
8733     else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL)
8734       outs() << " S_THREAD_LOCAL_ZEROFILL\n";
8735     else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES)
8736       outs() << " S_THREAD_LOCAL_VARIABLES\n";
8737     else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
8738       outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n";
8739     else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS)
8740       outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n";
8741     else
8742       outs() << format("0x%08" PRIx32, section_type) << "\n";
8743     outs() << "attributes";
8744     uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES;
8745     if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS)
8746       outs() << " PURE_INSTRUCTIONS";
8747     if (section_attributes & MachO::S_ATTR_NO_TOC)
8748       outs() << " NO_TOC";
8749     if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS)
8750       outs() << " STRIP_STATIC_SYMS";
8751     if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP)
8752       outs() << " NO_DEAD_STRIP";
8753     if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT)
8754       outs() << " LIVE_SUPPORT";
8755     if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE)
8756       outs() << " SELF_MODIFYING_CODE";
8757     if (section_attributes & MachO::S_ATTR_DEBUG)
8758       outs() << " DEBUG";
8759     if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS)
8760       outs() << " SOME_INSTRUCTIONS";
8761     if (section_attributes & MachO::S_ATTR_EXT_RELOC)
8762       outs() << " EXT_RELOC";
8763     if (section_attributes & MachO::S_ATTR_LOC_RELOC)
8764       outs() << " LOC_RELOC";
8765     if (section_attributes == 0)
8766       outs() << " (none)";
8767     outs() << "\n";
8768   } else
8769     outs() << "     flags " << format("0x%08" PRIx32, flags) << "\n";
8770   outs() << " reserved1 " << reserved1;
8771   if (section_type == MachO::S_SYMBOL_STUBS ||
8772       section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
8773       section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
8774       section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
8775       section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
8776     outs() << " (index into indirect symbol table)\n";
8777   else
8778     outs() << "\n";
8779   outs() << " reserved2 " << reserved2;
8780   if (section_type == MachO::S_SYMBOL_STUBS)
8781     outs() << " (size of stubs)\n";
8782   else
8783     outs() << "\n";
8784 }
8785 
PrintSymtabLoadCommand(MachO::symtab_command st,bool Is64Bit,uint32_t object_size)8786 static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit,
8787                                    uint32_t object_size) {
8788   outs() << "     cmd LC_SYMTAB\n";
8789   outs() << " cmdsize " << st.cmdsize;
8790   if (st.cmdsize != sizeof(struct MachO::symtab_command))
8791     outs() << " Incorrect size\n";
8792   else
8793     outs() << "\n";
8794   outs() << "  symoff " << st.symoff;
8795   if (st.symoff > object_size)
8796     outs() << " (past end of file)\n";
8797   else
8798     outs() << "\n";
8799   outs() << "   nsyms " << st.nsyms;
8800   uint64_t big_size;
8801   if (Is64Bit) {
8802     big_size = st.nsyms;
8803     big_size *= sizeof(struct MachO::nlist_64);
8804     big_size += st.symoff;
8805     if (big_size > object_size)
8806       outs() << " (past end of file)\n";
8807     else
8808       outs() << "\n";
8809   } else {
8810     big_size = st.nsyms;
8811     big_size *= sizeof(struct MachO::nlist);
8812     big_size += st.symoff;
8813     if (big_size > object_size)
8814       outs() << " (past end of file)\n";
8815     else
8816       outs() << "\n";
8817   }
8818   outs() << "  stroff " << st.stroff;
8819   if (st.stroff > object_size)
8820     outs() << " (past end of file)\n";
8821   else
8822     outs() << "\n";
8823   outs() << " strsize " << st.strsize;
8824   big_size = st.stroff;
8825   big_size += st.strsize;
8826   if (big_size > object_size)
8827     outs() << " (past end of file)\n";
8828   else
8829     outs() << "\n";
8830 }
8831 
PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,uint32_t nsyms,uint32_t object_size,bool Is64Bit)8832 static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,
8833                                      uint32_t nsyms, uint32_t object_size,
8834                                      bool Is64Bit) {
8835   outs() << "            cmd LC_DYSYMTAB\n";
8836   outs() << "        cmdsize " << dyst.cmdsize;
8837   if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command))
8838     outs() << " Incorrect size\n";
8839   else
8840     outs() << "\n";
8841   outs() << "      ilocalsym " << dyst.ilocalsym;
8842   if (dyst.ilocalsym > nsyms)
8843     outs() << " (greater than the number of symbols)\n";
8844   else
8845     outs() << "\n";
8846   outs() << "      nlocalsym " << dyst.nlocalsym;
8847   uint64_t big_size;
8848   big_size = dyst.ilocalsym;
8849   big_size += dyst.nlocalsym;
8850   if (big_size > nsyms)
8851     outs() << " (past the end of the symbol table)\n";
8852   else
8853     outs() << "\n";
8854   outs() << "     iextdefsym " << dyst.iextdefsym;
8855   if (dyst.iextdefsym > nsyms)
8856     outs() << " (greater than the number of symbols)\n";
8857   else
8858     outs() << "\n";
8859   outs() << "     nextdefsym " << dyst.nextdefsym;
8860   big_size = dyst.iextdefsym;
8861   big_size += dyst.nextdefsym;
8862   if (big_size > nsyms)
8863     outs() << " (past the end of the symbol table)\n";
8864   else
8865     outs() << "\n";
8866   outs() << "      iundefsym " << dyst.iundefsym;
8867   if (dyst.iundefsym > nsyms)
8868     outs() << " (greater than the number of symbols)\n";
8869   else
8870     outs() << "\n";
8871   outs() << "      nundefsym " << dyst.nundefsym;
8872   big_size = dyst.iundefsym;
8873   big_size += dyst.nundefsym;
8874   if (big_size > nsyms)
8875     outs() << " (past the end of the symbol table)\n";
8876   else
8877     outs() << "\n";
8878   outs() << "         tocoff " << dyst.tocoff;
8879   if (dyst.tocoff > object_size)
8880     outs() << " (past end of file)\n";
8881   else
8882     outs() << "\n";
8883   outs() << "           ntoc " << dyst.ntoc;
8884   big_size = dyst.ntoc;
8885   big_size *= sizeof(struct MachO::dylib_table_of_contents);
8886   big_size += dyst.tocoff;
8887   if (big_size > object_size)
8888     outs() << " (past end of file)\n";
8889   else
8890     outs() << "\n";
8891   outs() << "      modtaboff " << dyst.modtaboff;
8892   if (dyst.modtaboff > object_size)
8893     outs() << " (past end of file)\n";
8894   else
8895     outs() << "\n";
8896   outs() << "        nmodtab " << dyst.nmodtab;
8897   uint64_t modtabend;
8898   if (Is64Bit) {
8899     modtabend = dyst.nmodtab;
8900     modtabend *= sizeof(struct MachO::dylib_module_64);
8901     modtabend += dyst.modtaboff;
8902   } else {
8903     modtabend = dyst.nmodtab;
8904     modtabend *= sizeof(struct MachO::dylib_module);
8905     modtabend += dyst.modtaboff;
8906   }
8907   if (modtabend > object_size)
8908     outs() << " (past end of file)\n";
8909   else
8910     outs() << "\n";
8911   outs() << "   extrefsymoff " << dyst.extrefsymoff;
8912   if (dyst.extrefsymoff > object_size)
8913     outs() << " (past end of file)\n";
8914   else
8915     outs() << "\n";
8916   outs() << "    nextrefsyms " << dyst.nextrefsyms;
8917   big_size = dyst.nextrefsyms;
8918   big_size *= sizeof(struct MachO::dylib_reference);
8919   big_size += dyst.extrefsymoff;
8920   if (big_size > object_size)
8921     outs() << " (past end of file)\n";
8922   else
8923     outs() << "\n";
8924   outs() << " indirectsymoff " << dyst.indirectsymoff;
8925   if (dyst.indirectsymoff > object_size)
8926     outs() << " (past end of file)\n";
8927   else
8928     outs() << "\n";
8929   outs() << "  nindirectsyms " << dyst.nindirectsyms;
8930   big_size = dyst.nindirectsyms;
8931   big_size *= sizeof(uint32_t);
8932   big_size += dyst.indirectsymoff;
8933   if (big_size > object_size)
8934     outs() << " (past end of file)\n";
8935   else
8936     outs() << "\n";
8937   outs() << "      extreloff " << dyst.extreloff;
8938   if (dyst.extreloff > object_size)
8939     outs() << " (past end of file)\n";
8940   else
8941     outs() << "\n";
8942   outs() << "        nextrel " << dyst.nextrel;
8943   big_size = dyst.nextrel;
8944   big_size *= sizeof(struct MachO::relocation_info);
8945   big_size += dyst.extreloff;
8946   if (big_size > object_size)
8947     outs() << " (past end of file)\n";
8948   else
8949     outs() << "\n";
8950   outs() << "      locreloff " << dyst.locreloff;
8951   if (dyst.locreloff > object_size)
8952     outs() << " (past end of file)\n";
8953   else
8954     outs() << "\n";
8955   outs() << "        nlocrel " << dyst.nlocrel;
8956   big_size = dyst.nlocrel;
8957   big_size *= sizeof(struct MachO::relocation_info);
8958   big_size += dyst.locreloff;
8959   if (big_size > object_size)
8960     outs() << " (past end of file)\n";
8961   else
8962     outs() << "\n";
8963 }
8964 
PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,uint32_t object_size)8965 static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,
8966                                      uint32_t object_size) {
8967   if (dc.cmd == MachO::LC_DYLD_INFO)
8968     outs() << "            cmd LC_DYLD_INFO\n";
8969   else
8970     outs() << "            cmd LC_DYLD_INFO_ONLY\n";
8971   outs() << "        cmdsize " << dc.cmdsize;
8972   if (dc.cmdsize != sizeof(struct MachO::dyld_info_command))
8973     outs() << " Incorrect size\n";
8974   else
8975     outs() << "\n";
8976   outs() << "     rebase_off " << dc.rebase_off;
8977   if (dc.rebase_off > object_size)
8978     outs() << " (past end of file)\n";
8979   else
8980     outs() << "\n";
8981   outs() << "    rebase_size " << dc.rebase_size;
8982   uint64_t big_size;
8983   big_size = dc.rebase_off;
8984   big_size += dc.rebase_size;
8985   if (big_size > object_size)
8986     outs() << " (past end of file)\n";
8987   else
8988     outs() << "\n";
8989   outs() << "       bind_off " << dc.bind_off;
8990   if (dc.bind_off > object_size)
8991     outs() << " (past end of file)\n";
8992   else
8993     outs() << "\n";
8994   outs() << "      bind_size " << dc.bind_size;
8995   big_size = dc.bind_off;
8996   big_size += dc.bind_size;
8997   if (big_size > object_size)
8998     outs() << " (past end of file)\n";
8999   else
9000     outs() << "\n";
9001   outs() << "  weak_bind_off " << dc.weak_bind_off;
9002   if (dc.weak_bind_off > object_size)
9003     outs() << " (past end of file)\n";
9004   else
9005     outs() << "\n";
9006   outs() << " weak_bind_size " << dc.weak_bind_size;
9007   big_size = dc.weak_bind_off;
9008   big_size += dc.weak_bind_size;
9009   if (big_size > object_size)
9010     outs() << " (past end of file)\n";
9011   else
9012     outs() << "\n";
9013   outs() << "  lazy_bind_off " << dc.lazy_bind_off;
9014   if (dc.lazy_bind_off > object_size)
9015     outs() << " (past end of file)\n";
9016   else
9017     outs() << "\n";
9018   outs() << " lazy_bind_size " << dc.lazy_bind_size;
9019   big_size = dc.lazy_bind_off;
9020   big_size += dc.lazy_bind_size;
9021   if (big_size > object_size)
9022     outs() << " (past end of file)\n";
9023   else
9024     outs() << "\n";
9025   outs() << "     export_off " << dc.export_off;
9026   if (dc.export_off > object_size)
9027     outs() << " (past end of file)\n";
9028   else
9029     outs() << "\n";
9030   outs() << "    export_size " << dc.export_size;
9031   big_size = dc.export_off;
9032   big_size += dc.export_size;
9033   if (big_size > object_size)
9034     outs() << " (past end of file)\n";
9035   else
9036     outs() << "\n";
9037 }
9038 
PrintDyldLoadCommand(MachO::dylinker_command dyld,const char * Ptr)9039 static void PrintDyldLoadCommand(MachO::dylinker_command dyld,
9040                                  const char *Ptr) {
9041   if (dyld.cmd == MachO::LC_ID_DYLINKER)
9042     outs() << "          cmd LC_ID_DYLINKER\n";
9043   else if (dyld.cmd == MachO::LC_LOAD_DYLINKER)
9044     outs() << "          cmd LC_LOAD_DYLINKER\n";
9045   else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT)
9046     outs() << "          cmd LC_DYLD_ENVIRONMENT\n";
9047   else
9048     outs() << "          cmd ?(" << dyld.cmd << ")\n";
9049   outs() << "      cmdsize " << dyld.cmdsize;
9050   if (dyld.cmdsize < sizeof(struct MachO::dylinker_command))
9051     outs() << " Incorrect size\n";
9052   else
9053     outs() << "\n";
9054   if (dyld.name >= dyld.cmdsize)
9055     outs() << "         name ?(bad offset " << dyld.name << ")\n";
9056   else {
9057     const char *P = (const char *)(Ptr) + dyld.name;
9058     outs() << "         name " << P << " (offset " << dyld.name << ")\n";
9059   }
9060 }
9061 
PrintUuidLoadCommand(MachO::uuid_command uuid)9062 static void PrintUuidLoadCommand(MachO::uuid_command uuid) {
9063   outs() << "     cmd LC_UUID\n";
9064   outs() << " cmdsize " << uuid.cmdsize;
9065   if (uuid.cmdsize != sizeof(struct MachO::uuid_command))
9066     outs() << " Incorrect size\n";
9067   else
9068     outs() << "\n";
9069   outs() << "    uuid ";
9070   for (int i = 0; i < 16; ++i) {
9071     outs() << format("%02" PRIX32, uuid.uuid[i]);
9072     if (i == 3 || i == 5 || i == 7 || i == 9)
9073       outs() << "-";
9074   }
9075   outs() << "\n";
9076 }
9077 
PrintRpathLoadCommand(MachO::rpath_command rpath,const char * Ptr)9078 static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) {
9079   outs() << "          cmd LC_RPATH\n";
9080   outs() << "      cmdsize " << rpath.cmdsize;
9081   if (rpath.cmdsize < sizeof(struct MachO::rpath_command))
9082     outs() << " Incorrect size\n";
9083   else
9084     outs() << "\n";
9085   if (rpath.path >= rpath.cmdsize)
9086     outs() << "         path ?(bad offset " << rpath.path << ")\n";
9087   else {
9088     const char *P = (const char *)(Ptr) + rpath.path;
9089     outs() << "         path " << P << " (offset " << rpath.path << ")\n";
9090   }
9091 }
9092 
PrintVersionMinLoadCommand(MachO::version_min_command vd)9093 static void PrintVersionMinLoadCommand(MachO::version_min_command vd) {
9094   StringRef LoadCmdName;
9095   switch (vd.cmd) {
9096   case MachO::LC_VERSION_MIN_MACOSX:
9097     LoadCmdName = "LC_VERSION_MIN_MACOSX";
9098     break;
9099   case MachO::LC_VERSION_MIN_IPHONEOS:
9100     LoadCmdName = "LC_VERSION_MIN_IPHONEOS";
9101     break;
9102   case MachO::LC_VERSION_MIN_TVOS:
9103     LoadCmdName = "LC_VERSION_MIN_TVOS";
9104     break;
9105   case MachO::LC_VERSION_MIN_WATCHOS:
9106     LoadCmdName = "LC_VERSION_MIN_WATCHOS";
9107     break;
9108   default:
9109     llvm_unreachable("Unknown version min load command");
9110   }
9111 
9112   outs() << "      cmd " << LoadCmdName << '\n';
9113   outs() << "  cmdsize " << vd.cmdsize;
9114   if (vd.cmdsize != sizeof(struct MachO::version_min_command))
9115     outs() << " Incorrect size\n";
9116   else
9117     outs() << "\n";
9118   outs() << "  version "
9119          << MachOObjectFile::getVersionMinMajor(vd, false) << "."
9120          << MachOObjectFile::getVersionMinMinor(vd, false);
9121   uint32_t Update = MachOObjectFile::getVersionMinUpdate(vd, false);
9122   if (Update != 0)
9123     outs() << "." << Update;
9124   outs() << "\n";
9125   if (vd.sdk == 0)
9126     outs() << "      sdk n/a";
9127   else {
9128     outs() << "      sdk "
9129            << MachOObjectFile::getVersionMinMajor(vd, true) << "."
9130            << MachOObjectFile::getVersionMinMinor(vd, true);
9131   }
9132   Update = MachOObjectFile::getVersionMinUpdate(vd, true);
9133   if (Update != 0)
9134     outs() << "." << Update;
9135   outs() << "\n";
9136 }
9137 
PrintNoteLoadCommand(MachO::note_command Nt)9138 static void PrintNoteLoadCommand(MachO::note_command Nt) {
9139   outs() << "       cmd LC_NOTE\n";
9140   outs() << "   cmdsize " << Nt.cmdsize;
9141   if (Nt.cmdsize != sizeof(struct MachO::note_command))
9142     outs() << " Incorrect size\n";
9143   else
9144     outs() << "\n";
9145   const char *d = Nt.data_owner;
9146   outs() << "data_owner " << format("%.16s\n", d);
9147   outs() << "    offset " << Nt.offset << "\n";
9148   outs() << "      size " << Nt.size << "\n";
9149 }
9150 
PrintBuildToolVersion(MachO::build_tool_version bv)9151 static void PrintBuildToolVersion(MachO::build_tool_version bv) {
9152   outs() << "      tool " << MachOObjectFile::getBuildTool(bv.tool) << "\n";
9153   outs() << "   version " << MachOObjectFile::getVersionString(bv.version)
9154          << "\n";
9155 }
9156 
PrintBuildVersionLoadCommand(const MachOObjectFile * obj,MachO::build_version_command bd)9157 static void PrintBuildVersionLoadCommand(const MachOObjectFile *obj,
9158                                          MachO::build_version_command bd) {
9159   outs() << "       cmd LC_BUILD_VERSION\n";
9160   outs() << "   cmdsize " << bd.cmdsize;
9161   if (bd.cmdsize !=
9162       sizeof(struct MachO::build_version_command) +
9163           bd.ntools * sizeof(struct MachO::build_tool_version))
9164     outs() << " Incorrect size\n";
9165   else
9166     outs() << "\n";
9167   outs() << "  platform " << MachOObjectFile::getBuildPlatform(bd.platform)
9168          << "\n";
9169   if (bd.sdk)
9170     outs() << "       sdk " << MachOObjectFile::getVersionString(bd.sdk)
9171            << "\n";
9172   else
9173     outs() << "       sdk n/a\n";
9174   outs() << "     minos " << MachOObjectFile::getVersionString(bd.minos)
9175          << "\n";
9176   outs() << "    ntools " << bd.ntools << "\n";
9177   for (unsigned i = 0; i < bd.ntools; ++i) {
9178     MachO::build_tool_version bv = obj->getBuildToolVersion(i);
9179     PrintBuildToolVersion(bv);
9180   }
9181 }
9182 
PrintSourceVersionCommand(MachO::source_version_command sd)9183 static void PrintSourceVersionCommand(MachO::source_version_command sd) {
9184   outs() << "      cmd LC_SOURCE_VERSION\n";
9185   outs() << "  cmdsize " << sd.cmdsize;
9186   if (sd.cmdsize != sizeof(struct MachO::source_version_command))
9187     outs() << " Incorrect size\n";
9188   else
9189     outs() << "\n";
9190   uint64_t a = (sd.version >> 40) & 0xffffff;
9191   uint64_t b = (sd.version >> 30) & 0x3ff;
9192   uint64_t c = (sd.version >> 20) & 0x3ff;
9193   uint64_t d = (sd.version >> 10) & 0x3ff;
9194   uint64_t e = sd.version & 0x3ff;
9195   outs() << "  version " << a << "." << b;
9196   if (e != 0)
9197     outs() << "." << c << "." << d << "." << e;
9198   else if (d != 0)
9199     outs() << "." << c << "." << d;
9200   else if (c != 0)
9201     outs() << "." << c;
9202   outs() << "\n";
9203 }
9204 
PrintEntryPointCommand(MachO::entry_point_command ep)9205 static void PrintEntryPointCommand(MachO::entry_point_command ep) {
9206   outs() << "       cmd LC_MAIN\n";
9207   outs() << "   cmdsize " << ep.cmdsize;
9208   if (ep.cmdsize != sizeof(struct MachO::entry_point_command))
9209     outs() << " Incorrect size\n";
9210   else
9211     outs() << "\n";
9212   outs() << "  entryoff " << ep.entryoff << "\n";
9213   outs() << " stacksize " << ep.stacksize << "\n";
9214 }
9215 
PrintEncryptionInfoCommand(MachO::encryption_info_command ec,uint32_t object_size)9216 static void PrintEncryptionInfoCommand(MachO::encryption_info_command ec,
9217                                        uint32_t object_size) {
9218   outs() << "          cmd LC_ENCRYPTION_INFO\n";
9219   outs() << "      cmdsize " << ec.cmdsize;
9220   if (ec.cmdsize != sizeof(struct MachO::encryption_info_command))
9221     outs() << " Incorrect size\n";
9222   else
9223     outs() << "\n";
9224   outs() << "     cryptoff " << ec.cryptoff;
9225   if (ec.cryptoff > object_size)
9226     outs() << " (past end of file)\n";
9227   else
9228     outs() << "\n";
9229   outs() << "    cryptsize " << ec.cryptsize;
9230   if (ec.cryptsize > object_size)
9231     outs() << " (past end of file)\n";
9232   else
9233     outs() << "\n";
9234   outs() << "      cryptid " << ec.cryptid << "\n";
9235 }
9236 
PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,uint32_t object_size)9237 static void PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,
9238                                          uint32_t object_size) {
9239   outs() << "          cmd LC_ENCRYPTION_INFO_64\n";
9240   outs() << "      cmdsize " << ec.cmdsize;
9241   if (ec.cmdsize != sizeof(struct MachO::encryption_info_command_64))
9242     outs() << " Incorrect size\n";
9243   else
9244     outs() << "\n";
9245   outs() << "     cryptoff " << ec.cryptoff;
9246   if (ec.cryptoff > object_size)
9247     outs() << " (past end of file)\n";
9248   else
9249     outs() << "\n";
9250   outs() << "    cryptsize " << ec.cryptsize;
9251   if (ec.cryptsize > object_size)
9252     outs() << " (past end of file)\n";
9253   else
9254     outs() << "\n";
9255   outs() << "      cryptid " << ec.cryptid << "\n";
9256   outs() << "          pad " << ec.pad << "\n";
9257 }
9258 
PrintLinkerOptionCommand(MachO::linker_option_command lo,const char * Ptr)9259 static void PrintLinkerOptionCommand(MachO::linker_option_command lo,
9260                                      const char *Ptr) {
9261   outs() << "     cmd LC_LINKER_OPTION\n";
9262   outs() << " cmdsize " << lo.cmdsize;
9263   if (lo.cmdsize < sizeof(struct MachO::linker_option_command))
9264     outs() << " Incorrect size\n";
9265   else
9266     outs() << "\n";
9267   outs() << "   count " << lo.count << "\n";
9268   const char *string = Ptr + sizeof(struct MachO::linker_option_command);
9269   uint32_t left = lo.cmdsize - sizeof(struct MachO::linker_option_command);
9270   uint32_t i = 0;
9271   while (left > 0) {
9272     while (*string == '\0' && left > 0) {
9273       string++;
9274       left--;
9275     }
9276     if (left > 0) {
9277       i++;
9278       outs() << "  string #" << i << " " << format("%.*s\n", left, string);
9279       uint32_t NullPos = StringRef(string, left).find('\0');
9280       uint32_t len = std::min(NullPos, left) + 1;
9281       string += len;
9282       left -= len;
9283     }
9284   }
9285   if (lo.count != i)
9286     outs() << "   count " << lo.count << " does not match number of strings "
9287            << i << "\n";
9288 }
9289 
PrintSubFrameworkCommand(MachO::sub_framework_command sub,const char * Ptr)9290 static void PrintSubFrameworkCommand(MachO::sub_framework_command sub,
9291                                      const char *Ptr) {
9292   outs() << "          cmd LC_SUB_FRAMEWORK\n";
9293   outs() << "      cmdsize " << sub.cmdsize;
9294   if (sub.cmdsize < sizeof(struct MachO::sub_framework_command))
9295     outs() << " Incorrect size\n";
9296   else
9297     outs() << "\n";
9298   if (sub.umbrella < sub.cmdsize) {
9299     const char *P = Ptr + sub.umbrella;
9300     outs() << "     umbrella " << P << " (offset " << sub.umbrella << ")\n";
9301   } else {
9302     outs() << "     umbrella ?(bad offset " << sub.umbrella << ")\n";
9303   }
9304 }
9305 
PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,const char * Ptr)9306 static void PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,
9307                                     const char *Ptr) {
9308   outs() << "          cmd LC_SUB_UMBRELLA\n";
9309   outs() << "      cmdsize " << sub.cmdsize;
9310   if (sub.cmdsize < sizeof(struct MachO::sub_umbrella_command))
9311     outs() << " Incorrect size\n";
9312   else
9313     outs() << "\n";
9314   if (sub.sub_umbrella < sub.cmdsize) {
9315     const char *P = Ptr + sub.sub_umbrella;
9316     outs() << " sub_umbrella " << P << " (offset " << sub.sub_umbrella << ")\n";
9317   } else {
9318     outs() << " sub_umbrella ?(bad offset " << sub.sub_umbrella << ")\n";
9319   }
9320 }
9321 
PrintSubLibraryCommand(MachO::sub_library_command sub,const char * Ptr)9322 static void PrintSubLibraryCommand(MachO::sub_library_command sub,
9323                                    const char *Ptr) {
9324   outs() << "          cmd LC_SUB_LIBRARY\n";
9325   outs() << "      cmdsize " << sub.cmdsize;
9326   if (sub.cmdsize < sizeof(struct MachO::sub_library_command))
9327     outs() << " Incorrect size\n";
9328   else
9329     outs() << "\n";
9330   if (sub.sub_library < sub.cmdsize) {
9331     const char *P = Ptr + sub.sub_library;
9332     outs() << "  sub_library " << P << " (offset " << sub.sub_library << ")\n";
9333   } else {
9334     outs() << "  sub_library ?(bad offset " << sub.sub_library << ")\n";
9335   }
9336 }
9337 
PrintSubClientCommand(MachO::sub_client_command sub,const char * Ptr)9338 static void PrintSubClientCommand(MachO::sub_client_command sub,
9339                                   const char *Ptr) {
9340   outs() << "          cmd LC_SUB_CLIENT\n";
9341   outs() << "      cmdsize " << sub.cmdsize;
9342   if (sub.cmdsize < sizeof(struct MachO::sub_client_command))
9343     outs() << " Incorrect size\n";
9344   else
9345     outs() << "\n";
9346   if (sub.client < sub.cmdsize) {
9347     const char *P = Ptr + sub.client;
9348     outs() << "       client " << P << " (offset " << sub.client << ")\n";
9349   } else {
9350     outs() << "       client ?(bad offset " << sub.client << ")\n";
9351   }
9352 }
9353 
PrintRoutinesCommand(MachO::routines_command r)9354 static void PrintRoutinesCommand(MachO::routines_command r) {
9355   outs() << "          cmd LC_ROUTINES\n";
9356   outs() << "      cmdsize " << r.cmdsize;
9357   if (r.cmdsize != sizeof(struct MachO::routines_command))
9358     outs() << " Incorrect size\n";
9359   else
9360     outs() << "\n";
9361   outs() << " init_address " << format("0x%08" PRIx32, r.init_address) << "\n";
9362   outs() << "  init_module " << r.init_module << "\n";
9363   outs() << "    reserved1 " << r.reserved1 << "\n";
9364   outs() << "    reserved2 " << r.reserved2 << "\n";
9365   outs() << "    reserved3 " << r.reserved3 << "\n";
9366   outs() << "    reserved4 " << r.reserved4 << "\n";
9367   outs() << "    reserved5 " << r.reserved5 << "\n";
9368   outs() << "    reserved6 " << r.reserved6 << "\n";
9369 }
9370 
PrintRoutinesCommand64(MachO::routines_command_64 r)9371 static void PrintRoutinesCommand64(MachO::routines_command_64 r) {
9372   outs() << "          cmd LC_ROUTINES_64\n";
9373   outs() << "      cmdsize " << r.cmdsize;
9374   if (r.cmdsize != sizeof(struct MachO::routines_command_64))
9375     outs() << " Incorrect size\n";
9376   else
9377     outs() << "\n";
9378   outs() << " init_address " << format("0x%016" PRIx64, r.init_address) << "\n";
9379   outs() << "  init_module " << r.init_module << "\n";
9380   outs() << "    reserved1 " << r.reserved1 << "\n";
9381   outs() << "    reserved2 " << r.reserved2 << "\n";
9382   outs() << "    reserved3 " << r.reserved3 << "\n";
9383   outs() << "    reserved4 " << r.reserved4 << "\n";
9384   outs() << "    reserved5 " << r.reserved5 << "\n";
9385   outs() << "    reserved6 " << r.reserved6 << "\n";
9386 }
9387 
Print_x86_thread_state32_t(MachO::x86_thread_state32_t & cpu32)9388 static void Print_x86_thread_state32_t(MachO::x86_thread_state32_t &cpu32) {
9389   outs() << "\t    eax " << format("0x%08" PRIx32, cpu32.eax);
9390   outs() << " ebx    " << format("0x%08" PRIx32, cpu32.ebx);
9391   outs() << " ecx " << format("0x%08" PRIx32, cpu32.ecx);
9392   outs() << " edx " << format("0x%08" PRIx32, cpu32.edx) << "\n";
9393   outs() << "\t    edi " << format("0x%08" PRIx32, cpu32.edi);
9394   outs() << " esi    " << format("0x%08" PRIx32, cpu32.esi);
9395   outs() << " ebp " << format("0x%08" PRIx32, cpu32.ebp);
9396   outs() << " esp " << format("0x%08" PRIx32, cpu32.esp) << "\n";
9397   outs() << "\t    ss  " << format("0x%08" PRIx32, cpu32.ss);
9398   outs() << " eflags " << format("0x%08" PRIx32, cpu32.eflags);
9399   outs() << " eip " << format("0x%08" PRIx32, cpu32.eip);
9400   outs() << " cs  " << format("0x%08" PRIx32, cpu32.cs) << "\n";
9401   outs() << "\t    ds  " << format("0x%08" PRIx32, cpu32.ds);
9402   outs() << " es     " << format("0x%08" PRIx32, cpu32.es);
9403   outs() << " fs  " << format("0x%08" PRIx32, cpu32.fs);
9404   outs() << " gs  " << format("0x%08" PRIx32, cpu32.gs) << "\n";
9405 }
9406 
Print_x86_thread_state64_t(MachO::x86_thread_state64_t & cpu64)9407 static void Print_x86_thread_state64_t(MachO::x86_thread_state64_t &cpu64) {
9408   outs() << "   rax  " << format("0x%016" PRIx64, cpu64.rax);
9409   outs() << " rbx " << format("0x%016" PRIx64, cpu64.rbx);
9410   outs() << " rcx  " << format("0x%016" PRIx64, cpu64.rcx) << "\n";
9411   outs() << "   rdx  " << format("0x%016" PRIx64, cpu64.rdx);
9412   outs() << " rdi " << format("0x%016" PRIx64, cpu64.rdi);
9413   outs() << " rsi  " << format("0x%016" PRIx64, cpu64.rsi) << "\n";
9414   outs() << "   rbp  " << format("0x%016" PRIx64, cpu64.rbp);
9415   outs() << " rsp " << format("0x%016" PRIx64, cpu64.rsp);
9416   outs() << " r8   " << format("0x%016" PRIx64, cpu64.r8) << "\n";
9417   outs() << "    r9  " << format("0x%016" PRIx64, cpu64.r9);
9418   outs() << " r10 " << format("0x%016" PRIx64, cpu64.r10);
9419   outs() << " r11  " << format("0x%016" PRIx64, cpu64.r11) << "\n";
9420   outs() << "   r12  " << format("0x%016" PRIx64, cpu64.r12);
9421   outs() << " r13 " << format("0x%016" PRIx64, cpu64.r13);
9422   outs() << " r14  " << format("0x%016" PRIx64, cpu64.r14) << "\n";
9423   outs() << "   r15  " << format("0x%016" PRIx64, cpu64.r15);
9424   outs() << " rip " << format("0x%016" PRIx64, cpu64.rip) << "\n";
9425   outs() << "rflags  " << format("0x%016" PRIx64, cpu64.rflags);
9426   outs() << " cs  " << format("0x%016" PRIx64, cpu64.cs);
9427   outs() << " fs   " << format("0x%016" PRIx64, cpu64.fs) << "\n";
9428   outs() << "    gs  " << format("0x%016" PRIx64, cpu64.gs) << "\n";
9429 }
9430 
Print_mmst_reg(MachO::mmst_reg_t & r)9431 static void Print_mmst_reg(MachO::mmst_reg_t &r) {
9432   uint32_t f;
9433   outs() << "\t      mmst_reg  ";
9434   for (f = 0; f < 10; f++)
9435     outs() << format("%02" PRIx32, (r.mmst_reg[f] & 0xff)) << " ";
9436   outs() << "\n";
9437   outs() << "\t      mmst_rsrv ";
9438   for (f = 0; f < 6; f++)
9439     outs() << format("%02" PRIx32, (r.mmst_rsrv[f] & 0xff)) << " ";
9440   outs() << "\n";
9441 }
9442 
Print_xmm_reg(MachO::xmm_reg_t & r)9443 static void Print_xmm_reg(MachO::xmm_reg_t &r) {
9444   uint32_t f;
9445   outs() << "\t      xmm_reg ";
9446   for (f = 0; f < 16; f++)
9447     outs() << format("%02" PRIx32, (r.xmm_reg[f] & 0xff)) << " ";
9448   outs() << "\n";
9449 }
9450 
Print_x86_float_state_t(MachO::x86_float_state64_t & fpu)9451 static void Print_x86_float_state_t(MachO::x86_float_state64_t &fpu) {
9452   outs() << "\t    fpu_reserved[0] " << fpu.fpu_reserved[0];
9453   outs() << " fpu_reserved[1] " << fpu.fpu_reserved[1] << "\n";
9454   outs() << "\t    control: invalid " << fpu.fpu_fcw.invalid;
9455   outs() << " denorm " << fpu.fpu_fcw.denorm;
9456   outs() << " zdiv " << fpu.fpu_fcw.zdiv;
9457   outs() << " ovrfl " << fpu.fpu_fcw.ovrfl;
9458   outs() << " undfl " << fpu.fpu_fcw.undfl;
9459   outs() << " precis " << fpu.fpu_fcw.precis << "\n";
9460   outs() << "\t\t     pc ";
9461   if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_24B)
9462     outs() << "FP_PREC_24B ";
9463   else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_53B)
9464     outs() << "FP_PREC_53B ";
9465   else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_64B)
9466     outs() << "FP_PREC_64B ";
9467   else
9468     outs() << fpu.fpu_fcw.pc << " ";
9469   outs() << "rc ";
9470   if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_NEAR)
9471     outs() << "FP_RND_NEAR ";
9472   else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_DOWN)
9473     outs() << "FP_RND_DOWN ";
9474   else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_UP)
9475     outs() << "FP_RND_UP ";
9476   else if (fpu.fpu_fcw.rc == MachO::x86_FP_CHOP)
9477     outs() << "FP_CHOP ";
9478   outs() << "\n";
9479   outs() << "\t    status: invalid " << fpu.fpu_fsw.invalid;
9480   outs() << " denorm " << fpu.fpu_fsw.denorm;
9481   outs() << " zdiv " << fpu.fpu_fsw.zdiv;
9482   outs() << " ovrfl " << fpu.fpu_fsw.ovrfl;
9483   outs() << " undfl " << fpu.fpu_fsw.undfl;
9484   outs() << " precis " << fpu.fpu_fsw.precis;
9485   outs() << " stkflt " << fpu.fpu_fsw.stkflt << "\n";
9486   outs() << "\t            errsumm " << fpu.fpu_fsw.errsumm;
9487   outs() << " c0 " << fpu.fpu_fsw.c0;
9488   outs() << " c1 " << fpu.fpu_fsw.c1;
9489   outs() << " c2 " << fpu.fpu_fsw.c2;
9490   outs() << " tos " << fpu.fpu_fsw.tos;
9491   outs() << " c3 " << fpu.fpu_fsw.c3;
9492   outs() << " busy " << fpu.fpu_fsw.busy << "\n";
9493   outs() << "\t    fpu_ftw " << format("0x%02" PRIx32, fpu.fpu_ftw);
9494   outs() << " fpu_rsrv1 " << format("0x%02" PRIx32, fpu.fpu_rsrv1);
9495   outs() << " fpu_fop " << format("0x%04" PRIx32, fpu.fpu_fop);
9496   outs() << " fpu_ip " << format("0x%08" PRIx32, fpu.fpu_ip) << "\n";
9497   outs() << "\t    fpu_cs " << format("0x%04" PRIx32, fpu.fpu_cs);
9498   outs() << " fpu_rsrv2 " << format("0x%04" PRIx32, fpu.fpu_rsrv2);
9499   outs() << " fpu_dp " << format("0x%08" PRIx32, fpu.fpu_dp);
9500   outs() << " fpu_ds " << format("0x%04" PRIx32, fpu.fpu_ds) << "\n";
9501   outs() << "\t    fpu_rsrv3 " << format("0x%04" PRIx32, fpu.fpu_rsrv3);
9502   outs() << " fpu_mxcsr " << format("0x%08" PRIx32, fpu.fpu_mxcsr);
9503   outs() << " fpu_mxcsrmask " << format("0x%08" PRIx32, fpu.fpu_mxcsrmask);
9504   outs() << "\n";
9505   outs() << "\t    fpu_stmm0:\n";
9506   Print_mmst_reg(fpu.fpu_stmm0);
9507   outs() << "\t    fpu_stmm1:\n";
9508   Print_mmst_reg(fpu.fpu_stmm1);
9509   outs() << "\t    fpu_stmm2:\n";
9510   Print_mmst_reg(fpu.fpu_stmm2);
9511   outs() << "\t    fpu_stmm3:\n";
9512   Print_mmst_reg(fpu.fpu_stmm3);
9513   outs() << "\t    fpu_stmm4:\n";
9514   Print_mmst_reg(fpu.fpu_stmm4);
9515   outs() << "\t    fpu_stmm5:\n";
9516   Print_mmst_reg(fpu.fpu_stmm5);
9517   outs() << "\t    fpu_stmm6:\n";
9518   Print_mmst_reg(fpu.fpu_stmm6);
9519   outs() << "\t    fpu_stmm7:\n";
9520   Print_mmst_reg(fpu.fpu_stmm7);
9521   outs() << "\t    fpu_xmm0:\n";
9522   Print_xmm_reg(fpu.fpu_xmm0);
9523   outs() << "\t    fpu_xmm1:\n";
9524   Print_xmm_reg(fpu.fpu_xmm1);
9525   outs() << "\t    fpu_xmm2:\n";
9526   Print_xmm_reg(fpu.fpu_xmm2);
9527   outs() << "\t    fpu_xmm3:\n";
9528   Print_xmm_reg(fpu.fpu_xmm3);
9529   outs() << "\t    fpu_xmm4:\n";
9530   Print_xmm_reg(fpu.fpu_xmm4);
9531   outs() << "\t    fpu_xmm5:\n";
9532   Print_xmm_reg(fpu.fpu_xmm5);
9533   outs() << "\t    fpu_xmm6:\n";
9534   Print_xmm_reg(fpu.fpu_xmm6);
9535   outs() << "\t    fpu_xmm7:\n";
9536   Print_xmm_reg(fpu.fpu_xmm7);
9537   outs() << "\t    fpu_xmm8:\n";
9538   Print_xmm_reg(fpu.fpu_xmm8);
9539   outs() << "\t    fpu_xmm9:\n";
9540   Print_xmm_reg(fpu.fpu_xmm9);
9541   outs() << "\t    fpu_xmm10:\n";
9542   Print_xmm_reg(fpu.fpu_xmm10);
9543   outs() << "\t    fpu_xmm11:\n";
9544   Print_xmm_reg(fpu.fpu_xmm11);
9545   outs() << "\t    fpu_xmm12:\n";
9546   Print_xmm_reg(fpu.fpu_xmm12);
9547   outs() << "\t    fpu_xmm13:\n";
9548   Print_xmm_reg(fpu.fpu_xmm13);
9549   outs() << "\t    fpu_xmm14:\n";
9550   Print_xmm_reg(fpu.fpu_xmm14);
9551   outs() << "\t    fpu_xmm15:\n";
9552   Print_xmm_reg(fpu.fpu_xmm15);
9553   outs() << "\t    fpu_rsrv4:\n";
9554   for (uint32_t f = 0; f < 6; f++) {
9555     outs() << "\t            ";
9556     for (uint32_t g = 0; g < 16; g++)
9557       outs() << format("%02" PRIx32, fpu.fpu_rsrv4[f * g]) << " ";
9558     outs() << "\n";
9559   }
9560   outs() << "\t    fpu_reserved1 " << format("0x%08" PRIx32, fpu.fpu_reserved1);
9561   outs() << "\n";
9562 }
9563 
Print_x86_exception_state_t(MachO::x86_exception_state64_t & exc64)9564 static void Print_x86_exception_state_t(MachO::x86_exception_state64_t &exc64) {
9565   outs() << "\t    trapno " << format("0x%08" PRIx32, exc64.trapno);
9566   outs() << " err " << format("0x%08" PRIx32, exc64.err);
9567   outs() << " faultvaddr " << format("0x%016" PRIx64, exc64.faultvaddr) << "\n";
9568 }
9569 
Print_arm_thread_state32_t(MachO::arm_thread_state32_t & cpu32)9570 static void Print_arm_thread_state32_t(MachO::arm_thread_state32_t &cpu32) {
9571   outs() << "\t    r0  " << format("0x%08" PRIx32, cpu32.r[0]);
9572   outs() << " r1     "   << format("0x%08" PRIx32, cpu32.r[1]);
9573   outs() << " r2  "      << format("0x%08" PRIx32, cpu32.r[2]);
9574   outs() << " r3  "      << format("0x%08" PRIx32, cpu32.r[3]) << "\n";
9575   outs() << "\t    r4  " << format("0x%08" PRIx32, cpu32.r[4]);
9576   outs() << " r5     "   << format("0x%08" PRIx32, cpu32.r[5]);
9577   outs() << " r6  "      << format("0x%08" PRIx32, cpu32.r[6]);
9578   outs() << " r7  "      << format("0x%08" PRIx32, cpu32.r[7]) << "\n";
9579   outs() << "\t    r8  " << format("0x%08" PRIx32, cpu32.r[8]);
9580   outs() << " r9     "   << format("0x%08" PRIx32, cpu32.r[9]);
9581   outs() << " r10 "      << format("0x%08" PRIx32, cpu32.r[10]);
9582   outs() << " r11 "      << format("0x%08" PRIx32, cpu32.r[11]) << "\n";
9583   outs() << "\t    r12 " << format("0x%08" PRIx32, cpu32.r[12]);
9584   outs() << " sp     "   << format("0x%08" PRIx32, cpu32.sp);
9585   outs() << " lr  "      << format("0x%08" PRIx32, cpu32.lr);
9586   outs() << " pc  "      << format("0x%08" PRIx32, cpu32.pc) << "\n";
9587   outs() << "\t   cpsr " << format("0x%08" PRIx32, cpu32.cpsr) << "\n";
9588 }
9589 
Print_arm_thread_state64_t(MachO::arm_thread_state64_t & cpu64)9590 static void Print_arm_thread_state64_t(MachO::arm_thread_state64_t &cpu64) {
9591   outs() << "\t    x0  " << format("0x%016" PRIx64, cpu64.x[0]);
9592   outs() << " x1  "      << format("0x%016" PRIx64, cpu64.x[1]);
9593   outs() << " x2  "      << format("0x%016" PRIx64, cpu64.x[2]) << "\n";
9594   outs() << "\t    x3  " << format("0x%016" PRIx64, cpu64.x[3]);
9595   outs() << " x4  "      << format("0x%016" PRIx64, cpu64.x[4]);
9596   outs() << " x5  "      << format("0x%016" PRIx64, cpu64.x[5]) << "\n";
9597   outs() << "\t    x6  " << format("0x%016" PRIx64, cpu64.x[6]);
9598   outs() << " x7  "      << format("0x%016" PRIx64, cpu64.x[7]);
9599   outs() << " x8  "      << format("0x%016" PRIx64, cpu64.x[8]) << "\n";
9600   outs() << "\t    x9  " << format("0x%016" PRIx64, cpu64.x[9]);
9601   outs() << " x10 "      << format("0x%016" PRIx64, cpu64.x[10]);
9602   outs() << " x11 "      << format("0x%016" PRIx64, cpu64.x[11]) << "\n";
9603   outs() << "\t    x12 " << format("0x%016" PRIx64, cpu64.x[12]);
9604   outs() << " x13 "      << format("0x%016" PRIx64, cpu64.x[13]);
9605   outs() << " x14 "      << format("0x%016" PRIx64, cpu64.x[14]) << "\n";
9606   outs() << "\t    x15 " << format("0x%016" PRIx64, cpu64.x[15]);
9607   outs() << " x16 "      << format("0x%016" PRIx64, cpu64.x[16]);
9608   outs() << " x17 "      << format("0x%016" PRIx64, cpu64.x[17]) << "\n";
9609   outs() << "\t    x18 " << format("0x%016" PRIx64, cpu64.x[18]);
9610   outs() << " x19 "      << format("0x%016" PRIx64, cpu64.x[19]);
9611   outs() << " x20 "      << format("0x%016" PRIx64, cpu64.x[20]) << "\n";
9612   outs() << "\t    x21 " << format("0x%016" PRIx64, cpu64.x[21]);
9613   outs() << " x22 "      << format("0x%016" PRIx64, cpu64.x[22]);
9614   outs() << " x23 "      << format("0x%016" PRIx64, cpu64.x[23]) << "\n";
9615   outs() << "\t    x24 " << format("0x%016" PRIx64, cpu64.x[24]);
9616   outs() << " x25 "      << format("0x%016" PRIx64, cpu64.x[25]);
9617   outs() << " x26 "      << format("0x%016" PRIx64, cpu64.x[26]) << "\n";
9618   outs() << "\t    x27 " << format("0x%016" PRIx64, cpu64.x[27]);
9619   outs() << " x28 "      << format("0x%016" PRIx64, cpu64.x[28]);
9620   outs() << "  fp "      << format("0x%016" PRIx64, cpu64.fp) << "\n";
9621   outs() << "\t     lr " << format("0x%016" PRIx64, cpu64.lr);
9622   outs() << " sp  "      << format("0x%016" PRIx64, cpu64.sp);
9623   outs() << "  pc "      << format("0x%016" PRIx64, cpu64.pc) << "\n";
9624   outs() << "\t   cpsr " << format("0x%08"  PRIx32, cpu64.cpsr) << "\n";
9625 }
9626 
PrintThreadCommand(MachO::thread_command t,const char * Ptr,bool isLittleEndian,uint32_t cputype)9627 static void PrintThreadCommand(MachO::thread_command t, const char *Ptr,
9628                                bool isLittleEndian, uint32_t cputype) {
9629   if (t.cmd == MachO::LC_THREAD)
9630     outs() << "        cmd LC_THREAD\n";
9631   else if (t.cmd == MachO::LC_UNIXTHREAD)
9632     outs() << "        cmd LC_UNIXTHREAD\n";
9633   else
9634     outs() << "        cmd " << t.cmd << " (unknown)\n";
9635   outs() << "    cmdsize " << t.cmdsize;
9636   if (t.cmdsize < sizeof(struct MachO::thread_command) + 2 * sizeof(uint32_t))
9637     outs() << " Incorrect size\n";
9638   else
9639     outs() << "\n";
9640 
9641   const char *begin = Ptr + sizeof(struct MachO::thread_command);
9642   const char *end = Ptr + t.cmdsize;
9643   uint32_t flavor, count, left;
9644   if (cputype == MachO::CPU_TYPE_I386) {
9645     while (begin < end) {
9646       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9647         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9648         begin += sizeof(uint32_t);
9649       } else {
9650         flavor = 0;
9651         begin = end;
9652       }
9653       if (isLittleEndian != sys::IsLittleEndianHost)
9654         sys::swapByteOrder(flavor);
9655       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9656         memcpy((char *)&count, begin, sizeof(uint32_t));
9657         begin += sizeof(uint32_t);
9658       } else {
9659         count = 0;
9660         begin = end;
9661       }
9662       if (isLittleEndian != sys::IsLittleEndianHost)
9663         sys::swapByteOrder(count);
9664       if (flavor == MachO::x86_THREAD_STATE32) {
9665         outs() << "     flavor i386_THREAD_STATE\n";
9666         if (count == MachO::x86_THREAD_STATE32_COUNT)
9667           outs() << "      count i386_THREAD_STATE_COUNT\n";
9668         else
9669           outs() << "      count " << count
9670                  << " (not x86_THREAD_STATE32_COUNT)\n";
9671         MachO::x86_thread_state32_t cpu32;
9672         left = end - begin;
9673         if (left >= sizeof(MachO::x86_thread_state32_t)) {
9674           memcpy(&cpu32, begin, sizeof(MachO::x86_thread_state32_t));
9675           begin += sizeof(MachO::x86_thread_state32_t);
9676         } else {
9677           memset(&cpu32, '\0', sizeof(MachO::x86_thread_state32_t));
9678           memcpy(&cpu32, begin, left);
9679           begin += left;
9680         }
9681         if (isLittleEndian != sys::IsLittleEndianHost)
9682           swapStruct(cpu32);
9683         Print_x86_thread_state32_t(cpu32);
9684       } else if (flavor == MachO::x86_THREAD_STATE) {
9685         outs() << "     flavor x86_THREAD_STATE\n";
9686         if (count == MachO::x86_THREAD_STATE_COUNT)
9687           outs() << "      count x86_THREAD_STATE_COUNT\n";
9688         else
9689           outs() << "      count " << count
9690                  << " (not x86_THREAD_STATE_COUNT)\n";
9691         struct MachO::x86_thread_state_t ts;
9692         left = end - begin;
9693         if (left >= sizeof(MachO::x86_thread_state_t)) {
9694           memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
9695           begin += sizeof(MachO::x86_thread_state_t);
9696         } else {
9697           memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
9698           memcpy(&ts, begin, left);
9699           begin += left;
9700         }
9701         if (isLittleEndian != sys::IsLittleEndianHost)
9702           swapStruct(ts);
9703         if (ts.tsh.flavor == MachO::x86_THREAD_STATE32) {
9704           outs() << "\t    tsh.flavor x86_THREAD_STATE32 ";
9705           if (ts.tsh.count == MachO::x86_THREAD_STATE32_COUNT)
9706             outs() << "tsh.count x86_THREAD_STATE32_COUNT\n";
9707           else
9708             outs() << "tsh.count " << ts.tsh.count
9709                    << " (not x86_THREAD_STATE32_COUNT\n";
9710           Print_x86_thread_state32_t(ts.uts.ts32);
9711         } else {
9712           outs() << "\t    tsh.flavor " << ts.tsh.flavor << "  tsh.count "
9713                  << ts.tsh.count << "\n";
9714         }
9715       } else {
9716         outs() << "     flavor " << flavor << " (unknown)\n";
9717         outs() << "      count " << count << "\n";
9718         outs() << "      state (unknown)\n";
9719         begin += count * sizeof(uint32_t);
9720       }
9721     }
9722   } else if (cputype == MachO::CPU_TYPE_X86_64) {
9723     while (begin < end) {
9724       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9725         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9726         begin += sizeof(uint32_t);
9727       } else {
9728         flavor = 0;
9729         begin = end;
9730       }
9731       if (isLittleEndian != sys::IsLittleEndianHost)
9732         sys::swapByteOrder(flavor);
9733       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9734         memcpy((char *)&count, begin, sizeof(uint32_t));
9735         begin += sizeof(uint32_t);
9736       } else {
9737         count = 0;
9738         begin = end;
9739       }
9740       if (isLittleEndian != sys::IsLittleEndianHost)
9741         sys::swapByteOrder(count);
9742       if (flavor == MachO::x86_THREAD_STATE64) {
9743         outs() << "     flavor x86_THREAD_STATE64\n";
9744         if (count == MachO::x86_THREAD_STATE64_COUNT)
9745           outs() << "      count x86_THREAD_STATE64_COUNT\n";
9746         else
9747           outs() << "      count " << count
9748                  << " (not x86_THREAD_STATE64_COUNT)\n";
9749         MachO::x86_thread_state64_t cpu64;
9750         left = end - begin;
9751         if (left >= sizeof(MachO::x86_thread_state64_t)) {
9752           memcpy(&cpu64, begin, sizeof(MachO::x86_thread_state64_t));
9753           begin += sizeof(MachO::x86_thread_state64_t);
9754         } else {
9755           memset(&cpu64, '\0', sizeof(MachO::x86_thread_state64_t));
9756           memcpy(&cpu64, begin, left);
9757           begin += left;
9758         }
9759         if (isLittleEndian != sys::IsLittleEndianHost)
9760           swapStruct(cpu64);
9761         Print_x86_thread_state64_t(cpu64);
9762       } else if (flavor == MachO::x86_THREAD_STATE) {
9763         outs() << "     flavor x86_THREAD_STATE\n";
9764         if (count == MachO::x86_THREAD_STATE_COUNT)
9765           outs() << "      count x86_THREAD_STATE_COUNT\n";
9766         else
9767           outs() << "      count " << count
9768                  << " (not x86_THREAD_STATE_COUNT)\n";
9769         struct MachO::x86_thread_state_t ts;
9770         left = end - begin;
9771         if (left >= sizeof(MachO::x86_thread_state_t)) {
9772           memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
9773           begin += sizeof(MachO::x86_thread_state_t);
9774         } else {
9775           memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
9776           memcpy(&ts, begin, left);
9777           begin += left;
9778         }
9779         if (isLittleEndian != sys::IsLittleEndianHost)
9780           swapStruct(ts);
9781         if (ts.tsh.flavor == MachO::x86_THREAD_STATE64) {
9782           outs() << "\t    tsh.flavor x86_THREAD_STATE64 ";
9783           if (ts.tsh.count == MachO::x86_THREAD_STATE64_COUNT)
9784             outs() << "tsh.count x86_THREAD_STATE64_COUNT\n";
9785           else
9786             outs() << "tsh.count " << ts.tsh.count
9787                    << " (not x86_THREAD_STATE64_COUNT\n";
9788           Print_x86_thread_state64_t(ts.uts.ts64);
9789         } else {
9790           outs() << "\t    tsh.flavor " << ts.tsh.flavor << "  tsh.count "
9791                  << ts.tsh.count << "\n";
9792         }
9793       } else if (flavor == MachO::x86_FLOAT_STATE) {
9794         outs() << "     flavor x86_FLOAT_STATE\n";
9795         if (count == MachO::x86_FLOAT_STATE_COUNT)
9796           outs() << "      count x86_FLOAT_STATE_COUNT\n";
9797         else
9798           outs() << "      count " << count << " (not x86_FLOAT_STATE_COUNT)\n";
9799         struct MachO::x86_float_state_t fs;
9800         left = end - begin;
9801         if (left >= sizeof(MachO::x86_float_state_t)) {
9802           memcpy(&fs, begin, sizeof(MachO::x86_float_state_t));
9803           begin += sizeof(MachO::x86_float_state_t);
9804         } else {
9805           memset(&fs, '\0', sizeof(MachO::x86_float_state_t));
9806           memcpy(&fs, begin, left);
9807           begin += left;
9808         }
9809         if (isLittleEndian != sys::IsLittleEndianHost)
9810           swapStruct(fs);
9811         if (fs.fsh.flavor == MachO::x86_FLOAT_STATE64) {
9812           outs() << "\t    fsh.flavor x86_FLOAT_STATE64 ";
9813           if (fs.fsh.count == MachO::x86_FLOAT_STATE64_COUNT)
9814             outs() << "fsh.count x86_FLOAT_STATE64_COUNT\n";
9815           else
9816             outs() << "fsh.count " << fs.fsh.count
9817                    << " (not x86_FLOAT_STATE64_COUNT\n";
9818           Print_x86_float_state_t(fs.ufs.fs64);
9819         } else {
9820           outs() << "\t    fsh.flavor " << fs.fsh.flavor << "  fsh.count "
9821                  << fs.fsh.count << "\n";
9822         }
9823       } else if (flavor == MachO::x86_EXCEPTION_STATE) {
9824         outs() << "     flavor x86_EXCEPTION_STATE\n";
9825         if (count == MachO::x86_EXCEPTION_STATE_COUNT)
9826           outs() << "      count x86_EXCEPTION_STATE_COUNT\n";
9827         else
9828           outs() << "      count " << count
9829                  << " (not x86_EXCEPTION_STATE_COUNT)\n";
9830         struct MachO::x86_exception_state_t es;
9831         left = end - begin;
9832         if (left >= sizeof(MachO::x86_exception_state_t)) {
9833           memcpy(&es, begin, sizeof(MachO::x86_exception_state_t));
9834           begin += sizeof(MachO::x86_exception_state_t);
9835         } else {
9836           memset(&es, '\0', sizeof(MachO::x86_exception_state_t));
9837           memcpy(&es, begin, left);
9838           begin += left;
9839         }
9840         if (isLittleEndian != sys::IsLittleEndianHost)
9841           swapStruct(es);
9842         if (es.esh.flavor == MachO::x86_EXCEPTION_STATE64) {
9843           outs() << "\t    esh.flavor x86_EXCEPTION_STATE64\n";
9844           if (es.esh.count == MachO::x86_EXCEPTION_STATE64_COUNT)
9845             outs() << "\t    esh.count x86_EXCEPTION_STATE64_COUNT\n";
9846           else
9847             outs() << "\t    esh.count " << es.esh.count
9848                    << " (not x86_EXCEPTION_STATE64_COUNT\n";
9849           Print_x86_exception_state_t(es.ues.es64);
9850         } else {
9851           outs() << "\t    esh.flavor " << es.esh.flavor << "  esh.count "
9852                  << es.esh.count << "\n";
9853         }
9854       } else if (flavor == MachO::x86_EXCEPTION_STATE64) {
9855         outs() << "     flavor x86_EXCEPTION_STATE64\n";
9856         if (count == MachO::x86_EXCEPTION_STATE64_COUNT)
9857           outs() << "      count x86_EXCEPTION_STATE64_COUNT\n";
9858         else
9859           outs() << "      count " << count
9860                  << " (not x86_EXCEPTION_STATE64_COUNT)\n";
9861         struct MachO::x86_exception_state64_t es64;
9862         left = end - begin;
9863         if (left >= sizeof(MachO::x86_exception_state64_t)) {
9864           memcpy(&es64, begin, sizeof(MachO::x86_exception_state64_t));
9865           begin += sizeof(MachO::x86_exception_state64_t);
9866         } else {
9867           memset(&es64, '\0', sizeof(MachO::x86_exception_state64_t));
9868           memcpy(&es64, begin, left);
9869           begin += left;
9870         }
9871         if (isLittleEndian != sys::IsLittleEndianHost)
9872           swapStruct(es64);
9873         Print_x86_exception_state_t(es64);
9874       } else {
9875         outs() << "     flavor " << flavor << " (unknown)\n";
9876         outs() << "      count " << count << "\n";
9877         outs() << "      state (unknown)\n";
9878         begin += count * sizeof(uint32_t);
9879       }
9880     }
9881   } else if (cputype == MachO::CPU_TYPE_ARM) {
9882     while (begin < end) {
9883       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9884         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9885         begin += sizeof(uint32_t);
9886       } else {
9887         flavor = 0;
9888         begin = end;
9889       }
9890       if (isLittleEndian != sys::IsLittleEndianHost)
9891         sys::swapByteOrder(flavor);
9892       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9893         memcpy((char *)&count, begin, sizeof(uint32_t));
9894         begin += sizeof(uint32_t);
9895       } else {
9896         count = 0;
9897         begin = end;
9898       }
9899       if (isLittleEndian != sys::IsLittleEndianHost)
9900         sys::swapByteOrder(count);
9901       if (flavor == MachO::ARM_THREAD_STATE) {
9902         outs() << "     flavor ARM_THREAD_STATE\n";
9903         if (count == MachO::ARM_THREAD_STATE_COUNT)
9904           outs() << "      count ARM_THREAD_STATE_COUNT\n";
9905         else
9906           outs() << "      count " << count
9907                  << " (not ARM_THREAD_STATE_COUNT)\n";
9908         MachO::arm_thread_state32_t cpu32;
9909         left = end - begin;
9910         if (left >= sizeof(MachO::arm_thread_state32_t)) {
9911           memcpy(&cpu32, begin, sizeof(MachO::arm_thread_state32_t));
9912           begin += sizeof(MachO::arm_thread_state32_t);
9913         } else {
9914           memset(&cpu32, '\0', sizeof(MachO::arm_thread_state32_t));
9915           memcpy(&cpu32, begin, left);
9916           begin += left;
9917         }
9918         if (isLittleEndian != sys::IsLittleEndianHost)
9919           swapStruct(cpu32);
9920         Print_arm_thread_state32_t(cpu32);
9921       } else {
9922         outs() << "     flavor " << flavor << " (unknown)\n";
9923         outs() << "      count " << count << "\n";
9924         outs() << "      state (unknown)\n";
9925         begin += count * sizeof(uint32_t);
9926       }
9927     }
9928   } else if (cputype == MachO::CPU_TYPE_ARM64 ||
9929              cputype == MachO::CPU_TYPE_ARM64_32) {
9930     while (begin < end) {
9931       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9932         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9933         begin += sizeof(uint32_t);
9934       } else {
9935         flavor = 0;
9936         begin = end;
9937       }
9938       if (isLittleEndian != sys::IsLittleEndianHost)
9939         sys::swapByteOrder(flavor);
9940       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9941         memcpy((char *)&count, begin, sizeof(uint32_t));
9942         begin += sizeof(uint32_t);
9943       } else {
9944         count = 0;
9945         begin = end;
9946       }
9947       if (isLittleEndian != sys::IsLittleEndianHost)
9948         sys::swapByteOrder(count);
9949       if (flavor == MachO::ARM_THREAD_STATE64) {
9950         outs() << "     flavor ARM_THREAD_STATE64\n";
9951         if (count == MachO::ARM_THREAD_STATE64_COUNT)
9952           outs() << "      count ARM_THREAD_STATE64_COUNT\n";
9953         else
9954           outs() << "      count " << count
9955                  << " (not ARM_THREAD_STATE64_COUNT)\n";
9956         MachO::arm_thread_state64_t cpu64;
9957         left = end - begin;
9958         if (left >= sizeof(MachO::arm_thread_state64_t)) {
9959           memcpy(&cpu64, begin, sizeof(MachO::arm_thread_state64_t));
9960           begin += sizeof(MachO::arm_thread_state64_t);
9961         } else {
9962           memset(&cpu64, '\0', sizeof(MachO::arm_thread_state64_t));
9963           memcpy(&cpu64, begin, left);
9964           begin += left;
9965         }
9966         if (isLittleEndian != sys::IsLittleEndianHost)
9967           swapStruct(cpu64);
9968         Print_arm_thread_state64_t(cpu64);
9969       } else {
9970         outs() << "     flavor " << flavor << " (unknown)\n";
9971         outs() << "      count " << count << "\n";
9972         outs() << "      state (unknown)\n";
9973         begin += count * sizeof(uint32_t);
9974       }
9975     }
9976   } else {
9977     while (begin < end) {
9978       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9979         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9980         begin += sizeof(uint32_t);
9981       } else {
9982         flavor = 0;
9983         begin = end;
9984       }
9985       if (isLittleEndian != sys::IsLittleEndianHost)
9986         sys::swapByteOrder(flavor);
9987       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9988         memcpy((char *)&count, begin, sizeof(uint32_t));
9989         begin += sizeof(uint32_t);
9990       } else {
9991         count = 0;
9992         begin = end;
9993       }
9994       if (isLittleEndian != sys::IsLittleEndianHost)
9995         sys::swapByteOrder(count);
9996       outs() << "     flavor " << flavor << "\n";
9997       outs() << "      count " << count << "\n";
9998       outs() << "      state (Unknown cputype/cpusubtype)\n";
9999       begin += count * sizeof(uint32_t);
10000     }
10001   }
10002 }
10003 
PrintDylibCommand(MachO::dylib_command dl,const char * Ptr)10004 static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) {
10005   if (dl.cmd == MachO::LC_ID_DYLIB)
10006     outs() << "          cmd LC_ID_DYLIB\n";
10007   else if (dl.cmd == MachO::LC_LOAD_DYLIB)
10008     outs() << "          cmd LC_LOAD_DYLIB\n";
10009   else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB)
10010     outs() << "          cmd LC_LOAD_WEAK_DYLIB\n";
10011   else if (dl.cmd == MachO::LC_REEXPORT_DYLIB)
10012     outs() << "          cmd LC_REEXPORT_DYLIB\n";
10013   else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB)
10014     outs() << "          cmd LC_LAZY_LOAD_DYLIB\n";
10015   else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
10016     outs() << "          cmd LC_LOAD_UPWARD_DYLIB\n";
10017   else
10018     outs() << "          cmd " << dl.cmd << " (unknown)\n";
10019   outs() << "      cmdsize " << dl.cmdsize;
10020   if (dl.cmdsize < sizeof(struct MachO::dylib_command))
10021     outs() << " Incorrect size\n";
10022   else
10023     outs() << "\n";
10024   if (dl.dylib.name < dl.cmdsize) {
10025     const char *P = (const char *)(Ptr) + dl.dylib.name;
10026     outs() << "         name " << P << " (offset " << dl.dylib.name << ")\n";
10027   } else {
10028     outs() << "         name ?(bad offset " << dl.dylib.name << ")\n";
10029   }
10030   outs() << "   time stamp " << dl.dylib.timestamp << " ";
10031   time_t t = dl.dylib.timestamp;
10032   outs() << ctime(&t);
10033   outs() << "      current version ";
10034   if (dl.dylib.current_version == 0xffffffff)
10035     outs() << "n/a\n";
10036   else
10037     outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "."
10038            << ((dl.dylib.current_version >> 8) & 0xff) << "."
10039            << (dl.dylib.current_version & 0xff) << "\n";
10040   outs() << "compatibility version ";
10041   if (dl.dylib.compatibility_version == 0xffffffff)
10042     outs() << "n/a\n";
10043   else
10044     outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
10045            << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
10046            << (dl.dylib.compatibility_version & 0xff) << "\n";
10047 }
10048 
PrintLinkEditDataCommand(MachO::linkedit_data_command ld,uint32_t object_size)10049 static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld,
10050                                      uint32_t object_size) {
10051   if (ld.cmd == MachO::LC_CODE_SIGNATURE)
10052     outs() << "      cmd LC_CODE_SIGNATURE\n";
10053   else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO)
10054     outs() << "      cmd LC_SEGMENT_SPLIT_INFO\n";
10055   else if (ld.cmd == MachO::LC_FUNCTION_STARTS)
10056     outs() << "      cmd LC_FUNCTION_STARTS\n";
10057   else if (ld.cmd == MachO::LC_DATA_IN_CODE)
10058     outs() << "      cmd LC_DATA_IN_CODE\n";
10059   else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS)
10060     outs() << "      cmd LC_DYLIB_CODE_SIGN_DRS\n";
10061   else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT)
10062     outs() << "      cmd LC_LINKER_OPTIMIZATION_HINT\n";
10063   else
10064     outs() << "      cmd " << ld.cmd << " (?)\n";
10065   outs() << "  cmdsize " << ld.cmdsize;
10066   if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command))
10067     outs() << " Incorrect size\n";
10068   else
10069     outs() << "\n";
10070   outs() << "  dataoff " << ld.dataoff;
10071   if (ld.dataoff > object_size)
10072     outs() << " (past end of file)\n";
10073   else
10074     outs() << "\n";
10075   outs() << " datasize " << ld.datasize;
10076   uint64_t big_size = ld.dataoff;
10077   big_size += ld.datasize;
10078   if (big_size > object_size)
10079     outs() << " (past end of file)\n";
10080   else
10081     outs() << "\n";
10082 }
10083 
PrintLoadCommands(const MachOObjectFile * Obj,uint32_t filetype,uint32_t cputype,bool verbose)10084 static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t filetype,
10085                               uint32_t cputype, bool verbose) {
10086   StringRef Buf = Obj->getData();
10087   unsigned Index = 0;
10088   for (const auto &Command : Obj->load_commands()) {
10089     outs() << "Load command " << Index++ << "\n";
10090     if (Command.C.cmd == MachO::LC_SEGMENT) {
10091       MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command);
10092       const char *sg_segname = SLC.segname;
10093       PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr,
10094                           SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot,
10095                           SLC.initprot, SLC.nsects, SLC.flags, Buf.size(),
10096                           verbose);
10097       for (unsigned j = 0; j < SLC.nsects; j++) {
10098         MachO::section S = Obj->getSection(Command, j);
10099         PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align,
10100                      S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2,
10101                      SLC.cmd, sg_segname, filetype, Buf.size(), verbose);
10102       }
10103     } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
10104       MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command);
10105       const char *sg_segname = SLC_64.segname;
10106       PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname,
10107                           SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff,
10108                           SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot,
10109                           SLC_64.nsects, SLC_64.flags, Buf.size(), verbose);
10110       for (unsigned j = 0; j < SLC_64.nsects; j++) {
10111         MachO::section_64 S_64 = Obj->getSection64(Command, j);
10112         PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size,
10113                      S_64.offset, S_64.align, S_64.reloff, S_64.nreloc,
10114                      S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd,
10115                      sg_segname, filetype, Buf.size(), verbose);
10116       }
10117     } else if (Command.C.cmd == MachO::LC_SYMTAB) {
10118       MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
10119       PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size());
10120     } else if (Command.C.cmd == MachO::LC_DYSYMTAB) {
10121       MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand();
10122       MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
10123       PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(),
10124                                Obj->is64Bit());
10125     } else if (Command.C.cmd == MachO::LC_DYLD_INFO ||
10126                Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) {
10127       MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command);
10128       PrintDyldInfoLoadCommand(DyldInfo, Buf.size());
10129     } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER ||
10130                Command.C.cmd == MachO::LC_ID_DYLINKER ||
10131                Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) {
10132       MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command);
10133       PrintDyldLoadCommand(Dyld, Command.Ptr);
10134     } else if (Command.C.cmd == MachO::LC_UUID) {
10135       MachO::uuid_command Uuid = Obj->getUuidCommand(Command);
10136       PrintUuidLoadCommand(Uuid);
10137     } else if (Command.C.cmd == MachO::LC_RPATH) {
10138       MachO::rpath_command Rpath = Obj->getRpathCommand(Command);
10139       PrintRpathLoadCommand(Rpath, Command.Ptr);
10140     } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX ||
10141                Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS ||
10142                Command.C.cmd == MachO::LC_VERSION_MIN_TVOS ||
10143                Command.C.cmd == MachO::LC_VERSION_MIN_WATCHOS) {
10144       MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command);
10145       PrintVersionMinLoadCommand(Vd);
10146     } else if (Command.C.cmd == MachO::LC_NOTE) {
10147       MachO::note_command Nt = Obj->getNoteLoadCommand(Command);
10148       PrintNoteLoadCommand(Nt);
10149     } else if (Command.C.cmd == MachO::LC_BUILD_VERSION) {
10150       MachO::build_version_command Bv =
10151           Obj->getBuildVersionLoadCommand(Command);
10152       PrintBuildVersionLoadCommand(Obj, Bv);
10153     } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) {
10154       MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command);
10155       PrintSourceVersionCommand(Sd);
10156     } else if (Command.C.cmd == MachO::LC_MAIN) {
10157       MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command);
10158       PrintEntryPointCommand(Ep);
10159     } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO) {
10160       MachO::encryption_info_command Ei =
10161           Obj->getEncryptionInfoCommand(Command);
10162       PrintEncryptionInfoCommand(Ei, Buf.size());
10163     } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO_64) {
10164       MachO::encryption_info_command_64 Ei =
10165           Obj->getEncryptionInfoCommand64(Command);
10166       PrintEncryptionInfoCommand64(Ei, Buf.size());
10167     } else if (Command.C.cmd == MachO::LC_LINKER_OPTION) {
10168       MachO::linker_option_command Lo =
10169           Obj->getLinkerOptionLoadCommand(Command);
10170       PrintLinkerOptionCommand(Lo, Command.Ptr);
10171     } else if (Command.C.cmd == MachO::LC_SUB_FRAMEWORK) {
10172       MachO::sub_framework_command Sf = Obj->getSubFrameworkCommand(Command);
10173       PrintSubFrameworkCommand(Sf, Command.Ptr);
10174     } else if (Command.C.cmd == MachO::LC_SUB_UMBRELLA) {
10175       MachO::sub_umbrella_command Sf = Obj->getSubUmbrellaCommand(Command);
10176       PrintSubUmbrellaCommand(Sf, Command.Ptr);
10177     } else if (Command.C.cmd == MachO::LC_SUB_LIBRARY) {
10178       MachO::sub_library_command Sl = Obj->getSubLibraryCommand(Command);
10179       PrintSubLibraryCommand(Sl, Command.Ptr);
10180     } else if (Command.C.cmd == MachO::LC_SUB_CLIENT) {
10181       MachO::sub_client_command Sc = Obj->getSubClientCommand(Command);
10182       PrintSubClientCommand(Sc, Command.Ptr);
10183     } else if (Command.C.cmd == MachO::LC_ROUTINES) {
10184       MachO::routines_command Rc = Obj->getRoutinesCommand(Command);
10185       PrintRoutinesCommand(Rc);
10186     } else if (Command.C.cmd == MachO::LC_ROUTINES_64) {
10187       MachO::routines_command_64 Rc = Obj->getRoutinesCommand64(Command);
10188       PrintRoutinesCommand64(Rc);
10189     } else if (Command.C.cmd == MachO::LC_THREAD ||
10190                Command.C.cmd == MachO::LC_UNIXTHREAD) {
10191       MachO::thread_command Tc = Obj->getThreadCommand(Command);
10192       PrintThreadCommand(Tc, Command.Ptr, Obj->isLittleEndian(), cputype);
10193     } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB ||
10194                Command.C.cmd == MachO::LC_ID_DYLIB ||
10195                Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
10196                Command.C.cmd == MachO::LC_REEXPORT_DYLIB ||
10197                Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
10198                Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
10199       MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command);
10200       PrintDylibCommand(Dl, Command.Ptr);
10201     } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE ||
10202                Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO ||
10203                Command.C.cmd == MachO::LC_FUNCTION_STARTS ||
10204                Command.C.cmd == MachO::LC_DATA_IN_CODE ||
10205                Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS ||
10206                Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) {
10207       MachO::linkedit_data_command Ld =
10208           Obj->getLinkeditDataLoadCommand(Command);
10209       PrintLinkEditDataCommand(Ld, Buf.size());
10210     } else {
10211       outs() << "      cmd ?(" << format("0x%08" PRIx32, Command.C.cmd)
10212              << ")\n";
10213       outs() << "  cmdsize " << Command.C.cmdsize << "\n";
10214       // TODO: get and print the raw bytes of the load command.
10215     }
10216     // TODO: print all the other kinds of load commands.
10217   }
10218 }
10219 
PrintMachHeader(const MachOObjectFile * Obj,bool verbose)10220 static void PrintMachHeader(const MachOObjectFile *Obj, bool verbose) {
10221   if (Obj->is64Bit()) {
10222     MachO::mach_header_64 H_64;
10223     H_64 = Obj->getHeader64();
10224     PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype,
10225                     H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose);
10226   } else {
10227     MachO::mach_header H;
10228     H = Obj->getHeader();
10229     PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds,
10230                     H.sizeofcmds, H.flags, verbose);
10231   }
10232 }
10233 
printMachOFileHeader(const object::ObjectFile * Obj)10234 void objdump::printMachOFileHeader(const object::ObjectFile *Obj) {
10235   const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
10236   PrintMachHeader(file, !NonVerbose);
10237 }
10238 
printMachOLoadCommands(const object::ObjectFile * Obj)10239 void objdump::printMachOLoadCommands(const object::ObjectFile *Obj) {
10240   const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
10241   uint32_t filetype = 0;
10242   uint32_t cputype = 0;
10243   if (file->is64Bit()) {
10244     MachO::mach_header_64 H_64;
10245     H_64 = file->getHeader64();
10246     filetype = H_64.filetype;
10247     cputype = H_64.cputype;
10248   } else {
10249     MachO::mach_header H;
10250     H = file->getHeader();
10251     filetype = H.filetype;
10252     cputype = H.cputype;
10253   }
10254   PrintLoadCommands(file, filetype, cputype, !NonVerbose);
10255 }
10256 
10257 //===----------------------------------------------------------------------===//
10258 // export trie dumping
10259 //===----------------------------------------------------------------------===//
10260 
printMachOExportsTrie(const object::MachOObjectFile * Obj)10261 static void printMachOExportsTrie(const object::MachOObjectFile *Obj) {
10262   uint64_t BaseSegmentAddress = 0;
10263   for (const auto &Command : Obj->load_commands()) {
10264     if (Command.C.cmd == MachO::LC_SEGMENT) {
10265       MachO::segment_command Seg = Obj->getSegmentLoadCommand(Command);
10266       if (Seg.fileoff == 0 && Seg.filesize != 0) {
10267         BaseSegmentAddress = Seg.vmaddr;
10268         break;
10269       }
10270     } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
10271       MachO::segment_command_64 Seg = Obj->getSegment64LoadCommand(Command);
10272       if (Seg.fileoff == 0 && Seg.filesize != 0) {
10273         BaseSegmentAddress = Seg.vmaddr;
10274         break;
10275       }
10276     }
10277   }
10278   Error Err = Error::success();
10279   for (const object::ExportEntry &Entry : Obj->exports(Err)) {
10280     uint64_t Flags = Entry.flags();
10281     bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT);
10282     bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION);
10283     bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
10284                         MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL);
10285     bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
10286                 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE);
10287     bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER);
10288     if (ReExport)
10289       outs() << "[re-export] ";
10290     else
10291       outs() << format("0x%08llX  ",
10292                        Entry.address() + BaseSegmentAddress);
10293     outs() << Entry.name();
10294     if (WeakDef || ThreadLocal || Resolver || Abs) {
10295       bool NeedsComma = false;
10296       outs() << " [";
10297       if (WeakDef) {
10298         outs() << "weak_def";
10299         NeedsComma = true;
10300       }
10301       if (ThreadLocal) {
10302         if (NeedsComma)
10303           outs() << ", ";
10304         outs() << "per-thread";
10305         NeedsComma = true;
10306       }
10307       if (Abs) {
10308         if (NeedsComma)
10309           outs() << ", ";
10310         outs() << "absolute";
10311         NeedsComma = true;
10312       }
10313       if (Resolver) {
10314         if (NeedsComma)
10315           outs() << ", ";
10316         outs() << format("resolver=0x%08llX", Entry.other());
10317         NeedsComma = true;
10318       }
10319       outs() << "]";
10320     }
10321     if (ReExport) {
10322       StringRef DylibName = "unknown";
10323       int Ordinal = Entry.other() - 1;
10324       Obj->getLibraryShortNameByIndex(Ordinal, DylibName);
10325       if (Entry.otherName().empty())
10326         outs() << " (from " << DylibName << ")";
10327       else
10328         outs() << " (" << Entry.otherName() << " from " << DylibName << ")";
10329     }
10330     outs() << "\n";
10331   }
10332   if (Err)
10333     reportError(std::move(Err), Obj->getFileName());
10334 }
10335 
10336 //===----------------------------------------------------------------------===//
10337 // rebase table dumping
10338 //===----------------------------------------------------------------------===//
10339 
printMachORebaseTable(object::MachOObjectFile * Obj)10340 static void printMachORebaseTable(object::MachOObjectFile *Obj) {
10341   outs() << "segment  section            address     type\n";
10342   Error Err = Error::success();
10343   for (const object::MachORebaseEntry &Entry : Obj->rebaseTable(Err)) {
10344     StringRef SegmentName = Entry.segmentName();
10345     StringRef SectionName = Entry.sectionName();
10346     uint64_t Address = Entry.address();
10347 
10348     // Table lines look like: __DATA  __nl_symbol_ptr  0x0000F00C  pointer
10349     outs() << format("%-8s %-18s 0x%08" PRIX64 "  %s\n",
10350                      SegmentName.str().c_str(), SectionName.str().c_str(),
10351                      Address, Entry.typeName().str().c_str());
10352   }
10353   if (Err)
10354     reportError(std::move(Err), Obj->getFileName());
10355 }
10356 
ordinalName(const object::MachOObjectFile * Obj,int Ordinal)10357 static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) {
10358   StringRef DylibName;
10359   switch (Ordinal) {
10360   case MachO::BIND_SPECIAL_DYLIB_SELF:
10361     return "this-image";
10362   case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE:
10363     return "main-executable";
10364   case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP:
10365     return "flat-namespace";
10366   default:
10367     if (Ordinal > 0) {
10368       std::error_code EC =
10369           Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName);
10370       if (EC)
10371         return "<<bad library ordinal>>";
10372       return DylibName;
10373     }
10374   }
10375   return "<<unknown special ordinal>>";
10376 }
10377 
10378 //===----------------------------------------------------------------------===//
10379 // bind table dumping
10380 //===----------------------------------------------------------------------===//
10381 
printMachOBindTable(object::MachOObjectFile * Obj)10382 static void printMachOBindTable(object::MachOObjectFile *Obj) {
10383   // Build table of sections so names can used in final output.
10384   outs() << "segment  section            address    type       "
10385             "addend dylib            symbol\n";
10386   Error Err = Error::success();
10387   for (const object::MachOBindEntry &Entry : Obj->bindTable(Err)) {
10388     StringRef SegmentName = Entry.segmentName();
10389     StringRef SectionName = Entry.sectionName();
10390     uint64_t Address = Entry.address();
10391 
10392     // Table lines look like:
10393     //  __DATA  __got  0x00012010    pointer   0 libSystem ___stack_chk_guard
10394     StringRef Attr;
10395     if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT)
10396       Attr = " (weak_import)";
10397     outs() << left_justify(SegmentName, 8) << " "
10398            << left_justify(SectionName, 18) << " "
10399            << format_hex(Address, 10, true) << " "
10400            << left_justify(Entry.typeName(), 8) << " "
10401            << format_decimal(Entry.addend(), 8) << " "
10402            << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
10403            << Entry.symbolName() << Attr << "\n";
10404   }
10405   if (Err)
10406     reportError(std::move(Err), Obj->getFileName());
10407 }
10408 
10409 //===----------------------------------------------------------------------===//
10410 // lazy bind table dumping
10411 //===----------------------------------------------------------------------===//
10412 
printMachOLazyBindTable(object::MachOObjectFile * Obj)10413 static void printMachOLazyBindTable(object::MachOObjectFile *Obj) {
10414   outs() << "segment  section            address     "
10415             "dylib            symbol\n";
10416   Error Err = Error::success();
10417   for (const object::MachOBindEntry &Entry : Obj->lazyBindTable(Err)) {
10418     StringRef SegmentName = Entry.segmentName();
10419     StringRef SectionName = Entry.sectionName();
10420     uint64_t Address = Entry.address();
10421 
10422     // Table lines look like:
10423     //  __DATA  __got  0x00012010 libSystem ___stack_chk_guard
10424     outs() << left_justify(SegmentName, 8) << " "
10425            << left_justify(SectionName, 18) << " "
10426            << format_hex(Address, 10, true) << " "
10427            << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
10428            << Entry.symbolName() << "\n";
10429   }
10430   if (Err)
10431     reportError(std::move(Err), Obj->getFileName());
10432 }
10433 
10434 //===----------------------------------------------------------------------===//
10435 // weak bind table dumping
10436 //===----------------------------------------------------------------------===//
10437 
printMachOWeakBindTable(object::MachOObjectFile * Obj)10438 static void printMachOWeakBindTable(object::MachOObjectFile *Obj) {
10439   outs() << "segment  section            address     "
10440             "type       addend   symbol\n";
10441   Error Err = Error::success();
10442   for (const object::MachOBindEntry &Entry : Obj->weakBindTable(Err)) {
10443     // Strong symbols don't have a location to update.
10444     if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) {
10445       outs() << "                                        strong              "
10446              << Entry.symbolName() << "\n";
10447       continue;
10448     }
10449     StringRef SegmentName = Entry.segmentName();
10450     StringRef SectionName = Entry.sectionName();
10451     uint64_t Address = Entry.address();
10452 
10453     // Table lines look like:
10454     // __DATA  __data  0x00001000  pointer    0   _foo
10455     outs() << left_justify(SegmentName, 8) << " "
10456            << left_justify(SectionName, 18) << " "
10457            << format_hex(Address, 10, true) << " "
10458            << left_justify(Entry.typeName(), 8) << " "
10459            << format_decimal(Entry.addend(), 8) << "   " << Entry.symbolName()
10460            << "\n";
10461   }
10462   if (Err)
10463     reportError(std::move(Err), Obj->getFileName());
10464 }
10465 
10466 // get_dyld_bind_info_symbolname() is used for disassembly and passed an
10467 // address, ReferenceValue, in the Mach-O file and looks in the dyld bind
10468 // information for that address. If the address is found its binding symbol
10469 // name is returned.  If not nullptr is returned.
get_dyld_bind_info_symbolname(uint64_t ReferenceValue,struct DisassembleInfo * info)10470 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
10471                                                  struct DisassembleInfo *info) {
10472   if (info->bindtable == nullptr) {
10473     info->bindtable = std::make_unique<SymbolAddressMap>();
10474     Error Err = Error::success();
10475     for (const object::MachOBindEntry &Entry : info->O->bindTable(Err)) {
10476       uint64_t Address = Entry.address();
10477       StringRef name = Entry.symbolName();
10478       if (!name.empty())
10479         (*info->bindtable)[Address] = name;
10480     }
10481     if (Err)
10482       reportError(std::move(Err), info->O->getFileName());
10483   }
10484   auto name = info->bindtable->lookup(ReferenceValue);
10485   return !name.empty() ? name.data() : nullptr;
10486 }
10487 
printLazyBindTable(ObjectFile * o)10488 void objdump::printLazyBindTable(ObjectFile *o) {
10489   outs() << "Lazy bind table:\n";
10490   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10491     printMachOLazyBindTable(MachO);
10492   else
10493     WithColor::error()
10494         << "This operation is only currently supported "
10495            "for Mach-O executable files.\n";
10496 }
10497 
printWeakBindTable(ObjectFile * o)10498 void objdump::printWeakBindTable(ObjectFile *o) {
10499   outs() << "Weak bind table:\n";
10500   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10501     printMachOWeakBindTable(MachO);
10502   else
10503     WithColor::error()
10504         << "This operation is only currently supported "
10505            "for Mach-O executable files.\n";
10506 }
10507 
printExportsTrie(const ObjectFile * o)10508 void objdump::printExportsTrie(const ObjectFile *o) {
10509   outs() << "Exports trie:\n";
10510   if (const MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10511     printMachOExportsTrie(MachO);
10512   else
10513     WithColor::error()
10514         << "This operation is only currently supported "
10515            "for Mach-O executable files.\n";
10516 }
10517 
printRebaseTable(ObjectFile * o)10518 void objdump::printRebaseTable(ObjectFile *o) {
10519   outs() << "Rebase table:\n";
10520   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10521     printMachORebaseTable(MachO);
10522   else
10523     WithColor::error()
10524         << "This operation is only currently supported "
10525            "for Mach-O executable files.\n";
10526 }
10527 
printBindTable(ObjectFile * o)10528 void objdump::printBindTable(ObjectFile *o) {
10529   outs() << "Bind table:\n";
10530   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10531     printMachOBindTable(MachO);
10532   else
10533     WithColor::error()
10534         << "This operation is only currently supported "
10535            "for Mach-O executable files.\n";
10536 }
10537