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 ()__anon30756ae00111::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__anon30756ae00211::ScopedXarFile252   ScopedXarFile(const char *filename, int32_t flags)
253       : xar(xar_open(filename, flags)) {}
~ScopedXarFile__anon30756ae00211::ScopedXarFile254   ~ScopedXarFile() {
255     if (xar)
256       xar_close(xar);
257   }
258   ScopedXarFile(const ScopedXarFile &) = delete;
259   ScopedXarFile &operator=(const ScopedXarFile &) = delete;
operator xar_t__anon30756ae00211::ScopedXarFile260   operator xar_t() { return xar; }
261 };
262 
263 struct ScopedXarIter {
264   xar_iter_t iter;
ScopedXarIter__anon30756ae00211::ScopedXarIter265   ScopedXarIter() : iter(xar_iter_new()) {}
~ScopedXarIter__anon30756ae00211::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__anon30756ae00211::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 (none_of(ArchFlags, [&](const std::string &Name) {
1893         return Name == ArchFlagName;
1894       })) {
1895     WithColor::error(errs(), "llvm-objdump")
1896         << Filename << ": no architecture specified.\n";
1897     return false;
1898   }
1899   return true;
1900 }
1901 
1902 static void printObjcMetaData(MachOObjectFile *O, bool verbose);
1903 
1904 // ProcessMachO() is passed a single opened Mach-O file, which may be an
1905 // archive member and or in a slice of a universal file.  It prints the
1906 // the file name and header info and then processes it according to the
1907 // command line options.
ProcessMachO(StringRef Name,MachOObjectFile * MachOOF,StringRef ArchiveMemberName=StringRef (),StringRef ArchitectureName=StringRef ())1908 static void ProcessMachO(StringRef Name, MachOObjectFile *MachOOF,
1909                          StringRef ArchiveMemberName = StringRef(),
1910                          StringRef ArchitectureName = StringRef()) {
1911   // If we are doing some processing here on the Mach-O file print the header
1912   // info.  And don't print it otherwise like in the case of printing the
1913   // UniversalHeaders or ArchiveHeaders.
1914   if (Disassemble || Relocations || PrivateHeaders || ExportsTrie || Rebase ||
1915       Bind || SymbolTable || LazyBind || WeakBind || IndirectSymbols ||
1916       DataInCode || LinkOptHints || DylibsUsed || DylibId || ObjcMetaData ||
1917       (!FilterSections.empty())) {
1918     if (!NoLeadingHeaders) {
1919       outs() << Name;
1920       if (!ArchiveMemberName.empty())
1921         outs() << '(' << ArchiveMemberName << ')';
1922       if (!ArchitectureName.empty())
1923         outs() << " (architecture " << ArchitectureName << ")";
1924       outs() << ":\n";
1925     }
1926   }
1927   // To use the report_error() form with an ArchiveName and FileName set
1928   // these up based on what is passed for Name and ArchiveMemberName.
1929   StringRef ArchiveName;
1930   StringRef FileName;
1931   if (!ArchiveMemberName.empty()) {
1932     ArchiveName = Name;
1933     FileName = ArchiveMemberName;
1934   } else {
1935     ArchiveName = StringRef();
1936     FileName = Name;
1937   }
1938 
1939   // If we need the symbol table to do the operation then check it here to
1940   // produce a good error message as to where the Mach-O file comes from in
1941   // the error message.
1942   if (Disassemble || IndirectSymbols || !FilterSections.empty() || UnwindInfo)
1943     if (Error Err = MachOOF->checkSymbolTable())
1944       reportError(std::move(Err), FileName, ArchiveName, ArchitectureName);
1945 
1946   if (DisassembleAll) {
1947     for (const SectionRef &Section : MachOOF->sections()) {
1948       StringRef SectName;
1949       if (Expected<StringRef> NameOrErr = Section.getName())
1950         SectName = *NameOrErr;
1951       else
1952         consumeError(NameOrErr.takeError());
1953 
1954       if (SectName.equals("__text")) {
1955         DataRefImpl Ref = Section.getRawDataRefImpl();
1956         StringRef SegName = MachOOF->getSectionFinalSegmentName(Ref);
1957         DisassembleMachO(FileName, MachOOF, SegName, SectName);
1958       }
1959     }
1960   }
1961   else if (Disassemble) {
1962     if (MachOOF->getHeader().filetype == MachO::MH_KEXT_BUNDLE &&
1963         MachOOF->getHeader().cputype == MachO::CPU_TYPE_ARM64)
1964       DisassembleMachO(FileName, MachOOF, "__TEXT_EXEC", "__text");
1965     else
1966       DisassembleMachO(FileName, MachOOF, "__TEXT", "__text");
1967   }
1968   if (IndirectSymbols)
1969     PrintIndirectSymbols(MachOOF, !NonVerbose);
1970   if (DataInCode)
1971     PrintDataInCodeTable(MachOOF, !NonVerbose);
1972   if (LinkOptHints)
1973     PrintLinkOptHints(MachOOF);
1974   if (Relocations)
1975     PrintRelocations(MachOOF, !NonVerbose);
1976   if (SectionHeaders)
1977     printSectionHeaders(MachOOF);
1978   if (SectionContents)
1979     printSectionContents(MachOOF);
1980   if (!FilterSections.empty())
1981     DumpSectionContents(FileName, MachOOF, !NonVerbose);
1982   if (InfoPlist)
1983     DumpInfoPlistSectionContents(FileName, MachOOF);
1984   if (DylibsUsed)
1985     PrintDylibs(MachOOF, false);
1986   if (DylibId)
1987     PrintDylibs(MachOOF, true);
1988   if (SymbolTable)
1989     printSymbolTable(MachOOF, ArchiveName, ArchitectureName);
1990   if (UnwindInfo)
1991     printMachOUnwindInfo(MachOOF);
1992   if (PrivateHeaders) {
1993     printMachOFileHeader(MachOOF);
1994     printMachOLoadCommands(MachOOF);
1995   }
1996   if (FirstPrivateHeader)
1997     printMachOFileHeader(MachOOF);
1998   if (ObjcMetaData)
1999     printObjcMetaData(MachOOF, !NonVerbose);
2000   if (ExportsTrie)
2001     printExportsTrie(MachOOF);
2002   if (Rebase)
2003     printRebaseTable(MachOOF);
2004   if (Bind)
2005     printBindTable(MachOOF);
2006   if (LazyBind)
2007     printLazyBindTable(MachOOF);
2008   if (WeakBind)
2009     printWeakBindTable(MachOOF);
2010 
2011   if (DwarfDumpType != DIDT_Null) {
2012     std::unique_ptr<DIContext> DICtx = DWARFContext::create(*MachOOF);
2013     // Dump the complete DWARF structure.
2014     DIDumpOptions DumpOpts;
2015     DumpOpts.DumpType = DwarfDumpType;
2016     DICtx->dump(outs(), DumpOpts);
2017   }
2018 }
2019 
2020 // printUnknownCPUType() helps print_fat_headers for unknown CPU's.
printUnknownCPUType(uint32_t cputype,uint32_t cpusubtype)2021 static void printUnknownCPUType(uint32_t cputype, uint32_t cpusubtype) {
2022   outs() << "    cputype (" << cputype << ")\n";
2023   outs() << "    cpusubtype (" << cpusubtype << ")\n";
2024 }
2025 
2026 // printCPUType() helps print_fat_headers by printing the cputype and
2027 // pusubtype (symbolically for the one's it knows about).
printCPUType(uint32_t cputype,uint32_t cpusubtype)2028 static void printCPUType(uint32_t cputype, uint32_t cpusubtype) {
2029   switch (cputype) {
2030   case MachO::CPU_TYPE_I386:
2031     switch (cpusubtype) {
2032     case MachO::CPU_SUBTYPE_I386_ALL:
2033       outs() << "    cputype CPU_TYPE_I386\n";
2034       outs() << "    cpusubtype CPU_SUBTYPE_I386_ALL\n";
2035       break;
2036     default:
2037       printUnknownCPUType(cputype, cpusubtype);
2038       break;
2039     }
2040     break;
2041   case MachO::CPU_TYPE_X86_64:
2042     switch (cpusubtype) {
2043     case MachO::CPU_SUBTYPE_X86_64_ALL:
2044       outs() << "    cputype CPU_TYPE_X86_64\n";
2045       outs() << "    cpusubtype CPU_SUBTYPE_X86_64_ALL\n";
2046       break;
2047     case MachO::CPU_SUBTYPE_X86_64_H:
2048       outs() << "    cputype CPU_TYPE_X86_64\n";
2049       outs() << "    cpusubtype CPU_SUBTYPE_X86_64_H\n";
2050       break;
2051     default:
2052       printUnknownCPUType(cputype, cpusubtype);
2053       break;
2054     }
2055     break;
2056   case MachO::CPU_TYPE_ARM:
2057     switch (cpusubtype) {
2058     case MachO::CPU_SUBTYPE_ARM_ALL:
2059       outs() << "    cputype CPU_TYPE_ARM\n";
2060       outs() << "    cpusubtype CPU_SUBTYPE_ARM_ALL\n";
2061       break;
2062     case MachO::CPU_SUBTYPE_ARM_V4T:
2063       outs() << "    cputype CPU_TYPE_ARM\n";
2064       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V4T\n";
2065       break;
2066     case MachO::CPU_SUBTYPE_ARM_V5TEJ:
2067       outs() << "    cputype CPU_TYPE_ARM\n";
2068       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V5TEJ\n";
2069       break;
2070     case MachO::CPU_SUBTYPE_ARM_XSCALE:
2071       outs() << "    cputype CPU_TYPE_ARM\n";
2072       outs() << "    cpusubtype CPU_SUBTYPE_ARM_XSCALE\n";
2073       break;
2074     case MachO::CPU_SUBTYPE_ARM_V6:
2075       outs() << "    cputype CPU_TYPE_ARM\n";
2076       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V6\n";
2077       break;
2078     case MachO::CPU_SUBTYPE_ARM_V6M:
2079       outs() << "    cputype CPU_TYPE_ARM\n";
2080       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V6M\n";
2081       break;
2082     case MachO::CPU_SUBTYPE_ARM_V7:
2083       outs() << "    cputype CPU_TYPE_ARM\n";
2084       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7\n";
2085       break;
2086     case MachO::CPU_SUBTYPE_ARM_V7EM:
2087       outs() << "    cputype CPU_TYPE_ARM\n";
2088       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7EM\n";
2089       break;
2090     case MachO::CPU_SUBTYPE_ARM_V7K:
2091       outs() << "    cputype CPU_TYPE_ARM\n";
2092       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7K\n";
2093       break;
2094     case MachO::CPU_SUBTYPE_ARM_V7M:
2095       outs() << "    cputype CPU_TYPE_ARM\n";
2096       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7M\n";
2097       break;
2098     case MachO::CPU_SUBTYPE_ARM_V7S:
2099       outs() << "    cputype CPU_TYPE_ARM\n";
2100       outs() << "    cpusubtype CPU_SUBTYPE_ARM_V7S\n";
2101       break;
2102     default:
2103       printUnknownCPUType(cputype, cpusubtype);
2104       break;
2105     }
2106     break;
2107   case MachO::CPU_TYPE_ARM64:
2108     switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2109     case MachO::CPU_SUBTYPE_ARM64_ALL:
2110       outs() << "    cputype CPU_TYPE_ARM64\n";
2111       outs() << "    cpusubtype CPU_SUBTYPE_ARM64_ALL\n";
2112       break;
2113     case MachO::CPU_SUBTYPE_ARM64_V8:
2114       outs() << "    cputype CPU_TYPE_ARM64\n";
2115       outs() << "    cpusubtype CPU_SUBTYPE_ARM64_V8\n";
2116       break;
2117     case MachO::CPU_SUBTYPE_ARM64E:
2118       outs() << "    cputype CPU_TYPE_ARM64\n";
2119       outs() << "    cpusubtype CPU_SUBTYPE_ARM64E\n";
2120       break;
2121     default:
2122       printUnknownCPUType(cputype, cpusubtype);
2123       break;
2124     }
2125     break;
2126   case MachO::CPU_TYPE_ARM64_32:
2127     switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2128     case MachO::CPU_SUBTYPE_ARM64_32_V8:
2129       outs() << "    cputype CPU_TYPE_ARM64_32\n";
2130       outs() << "    cpusubtype CPU_SUBTYPE_ARM64_32_V8\n";
2131       break;
2132     default:
2133       printUnknownCPUType(cputype, cpusubtype);
2134       break;
2135     }
2136     break;
2137   default:
2138     printUnknownCPUType(cputype, cpusubtype);
2139     break;
2140   }
2141 }
2142 
printMachOUniversalHeaders(const object::MachOUniversalBinary * UB,bool verbose)2143 static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB,
2144                                        bool verbose) {
2145   outs() << "Fat headers\n";
2146   if (verbose) {
2147     if (UB->getMagic() == MachO::FAT_MAGIC)
2148       outs() << "fat_magic FAT_MAGIC\n";
2149     else // UB->getMagic() == MachO::FAT_MAGIC_64
2150       outs() << "fat_magic FAT_MAGIC_64\n";
2151   } else
2152     outs() << "fat_magic " << format("0x%" PRIx32, MachO::FAT_MAGIC) << "\n";
2153 
2154   uint32_t nfat_arch = UB->getNumberOfObjects();
2155   StringRef Buf = UB->getData();
2156   uint64_t size = Buf.size();
2157   uint64_t big_size = sizeof(struct MachO::fat_header) +
2158                       nfat_arch * sizeof(struct MachO::fat_arch);
2159   outs() << "nfat_arch " << UB->getNumberOfObjects();
2160   if (nfat_arch == 0)
2161     outs() << " (malformed, contains zero architecture types)\n";
2162   else if (big_size > size)
2163     outs() << " (malformed, architectures past end of file)\n";
2164   else
2165     outs() << "\n";
2166 
2167   for (uint32_t i = 0; i < nfat_arch; ++i) {
2168     MachOUniversalBinary::ObjectForArch OFA(UB, i);
2169     uint32_t cputype = OFA.getCPUType();
2170     uint32_t cpusubtype = OFA.getCPUSubType();
2171     outs() << "architecture ";
2172     for (uint32_t j = 0; i != 0 && j <= i - 1; j++) {
2173       MachOUniversalBinary::ObjectForArch other_OFA(UB, j);
2174       uint32_t other_cputype = other_OFA.getCPUType();
2175       uint32_t other_cpusubtype = other_OFA.getCPUSubType();
2176       if (cputype != 0 && cpusubtype != 0 && cputype == other_cputype &&
2177           (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) ==
2178               (other_cpusubtype & ~MachO::CPU_SUBTYPE_MASK)) {
2179         outs() << "(illegal duplicate architecture) ";
2180         break;
2181       }
2182     }
2183     if (verbose) {
2184       outs() << OFA.getArchFlagName() << "\n";
2185       printCPUType(cputype, cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2186     } else {
2187       outs() << i << "\n";
2188       outs() << "    cputype " << cputype << "\n";
2189       outs() << "    cpusubtype " << (cpusubtype & ~MachO::CPU_SUBTYPE_MASK)
2190              << "\n";
2191     }
2192     if (verbose &&
2193         (cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64)
2194       outs() << "    capabilities CPU_SUBTYPE_LIB64\n";
2195     else
2196       outs() << "    capabilities "
2197              << format("0x%" PRIx32,
2198                        (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24) << "\n";
2199     outs() << "    offset " << OFA.getOffset();
2200     if (OFA.getOffset() > size)
2201       outs() << " (past end of file)";
2202     if (OFA.getOffset() % (1ull << OFA.getAlign()) != 0)
2203       outs() << " (not aligned on it's alignment (2^" << OFA.getAlign() << ")";
2204     outs() << "\n";
2205     outs() << "    size " << OFA.getSize();
2206     big_size = OFA.getOffset() + OFA.getSize();
2207     if (big_size > size)
2208       outs() << " (past end of file)";
2209     outs() << "\n";
2210     outs() << "    align 2^" << OFA.getAlign() << " (" << (1 << OFA.getAlign())
2211            << ")\n";
2212   }
2213 }
2214 
printArchiveChild(StringRef Filename,const Archive::Child & C,size_t ChildIndex,bool verbose,bool print_offset,StringRef ArchitectureName=StringRef ())2215 static void printArchiveChild(StringRef Filename, const Archive::Child &C,
2216                               size_t ChildIndex, bool verbose,
2217                               bool print_offset,
2218                               StringRef ArchitectureName = StringRef()) {
2219   if (print_offset)
2220     outs() << C.getChildOffset() << "\t";
2221   sys::fs::perms Mode =
2222       unwrapOrError(C.getAccessMode(), getFileNameForError(C, ChildIndex),
2223                     Filename, ArchitectureName);
2224   if (verbose) {
2225     // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG.
2226     // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG.
2227     outs() << "-";
2228     outs() << ((Mode & sys::fs::owner_read) ? "r" : "-");
2229     outs() << ((Mode & sys::fs::owner_write) ? "w" : "-");
2230     outs() << ((Mode & sys::fs::owner_exe) ? "x" : "-");
2231     outs() << ((Mode & sys::fs::group_read) ? "r" : "-");
2232     outs() << ((Mode & sys::fs::group_write) ? "w" : "-");
2233     outs() << ((Mode & sys::fs::group_exe) ? "x" : "-");
2234     outs() << ((Mode & sys::fs::others_read) ? "r" : "-");
2235     outs() << ((Mode & sys::fs::others_write) ? "w" : "-");
2236     outs() << ((Mode & sys::fs::others_exe) ? "x" : "-");
2237   } else {
2238     outs() << format("0%o ", Mode);
2239   }
2240 
2241   outs() << format("%3d/%-3d %5" PRId64 " ",
2242                    unwrapOrError(C.getUID(), getFileNameForError(C, ChildIndex),
2243                                  Filename, ArchitectureName),
2244                    unwrapOrError(C.getGID(), getFileNameForError(C, ChildIndex),
2245                                  Filename, ArchitectureName),
2246                    unwrapOrError(C.getRawSize(),
2247                                  getFileNameForError(C, ChildIndex), Filename,
2248                                  ArchitectureName));
2249 
2250   StringRef RawLastModified = C.getRawLastModified();
2251   if (verbose) {
2252     unsigned Seconds;
2253     if (RawLastModified.getAsInteger(10, Seconds))
2254       outs() << "(date: \"" << RawLastModified
2255              << "\" contains non-decimal chars) ";
2256     else {
2257       // Since cime(3) returns a 26 character string of the form:
2258       // "Sun Sep 16 01:03:52 1973\n\0"
2259       // just print 24 characters.
2260       time_t t = Seconds;
2261       outs() << format("%.24s ", ctime(&t));
2262     }
2263   } else {
2264     outs() << RawLastModified << " ";
2265   }
2266 
2267   if (verbose) {
2268     Expected<StringRef> NameOrErr = C.getName();
2269     if (!NameOrErr) {
2270       consumeError(NameOrErr.takeError());
2271       outs() << unwrapOrError(C.getRawName(),
2272                               getFileNameForError(C, ChildIndex), Filename,
2273                               ArchitectureName)
2274              << "\n";
2275     } else {
2276       StringRef Name = NameOrErr.get();
2277       outs() << Name << "\n";
2278     }
2279   } else {
2280     outs() << unwrapOrError(C.getRawName(), getFileNameForError(C, ChildIndex),
2281                             Filename, ArchitectureName)
2282            << "\n";
2283   }
2284 }
2285 
printArchiveHeaders(StringRef Filename,Archive * A,bool verbose,bool print_offset,StringRef ArchitectureName=StringRef ())2286 static void printArchiveHeaders(StringRef Filename, Archive *A, bool verbose,
2287                                 bool print_offset,
2288                                 StringRef ArchitectureName = StringRef()) {
2289   Error Err = Error::success();
2290   size_t I = 0;
2291   for (const auto &C : A->children(Err, false))
2292     printArchiveChild(Filename, C, I++, verbose, print_offset,
2293                       ArchitectureName);
2294 
2295   if (Err)
2296     reportError(std::move(Err), Filename, "", ArchitectureName);
2297 }
2298 
ValidateArchFlags()2299 static bool ValidateArchFlags() {
2300   // Check for -arch all and verifiy the -arch flags are valid.
2301   for (unsigned i = 0; i < ArchFlags.size(); ++i) {
2302     if (ArchFlags[i] == "all") {
2303       ArchAll = true;
2304     } else {
2305       if (!MachOObjectFile::isValidArch(ArchFlags[i])) {
2306         WithColor::error(errs(), "llvm-objdump")
2307             << "unknown architecture named '" + ArchFlags[i] +
2308                    "'for the -arch option\n";
2309         return false;
2310       }
2311     }
2312   }
2313   return true;
2314 }
2315 
2316 // ParseInputMachO() parses the named Mach-O file in Filename and handles the
2317 // -arch flags selecting just those slices as specified by them and also parses
2318 // archive files.  Then for each individual Mach-O file ProcessMachO() is
2319 // called to process the file based on the command line options.
parseInputMachO(StringRef Filename)2320 void objdump::parseInputMachO(StringRef Filename) {
2321   if (!ValidateArchFlags())
2322     return;
2323 
2324   // Attempt to open the binary.
2325   Expected<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename);
2326   if (!BinaryOrErr) {
2327     if (Error E = isNotObjectErrorInvalidFileType(BinaryOrErr.takeError()))
2328       reportError(std::move(E), Filename);
2329     else
2330       outs() << Filename << ": is not an object file\n";
2331     return;
2332   }
2333   Binary &Bin = *BinaryOrErr.get().getBinary();
2334 
2335   if (Archive *A = dyn_cast<Archive>(&Bin)) {
2336     outs() << "Archive : " << Filename << "\n";
2337     if (ArchiveHeaders)
2338       printArchiveHeaders(Filename, A, !NonVerbose, ArchiveMemberOffsets);
2339 
2340     Error Err = Error::success();
2341     unsigned I = -1;
2342     for (auto &C : A->children(Err)) {
2343       ++I;
2344       Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2345       if (!ChildOrErr) {
2346         if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2347           reportError(std::move(E), getFileNameForError(C, I), Filename);
2348         continue;
2349       }
2350       if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
2351         if (!checkMachOAndArchFlags(O, Filename))
2352           return;
2353         ProcessMachO(Filename, O, O->getFileName());
2354       }
2355     }
2356     if (Err)
2357       reportError(std::move(Err), Filename);
2358     return;
2359   }
2360   if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) {
2361     parseInputMachO(UB);
2362     return;
2363   }
2364   if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) {
2365     if (!checkMachOAndArchFlags(O, Filename))
2366       return;
2367     if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O))
2368       ProcessMachO(Filename, MachOOF);
2369     else
2370       WithColor::error(errs(), "llvm-objdump")
2371           << Filename << "': "
2372           << "object is not a Mach-O file type.\n";
2373     return;
2374   }
2375   llvm_unreachable("Input object can't be invalid at this point");
2376 }
2377 
parseInputMachO(MachOUniversalBinary * UB)2378 void objdump::parseInputMachO(MachOUniversalBinary *UB) {
2379   if (!ValidateArchFlags())
2380     return;
2381 
2382   auto Filename = UB->getFileName();
2383 
2384   if (UniversalHeaders)
2385     printMachOUniversalHeaders(UB, !NonVerbose);
2386 
2387   // If we have a list of architecture flags specified dump only those.
2388   if (!ArchAll && !ArchFlags.empty()) {
2389     // Look for a slice in the universal binary that matches each ArchFlag.
2390     bool ArchFound;
2391     for (unsigned i = 0; i < ArchFlags.size(); ++i) {
2392       ArchFound = false;
2393       for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2394                                                   E = UB->end_objects();
2395             I != E; ++I) {
2396         if (ArchFlags[i] == I->getArchFlagName()) {
2397           ArchFound = true;
2398           Expected<std::unique_ptr<ObjectFile>> ObjOrErr =
2399               I->getAsObjectFile();
2400           std::string ArchitectureName = "";
2401           if (ArchFlags.size() > 1)
2402             ArchitectureName = I->getArchFlagName();
2403           if (ObjOrErr) {
2404             ObjectFile &O = *ObjOrErr.get();
2405             if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
2406               ProcessMachO(Filename, MachOOF, "", ArchitectureName);
2407           } else if (Error E = isNotObjectErrorInvalidFileType(
2408                          ObjOrErr.takeError())) {
2409             reportError(std::move(E), "", Filename, ArchitectureName);
2410             continue;
2411           } else if (Expected<std::unique_ptr<Archive>> AOrErr =
2412                          I->getAsArchive()) {
2413             std::unique_ptr<Archive> &A = *AOrErr;
2414             outs() << "Archive : " << Filename;
2415             if (!ArchitectureName.empty())
2416               outs() << " (architecture " << ArchitectureName << ")";
2417             outs() << "\n";
2418             if (ArchiveHeaders)
2419               printArchiveHeaders(Filename, A.get(), !NonVerbose,
2420                                   ArchiveMemberOffsets, ArchitectureName);
2421             Error Err = Error::success();
2422             unsigned I = -1;
2423             for (auto &C : A->children(Err)) {
2424               ++I;
2425               Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2426               if (!ChildOrErr) {
2427                 if (Error E =
2428                         isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2429                   reportError(std::move(E), getFileNameForError(C, I), Filename,
2430                               ArchitectureName);
2431                 continue;
2432               }
2433               if (MachOObjectFile *O =
2434                       dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
2435                 ProcessMachO(Filename, O, O->getFileName(), ArchitectureName);
2436             }
2437             if (Err)
2438               reportError(std::move(Err), Filename);
2439           } else {
2440             consumeError(AOrErr.takeError());
2441             reportError(Filename,
2442                         "Mach-O universal file for architecture " +
2443                             StringRef(I->getArchFlagName()) +
2444                             " is not a Mach-O file or an archive file");
2445           }
2446         }
2447       }
2448       if (!ArchFound) {
2449         WithColor::error(errs(), "llvm-objdump")
2450             << "file: " + Filename + " does not contain "
2451             << "architecture: " + ArchFlags[i] + "\n";
2452         return;
2453       }
2454     }
2455     return;
2456   }
2457   // No architecture flags were specified so if this contains a slice that
2458   // matches the host architecture dump only that.
2459   if (!ArchAll) {
2460     for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2461                                                 E = UB->end_objects();
2462           I != E; ++I) {
2463       if (MachOObjectFile::getHostArch().getArchName() ==
2464           I->getArchFlagName()) {
2465         Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
2466         std::string ArchiveName;
2467         ArchiveName.clear();
2468         if (ObjOrErr) {
2469           ObjectFile &O = *ObjOrErr.get();
2470           if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
2471             ProcessMachO(Filename, MachOOF);
2472         } else if (Error E =
2473                        isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) {
2474           reportError(std::move(E), Filename);
2475         } else if (Expected<std::unique_ptr<Archive>> AOrErr =
2476                        I->getAsArchive()) {
2477           std::unique_ptr<Archive> &A = *AOrErr;
2478           outs() << "Archive : " << Filename << "\n";
2479           if (ArchiveHeaders)
2480             printArchiveHeaders(Filename, A.get(), !NonVerbose,
2481                                 ArchiveMemberOffsets);
2482           Error Err = Error::success();
2483           unsigned I = -1;
2484           for (auto &C : A->children(Err)) {
2485             ++I;
2486             Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2487             if (!ChildOrErr) {
2488               if (Error E =
2489                       isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2490                 reportError(std::move(E), getFileNameForError(C, I), Filename);
2491               continue;
2492             }
2493             if (MachOObjectFile *O =
2494                     dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
2495               ProcessMachO(Filename, O, O->getFileName());
2496           }
2497           if (Err)
2498             reportError(std::move(Err), Filename);
2499         } else {
2500           consumeError(AOrErr.takeError());
2501           reportError(Filename, "Mach-O universal file for architecture " +
2502                                     StringRef(I->getArchFlagName()) +
2503                                     " is not a Mach-O file or an archive file");
2504         }
2505         return;
2506       }
2507     }
2508   }
2509   // Either all architectures have been specified or none have been specified
2510   // and this does not contain the host architecture so dump all the slices.
2511   bool moreThanOneArch = UB->getNumberOfObjects() > 1;
2512   for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
2513                                               E = UB->end_objects();
2514         I != E; ++I) {
2515     Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
2516     std::string ArchitectureName = "";
2517     if (moreThanOneArch)
2518       ArchitectureName = I->getArchFlagName();
2519     if (ObjOrErr) {
2520       ObjectFile &Obj = *ObjOrErr.get();
2521       if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj))
2522         ProcessMachO(Filename, MachOOF, "", ArchitectureName);
2523     } else if (Error E =
2524                    isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) {
2525       reportError(std::move(E), Filename, "", ArchitectureName);
2526     } else if (Expected<std::unique_ptr<Archive>> AOrErr = I->getAsArchive()) {
2527       std::unique_ptr<Archive> &A = *AOrErr;
2528       outs() << "Archive : " << Filename;
2529       if (!ArchitectureName.empty())
2530         outs() << " (architecture " << ArchitectureName << ")";
2531       outs() << "\n";
2532       if (ArchiveHeaders)
2533         printArchiveHeaders(Filename, A.get(), !NonVerbose,
2534                             ArchiveMemberOffsets, ArchitectureName);
2535       Error Err = Error::success();
2536       unsigned I = -1;
2537       for (auto &C : A->children(Err)) {
2538         ++I;
2539         Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary();
2540         if (!ChildOrErr) {
2541           if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError()))
2542             reportError(std::move(E), getFileNameForError(C, I), Filename,
2543                         ArchitectureName);
2544           continue;
2545         }
2546         if (MachOObjectFile *O =
2547                 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
2548           if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O))
2549             ProcessMachO(Filename, MachOOF, MachOOF->getFileName(),
2550                           ArchitectureName);
2551         }
2552       }
2553       if (Err)
2554         reportError(std::move(Err), Filename);
2555     } else {
2556       consumeError(AOrErr.takeError());
2557       reportError(Filename, "Mach-O universal file for architecture " +
2558                                 StringRef(I->getArchFlagName()) +
2559                                 " is not a Mach-O file or an archive file");
2560     }
2561   }
2562 }
2563 
2564 namespace {
2565 // The block of info used by the Symbolizer call backs.
2566 struct DisassembleInfo {
DisassembleInfo__anon30756ae00711::DisassembleInfo2567   DisassembleInfo(MachOObjectFile *O, SymbolAddressMap *AddrMap,
2568                   std::vector<SectionRef> *Sections, bool verbose)
2569     : verbose(verbose), O(O), AddrMap(AddrMap), Sections(Sections) {}
2570   bool verbose;
2571   MachOObjectFile *O;
2572   SectionRef S;
2573   SymbolAddressMap *AddrMap;
2574   std::vector<SectionRef> *Sections;
2575   const char *class_name = nullptr;
2576   const char *selector_name = nullptr;
2577   std::unique_ptr<char[]> method = nullptr;
2578   char *demangled_name = nullptr;
2579   uint64_t adrp_addr = 0;
2580   uint32_t adrp_inst = 0;
2581   std::unique_ptr<SymbolAddressMap> bindtable;
2582   uint32_t depth = 0;
2583 };
2584 } // namespace
2585 
2586 // SymbolizerGetOpInfo() is the operand information call back function.
2587 // This is called to get the symbolic information for operand(s) of an
2588 // instruction when it is being done.  This routine does this from
2589 // the relocation information, symbol table, etc. That block of information
2590 // is a pointer to the struct DisassembleInfo that was passed when the
2591 // disassembler context was created and passed to back to here when
2592 // called back by the disassembler for instruction operands that could have
2593 // relocation information. The address of the instruction containing operand is
2594 // at the Pc parameter.  The immediate value the operand has is passed in
2595 // op_info->Value and is at Offset past the start of the instruction and has a
2596 // byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the
2597 // LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol
2598 // names and addends of the symbolic expression to add for the operand.  The
2599 // value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic
2600 // 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)2601 static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset,
2602                                uint64_t Size, int TagType, void *TagBuf) {
2603   struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
2604   struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf;
2605   uint64_t value = op_info->Value;
2606 
2607   // Make sure all fields returned are zero if we don't set them.
2608   memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1));
2609   op_info->Value = value;
2610 
2611   // If the TagType is not the value 1 which it code knows about or if no
2612   // verbose symbolic information is wanted then just return 0, indicating no
2613   // information is being returned.
2614   if (TagType != 1 || !info->verbose)
2615     return 0;
2616 
2617   unsigned int Arch = info->O->getArch();
2618   if (Arch == Triple::x86) {
2619     if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
2620       return 0;
2621     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2622       // TODO:
2623       // Search the external relocation entries of a fully linked image
2624       // (if any) for an entry that matches this segment offset.
2625       // uint32_t seg_offset = (Pc + Offset);
2626       return 0;
2627     }
2628     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2629     // for an entry for this section offset.
2630     uint32_t sect_addr = info->S.getAddress();
2631     uint32_t sect_offset = (Pc + Offset) - sect_addr;
2632     bool reloc_found = false;
2633     DataRefImpl Rel;
2634     MachO::any_relocation_info RE;
2635     bool isExtern = false;
2636     SymbolRef Symbol;
2637     bool r_scattered = false;
2638     uint32_t r_value, pair_r_value, r_type;
2639     for (const RelocationRef &Reloc : info->S.relocations()) {
2640       uint64_t RelocOffset = Reloc.getOffset();
2641       if (RelocOffset == sect_offset) {
2642         Rel = Reloc.getRawDataRefImpl();
2643         RE = info->O->getRelocation(Rel);
2644         r_type = info->O->getAnyRelocationType(RE);
2645         r_scattered = info->O->isRelocationScattered(RE);
2646         if (r_scattered) {
2647           r_value = info->O->getScatteredRelocationValue(RE);
2648           if (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
2649               r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) {
2650             DataRefImpl RelNext = Rel;
2651             info->O->moveRelocationNext(RelNext);
2652             MachO::any_relocation_info RENext;
2653             RENext = info->O->getRelocation(RelNext);
2654             if (info->O->isRelocationScattered(RENext))
2655               pair_r_value = info->O->getScatteredRelocationValue(RENext);
2656             else
2657               return 0;
2658           }
2659         } else {
2660           isExtern = info->O->getPlainRelocationExternal(RE);
2661           if (isExtern) {
2662             symbol_iterator RelocSym = Reloc.getSymbol();
2663             Symbol = *RelocSym;
2664           }
2665         }
2666         reloc_found = true;
2667         break;
2668       }
2669     }
2670     if (reloc_found && isExtern) {
2671       op_info->AddSymbol.Present = 1;
2672       op_info->AddSymbol.Name =
2673           unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2674       // For i386 extern relocation entries the value in the instruction is
2675       // the offset from the symbol, and value is already set in op_info->Value.
2676       return 1;
2677     }
2678     if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
2679                         r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) {
2680       const char *add = GuessSymbolName(r_value, info->AddrMap);
2681       const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
2682       uint32_t offset = value - (r_value - pair_r_value);
2683       op_info->AddSymbol.Present = 1;
2684       if (add != nullptr)
2685         op_info->AddSymbol.Name = add;
2686       else
2687         op_info->AddSymbol.Value = r_value;
2688       op_info->SubtractSymbol.Present = 1;
2689       if (sub != nullptr)
2690         op_info->SubtractSymbol.Name = sub;
2691       else
2692         op_info->SubtractSymbol.Value = pair_r_value;
2693       op_info->Value = offset;
2694       return 1;
2695     }
2696     return 0;
2697   }
2698   if (Arch == Triple::x86_64) {
2699     if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
2700       return 0;
2701     // For non MH_OBJECT types, like MH_KEXT_BUNDLE, Search the external
2702     // relocation entries of a linked image (if any) for an entry that matches
2703     // this segment offset.
2704     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2705       uint64_t seg_offset = Pc + Offset;
2706       bool reloc_found = false;
2707       DataRefImpl Rel;
2708       MachO::any_relocation_info RE;
2709       bool isExtern = false;
2710       SymbolRef Symbol;
2711       for (const RelocationRef &Reloc : info->O->external_relocations()) {
2712         uint64_t RelocOffset = Reloc.getOffset();
2713         if (RelocOffset == seg_offset) {
2714           Rel = Reloc.getRawDataRefImpl();
2715           RE = info->O->getRelocation(Rel);
2716           // external relocation entries should always be external.
2717           isExtern = info->O->getPlainRelocationExternal(RE);
2718           if (isExtern) {
2719             symbol_iterator RelocSym = Reloc.getSymbol();
2720             Symbol = *RelocSym;
2721           }
2722           reloc_found = true;
2723           break;
2724         }
2725       }
2726       if (reloc_found && isExtern) {
2727         // The Value passed in will be adjusted by the Pc if the instruction
2728         // adds the Pc.  But for x86_64 external relocation entries the Value
2729         // is the offset from the external symbol.
2730         if (info->O->getAnyRelocationPCRel(RE))
2731           op_info->Value -= Pc + Offset + Size;
2732         const char *name =
2733             unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2734         op_info->AddSymbol.Present = 1;
2735         op_info->AddSymbol.Name = name;
2736         return 1;
2737       }
2738       return 0;
2739     }
2740     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2741     // for an entry for this section offset.
2742     uint64_t sect_addr = info->S.getAddress();
2743     uint64_t sect_offset = (Pc + Offset) - sect_addr;
2744     bool reloc_found = false;
2745     DataRefImpl Rel;
2746     MachO::any_relocation_info RE;
2747     bool isExtern = false;
2748     SymbolRef Symbol;
2749     for (const RelocationRef &Reloc : info->S.relocations()) {
2750       uint64_t RelocOffset = Reloc.getOffset();
2751       if (RelocOffset == sect_offset) {
2752         Rel = Reloc.getRawDataRefImpl();
2753         RE = info->O->getRelocation(Rel);
2754         // NOTE: Scattered relocations don't exist on x86_64.
2755         isExtern = info->O->getPlainRelocationExternal(RE);
2756         if (isExtern) {
2757           symbol_iterator RelocSym = Reloc.getSymbol();
2758           Symbol = *RelocSym;
2759         }
2760         reloc_found = true;
2761         break;
2762       }
2763     }
2764     if (reloc_found && isExtern) {
2765       // The Value passed in will be adjusted by the Pc if the instruction
2766       // adds the Pc.  But for x86_64 external relocation entries the Value
2767       // is the offset from the external symbol.
2768       if (info->O->getAnyRelocationPCRel(RE))
2769         op_info->Value -= Pc + Offset + Size;
2770       const char *name =
2771           unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2772       unsigned Type = info->O->getAnyRelocationType(RE);
2773       if (Type == MachO::X86_64_RELOC_SUBTRACTOR) {
2774         DataRefImpl RelNext = Rel;
2775         info->O->moveRelocationNext(RelNext);
2776         MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2777         unsigned TypeNext = info->O->getAnyRelocationType(RENext);
2778         bool isExternNext = info->O->getPlainRelocationExternal(RENext);
2779         unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext);
2780         if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) {
2781           op_info->SubtractSymbol.Present = 1;
2782           op_info->SubtractSymbol.Name = name;
2783           symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum);
2784           Symbol = *RelocSymNext;
2785           name = unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2786         }
2787       }
2788       // TODO: add the VariantKinds to op_info->VariantKind for relocation types
2789       // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT.
2790       op_info->AddSymbol.Present = 1;
2791       op_info->AddSymbol.Name = name;
2792       return 1;
2793     }
2794     return 0;
2795   }
2796   if (Arch == Triple::arm) {
2797     if (Offset != 0 || (Size != 4 && Size != 2))
2798       return 0;
2799     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2800       // TODO:
2801       // Search the external relocation entries of a fully linked image
2802       // (if any) for an entry that matches this segment offset.
2803       // uint32_t seg_offset = (Pc + Offset);
2804       return 0;
2805     }
2806     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2807     // for an entry for this section offset.
2808     uint32_t sect_addr = info->S.getAddress();
2809     uint32_t sect_offset = (Pc + Offset) - sect_addr;
2810     DataRefImpl Rel;
2811     MachO::any_relocation_info RE;
2812     bool isExtern = false;
2813     SymbolRef Symbol;
2814     bool r_scattered = false;
2815     uint32_t r_value, pair_r_value, r_type, r_length, other_half;
2816     auto Reloc =
2817         find_if(info->S.relocations(), [&](const RelocationRef &Reloc) {
2818           uint64_t RelocOffset = Reloc.getOffset();
2819           return RelocOffset == sect_offset;
2820         });
2821 
2822     if (Reloc == info->S.relocations().end())
2823       return 0;
2824 
2825     Rel = Reloc->getRawDataRefImpl();
2826     RE = info->O->getRelocation(Rel);
2827     r_length = info->O->getAnyRelocationLength(RE);
2828     r_scattered = info->O->isRelocationScattered(RE);
2829     if (r_scattered) {
2830       r_value = info->O->getScatteredRelocationValue(RE);
2831       r_type = info->O->getScatteredRelocationType(RE);
2832     } else {
2833       r_type = info->O->getAnyRelocationType(RE);
2834       isExtern = info->O->getPlainRelocationExternal(RE);
2835       if (isExtern) {
2836         symbol_iterator RelocSym = Reloc->getSymbol();
2837         Symbol = *RelocSym;
2838       }
2839     }
2840     if (r_type == MachO::ARM_RELOC_HALF ||
2841         r_type == MachO::ARM_RELOC_SECTDIFF ||
2842         r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
2843         r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2844       DataRefImpl RelNext = Rel;
2845       info->O->moveRelocationNext(RelNext);
2846       MachO::any_relocation_info RENext;
2847       RENext = info->O->getRelocation(RelNext);
2848       other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff;
2849       if (info->O->isRelocationScattered(RENext))
2850         pair_r_value = info->O->getScatteredRelocationValue(RENext);
2851     }
2852 
2853     if (isExtern) {
2854       const char *name =
2855           unwrapOrError(Symbol.getName(), info->O->getFileName()).data();
2856       op_info->AddSymbol.Present = 1;
2857       op_info->AddSymbol.Name = name;
2858       switch (r_type) {
2859       case MachO::ARM_RELOC_HALF:
2860         if ((r_length & 0x1) == 1) {
2861           op_info->Value = value << 16 | other_half;
2862           op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2863         } else {
2864           op_info->Value = other_half << 16 | value;
2865           op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2866         }
2867         break;
2868       default:
2869         break;
2870       }
2871       return 1;
2872     }
2873     // If we have a branch that is not an external relocation entry then
2874     // return 0 so the code in tryAddingSymbolicOperand() can use the
2875     // SymbolLookUp call back with the branch target address to look up the
2876     // symbol and possibility add an annotation for a symbol stub.
2877     if (isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 ||
2878                           r_type == MachO::ARM_THUMB_RELOC_BR22))
2879       return 0;
2880 
2881     uint32_t offset = 0;
2882     if (r_type == MachO::ARM_RELOC_HALF ||
2883         r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2884       if ((r_length & 0x1) == 1)
2885         value = value << 16 | other_half;
2886       else
2887         value = other_half << 16 | value;
2888     }
2889     if (r_scattered && (r_type != MachO::ARM_RELOC_HALF &&
2890                         r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) {
2891       offset = value - r_value;
2892       value = r_value;
2893     }
2894 
2895     if (r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
2896       if ((r_length & 0x1) == 1)
2897         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2898       else
2899         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2900       const char *add = GuessSymbolName(r_value, info->AddrMap);
2901       const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
2902       int32_t offset = value - (r_value - pair_r_value);
2903       op_info->AddSymbol.Present = 1;
2904       if (add != nullptr)
2905         op_info->AddSymbol.Name = add;
2906       else
2907         op_info->AddSymbol.Value = r_value;
2908       op_info->SubtractSymbol.Present = 1;
2909       if (sub != nullptr)
2910         op_info->SubtractSymbol.Name = sub;
2911       else
2912         op_info->SubtractSymbol.Value = pair_r_value;
2913       op_info->Value = offset;
2914       return 1;
2915     }
2916 
2917     op_info->AddSymbol.Present = 1;
2918     op_info->Value = offset;
2919     if (r_type == MachO::ARM_RELOC_HALF) {
2920       if ((r_length & 0x1) == 1)
2921         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2922       else
2923         op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2924     }
2925     const char *add = GuessSymbolName(value, info->AddrMap);
2926     if (add != nullptr) {
2927       op_info->AddSymbol.Name = add;
2928       return 1;
2929     }
2930     op_info->AddSymbol.Value = value;
2931     return 1;
2932   }
2933   if (Arch == Triple::aarch64) {
2934     if (Offset != 0 || Size != 4)
2935       return 0;
2936     if (info->O->getHeader().filetype != MachO::MH_OBJECT) {
2937       // TODO:
2938       // Search the external relocation entries of a fully linked image
2939       // (if any) for an entry that matches this segment offset.
2940       // uint64_t seg_offset = (Pc + Offset);
2941       return 0;
2942     }
2943     // In MH_OBJECT filetypes search the section's relocation entries (if any)
2944     // for an entry for this section offset.
2945     uint64_t sect_addr = info->S.getAddress();
2946     uint64_t sect_offset = (Pc + Offset) - sect_addr;
2947     auto Reloc =
2948         find_if(info->S.relocations(), [&](const RelocationRef &Reloc) {
2949           uint64_t RelocOffset = Reloc.getOffset();
2950           return RelocOffset == sect_offset;
2951         });
2952 
2953     if (Reloc == info->S.relocations().end())
2954       return 0;
2955 
2956     DataRefImpl Rel = Reloc->getRawDataRefImpl();
2957     MachO::any_relocation_info RE = info->O->getRelocation(Rel);
2958     uint32_t r_type = info->O->getAnyRelocationType(RE);
2959     if (r_type == MachO::ARM64_RELOC_ADDEND) {
2960       DataRefImpl RelNext = Rel;
2961       info->O->moveRelocationNext(RelNext);
2962       MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2963       if (value == 0) {
2964         value = info->O->getPlainRelocationSymbolNum(RENext);
2965         op_info->Value = value;
2966       }
2967     }
2968     // NOTE: Scattered relocations don't exist on arm64.
2969     if (!info->O->getPlainRelocationExternal(RE))
2970       return 0;
2971     const char *name =
2972         unwrapOrError(Reloc->getSymbol()->getName(), info->O->getFileName())
2973             .data();
2974     op_info->AddSymbol.Present = 1;
2975     op_info->AddSymbol.Name = name;
2976 
2977     switch (r_type) {
2978     case MachO::ARM64_RELOC_PAGE21:
2979       /* @page */
2980       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE;
2981       break;
2982     case MachO::ARM64_RELOC_PAGEOFF12:
2983       /* @pageoff */
2984       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF;
2985       break;
2986     case MachO::ARM64_RELOC_GOT_LOAD_PAGE21:
2987       /* @gotpage */
2988       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE;
2989       break;
2990     case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12:
2991       /* @gotpageoff */
2992       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF;
2993       break;
2994     case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21:
2995       /* @tvlppage is not implemented in llvm-mc */
2996       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP;
2997       break;
2998     case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12:
2999       /* @tvlppageoff is not implemented in llvm-mc */
3000       op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF;
3001       break;
3002     default:
3003     case MachO::ARM64_RELOC_BRANCH26:
3004       op_info->VariantKind = LLVMDisassembler_VariantKind_None;
3005       break;
3006     }
3007     return 1;
3008   }
3009   return 0;
3010 }
3011 
3012 // GuessCstringPointer is passed the address of what might be a pointer to a
3013 // literal string in a cstring section.  If that address is in a cstring section
3014 // it returns a pointer to that string.  Else it returns nullptr.
GuessCstringPointer(uint64_t ReferenceValue,struct DisassembleInfo * info)3015 static const char *GuessCstringPointer(uint64_t ReferenceValue,
3016                                        struct DisassembleInfo *info) {
3017   for (const auto &Load : info->O->load_commands()) {
3018     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3019       MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3020       for (unsigned J = 0; J < Seg.nsects; ++J) {
3021         MachO::section_64 Sec = info->O->getSection64(Load, J);
3022         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3023         if (section_type == MachO::S_CSTRING_LITERALS &&
3024             ReferenceValue >= Sec.addr &&
3025             ReferenceValue < Sec.addr + Sec.size) {
3026           uint64_t sect_offset = ReferenceValue - Sec.addr;
3027           uint64_t object_offset = Sec.offset + sect_offset;
3028           StringRef MachOContents = info->O->getData();
3029           uint64_t object_size = MachOContents.size();
3030           const char *object_addr = (const char *)MachOContents.data();
3031           if (object_offset < object_size) {
3032             const char *name = object_addr + object_offset;
3033             return name;
3034           } else {
3035             return nullptr;
3036           }
3037         }
3038       }
3039     } else if (Load.C.cmd == MachO::LC_SEGMENT) {
3040       MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
3041       for (unsigned J = 0; J < Seg.nsects; ++J) {
3042         MachO::section Sec = info->O->getSection(Load, J);
3043         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3044         if (section_type == MachO::S_CSTRING_LITERALS &&
3045             ReferenceValue >= Sec.addr &&
3046             ReferenceValue < Sec.addr + Sec.size) {
3047           uint64_t sect_offset = ReferenceValue - Sec.addr;
3048           uint64_t object_offset = Sec.offset + sect_offset;
3049           StringRef MachOContents = info->O->getData();
3050           uint64_t object_size = MachOContents.size();
3051           const char *object_addr = (const char *)MachOContents.data();
3052           if (object_offset < object_size) {
3053             const char *name = object_addr + object_offset;
3054             return name;
3055           } else {
3056             return nullptr;
3057           }
3058         }
3059       }
3060     }
3061   }
3062   return nullptr;
3063 }
3064 
3065 // GuessIndirectSymbol returns the name of the indirect symbol for the
3066 // ReferenceValue passed in or nullptr.  This is used when ReferenceValue maybe
3067 // an address of a symbol stub or a lazy or non-lazy pointer to associate the
3068 // symbol name being referenced by the stub or pointer.
GuessIndirectSymbol(uint64_t ReferenceValue,struct DisassembleInfo * info)3069 static const char *GuessIndirectSymbol(uint64_t ReferenceValue,
3070                                        struct DisassembleInfo *info) {
3071   MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand();
3072   MachO::symtab_command Symtab = info->O->getSymtabLoadCommand();
3073   for (const auto &Load : info->O->load_commands()) {
3074     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3075       MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3076       for (unsigned J = 0; J < Seg.nsects; ++J) {
3077         MachO::section_64 Sec = info->O->getSection64(Load, J);
3078         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3079         if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3080              section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3081              section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3082              section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
3083              section_type == MachO::S_SYMBOL_STUBS) &&
3084             ReferenceValue >= Sec.addr &&
3085             ReferenceValue < Sec.addr + Sec.size) {
3086           uint32_t stride;
3087           if (section_type == MachO::S_SYMBOL_STUBS)
3088             stride = Sec.reserved2;
3089           else
3090             stride = 8;
3091           if (stride == 0)
3092             return nullptr;
3093           uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
3094           if (index < Dysymtab.nindirectsyms) {
3095             uint32_t indirect_symbol =
3096                 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
3097             if (indirect_symbol < Symtab.nsyms) {
3098               symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
3099               return unwrapOrError(Sym->getName(), info->O->getFileName())
3100                   .data();
3101             }
3102           }
3103         }
3104       }
3105     } else if (Load.C.cmd == MachO::LC_SEGMENT) {
3106       MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
3107       for (unsigned J = 0; J < Seg.nsects; ++J) {
3108         MachO::section Sec = info->O->getSection(Load, J);
3109         uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
3110         if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3111              section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3112              section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3113              section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
3114              section_type == MachO::S_SYMBOL_STUBS) &&
3115             ReferenceValue >= Sec.addr &&
3116             ReferenceValue < Sec.addr + Sec.size) {
3117           uint32_t stride;
3118           if (section_type == MachO::S_SYMBOL_STUBS)
3119             stride = Sec.reserved2;
3120           else
3121             stride = 4;
3122           if (stride == 0)
3123             return nullptr;
3124           uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
3125           if (index < Dysymtab.nindirectsyms) {
3126             uint32_t indirect_symbol =
3127                 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
3128             if (indirect_symbol < Symtab.nsyms) {
3129               symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
3130               return unwrapOrError(Sym->getName(), info->O->getFileName())
3131                   .data();
3132             }
3133           }
3134         }
3135       }
3136     }
3137   }
3138   return nullptr;
3139 }
3140 
3141 // method_reference() is called passing it the ReferenceName that might be
3142 // a reference it to an Objective-C method call.  If so then it allocates and
3143 // assembles a method call string with the values last seen and saved in
3144 // the DisassembleInfo's class_name and selector_name fields.  This is saved
3145 // into the method field of the info and any previous string is free'ed.
3146 // Then the class_name field in the info is set to nullptr.  The method call
3147 // string is set into ReferenceName and ReferenceType is set to
3148 // LLVMDisassembler_ReferenceType_Out_Objc_Message.  If this not a method call
3149 // then both ReferenceType and ReferenceName are left unchanged.
method_reference(struct DisassembleInfo * info,uint64_t * ReferenceType,const char ** ReferenceName)3150 static void method_reference(struct DisassembleInfo *info,
3151                              uint64_t *ReferenceType,
3152                              const char **ReferenceName) {
3153   unsigned int Arch = info->O->getArch();
3154   if (*ReferenceName != nullptr) {
3155     if (strcmp(*ReferenceName, "_objc_msgSend") == 0) {
3156       if (info->selector_name != nullptr) {
3157         if (info->class_name != nullptr) {
3158           info->method = std::make_unique<char[]>(
3159               5 + strlen(info->class_name) + strlen(info->selector_name));
3160           char *method = info->method.get();
3161           if (method != nullptr) {
3162             strcpy(method, "+[");
3163             strcat(method, info->class_name);
3164             strcat(method, " ");
3165             strcat(method, info->selector_name);
3166             strcat(method, "]");
3167             *ReferenceName = method;
3168             *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3169           }
3170         } else {
3171           info->method =
3172               std::make_unique<char[]>(9 + strlen(info->selector_name));
3173           char *method = info->method.get();
3174           if (method != nullptr) {
3175             if (Arch == Triple::x86_64)
3176               strcpy(method, "-[%rdi ");
3177             else if (Arch == Triple::aarch64)
3178               strcpy(method, "-[x0 ");
3179             else
3180               strcpy(method, "-[r? ");
3181             strcat(method, info->selector_name);
3182             strcat(method, "]");
3183             *ReferenceName = method;
3184             *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3185           }
3186         }
3187         info->class_name = nullptr;
3188       }
3189     } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) {
3190       if (info->selector_name != nullptr) {
3191         info->method =
3192             std::make_unique<char[]>(17 + strlen(info->selector_name));
3193         char *method = info->method.get();
3194         if (method != nullptr) {
3195           if (Arch == Triple::x86_64)
3196             strcpy(method, "-[[%rdi super] ");
3197           else if (Arch == Triple::aarch64)
3198             strcpy(method, "-[[x0 super] ");
3199           else
3200             strcpy(method, "-[[r? super] ");
3201           strcat(method, info->selector_name);
3202           strcat(method, "]");
3203           *ReferenceName = method;
3204           *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
3205         }
3206         info->class_name = nullptr;
3207       }
3208     }
3209   }
3210 }
3211 
3212 // GuessPointerPointer() is passed the address of what might be a pointer to
3213 // a reference to an Objective-C class, selector, message ref or cfstring.
3214 // If so the value of the pointer is returned and one of the booleans are set
3215 // 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)3216 static uint64_t GuessPointerPointer(uint64_t ReferenceValue,
3217                                     struct DisassembleInfo *info,
3218                                     bool &classref, bool &selref, bool &msgref,
3219                                     bool &cfstring) {
3220   classref = false;
3221   selref = false;
3222   msgref = false;
3223   cfstring = false;
3224   for (const auto &Load : info->O->load_commands()) {
3225     if (Load.C.cmd == MachO::LC_SEGMENT_64) {
3226       MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
3227       for (unsigned J = 0; J < Seg.nsects; ++J) {
3228         MachO::section_64 Sec = info->O->getSection64(Load, J);
3229         if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 ||
3230              strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
3231              strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 ||
3232              strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 ||
3233              strncmp(Sec.sectname, "__cfstring", 16) == 0) &&
3234             ReferenceValue >= Sec.addr &&
3235             ReferenceValue < Sec.addr + Sec.size) {
3236           uint64_t sect_offset = ReferenceValue - Sec.addr;
3237           uint64_t object_offset = Sec.offset + sect_offset;
3238           StringRef MachOContents = info->O->getData();
3239           uint64_t object_size = MachOContents.size();
3240           const char *object_addr = (const char *)MachOContents.data();
3241           if (object_offset < object_size) {
3242             uint64_t pointer_value;
3243             memcpy(&pointer_value, object_addr + object_offset,
3244                    sizeof(uint64_t));
3245             if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
3246               sys::swapByteOrder(pointer_value);
3247             if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0)
3248               selref = true;
3249             else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
3250                      strncmp(Sec.sectname, "__objc_superrefs", 16) == 0)
3251               classref = true;
3252             else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 &&
3253                      ReferenceValue + 8 < Sec.addr + Sec.size) {
3254               msgref = true;
3255               memcpy(&pointer_value, object_addr + object_offset + 8,
3256                      sizeof(uint64_t));
3257               if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
3258                 sys::swapByteOrder(pointer_value);
3259             } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0)
3260               cfstring = true;
3261             return pointer_value;
3262           } else {
3263             return 0;
3264           }
3265         }
3266       }
3267     }
3268     // TODO: Look for LC_SEGMENT for 32-bit Mach-O files.
3269   }
3270   return 0;
3271 }
3272 
3273 // get_pointer_64 returns a pointer to the bytes in the object file at the
3274 // Address from a section in the Mach-O file.  And indirectly returns the
3275 // offset into the section, number of bytes left in the section past the offset
3276 // and which section is was being referenced.  If the Address is not in a
3277 // section nullptr is returned.
get_pointer_64(uint64_t Address,uint32_t & offset,uint32_t & left,SectionRef & S,DisassembleInfo * info,bool objc_only=false)3278 static const char *get_pointer_64(uint64_t Address, uint32_t &offset,
3279                                   uint32_t &left, SectionRef &S,
3280                                   DisassembleInfo *info,
3281                                   bool objc_only = false) {
3282   offset = 0;
3283   left = 0;
3284   S = SectionRef();
3285   for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) {
3286     uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress();
3287     uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize();
3288     if (SectSize == 0)
3289       continue;
3290     if (objc_only) {
3291       StringRef SectName;
3292       Expected<StringRef> SecNameOrErr =
3293           ((*(info->Sections))[SectIdx]).getName();
3294       if (SecNameOrErr)
3295         SectName = *SecNameOrErr;
3296       else
3297         consumeError(SecNameOrErr.takeError());
3298 
3299       DataRefImpl Ref = ((*(info->Sections))[SectIdx]).getRawDataRefImpl();
3300       StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
3301       if (SegName != "__OBJC" && SectName != "__cstring")
3302         continue;
3303     }
3304     if (Address >= SectAddress && Address < SectAddress + SectSize) {
3305       S = (*(info->Sections))[SectIdx];
3306       offset = Address - SectAddress;
3307       left = SectSize - offset;
3308       StringRef SectContents = unwrapOrError(
3309           ((*(info->Sections))[SectIdx]).getContents(), info->O->getFileName());
3310       return SectContents.data() + offset;
3311     }
3312   }
3313   return nullptr;
3314 }
3315 
get_pointer_32(uint32_t Address,uint32_t & offset,uint32_t & left,SectionRef & S,DisassembleInfo * info,bool objc_only=false)3316 static const char *get_pointer_32(uint32_t Address, uint32_t &offset,
3317                                   uint32_t &left, SectionRef &S,
3318                                   DisassembleInfo *info,
3319                                   bool objc_only = false) {
3320   return get_pointer_64(Address, offset, left, S, info, objc_only);
3321 }
3322 
3323 // get_symbol_64() returns the name of a symbol (or nullptr) and the address of
3324 // the symbol indirectly through n_value. Based on the relocation information
3325 // for the specified section offset in the specified section reference.
3326 // If no relocation information is found and a non-zero ReferenceValue for the
3327 // 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)3328 static const char *get_symbol_64(uint32_t sect_offset, SectionRef S,
3329                                  DisassembleInfo *info, uint64_t &n_value,
3330                                  uint64_t ReferenceValue = 0) {
3331   n_value = 0;
3332   if (!info->verbose)
3333     return nullptr;
3334 
3335   // See if there is an external relocation entry at the sect_offset.
3336   bool reloc_found = false;
3337   DataRefImpl Rel;
3338   MachO::any_relocation_info RE;
3339   bool isExtern = false;
3340   SymbolRef Symbol;
3341   for (const RelocationRef &Reloc : S.relocations()) {
3342     uint64_t RelocOffset = Reloc.getOffset();
3343     if (RelocOffset == sect_offset) {
3344       Rel = Reloc.getRawDataRefImpl();
3345       RE = info->O->getRelocation(Rel);
3346       if (info->O->isRelocationScattered(RE))
3347         continue;
3348       isExtern = info->O->getPlainRelocationExternal(RE);
3349       if (isExtern) {
3350         symbol_iterator RelocSym = Reloc.getSymbol();
3351         Symbol = *RelocSym;
3352       }
3353       reloc_found = true;
3354       break;
3355     }
3356   }
3357   // If there is an external relocation entry for a symbol in this section
3358   // at this section_offset then use that symbol's value for the n_value
3359   // and return its name.
3360   const char *SymbolName = nullptr;
3361   if (reloc_found && isExtern) {
3362     n_value = cantFail(Symbol.getValue());
3363     StringRef Name = unwrapOrError(Symbol.getName(), info->O->getFileName());
3364     if (!Name.empty()) {
3365       SymbolName = Name.data();
3366       return SymbolName;
3367     }
3368   }
3369 
3370   // TODO: For fully linked images, look through the external relocation
3371   // entries off the dynamic symtab command. For these the r_offset is from the
3372   // start of the first writeable segment in the Mach-O file.  So the offset
3373   // to this section from that segment is passed to this routine by the caller,
3374   // as the database_offset. Which is the difference of the section's starting
3375   // address and the first writable segment.
3376   //
3377   // NOTE: need add passing the database_offset to this routine.
3378 
3379   // We did not find an external relocation entry so look up the ReferenceValue
3380   // as an address of a symbol and if found return that symbol's name.
3381   SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
3382 
3383   return SymbolName;
3384 }
3385 
get_symbol_32(uint32_t sect_offset,SectionRef S,DisassembleInfo * info,uint32_t ReferenceValue)3386 static const char *get_symbol_32(uint32_t sect_offset, SectionRef S,
3387                                  DisassembleInfo *info,
3388                                  uint32_t ReferenceValue) {
3389   uint64_t n_value64;
3390   return get_symbol_64(sect_offset, S, info, n_value64, ReferenceValue);
3391 }
3392 
3393 namespace {
3394 
3395 // These are structs in the Objective-C meta data and read to produce the
3396 // comments for disassembly.  While these are part of the ABI they are no
3397 // public defintions.  So the are here not in include/llvm/BinaryFormat/MachO.h
3398 // .
3399 
3400 // The cfstring object in a 64-bit Mach-O file.
3401 struct cfstring64_t {
3402   uint64_t isa;        // class64_t * (64-bit pointer)
3403   uint64_t flags;      // flag bits
3404   uint64_t characters; // char * (64-bit pointer)
3405   uint64_t length;     // number of non-NULL characters in above
3406 };
3407 
3408 // The class object in a 64-bit Mach-O file.
3409 struct class64_t {
3410   uint64_t isa;        // class64_t * (64-bit pointer)
3411   uint64_t superclass; // class64_t * (64-bit pointer)
3412   uint64_t cache;      // Cache (64-bit pointer)
3413   uint64_t vtable;     // IMP * (64-bit pointer)
3414   uint64_t data;       // class_ro64_t * (64-bit pointer)
3415 };
3416 
3417 struct class32_t {
3418   uint32_t isa;        /* class32_t * (32-bit pointer) */
3419   uint32_t superclass; /* class32_t * (32-bit pointer) */
3420   uint32_t cache;      /* Cache (32-bit pointer) */
3421   uint32_t vtable;     /* IMP * (32-bit pointer) */
3422   uint32_t data;       /* class_ro32_t * (32-bit pointer) */
3423 };
3424 
3425 struct class_ro64_t {
3426   uint32_t flags;
3427   uint32_t instanceStart;
3428   uint32_t instanceSize;
3429   uint32_t reserved;
3430   uint64_t ivarLayout;     // const uint8_t * (64-bit pointer)
3431   uint64_t name;           // const char * (64-bit pointer)
3432   uint64_t baseMethods;    // const method_list_t * (64-bit pointer)
3433   uint64_t baseProtocols;  // const protocol_list_t * (64-bit pointer)
3434   uint64_t ivars;          // const ivar_list_t * (64-bit pointer)
3435   uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer)
3436   uint64_t baseProperties; // const struct objc_property_list (64-bit pointer)
3437 };
3438 
3439 struct class_ro32_t {
3440   uint32_t flags;
3441   uint32_t instanceStart;
3442   uint32_t instanceSize;
3443   uint32_t ivarLayout;     /* const uint8_t * (32-bit pointer) */
3444   uint32_t name;           /* const char * (32-bit pointer) */
3445   uint32_t baseMethods;    /* const method_list_t * (32-bit pointer) */
3446   uint32_t baseProtocols;  /* const protocol_list_t * (32-bit pointer) */
3447   uint32_t ivars;          /* const ivar_list_t * (32-bit pointer) */
3448   uint32_t weakIvarLayout; /* const uint8_t * (32-bit pointer) */
3449   uint32_t baseProperties; /* const struct objc_property_list *
3450                                                    (32-bit pointer) */
3451 };
3452 
3453 /* Values for class_ro{64,32}_t->flags */
3454 #define RO_META (1 << 0)
3455 #define RO_ROOT (1 << 1)
3456 #define RO_HAS_CXX_STRUCTORS (1 << 2)
3457 
3458 struct method_list64_t {
3459   uint32_t entsize;
3460   uint32_t count;
3461   /* struct method64_t first;  These structures follow inline */
3462 };
3463 
3464 struct method_list32_t {
3465   uint32_t entsize;
3466   uint32_t count;
3467   /* struct method32_t first;  These structures follow inline */
3468 };
3469 
3470 struct method64_t {
3471   uint64_t name;  /* SEL (64-bit pointer) */
3472   uint64_t types; /* const char * (64-bit pointer) */
3473   uint64_t imp;   /* IMP (64-bit pointer) */
3474 };
3475 
3476 struct method32_t {
3477   uint32_t name;  /* SEL (32-bit pointer) */
3478   uint32_t types; /* const char * (32-bit pointer) */
3479   uint32_t imp;   /* IMP (32-bit pointer) */
3480 };
3481 
3482 struct protocol_list64_t {
3483   uint64_t count; /* uintptr_t (a 64-bit value) */
3484   /* struct protocol64_t * list[0];  These pointers follow inline */
3485 };
3486 
3487 struct protocol_list32_t {
3488   uint32_t count; /* uintptr_t (a 32-bit value) */
3489   /* struct protocol32_t * list[0];  These pointers follow inline */
3490 };
3491 
3492 struct protocol64_t {
3493   uint64_t isa;                     /* id * (64-bit pointer) */
3494   uint64_t name;                    /* const char * (64-bit pointer) */
3495   uint64_t protocols;               /* struct protocol_list64_t *
3496                                                     (64-bit pointer) */
3497   uint64_t instanceMethods;         /* method_list_t * (64-bit pointer) */
3498   uint64_t classMethods;            /* method_list_t * (64-bit pointer) */
3499   uint64_t optionalInstanceMethods; /* method_list_t * (64-bit pointer) */
3500   uint64_t optionalClassMethods;    /* method_list_t * (64-bit pointer) */
3501   uint64_t instanceProperties;      /* struct objc_property_list *
3502                                                        (64-bit pointer) */
3503 };
3504 
3505 struct protocol32_t {
3506   uint32_t isa;                     /* id * (32-bit pointer) */
3507   uint32_t name;                    /* const char * (32-bit pointer) */
3508   uint32_t protocols;               /* struct protocol_list_t *
3509                                                     (32-bit pointer) */
3510   uint32_t instanceMethods;         /* method_list_t * (32-bit pointer) */
3511   uint32_t classMethods;            /* method_list_t * (32-bit pointer) */
3512   uint32_t optionalInstanceMethods; /* method_list_t * (32-bit pointer) */
3513   uint32_t optionalClassMethods;    /* method_list_t * (32-bit pointer) */
3514   uint32_t instanceProperties;      /* struct objc_property_list *
3515                                                        (32-bit pointer) */
3516 };
3517 
3518 struct ivar_list64_t {
3519   uint32_t entsize;
3520   uint32_t count;
3521   /* struct ivar64_t first;  These structures follow inline */
3522 };
3523 
3524 struct ivar_list32_t {
3525   uint32_t entsize;
3526   uint32_t count;
3527   /* struct ivar32_t first;  These structures follow inline */
3528 };
3529 
3530 struct ivar64_t {
3531   uint64_t offset; /* uintptr_t * (64-bit pointer) */
3532   uint64_t name;   /* const char * (64-bit pointer) */
3533   uint64_t type;   /* const char * (64-bit pointer) */
3534   uint32_t alignment;
3535   uint32_t size;
3536 };
3537 
3538 struct ivar32_t {
3539   uint32_t offset; /* uintptr_t * (32-bit pointer) */
3540   uint32_t name;   /* const char * (32-bit pointer) */
3541   uint32_t type;   /* const char * (32-bit pointer) */
3542   uint32_t alignment;
3543   uint32_t size;
3544 };
3545 
3546 struct objc_property_list64 {
3547   uint32_t entsize;
3548   uint32_t count;
3549   /* struct objc_property64 first;  These structures follow inline */
3550 };
3551 
3552 struct objc_property_list32 {
3553   uint32_t entsize;
3554   uint32_t count;
3555   /* struct objc_property32 first;  These structures follow inline */
3556 };
3557 
3558 struct objc_property64 {
3559   uint64_t name;       /* const char * (64-bit pointer) */
3560   uint64_t attributes; /* const char * (64-bit pointer) */
3561 };
3562 
3563 struct objc_property32 {
3564   uint32_t name;       /* const char * (32-bit pointer) */
3565   uint32_t attributes; /* const char * (32-bit pointer) */
3566 };
3567 
3568 struct category64_t {
3569   uint64_t name;               /* const char * (64-bit pointer) */
3570   uint64_t cls;                /* struct class_t * (64-bit pointer) */
3571   uint64_t instanceMethods;    /* struct method_list_t * (64-bit pointer) */
3572   uint64_t classMethods;       /* struct method_list_t * (64-bit pointer) */
3573   uint64_t protocols;          /* struct protocol_list_t * (64-bit pointer) */
3574   uint64_t instanceProperties; /* struct objc_property_list *
3575                                   (64-bit pointer) */
3576 };
3577 
3578 struct category32_t {
3579   uint32_t name;               /* const char * (32-bit pointer) */
3580   uint32_t cls;                /* struct class_t * (32-bit pointer) */
3581   uint32_t instanceMethods;    /* struct method_list_t * (32-bit pointer) */
3582   uint32_t classMethods;       /* struct method_list_t * (32-bit pointer) */
3583   uint32_t protocols;          /* struct protocol_list_t * (32-bit pointer) */
3584   uint32_t instanceProperties; /* struct objc_property_list *
3585                                   (32-bit pointer) */
3586 };
3587 
3588 struct objc_image_info64 {
3589   uint32_t version;
3590   uint32_t flags;
3591 };
3592 struct objc_image_info32 {
3593   uint32_t version;
3594   uint32_t flags;
3595 };
3596 struct imageInfo_t {
3597   uint32_t version;
3598   uint32_t flags;
3599 };
3600 /* masks for objc_image_info.flags */
3601 #define OBJC_IMAGE_IS_REPLACEMENT (1 << 0)
3602 #define OBJC_IMAGE_SUPPORTS_GC (1 << 1)
3603 #define OBJC_IMAGE_IS_SIMULATED (1 << 5)
3604 #define OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES (1 << 6)
3605 
3606 struct message_ref64 {
3607   uint64_t imp; /* IMP (64-bit pointer) */
3608   uint64_t sel; /* SEL (64-bit pointer) */
3609 };
3610 
3611 struct message_ref32 {
3612   uint32_t imp; /* IMP (32-bit pointer) */
3613   uint32_t sel; /* SEL (32-bit pointer) */
3614 };
3615 
3616 // Objective-C 1 (32-bit only) meta data structs.
3617 
3618 struct objc_module_t {
3619   uint32_t version;
3620   uint32_t size;
3621   uint32_t name;   /* char * (32-bit pointer) */
3622   uint32_t symtab; /* struct objc_symtab * (32-bit pointer) */
3623 };
3624 
3625 struct objc_symtab_t {
3626   uint32_t sel_ref_cnt;
3627   uint32_t refs; /* SEL * (32-bit pointer) */
3628   uint16_t cls_def_cnt;
3629   uint16_t cat_def_cnt;
3630   // uint32_t defs[1];        /* void * (32-bit pointer) variable size */
3631 };
3632 
3633 struct objc_class_t {
3634   uint32_t isa;         /* struct objc_class * (32-bit pointer) */
3635   uint32_t super_class; /* struct objc_class * (32-bit pointer) */
3636   uint32_t name;        /* const char * (32-bit pointer) */
3637   int32_t version;
3638   int32_t info;
3639   int32_t instance_size;
3640   uint32_t ivars;       /* struct objc_ivar_list * (32-bit pointer) */
3641   uint32_t methodLists; /* struct objc_method_list ** (32-bit pointer) */
3642   uint32_t cache;       /* struct objc_cache * (32-bit pointer) */
3643   uint32_t protocols;   /* struct objc_protocol_list * (32-bit pointer) */
3644 };
3645 
3646 #define CLS_GETINFO(cls, infomask) ((cls)->info & (infomask))
3647 // class is not a metaclass
3648 #define CLS_CLASS 0x1
3649 // class is a metaclass
3650 #define CLS_META 0x2
3651 
3652 struct objc_category_t {
3653   uint32_t category_name;    /* char * (32-bit pointer) */
3654   uint32_t class_name;       /* char * (32-bit pointer) */
3655   uint32_t instance_methods; /* struct objc_method_list * (32-bit pointer) */
3656   uint32_t class_methods;    /* struct objc_method_list * (32-bit pointer) */
3657   uint32_t protocols;        /* struct objc_protocol_list * (32-bit ptr) */
3658 };
3659 
3660 struct objc_ivar_t {
3661   uint32_t ivar_name; /* char * (32-bit pointer) */
3662   uint32_t ivar_type; /* char * (32-bit pointer) */
3663   int32_t ivar_offset;
3664 };
3665 
3666 struct objc_ivar_list_t {
3667   int32_t ivar_count;
3668   // struct objc_ivar_t ivar_list[1];          /* variable length structure */
3669 };
3670 
3671 struct objc_method_list_t {
3672   uint32_t obsolete; /* struct objc_method_list * (32-bit pointer) */
3673   int32_t method_count;
3674   // struct objc_method_t method_list[1];      /* variable length structure */
3675 };
3676 
3677 struct objc_method_t {
3678   uint32_t method_name;  /* SEL, aka struct objc_selector * (32-bit pointer) */
3679   uint32_t method_types; /* char * (32-bit pointer) */
3680   uint32_t method_imp;   /* IMP, aka function pointer, (*IMP)(id, SEL, ...)
3681                             (32-bit pointer) */
3682 };
3683 
3684 struct objc_protocol_list_t {
3685   uint32_t next; /* struct objc_protocol_list * (32-bit pointer) */
3686   int32_t count;
3687   // uint32_t list[1];   /* Protocol *, aka struct objc_protocol_t *
3688   //                        (32-bit pointer) */
3689 };
3690 
3691 struct objc_protocol_t {
3692   uint32_t isa;              /* struct objc_class * (32-bit pointer) */
3693   uint32_t protocol_name;    /* char * (32-bit pointer) */
3694   uint32_t protocol_list;    /* struct objc_protocol_list * (32-bit pointer) */
3695   uint32_t instance_methods; /* struct objc_method_description_list *
3696                                 (32-bit pointer) */
3697   uint32_t class_methods;    /* struct objc_method_description_list *
3698                                 (32-bit pointer) */
3699 };
3700 
3701 struct objc_method_description_list_t {
3702   int32_t count;
3703   // struct objc_method_description_t list[1];
3704 };
3705 
3706 struct objc_method_description_t {
3707   uint32_t name;  /* SEL, aka struct objc_selector * (32-bit pointer) */
3708   uint32_t types; /* char * (32-bit pointer) */
3709 };
3710 
swapStruct(struct cfstring64_t & cfs)3711 inline void swapStruct(struct cfstring64_t &cfs) {
3712   sys::swapByteOrder(cfs.isa);
3713   sys::swapByteOrder(cfs.flags);
3714   sys::swapByteOrder(cfs.characters);
3715   sys::swapByteOrder(cfs.length);
3716 }
3717 
swapStruct(struct class64_t & c)3718 inline void swapStruct(struct class64_t &c) {
3719   sys::swapByteOrder(c.isa);
3720   sys::swapByteOrder(c.superclass);
3721   sys::swapByteOrder(c.cache);
3722   sys::swapByteOrder(c.vtable);
3723   sys::swapByteOrder(c.data);
3724 }
3725 
swapStruct(struct class32_t & c)3726 inline void swapStruct(struct class32_t &c) {
3727   sys::swapByteOrder(c.isa);
3728   sys::swapByteOrder(c.superclass);
3729   sys::swapByteOrder(c.cache);
3730   sys::swapByteOrder(c.vtable);
3731   sys::swapByteOrder(c.data);
3732 }
3733 
swapStruct(struct class_ro64_t & cro)3734 inline void swapStruct(struct class_ro64_t &cro) {
3735   sys::swapByteOrder(cro.flags);
3736   sys::swapByteOrder(cro.instanceStart);
3737   sys::swapByteOrder(cro.instanceSize);
3738   sys::swapByteOrder(cro.reserved);
3739   sys::swapByteOrder(cro.ivarLayout);
3740   sys::swapByteOrder(cro.name);
3741   sys::swapByteOrder(cro.baseMethods);
3742   sys::swapByteOrder(cro.baseProtocols);
3743   sys::swapByteOrder(cro.ivars);
3744   sys::swapByteOrder(cro.weakIvarLayout);
3745   sys::swapByteOrder(cro.baseProperties);
3746 }
3747 
swapStruct(struct class_ro32_t & cro)3748 inline void swapStruct(struct class_ro32_t &cro) {
3749   sys::swapByteOrder(cro.flags);
3750   sys::swapByteOrder(cro.instanceStart);
3751   sys::swapByteOrder(cro.instanceSize);
3752   sys::swapByteOrder(cro.ivarLayout);
3753   sys::swapByteOrder(cro.name);
3754   sys::swapByteOrder(cro.baseMethods);
3755   sys::swapByteOrder(cro.baseProtocols);
3756   sys::swapByteOrder(cro.ivars);
3757   sys::swapByteOrder(cro.weakIvarLayout);
3758   sys::swapByteOrder(cro.baseProperties);
3759 }
3760 
swapStruct(struct method_list64_t & ml)3761 inline void swapStruct(struct method_list64_t &ml) {
3762   sys::swapByteOrder(ml.entsize);
3763   sys::swapByteOrder(ml.count);
3764 }
3765 
swapStruct(struct method_list32_t & ml)3766 inline void swapStruct(struct method_list32_t &ml) {
3767   sys::swapByteOrder(ml.entsize);
3768   sys::swapByteOrder(ml.count);
3769 }
3770 
swapStruct(struct method64_t & m)3771 inline void swapStruct(struct method64_t &m) {
3772   sys::swapByteOrder(m.name);
3773   sys::swapByteOrder(m.types);
3774   sys::swapByteOrder(m.imp);
3775 }
3776 
swapStruct(struct method32_t & m)3777 inline void swapStruct(struct method32_t &m) {
3778   sys::swapByteOrder(m.name);
3779   sys::swapByteOrder(m.types);
3780   sys::swapByteOrder(m.imp);
3781 }
3782 
swapStruct(struct protocol_list64_t & pl)3783 inline void swapStruct(struct protocol_list64_t &pl) {
3784   sys::swapByteOrder(pl.count);
3785 }
3786 
swapStruct(struct protocol_list32_t & pl)3787 inline void swapStruct(struct protocol_list32_t &pl) {
3788   sys::swapByteOrder(pl.count);
3789 }
3790 
swapStruct(struct protocol64_t & p)3791 inline void swapStruct(struct protocol64_t &p) {
3792   sys::swapByteOrder(p.isa);
3793   sys::swapByteOrder(p.name);
3794   sys::swapByteOrder(p.protocols);
3795   sys::swapByteOrder(p.instanceMethods);
3796   sys::swapByteOrder(p.classMethods);
3797   sys::swapByteOrder(p.optionalInstanceMethods);
3798   sys::swapByteOrder(p.optionalClassMethods);
3799   sys::swapByteOrder(p.instanceProperties);
3800 }
3801 
swapStruct(struct protocol32_t & p)3802 inline void swapStruct(struct protocol32_t &p) {
3803   sys::swapByteOrder(p.isa);
3804   sys::swapByteOrder(p.name);
3805   sys::swapByteOrder(p.protocols);
3806   sys::swapByteOrder(p.instanceMethods);
3807   sys::swapByteOrder(p.classMethods);
3808   sys::swapByteOrder(p.optionalInstanceMethods);
3809   sys::swapByteOrder(p.optionalClassMethods);
3810   sys::swapByteOrder(p.instanceProperties);
3811 }
3812 
swapStruct(struct ivar_list64_t & il)3813 inline void swapStruct(struct ivar_list64_t &il) {
3814   sys::swapByteOrder(il.entsize);
3815   sys::swapByteOrder(il.count);
3816 }
3817 
swapStruct(struct ivar_list32_t & il)3818 inline void swapStruct(struct ivar_list32_t &il) {
3819   sys::swapByteOrder(il.entsize);
3820   sys::swapByteOrder(il.count);
3821 }
3822 
swapStruct(struct ivar64_t & i)3823 inline void swapStruct(struct ivar64_t &i) {
3824   sys::swapByteOrder(i.offset);
3825   sys::swapByteOrder(i.name);
3826   sys::swapByteOrder(i.type);
3827   sys::swapByteOrder(i.alignment);
3828   sys::swapByteOrder(i.size);
3829 }
3830 
swapStruct(struct ivar32_t & i)3831 inline void swapStruct(struct ivar32_t &i) {
3832   sys::swapByteOrder(i.offset);
3833   sys::swapByteOrder(i.name);
3834   sys::swapByteOrder(i.type);
3835   sys::swapByteOrder(i.alignment);
3836   sys::swapByteOrder(i.size);
3837 }
3838 
swapStruct(struct objc_property_list64 & pl)3839 inline void swapStruct(struct objc_property_list64 &pl) {
3840   sys::swapByteOrder(pl.entsize);
3841   sys::swapByteOrder(pl.count);
3842 }
3843 
swapStruct(struct objc_property_list32 & pl)3844 inline void swapStruct(struct objc_property_list32 &pl) {
3845   sys::swapByteOrder(pl.entsize);
3846   sys::swapByteOrder(pl.count);
3847 }
3848 
swapStruct(struct objc_property64 & op)3849 inline void swapStruct(struct objc_property64 &op) {
3850   sys::swapByteOrder(op.name);
3851   sys::swapByteOrder(op.attributes);
3852 }
3853 
swapStruct(struct objc_property32 & op)3854 inline void swapStruct(struct objc_property32 &op) {
3855   sys::swapByteOrder(op.name);
3856   sys::swapByteOrder(op.attributes);
3857 }
3858 
swapStruct(struct category64_t & c)3859 inline void swapStruct(struct category64_t &c) {
3860   sys::swapByteOrder(c.name);
3861   sys::swapByteOrder(c.cls);
3862   sys::swapByteOrder(c.instanceMethods);
3863   sys::swapByteOrder(c.classMethods);
3864   sys::swapByteOrder(c.protocols);
3865   sys::swapByteOrder(c.instanceProperties);
3866 }
3867 
swapStruct(struct category32_t & c)3868 inline void swapStruct(struct category32_t &c) {
3869   sys::swapByteOrder(c.name);
3870   sys::swapByteOrder(c.cls);
3871   sys::swapByteOrder(c.instanceMethods);
3872   sys::swapByteOrder(c.classMethods);
3873   sys::swapByteOrder(c.protocols);
3874   sys::swapByteOrder(c.instanceProperties);
3875 }
3876 
swapStruct(struct objc_image_info64 & o)3877 inline void swapStruct(struct objc_image_info64 &o) {
3878   sys::swapByteOrder(o.version);
3879   sys::swapByteOrder(o.flags);
3880 }
3881 
swapStruct(struct objc_image_info32 & o)3882 inline void swapStruct(struct objc_image_info32 &o) {
3883   sys::swapByteOrder(o.version);
3884   sys::swapByteOrder(o.flags);
3885 }
3886 
swapStruct(struct imageInfo_t & o)3887 inline void swapStruct(struct imageInfo_t &o) {
3888   sys::swapByteOrder(o.version);
3889   sys::swapByteOrder(o.flags);
3890 }
3891 
swapStruct(struct message_ref64 & mr)3892 inline void swapStruct(struct message_ref64 &mr) {
3893   sys::swapByteOrder(mr.imp);
3894   sys::swapByteOrder(mr.sel);
3895 }
3896 
swapStruct(struct message_ref32 & mr)3897 inline void swapStruct(struct message_ref32 &mr) {
3898   sys::swapByteOrder(mr.imp);
3899   sys::swapByteOrder(mr.sel);
3900 }
3901 
swapStruct(struct objc_module_t & module)3902 inline void swapStruct(struct objc_module_t &module) {
3903   sys::swapByteOrder(module.version);
3904   sys::swapByteOrder(module.size);
3905   sys::swapByteOrder(module.name);
3906   sys::swapByteOrder(module.symtab);
3907 }
3908 
swapStruct(struct objc_symtab_t & symtab)3909 inline void swapStruct(struct objc_symtab_t &symtab) {
3910   sys::swapByteOrder(symtab.sel_ref_cnt);
3911   sys::swapByteOrder(symtab.refs);
3912   sys::swapByteOrder(symtab.cls_def_cnt);
3913   sys::swapByteOrder(symtab.cat_def_cnt);
3914 }
3915 
swapStruct(struct objc_class_t & objc_class)3916 inline void swapStruct(struct objc_class_t &objc_class) {
3917   sys::swapByteOrder(objc_class.isa);
3918   sys::swapByteOrder(objc_class.super_class);
3919   sys::swapByteOrder(objc_class.name);
3920   sys::swapByteOrder(objc_class.version);
3921   sys::swapByteOrder(objc_class.info);
3922   sys::swapByteOrder(objc_class.instance_size);
3923   sys::swapByteOrder(objc_class.ivars);
3924   sys::swapByteOrder(objc_class.methodLists);
3925   sys::swapByteOrder(objc_class.cache);
3926   sys::swapByteOrder(objc_class.protocols);
3927 }
3928 
swapStruct(struct objc_category_t & objc_category)3929 inline void swapStruct(struct objc_category_t &objc_category) {
3930   sys::swapByteOrder(objc_category.category_name);
3931   sys::swapByteOrder(objc_category.class_name);
3932   sys::swapByteOrder(objc_category.instance_methods);
3933   sys::swapByteOrder(objc_category.class_methods);
3934   sys::swapByteOrder(objc_category.protocols);
3935 }
3936 
swapStruct(struct objc_ivar_list_t & objc_ivar_list)3937 inline void swapStruct(struct objc_ivar_list_t &objc_ivar_list) {
3938   sys::swapByteOrder(objc_ivar_list.ivar_count);
3939 }
3940 
swapStruct(struct objc_ivar_t & objc_ivar)3941 inline void swapStruct(struct objc_ivar_t &objc_ivar) {
3942   sys::swapByteOrder(objc_ivar.ivar_name);
3943   sys::swapByteOrder(objc_ivar.ivar_type);
3944   sys::swapByteOrder(objc_ivar.ivar_offset);
3945 }
3946 
swapStruct(struct objc_method_list_t & method_list)3947 inline void swapStruct(struct objc_method_list_t &method_list) {
3948   sys::swapByteOrder(method_list.obsolete);
3949   sys::swapByteOrder(method_list.method_count);
3950 }
3951 
swapStruct(struct objc_method_t & method)3952 inline void swapStruct(struct objc_method_t &method) {
3953   sys::swapByteOrder(method.method_name);
3954   sys::swapByteOrder(method.method_types);
3955   sys::swapByteOrder(method.method_imp);
3956 }
3957 
swapStruct(struct objc_protocol_list_t & protocol_list)3958 inline void swapStruct(struct objc_protocol_list_t &protocol_list) {
3959   sys::swapByteOrder(protocol_list.next);
3960   sys::swapByteOrder(protocol_list.count);
3961 }
3962 
swapStruct(struct objc_protocol_t & protocol)3963 inline void swapStruct(struct objc_protocol_t &protocol) {
3964   sys::swapByteOrder(protocol.isa);
3965   sys::swapByteOrder(protocol.protocol_name);
3966   sys::swapByteOrder(protocol.protocol_list);
3967   sys::swapByteOrder(protocol.instance_methods);
3968   sys::swapByteOrder(protocol.class_methods);
3969 }
3970 
swapStruct(struct objc_method_description_list_t & mdl)3971 inline void swapStruct(struct objc_method_description_list_t &mdl) {
3972   sys::swapByteOrder(mdl.count);
3973 }
3974 
swapStruct(struct objc_method_description_t & md)3975 inline void swapStruct(struct objc_method_description_t &md) {
3976   sys::swapByteOrder(md.name);
3977   sys::swapByteOrder(md.types);
3978 }
3979 
3980 } // namespace
3981 
3982 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
3983                                                  struct DisassembleInfo *info);
3984 
3985 // get_objc2_64bit_class_name() is used for disassembly and is passed a pointer
3986 // to an Objective-C class and returns the class name.  It is also passed the
3987 // address of the pointer, so when the pointer is zero as it can be in an .o
3988 // file, that is used to look for an external relocation entry with a symbol
3989 // name.
get_objc2_64bit_class_name(uint64_t pointer_value,uint64_t ReferenceValue,struct DisassembleInfo * info)3990 static const char *get_objc2_64bit_class_name(uint64_t pointer_value,
3991                                               uint64_t ReferenceValue,
3992                                               struct DisassembleInfo *info) {
3993   const char *r;
3994   uint32_t offset, left;
3995   SectionRef S;
3996 
3997   // The pointer_value can be 0 in an object file and have a relocation
3998   // entry for the class symbol at the ReferenceValue (the address of the
3999   // pointer).
4000   if (pointer_value == 0) {
4001     r = get_pointer_64(ReferenceValue, offset, left, S, info);
4002     if (r == nullptr || left < sizeof(uint64_t))
4003       return nullptr;
4004     uint64_t n_value;
4005     const char *symbol_name = get_symbol_64(offset, S, info, n_value);
4006     if (symbol_name == nullptr)
4007       return nullptr;
4008     const char *class_name = strrchr(symbol_name, '$');
4009     if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0')
4010       return class_name + 2;
4011     else
4012       return nullptr;
4013   }
4014 
4015   // The case were the pointer_value is non-zero and points to a class defined
4016   // in this Mach-O file.
4017   r = get_pointer_64(pointer_value, offset, left, S, info);
4018   if (r == nullptr || left < sizeof(struct class64_t))
4019     return nullptr;
4020   struct class64_t c;
4021   memcpy(&c, r, sizeof(struct class64_t));
4022   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4023     swapStruct(c);
4024   if (c.data == 0)
4025     return nullptr;
4026   r = get_pointer_64(c.data, offset, left, S, info);
4027   if (r == nullptr || left < sizeof(struct class_ro64_t))
4028     return nullptr;
4029   struct class_ro64_t cro;
4030   memcpy(&cro, r, sizeof(struct class_ro64_t));
4031   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4032     swapStruct(cro);
4033   if (cro.name == 0)
4034     return nullptr;
4035   const char *name = get_pointer_64(cro.name, offset, left, S, info);
4036   return name;
4037 }
4038 
4039 // get_objc2_64bit_cfstring_name is used for disassembly and is passed a
4040 // pointer to a cfstring and returns its name or nullptr.
get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,struct DisassembleInfo * info)4041 static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,
4042                                                  struct DisassembleInfo *info) {
4043   const char *r, *name;
4044   uint32_t offset, left;
4045   SectionRef S;
4046   struct cfstring64_t cfs;
4047   uint64_t cfs_characters;
4048 
4049   r = get_pointer_64(ReferenceValue, offset, left, S, info);
4050   if (r == nullptr || left < sizeof(struct cfstring64_t))
4051     return nullptr;
4052   memcpy(&cfs, r, sizeof(struct cfstring64_t));
4053   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4054     swapStruct(cfs);
4055   if (cfs.characters == 0) {
4056     uint64_t n_value;
4057     const char *symbol_name = get_symbol_64(
4058         offset + offsetof(struct cfstring64_t, characters), S, info, n_value);
4059     if (symbol_name == nullptr)
4060       return nullptr;
4061     cfs_characters = n_value;
4062   } else
4063     cfs_characters = cfs.characters;
4064   name = get_pointer_64(cfs_characters, offset, left, S, info);
4065 
4066   return name;
4067 }
4068 
4069 // get_objc2_64bit_selref() is used for disassembly and is passed a the address
4070 // of a pointer to an Objective-C selector reference when the pointer value is
4071 // zero as in a .o file and is likely to have a external relocation entry with
4072 // who's symbol's n_value is the real pointer to the selector name.  If that is
4073 // the case the real pointer to the selector name is returned else 0 is
4074 // returned
get_objc2_64bit_selref(uint64_t ReferenceValue,struct DisassembleInfo * info)4075 static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue,
4076                                        struct DisassembleInfo *info) {
4077   uint32_t offset, left;
4078   SectionRef S;
4079 
4080   const char *r = get_pointer_64(ReferenceValue, offset, left, S, info);
4081   if (r == nullptr || left < sizeof(uint64_t))
4082     return 0;
4083   uint64_t n_value;
4084   const char *symbol_name = get_symbol_64(offset, S, info, n_value);
4085   if (symbol_name == nullptr)
4086     return 0;
4087   return n_value;
4088 }
4089 
get_section(MachOObjectFile * O,const char * segname,const char * sectname)4090 static const SectionRef get_section(MachOObjectFile *O, const char *segname,
4091                                     const char *sectname) {
4092   for (const SectionRef &Section : O->sections()) {
4093     StringRef SectName;
4094     Expected<StringRef> SecNameOrErr = Section.getName();
4095     if (SecNameOrErr)
4096       SectName = *SecNameOrErr;
4097     else
4098       consumeError(SecNameOrErr.takeError());
4099 
4100     DataRefImpl Ref = Section.getRawDataRefImpl();
4101     StringRef SegName = O->getSectionFinalSegmentName(Ref);
4102     if (SegName == segname && SectName == sectname)
4103       return Section;
4104   }
4105   return SectionRef();
4106 }
4107 
4108 static void
walk_pointer_list_64(const char * listname,const SectionRef S,MachOObjectFile * O,struct DisassembleInfo * info,void (* func)(uint64_t,struct DisassembleInfo * info))4109 walk_pointer_list_64(const char *listname, const SectionRef S,
4110                      MachOObjectFile *O, struct DisassembleInfo *info,
4111                      void (*func)(uint64_t, struct DisassembleInfo *info)) {
4112   if (S == SectionRef())
4113     return;
4114 
4115   StringRef SectName;
4116   Expected<StringRef> SecNameOrErr = S.getName();
4117   if (SecNameOrErr)
4118     SectName = *SecNameOrErr;
4119   else
4120     consumeError(SecNameOrErr.takeError());
4121 
4122   DataRefImpl Ref = S.getRawDataRefImpl();
4123   StringRef SegName = O->getSectionFinalSegmentName(Ref);
4124   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
4125 
4126   StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName());
4127   const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
4128 
4129   for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint64_t)) {
4130     uint32_t left = S.getSize() - i;
4131     uint32_t size = left < sizeof(uint64_t) ? left : sizeof(uint64_t);
4132     uint64_t p = 0;
4133     memcpy(&p, Contents + i, size);
4134     if (i + sizeof(uint64_t) > S.getSize())
4135       outs() << listname << " list pointer extends past end of (" << SegName
4136              << "," << SectName << ") section\n";
4137     outs() << format("%016" PRIx64, S.getAddress() + i) << " ";
4138 
4139     if (O->isLittleEndian() != sys::IsLittleEndianHost)
4140       sys::swapByteOrder(p);
4141 
4142     uint64_t n_value = 0;
4143     const char *name = get_symbol_64(i, S, info, n_value, p);
4144     if (name == nullptr)
4145       name = get_dyld_bind_info_symbolname(S.getAddress() + i, info);
4146 
4147     if (n_value != 0) {
4148       outs() << format("0x%" PRIx64, n_value);
4149       if (p != 0)
4150         outs() << " + " << format("0x%" PRIx64, p);
4151     } else
4152       outs() << format("0x%" PRIx64, p);
4153     if (name != nullptr)
4154       outs() << " " << name;
4155     outs() << "\n";
4156 
4157     p += n_value;
4158     if (func)
4159       func(p, info);
4160   }
4161 }
4162 
4163 static void
walk_pointer_list_32(const char * listname,const SectionRef S,MachOObjectFile * O,struct DisassembleInfo * info,void (* func)(uint32_t,struct DisassembleInfo * info))4164 walk_pointer_list_32(const char *listname, const SectionRef S,
4165                      MachOObjectFile *O, struct DisassembleInfo *info,
4166                      void (*func)(uint32_t, struct DisassembleInfo *info)) {
4167   if (S == SectionRef())
4168     return;
4169 
4170   StringRef SectName = unwrapOrError(S.getName(), O->getFileName());
4171   DataRefImpl Ref = S.getRawDataRefImpl();
4172   StringRef SegName = O->getSectionFinalSegmentName(Ref);
4173   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
4174 
4175   StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName());
4176   const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
4177 
4178   for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint32_t)) {
4179     uint32_t left = S.getSize() - i;
4180     uint32_t size = left < sizeof(uint32_t) ? left : sizeof(uint32_t);
4181     uint32_t p = 0;
4182     memcpy(&p, Contents + i, size);
4183     if (i + sizeof(uint32_t) > S.getSize())
4184       outs() << listname << " list pointer extends past end of (" << SegName
4185              << "," << SectName << ") section\n";
4186     uint32_t Address = S.getAddress() + i;
4187     outs() << format("%08" PRIx32, Address) << " ";
4188 
4189     if (O->isLittleEndian() != sys::IsLittleEndianHost)
4190       sys::swapByteOrder(p);
4191     outs() << format("0x%" PRIx32, p);
4192 
4193     const char *name = get_symbol_32(i, S, info, p);
4194     if (name != nullptr)
4195       outs() << " " << name;
4196     outs() << "\n";
4197 
4198     if (func)
4199       func(p, info);
4200   }
4201 }
4202 
print_layout_map(const char * layout_map,uint32_t left)4203 static void print_layout_map(const char *layout_map, uint32_t left) {
4204   if (layout_map == nullptr)
4205     return;
4206   outs() << "                layout map: ";
4207   do {
4208     outs() << format("0x%02" PRIx32, (*layout_map) & 0xff) << " ";
4209     left--;
4210     layout_map++;
4211   } while (*layout_map != '\0' && left != 0);
4212   outs() << "\n";
4213 }
4214 
print_layout_map64(uint64_t p,struct DisassembleInfo * info)4215 static void print_layout_map64(uint64_t p, struct DisassembleInfo *info) {
4216   uint32_t offset, left;
4217   SectionRef S;
4218   const char *layout_map;
4219 
4220   if (p == 0)
4221     return;
4222   layout_map = get_pointer_64(p, offset, left, S, info);
4223   print_layout_map(layout_map, left);
4224 }
4225 
print_layout_map32(uint32_t p,struct DisassembleInfo * info)4226 static void print_layout_map32(uint32_t p, struct DisassembleInfo *info) {
4227   uint32_t offset, left;
4228   SectionRef S;
4229   const char *layout_map;
4230 
4231   if (p == 0)
4232     return;
4233   layout_map = get_pointer_32(p, offset, left, S, info);
4234   print_layout_map(layout_map, left);
4235 }
4236 
print_method_list64_t(uint64_t p,struct DisassembleInfo * info,const char * indent)4237 static void print_method_list64_t(uint64_t p, struct DisassembleInfo *info,
4238                                   const char *indent) {
4239   struct method_list64_t ml;
4240   struct method64_t m;
4241   const char *r;
4242   uint32_t offset, xoffset, left, i;
4243   SectionRef S, xS;
4244   const char *name, *sym_name;
4245   uint64_t n_value;
4246 
4247   r = get_pointer_64(p, offset, left, S, info);
4248   if (r == nullptr)
4249     return;
4250   memset(&ml, '\0', sizeof(struct method_list64_t));
4251   if (left < sizeof(struct method_list64_t)) {
4252     memcpy(&ml, r, left);
4253     outs() << "   (method_list_t entends past the end of the section)\n";
4254   } else
4255     memcpy(&ml, r, sizeof(struct method_list64_t));
4256   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4257     swapStruct(ml);
4258   outs() << indent << "\t\t   entsize " << ml.entsize << "\n";
4259   outs() << indent << "\t\t     count " << ml.count << "\n";
4260 
4261   p += sizeof(struct method_list64_t);
4262   offset += sizeof(struct method_list64_t);
4263   for (i = 0; i < ml.count; i++) {
4264     r = get_pointer_64(p, offset, left, S, info);
4265     if (r == nullptr)
4266       return;
4267     memset(&m, '\0', sizeof(struct method64_t));
4268     if (left < sizeof(struct method64_t)) {
4269       memcpy(&m, r, left);
4270       outs() << indent << "   (method_t extends past the end of the section)\n";
4271     } else
4272       memcpy(&m, r, sizeof(struct method64_t));
4273     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4274       swapStruct(m);
4275 
4276     outs() << indent << "\t\t      name ";
4277     sym_name = get_symbol_64(offset + offsetof(struct method64_t, name), S,
4278                              info, n_value, m.name);
4279     if (n_value != 0) {
4280       if (info->verbose && sym_name != nullptr)
4281         outs() << sym_name;
4282       else
4283         outs() << format("0x%" PRIx64, n_value);
4284       if (m.name != 0)
4285         outs() << " + " << format("0x%" PRIx64, m.name);
4286     } else
4287       outs() << format("0x%" PRIx64, m.name);
4288     name = get_pointer_64(m.name + n_value, xoffset, left, xS, info);
4289     if (name != nullptr)
4290       outs() << format(" %.*s", left, name);
4291     outs() << "\n";
4292 
4293     outs() << indent << "\t\t     types ";
4294     sym_name = get_symbol_64(offset + offsetof(struct method64_t, types), S,
4295                              info, n_value, m.types);
4296     if (n_value != 0) {
4297       if (info->verbose && sym_name != nullptr)
4298         outs() << sym_name;
4299       else
4300         outs() << format("0x%" PRIx64, n_value);
4301       if (m.types != 0)
4302         outs() << " + " << format("0x%" PRIx64, m.types);
4303     } else
4304       outs() << format("0x%" PRIx64, m.types);
4305     name = get_pointer_64(m.types + n_value, xoffset, left, xS, info);
4306     if (name != nullptr)
4307       outs() << format(" %.*s", left, name);
4308     outs() << "\n";
4309 
4310     outs() << indent << "\t\t       imp ";
4311     name = get_symbol_64(offset + offsetof(struct method64_t, imp), S, info,
4312                          n_value, m.imp);
4313     if (info->verbose && name == nullptr) {
4314       if (n_value != 0) {
4315         outs() << format("0x%" PRIx64, n_value) << " ";
4316         if (m.imp != 0)
4317           outs() << "+ " << format("0x%" PRIx64, m.imp) << " ";
4318       } else
4319         outs() << format("0x%" PRIx64, m.imp) << " ";
4320     }
4321     if (name != nullptr)
4322       outs() << name;
4323     outs() << "\n";
4324 
4325     p += sizeof(struct method64_t);
4326     offset += sizeof(struct method64_t);
4327   }
4328 }
4329 
print_method_list32_t(uint64_t p,struct DisassembleInfo * info,const char * indent)4330 static void print_method_list32_t(uint64_t p, struct DisassembleInfo *info,
4331                                   const char *indent) {
4332   struct method_list32_t ml;
4333   struct method32_t m;
4334   const char *r, *name;
4335   uint32_t offset, xoffset, left, i;
4336   SectionRef S, xS;
4337 
4338   r = get_pointer_32(p, offset, left, S, info);
4339   if (r == nullptr)
4340     return;
4341   memset(&ml, '\0', sizeof(struct method_list32_t));
4342   if (left < sizeof(struct method_list32_t)) {
4343     memcpy(&ml, r, left);
4344     outs() << "   (method_list_t entends past the end of the section)\n";
4345   } else
4346     memcpy(&ml, r, sizeof(struct method_list32_t));
4347   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4348     swapStruct(ml);
4349   outs() << indent << "\t\t   entsize " << ml.entsize << "\n";
4350   outs() << indent << "\t\t     count " << ml.count << "\n";
4351 
4352   p += sizeof(struct method_list32_t);
4353   offset += sizeof(struct method_list32_t);
4354   for (i = 0; i < ml.count; i++) {
4355     r = get_pointer_32(p, offset, left, S, info);
4356     if (r == nullptr)
4357       return;
4358     memset(&m, '\0', sizeof(struct method32_t));
4359     if (left < sizeof(struct method32_t)) {
4360       memcpy(&ml, r, left);
4361       outs() << indent << "   (method_t entends past the end of the section)\n";
4362     } else
4363       memcpy(&m, r, sizeof(struct method32_t));
4364     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4365       swapStruct(m);
4366 
4367     outs() << indent << "\t\t      name " << format("0x%" PRIx32, m.name);
4368     name = get_pointer_32(m.name, xoffset, left, xS, info);
4369     if (name != nullptr)
4370       outs() << format(" %.*s", left, name);
4371     outs() << "\n";
4372 
4373     outs() << indent << "\t\t     types " << format("0x%" PRIx32, m.types);
4374     name = get_pointer_32(m.types, xoffset, left, xS, info);
4375     if (name != nullptr)
4376       outs() << format(" %.*s", left, name);
4377     outs() << "\n";
4378 
4379     outs() << indent << "\t\t       imp " << format("0x%" PRIx32, m.imp);
4380     name = get_symbol_32(offset + offsetof(struct method32_t, imp), S, info,
4381                          m.imp);
4382     if (name != nullptr)
4383       outs() << " " << name;
4384     outs() << "\n";
4385 
4386     p += sizeof(struct method32_t);
4387     offset += sizeof(struct method32_t);
4388   }
4389 }
4390 
print_method_list(uint32_t p,struct DisassembleInfo * info)4391 static bool print_method_list(uint32_t p, struct DisassembleInfo *info) {
4392   uint32_t offset, left, xleft;
4393   SectionRef S;
4394   struct objc_method_list_t method_list;
4395   struct objc_method_t method;
4396   const char *r, *methods, *name, *SymbolName;
4397   int32_t i;
4398 
4399   r = get_pointer_32(p, offset, left, S, info, true);
4400   if (r == nullptr)
4401     return true;
4402 
4403   outs() << "\n";
4404   if (left > sizeof(struct objc_method_list_t)) {
4405     memcpy(&method_list, r, sizeof(struct objc_method_list_t));
4406   } else {
4407     outs() << "\t\t objc_method_list extends past end of the section\n";
4408     memset(&method_list, '\0', sizeof(struct objc_method_list_t));
4409     memcpy(&method_list, r, left);
4410   }
4411   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4412     swapStruct(method_list);
4413 
4414   outs() << "\t\t         obsolete "
4415          << format("0x%08" PRIx32, method_list.obsolete) << "\n";
4416   outs() << "\t\t     method_count " << method_list.method_count << "\n";
4417 
4418   methods = r + sizeof(struct objc_method_list_t);
4419   for (i = 0; i < method_list.method_count; i++) {
4420     if ((i + 1) * sizeof(struct objc_method_t) > left) {
4421       outs() << "\t\t remaining method's extend past the of the section\n";
4422       break;
4423     }
4424     memcpy(&method, methods + i * sizeof(struct objc_method_t),
4425            sizeof(struct objc_method_t));
4426     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4427       swapStruct(method);
4428 
4429     outs() << "\t\t      method_name "
4430            << format("0x%08" PRIx32, method.method_name);
4431     if (info->verbose) {
4432       name = get_pointer_32(method.method_name, offset, xleft, S, info, true);
4433       if (name != nullptr)
4434         outs() << format(" %.*s", xleft, name);
4435       else
4436         outs() << " (not in an __OBJC section)";
4437     }
4438     outs() << "\n";
4439 
4440     outs() << "\t\t     method_types "
4441            << format("0x%08" PRIx32, method.method_types);
4442     if (info->verbose) {
4443       name = get_pointer_32(method.method_types, offset, xleft, S, info, true);
4444       if (name != nullptr)
4445         outs() << format(" %.*s", xleft, name);
4446       else
4447         outs() << " (not in an __OBJC section)";
4448     }
4449     outs() << "\n";
4450 
4451     outs() << "\t\t       method_imp "
4452            << format("0x%08" PRIx32, method.method_imp) << " ";
4453     if (info->verbose) {
4454       SymbolName = GuessSymbolName(method.method_imp, info->AddrMap);
4455       if (SymbolName != nullptr)
4456         outs() << SymbolName;
4457     }
4458     outs() << "\n";
4459   }
4460   return false;
4461 }
4462 
print_protocol_list64_t(uint64_t p,struct DisassembleInfo * info)4463 static void print_protocol_list64_t(uint64_t p, struct DisassembleInfo *info) {
4464   struct protocol_list64_t pl;
4465   uint64_t q, n_value;
4466   struct protocol64_t pc;
4467   const char *r;
4468   uint32_t offset, xoffset, left, i;
4469   SectionRef S, xS;
4470   const char *name, *sym_name;
4471 
4472   r = get_pointer_64(p, offset, left, S, info);
4473   if (r == nullptr)
4474     return;
4475   memset(&pl, '\0', sizeof(struct protocol_list64_t));
4476   if (left < sizeof(struct protocol_list64_t)) {
4477     memcpy(&pl, r, left);
4478     outs() << "   (protocol_list_t entends past the end of the section)\n";
4479   } else
4480     memcpy(&pl, r, sizeof(struct protocol_list64_t));
4481   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4482     swapStruct(pl);
4483   outs() << "                      count " << pl.count << "\n";
4484 
4485   p += sizeof(struct protocol_list64_t);
4486   offset += sizeof(struct protocol_list64_t);
4487   for (i = 0; i < pl.count; i++) {
4488     r = get_pointer_64(p, offset, left, S, info);
4489     if (r == nullptr)
4490       return;
4491     q = 0;
4492     if (left < sizeof(uint64_t)) {
4493       memcpy(&q, r, left);
4494       outs() << "   (protocol_t * entends past the end of the section)\n";
4495     } else
4496       memcpy(&q, r, sizeof(uint64_t));
4497     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4498       sys::swapByteOrder(q);
4499 
4500     outs() << "\t\t      list[" << i << "] ";
4501     sym_name = get_symbol_64(offset, S, info, n_value, q);
4502     if (n_value != 0) {
4503       if (info->verbose && sym_name != nullptr)
4504         outs() << sym_name;
4505       else
4506         outs() << format("0x%" PRIx64, n_value);
4507       if (q != 0)
4508         outs() << " + " << format("0x%" PRIx64, q);
4509     } else
4510       outs() << format("0x%" PRIx64, q);
4511     outs() << " (struct protocol_t *)\n";
4512 
4513     r = get_pointer_64(q + n_value, offset, left, S, info);
4514     if (r == nullptr)
4515       return;
4516     memset(&pc, '\0', sizeof(struct protocol64_t));
4517     if (left < sizeof(struct protocol64_t)) {
4518       memcpy(&pc, r, left);
4519       outs() << "   (protocol_t entends past the end of the section)\n";
4520     } else
4521       memcpy(&pc, r, sizeof(struct protocol64_t));
4522     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4523       swapStruct(pc);
4524 
4525     outs() << "\t\t\t      isa " << format("0x%" PRIx64, pc.isa) << "\n";
4526 
4527     outs() << "\t\t\t     name ";
4528     sym_name = get_symbol_64(offset + offsetof(struct protocol64_t, name), S,
4529                              info, n_value, pc.name);
4530     if (n_value != 0) {
4531       if (info->verbose && sym_name != nullptr)
4532         outs() << sym_name;
4533       else
4534         outs() << format("0x%" PRIx64, n_value);
4535       if (pc.name != 0)
4536         outs() << " + " << format("0x%" PRIx64, pc.name);
4537     } else
4538       outs() << format("0x%" PRIx64, pc.name);
4539     name = get_pointer_64(pc.name + n_value, xoffset, left, xS, info);
4540     if (name != nullptr)
4541       outs() << format(" %.*s", left, name);
4542     outs() << "\n";
4543 
4544     outs() << "\t\t\tprotocols " << format("0x%" PRIx64, pc.protocols) << "\n";
4545 
4546     outs() << "\t\t  instanceMethods ";
4547     sym_name =
4548         get_symbol_64(offset + offsetof(struct protocol64_t, instanceMethods),
4549                       S, info, n_value, pc.instanceMethods);
4550     if (n_value != 0) {
4551       if (info->verbose && sym_name != nullptr)
4552         outs() << sym_name;
4553       else
4554         outs() << format("0x%" PRIx64, n_value);
4555       if (pc.instanceMethods != 0)
4556         outs() << " + " << format("0x%" PRIx64, pc.instanceMethods);
4557     } else
4558       outs() << format("0x%" PRIx64, pc.instanceMethods);
4559     outs() << " (struct method_list_t *)\n";
4560     if (pc.instanceMethods + n_value != 0)
4561       print_method_list64_t(pc.instanceMethods + n_value, info, "\t");
4562 
4563     outs() << "\t\t     classMethods ";
4564     sym_name =
4565         get_symbol_64(offset + offsetof(struct protocol64_t, classMethods), S,
4566                       info, n_value, pc.classMethods);
4567     if (n_value != 0) {
4568       if (info->verbose && sym_name != nullptr)
4569         outs() << sym_name;
4570       else
4571         outs() << format("0x%" PRIx64, n_value);
4572       if (pc.classMethods != 0)
4573         outs() << " + " << format("0x%" PRIx64, pc.classMethods);
4574     } else
4575       outs() << format("0x%" PRIx64, pc.classMethods);
4576     outs() << " (struct method_list_t *)\n";
4577     if (pc.classMethods + n_value != 0)
4578       print_method_list64_t(pc.classMethods + n_value, info, "\t");
4579 
4580     outs() << "\t  optionalInstanceMethods "
4581            << format("0x%" PRIx64, pc.optionalInstanceMethods) << "\n";
4582     outs() << "\t     optionalClassMethods "
4583            << format("0x%" PRIx64, pc.optionalClassMethods) << "\n";
4584     outs() << "\t       instanceProperties "
4585            << format("0x%" PRIx64, pc.instanceProperties) << "\n";
4586 
4587     p += sizeof(uint64_t);
4588     offset += sizeof(uint64_t);
4589   }
4590 }
4591 
print_protocol_list32_t(uint32_t p,struct DisassembleInfo * info)4592 static void print_protocol_list32_t(uint32_t p, struct DisassembleInfo *info) {
4593   struct protocol_list32_t pl;
4594   uint32_t q;
4595   struct protocol32_t pc;
4596   const char *r;
4597   uint32_t offset, xoffset, left, i;
4598   SectionRef S, xS;
4599   const char *name;
4600 
4601   r = get_pointer_32(p, offset, left, S, info);
4602   if (r == nullptr)
4603     return;
4604   memset(&pl, '\0', sizeof(struct protocol_list32_t));
4605   if (left < sizeof(struct protocol_list32_t)) {
4606     memcpy(&pl, r, left);
4607     outs() << "   (protocol_list_t entends past the end of the section)\n";
4608   } else
4609     memcpy(&pl, r, sizeof(struct protocol_list32_t));
4610   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4611     swapStruct(pl);
4612   outs() << "                      count " << pl.count << "\n";
4613 
4614   p += sizeof(struct protocol_list32_t);
4615   offset += sizeof(struct protocol_list32_t);
4616   for (i = 0; i < pl.count; i++) {
4617     r = get_pointer_32(p, offset, left, S, info);
4618     if (r == nullptr)
4619       return;
4620     q = 0;
4621     if (left < sizeof(uint32_t)) {
4622       memcpy(&q, r, left);
4623       outs() << "   (protocol_t * entends past the end of the section)\n";
4624     } else
4625       memcpy(&q, r, sizeof(uint32_t));
4626     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4627       sys::swapByteOrder(q);
4628     outs() << "\t\t      list[" << i << "] " << format("0x%" PRIx32, q)
4629            << " (struct protocol_t *)\n";
4630     r = get_pointer_32(q, offset, left, S, info);
4631     if (r == nullptr)
4632       return;
4633     memset(&pc, '\0', sizeof(struct protocol32_t));
4634     if (left < sizeof(struct protocol32_t)) {
4635       memcpy(&pc, r, left);
4636       outs() << "   (protocol_t entends past the end of the section)\n";
4637     } else
4638       memcpy(&pc, r, sizeof(struct protocol32_t));
4639     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4640       swapStruct(pc);
4641     outs() << "\t\t\t      isa " << format("0x%" PRIx32, pc.isa) << "\n";
4642     outs() << "\t\t\t     name " << format("0x%" PRIx32, pc.name);
4643     name = get_pointer_32(pc.name, xoffset, left, xS, info);
4644     if (name != nullptr)
4645       outs() << format(" %.*s", left, name);
4646     outs() << "\n";
4647     outs() << "\t\t\tprotocols " << format("0x%" PRIx32, pc.protocols) << "\n";
4648     outs() << "\t\t  instanceMethods "
4649            << format("0x%" PRIx32, pc.instanceMethods)
4650            << " (struct method_list_t *)\n";
4651     if (pc.instanceMethods != 0)
4652       print_method_list32_t(pc.instanceMethods, info, "\t");
4653     outs() << "\t\t     classMethods " << format("0x%" PRIx32, pc.classMethods)
4654            << " (struct method_list_t *)\n";
4655     if (pc.classMethods != 0)
4656       print_method_list32_t(pc.classMethods, info, "\t");
4657     outs() << "\t  optionalInstanceMethods "
4658            << format("0x%" PRIx32, pc.optionalInstanceMethods) << "\n";
4659     outs() << "\t     optionalClassMethods "
4660            << format("0x%" PRIx32, pc.optionalClassMethods) << "\n";
4661     outs() << "\t       instanceProperties "
4662            << format("0x%" PRIx32, pc.instanceProperties) << "\n";
4663     p += sizeof(uint32_t);
4664     offset += sizeof(uint32_t);
4665   }
4666 }
4667 
print_indent(uint32_t indent)4668 static void print_indent(uint32_t indent) {
4669   for (uint32_t i = 0; i < indent;) {
4670     if (indent - i >= 8) {
4671       outs() << "\t";
4672       i += 8;
4673     } else {
4674       for (uint32_t j = i; j < indent; j++)
4675         outs() << " ";
4676       return;
4677     }
4678   }
4679 }
4680 
print_method_description_list(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4681 static bool print_method_description_list(uint32_t p, uint32_t indent,
4682                                           struct DisassembleInfo *info) {
4683   uint32_t offset, left, xleft;
4684   SectionRef S;
4685   struct objc_method_description_list_t mdl;
4686   struct objc_method_description_t md;
4687   const char *r, *list, *name;
4688   int32_t i;
4689 
4690   r = get_pointer_32(p, offset, left, S, info, true);
4691   if (r == nullptr)
4692     return true;
4693 
4694   outs() << "\n";
4695   if (left > sizeof(struct objc_method_description_list_t)) {
4696     memcpy(&mdl, r, sizeof(struct objc_method_description_list_t));
4697   } else {
4698     print_indent(indent);
4699     outs() << " objc_method_description_list extends past end of the section\n";
4700     memset(&mdl, '\0', sizeof(struct objc_method_description_list_t));
4701     memcpy(&mdl, r, left);
4702   }
4703   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4704     swapStruct(mdl);
4705 
4706   print_indent(indent);
4707   outs() << "        count " << mdl.count << "\n";
4708 
4709   list = r + sizeof(struct objc_method_description_list_t);
4710   for (i = 0; i < mdl.count; i++) {
4711     if ((i + 1) * sizeof(struct objc_method_description_t) > left) {
4712       print_indent(indent);
4713       outs() << " remaining list entries extend past the of the section\n";
4714       break;
4715     }
4716     print_indent(indent);
4717     outs() << "        list[" << i << "]\n";
4718     memcpy(&md, list + i * sizeof(struct objc_method_description_t),
4719            sizeof(struct objc_method_description_t));
4720     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4721       swapStruct(md);
4722 
4723     print_indent(indent);
4724     outs() << "             name " << format("0x%08" PRIx32, md.name);
4725     if (info->verbose) {
4726       name = get_pointer_32(md.name, offset, xleft, S, info, true);
4727       if (name != nullptr)
4728         outs() << format(" %.*s", xleft, name);
4729       else
4730         outs() << " (not in an __OBJC section)";
4731     }
4732     outs() << "\n";
4733 
4734     print_indent(indent);
4735     outs() << "            types " << format("0x%08" PRIx32, md.types);
4736     if (info->verbose) {
4737       name = get_pointer_32(md.types, offset, xleft, S, info, true);
4738       if (name != nullptr)
4739         outs() << format(" %.*s", xleft, name);
4740       else
4741         outs() << " (not in an __OBJC section)";
4742     }
4743     outs() << "\n";
4744   }
4745   return false;
4746 }
4747 
4748 static bool print_protocol_list(uint32_t p, uint32_t indent,
4749                                 struct DisassembleInfo *info);
4750 
print_protocol(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4751 static bool print_protocol(uint32_t p, uint32_t indent,
4752                            struct DisassembleInfo *info) {
4753   uint32_t offset, left;
4754   SectionRef S;
4755   struct objc_protocol_t protocol;
4756   const char *r, *name;
4757 
4758   r = get_pointer_32(p, offset, left, S, info, true);
4759   if (r == nullptr)
4760     return true;
4761 
4762   outs() << "\n";
4763   if (left >= sizeof(struct objc_protocol_t)) {
4764     memcpy(&protocol, r, sizeof(struct objc_protocol_t));
4765   } else {
4766     print_indent(indent);
4767     outs() << "            Protocol extends past end of the section\n";
4768     memset(&protocol, '\0', sizeof(struct objc_protocol_t));
4769     memcpy(&protocol, r, left);
4770   }
4771   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4772     swapStruct(protocol);
4773 
4774   print_indent(indent);
4775   outs() << "              isa " << format("0x%08" PRIx32, protocol.isa)
4776          << "\n";
4777 
4778   print_indent(indent);
4779   outs() << "    protocol_name "
4780          << format("0x%08" PRIx32, protocol.protocol_name);
4781   if (info->verbose) {
4782     name = get_pointer_32(protocol.protocol_name, offset, left, S, info, true);
4783     if (name != nullptr)
4784       outs() << format(" %.*s", left, name);
4785     else
4786       outs() << " (not in an __OBJC section)";
4787   }
4788   outs() << "\n";
4789 
4790   print_indent(indent);
4791   outs() << "    protocol_list "
4792          << format("0x%08" PRIx32, protocol.protocol_list);
4793   if (print_protocol_list(protocol.protocol_list, indent + 4, info))
4794     outs() << " (not in an __OBJC section)\n";
4795 
4796   print_indent(indent);
4797   outs() << " instance_methods "
4798          << format("0x%08" PRIx32, protocol.instance_methods);
4799   if (print_method_description_list(protocol.instance_methods, indent, info))
4800     outs() << " (not in an __OBJC section)\n";
4801 
4802   print_indent(indent);
4803   outs() << "    class_methods "
4804          << format("0x%08" PRIx32, protocol.class_methods);
4805   if (print_method_description_list(protocol.class_methods, indent, info))
4806     outs() << " (not in an __OBJC section)\n";
4807 
4808   return false;
4809 }
4810 
print_protocol_list(uint32_t p,uint32_t indent,struct DisassembleInfo * info)4811 static bool print_protocol_list(uint32_t p, uint32_t indent,
4812                                 struct DisassembleInfo *info) {
4813   uint32_t offset, left, l;
4814   SectionRef S;
4815   struct objc_protocol_list_t protocol_list;
4816   const char *r, *list;
4817   int32_t i;
4818 
4819   r = get_pointer_32(p, offset, left, S, info, true);
4820   if (r == nullptr)
4821     return true;
4822 
4823   outs() << "\n";
4824   if (left > sizeof(struct objc_protocol_list_t)) {
4825     memcpy(&protocol_list, r, sizeof(struct objc_protocol_list_t));
4826   } else {
4827     outs() << "\t\t objc_protocol_list_t extends past end of the section\n";
4828     memset(&protocol_list, '\0', sizeof(struct objc_protocol_list_t));
4829     memcpy(&protocol_list, r, left);
4830   }
4831   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4832     swapStruct(protocol_list);
4833 
4834   print_indent(indent);
4835   outs() << "         next " << format("0x%08" PRIx32, protocol_list.next)
4836          << "\n";
4837   print_indent(indent);
4838   outs() << "        count " << protocol_list.count << "\n";
4839 
4840   list = r + sizeof(struct objc_protocol_list_t);
4841   for (i = 0; i < protocol_list.count; i++) {
4842     if ((i + 1) * sizeof(uint32_t) > left) {
4843       outs() << "\t\t remaining list entries extend past the of the section\n";
4844       break;
4845     }
4846     memcpy(&l, list + i * sizeof(uint32_t), sizeof(uint32_t));
4847     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4848       sys::swapByteOrder(l);
4849 
4850     print_indent(indent);
4851     outs() << "      list[" << i << "] " << format("0x%08" PRIx32, l);
4852     if (print_protocol(l, indent, info))
4853       outs() << "(not in an __OBJC section)\n";
4854   }
4855   return false;
4856 }
4857 
print_ivar_list64_t(uint64_t p,struct DisassembleInfo * info)4858 static void print_ivar_list64_t(uint64_t p, struct DisassembleInfo *info) {
4859   struct ivar_list64_t il;
4860   struct ivar64_t i;
4861   const char *r;
4862   uint32_t offset, xoffset, left, j;
4863   SectionRef S, xS;
4864   const char *name, *sym_name, *ivar_offset_p;
4865   uint64_t ivar_offset, n_value;
4866 
4867   r = get_pointer_64(p, offset, left, S, info);
4868   if (r == nullptr)
4869     return;
4870   memset(&il, '\0', sizeof(struct ivar_list64_t));
4871   if (left < sizeof(struct ivar_list64_t)) {
4872     memcpy(&il, r, left);
4873     outs() << "   (ivar_list_t entends past the end of the section)\n";
4874   } else
4875     memcpy(&il, r, sizeof(struct ivar_list64_t));
4876   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4877     swapStruct(il);
4878   outs() << "                    entsize " << il.entsize << "\n";
4879   outs() << "                      count " << il.count << "\n";
4880 
4881   p += sizeof(struct ivar_list64_t);
4882   offset += sizeof(struct ivar_list64_t);
4883   for (j = 0; j < il.count; j++) {
4884     r = get_pointer_64(p, offset, left, S, info);
4885     if (r == nullptr)
4886       return;
4887     memset(&i, '\0', sizeof(struct ivar64_t));
4888     if (left < sizeof(struct ivar64_t)) {
4889       memcpy(&i, r, left);
4890       outs() << "   (ivar_t entends past the end of the section)\n";
4891     } else
4892       memcpy(&i, r, sizeof(struct ivar64_t));
4893     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4894       swapStruct(i);
4895 
4896     outs() << "\t\t\t   offset ";
4897     sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, offset), S,
4898                              info, n_value, i.offset);
4899     if (n_value != 0) {
4900       if (info->verbose && sym_name != nullptr)
4901         outs() << sym_name;
4902       else
4903         outs() << format("0x%" PRIx64, n_value);
4904       if (i.offset != 0)
4905         outs() << " + " << format("0x%" PRIx64, i.offset);
4906     } else
4907       outs() << format("0x%" PRIx64, i.offset);
4908     ivar_offset_p = get_pointer_64(i.offset + n_value, xoffset, left, xS, info);
4909     if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) {
4910       memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset));
4911       if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4912         sys::swapByteOrder(ivar_offset);
4913       outs() << " " << ivar_offset << "\n";
4914     } else
4915       outs() << "\n";
4916 
4917     outs() << "\t\t\t     name ";
4918     sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, name), S, info,
4919                              n_value, i.name);
4920     if (n_value != 0) {
4921       if (info->verbose && sym_name != nullptr)
4922         outs() << sym_name;
4923       else
4924         outs() << format("0x%" PRIx64, n_value);
4925       if (i.name != 0)
4926         outs() << " + " << format("0x%" PRIx64, i.name);
4927     } else
4928       outs() << format("0x%" PRIx64, i.name);
4929     name = get_pointer_64(i.name + n_value, xoffset, left, xS, info);
4930     if (name != nullptr)
4931       outs() << format(" %.*s", left, name);
4932     outs() << "\n";
4933 
4934     outs() << "\t\t\t     type ";
4935     sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, type), S, info,
4936                              n_value, i.name);
4937     name = get_pointer_64(i.type + n_value, xoffset, left, xS, info);
4938     if (n_value != 0) {
4939       if (info->verbose && sym_name != nullptr)
4940         outs() << sym_name;
4941       else
4942         outs() << format("0x%" PRIx64, n_value);
4943       if (i.type != 0)
4944         outs() << " + " << format("0x%" PRIx64, i.type);
4945     } else
4946       outs() << format("0x%" PRIx64, i.type);
4947     if (name != nullptr)
4948       outs() << format(" %.*s", left, name);
4949     outs() << "\n";
4950 
4951     outs() << "\t\t\talignment " << i.alignment << "\n";
4952     outs() << "\t\t\t     size " << i.size << "\n";
4953 
4954     p += sizeof(struct ivar64_t);
4955     offset += sizeof(struct ivar64_t);
4956   }
4957 }
4958 
print_ivar_list32_t(uint32_t p,struct DisassembleInfo * info)4959 static void print_ivar_list32_t(uint32_t p, struct DisassembleInfo *info) {
4960   struct ivar_list32_t il;
4961   struct ivar32_t i;
4962   const char *r;
4963   uint32_t offset, xoffset, left, j;
4964   SectionRef S, xS;
4965   const char *name, *ivar_offset_p;
4966   uint32_t ivar_offset;
4967 
4968   r = get_pointer_32(p, offset, left, S, info);
4969   if (r == nullptr)
4970     return;
4971   memset(&il, '\0', sizeof(struct ivar_list32_t));
4972   if (left < sizeof(struct ivar_list32_t)) {
4973     memcpy(&il, r, left);
4974     outs() << "   (ivar_list_t entends past the end of the section)\n";
4975   } else
4976     memcpy(&il, r, sizeof(struct ivar_list32_t));
4977   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4978     swapStruct(il);
4979   outs() << "                    entsize " << il.entsize << "\n";
4980   outs() << "                      count " << il.count << "\n";
4981 
4982   p += sizeof(struct ivar_list32_t);
4983   offset += sizeof(struct ivar_list32_t);
4984   for (j = 0; j < il.count; j++) {
4985     r = get_pointer_32(p, offset, left, S, info);
4986     if (r == nullptr)
4987       return;
4988     memset(&i, '\0', sizeof(struct ivar32_t));
4989     if (left < sizeof(struct ivar32_t)) {
4990       memcpy(&i, r, left);
4991       outs() << "   (ivar_t entends past the end of the section)\n";
4992     } else
4993       memcpy(&i, r, sizeof(struct ivar32_t));
4994     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
4995       swapStruct(i);
4996 
4997     outs() << "\t\t\t   offset " << format("0x%" PRIx32, i.offset);
4998     ivar_offset_p = get_pointer_32(i.offset, xoffset, left, xS, info);
4999     if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) {
5000       memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset));
5001       if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5002         sys::swapByteOrder(ivar_offset);
5003       outs() << " " << ivar_offset << "\n";
5004     } else
5005       outs() << "\n";
5006 
5007     outs() << "\t\t\t     name " << format("0x%" PRIx32, i.name);
5008     name = get_pointer_32(i.name, xoffset, left, xS, info);
5009     if (name != nullptr)
5010       outs() << format(" %.*s", left, name);
5011     outs() << "\n";
5012 
5013     outs() << "\t\t\t     type " << format("0x%" PRIx32, i.type);
5014     name = get_pointer_32(i.type, xoffset, left, xS, info);
5015     if (name != nullptr)
5016       outs() << format(" %.*s", left, name);
5017     outs() << "\n";
5018 
5019     outs() << "\t\t\talignment " << i.alignment << "\n";
5020     outs() << "\t\t\t     size " << i.size << "\n";
5021 
5022     p += sizeof(struct ivar32_t);
5023     offset += sizeof(struct ivar32_t);
5024   }
5025 }
5026 
print_objc_property_list64(uint64_t p,struct DisassembleInfo * info)5027 static void print_objc_property_list64(uint64_t p,
5028                                        struct DisassembleInfo *info) {
5029   struct objc_property_list64 opl;
5030   struct objc_property64 op;
5031   const char *r;
5032   uint32_t offset, xoffset, left, j;
5033   SectionRef S, xS;
5034   const char *name, *sym_name;
5035   uint64_t n_value;
5036 
5037   r = get_pointer_64(p, offset, left, S, info);
5038   if (r == nullptr)
5039     return;
5040   memset(&opl, '\0', sizeof(struct objc_property_list64));
5041   if (left < sizeof(struct objc_property_list64)) {
5042     memcpy(&opl, r, left);
5043     outs() << "   (objc_property_list entends past the end of the section)\n";
5044   } else
5045     memcpy(&opl, r, sizeof(struct objc_property_list64));
5046   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5047     swapStruct(opl);
5048   outs() << "                    entsize " << opl.entsize << "\n";
5049   outs() << "                      count " << opl.count << "\n";
5050 
5051   p += sizeof(struct objc_property_list64);
5052   offset += sizeof(struct objc_property_list64);
5053   for (j = 0; j < opl.count; j++) {
5054     r = get_pointer_64(p, offset, left, S, info);
5055     if (r == nullptr)
5056       return;
5057     memset(&op, '\0', sizeof(struct objc_property64));
5058     if (left < sizeof(struct objc_property64)) {
5059       memcpy(&op, r, left);
5060       outs() << "   (objc_property entends past the end of the section)\n";
5061     } else
5062       memcpy(&op, r, sizeof(struct objc_property64));
5063     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5064       swapStruct(op);
5065 
5066     outs() << "\t\t\t     name ";
5067     sym_name = get_symbol_64(offset + offsetof(struct objc_property64, name), S,
5068                              info, n_value, op.name);
5069     if (n_value != 0) {
5070       if (info->verbose && sym_name != nullptr)
5071         outs() << sym_name;
5072       else
5073         outs() << format("0x%" PRIx64, n_value);
5074       if (op.name != 0)
5075         outs() << " + " << format("0x%" PRIx64, op.name);
5076     } else
5077       outs() << format("0x%" PRIx64, op.name);
5078     name = get_pointer_64(op.name + n_value, xoffset, left, xS, info);
5079     if (name != nullptr)
5080       outs() << format(" %.*s", left, name);
5081     outs() << "\n";
5082 
5083     outs() << "\t\t\tattributes ";
5084     sym_name =
5085         get_symbol_64(offset + offsetof(struct objc_property64, attributes), S,
5086                       info, n_value, op.attributes);
5087     if (n_value != 0) {
5088       if (info->verbose && sym_name != nullptr)
5089         outs() << sym_name;
5090       else
5091         outs() << format("0x%" PRIx64, n_value);
5092       if (op.attributes != 0)
5093         outs() << " + " << format("0x%" PRIx64, op.attributes);
5094     } else
5095       outs() << format("0x%" PRIx64, op.attributes);
5096     name = get_pointer_64(op.attributes + n_value, xoffset, left, xS, info);
5097     if (name != nullptr)
5098       outs() << format(" %.*s", left, name);
5099     outs() << "\n";
5100 
5101     p += sizeof(struct objc_property64);
5102     offset += sizeof(struct objc_property64);
5103   }
5104 }
5105 
print_objc_property_list32(uint32_t p,struct DisassembleInfo * info)5106 static void print_objc_property_list32(uint32_t p,
5107                                        struct DisassembleInfo *info) {
5108   struct objc_property_list32 opl;
5109   struct objc_property32 op;
5110   const char *r;
5111   uint32_t offset, xoffset, left, j;
5112   SectionRef S, xS;
5113   const char *name;
5114 
5115   r = get_pointer_32(p, offset, left, S, info);
5116   if (r == nullptr)
5117     return;
5118   memset(&opl, '\0', sizeof(struct objc_property_list32));
5119   if (left < sizeof(struct objc_property_list32)) {
5120     memcpy(&opl, r, left);
5121     outs() << "   (objc_property_list entends past the end of the section)\n";
5122   } else
5123     memcpy(&opl, r, sizeof(struct objc_property_list32));
5124   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5125     swapStruct(opl);
5126   outs() << "                    entsize " << opl.entsize << "\n";
5127   outs() << "                      count " << opl.count << "\n";
5128 
5129   p += sizeof(struct objc_property_list32);
5130   offset += sizeof(struct objc_property_list32);
5131   for (j = 0; j < opl.count; j++) {
5132     r = get_pointer_32(p, offset, left, S, info);
5133     if (r == nullptr)
5134       return;
5135     memset(&op, '\0', sizeof(struct objc_property32));
5136     if (left < sizeof(struct objc_property32)) {
5137       memcpy(&op, r, left);
5138       outs() << "   (objc_property entends past the end of the section)\n";
5139     } else
5140       memcpy(&op, r, sizeof(struct objc_property32));
5141     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5142       swapStruct(op);
5143 
5144     outs() << "\t\t\t     name " << format("0x%" PRIx32, op.name);
5145     name = get_pointer_32(op.name, xoffset, left, xS, info);
5146     if (name != nullptr)
5147       outs() << format(" %.*s", left, name);
5148     outs() << "\n";
5149 
5150     outs() << "\t\t\tattributes " << format("0x%" PRIx32, op.attributes);
5151     name = get_pointer_32(op.attributes, xoffset, left, xS, info);
5152     if (name != nullptr)
5153       outs() << format(" %.*s", left, name);
5154     outs() << "\n";
5155 
5156     p += sizeof(struct objc_property32);
5157     offset += sizeof(struct objc_property32);
5158   }
5159 }
5160 
print_class_ro64_t(uint64_t p,struct DisassembleInfo * info,bool & is_meta_class)5161 static bool print_class_ro64_t(uint64_t p, struct DisassembleInfo *info,
5162                                bool &is_meta_class) {
5163   struct class_ro64_t cro;
5164   const char *r;
5165   uint32_t offset, xoffset, left;
5166   SectionRef S, xS;
5167   const char *name, *sym_name;
5168   uint64_t n_value;
5169 
5170   r = get_pointer_64(p, offset, left, S, info);
5171   if (r == nullptr || left < sizeof(struct class_ro64_t))
5172     return false;
5173   memcpy(&cro, r, sizeof(struct class_ro64_t));
5174   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5175     swapStruct(cro);
5176   outs() << "                    flags " << format("0x%" PRIx32, cro.flags);
5177   if (cro.flags & RO_META)
5178     outs() << " RO_META";
5179   if (cro.flags & RO_ROOT)
5180     outs() << " RO_ROOT";
5181   if (cro.flags & RO_HAS_CXX_STRUCTORS)
5182     outs() << " RO_HAS_CXX_STRUCTORS";
5183   outs() << "\n";
5184   outs() << "            instanceStart " << cro.instanceStart << "\n";
5185   outs() << "             instanceSize " << cro.instanceSize << "\n";
5186   outs() << "                 reserved " << format("0x%" PRIx32, cro.reserved)
5187          << "\n";
5188   outs() << "               ivarLayout " << format("0x%" PRIx64, cro.ivarLayout)
5189          << "\n";
5190   print_layout_map64(cro.ivarLayout, info);
5191 
5192   outs() << "                     name ";
5193   sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, name), S,
5194                            info, n_value, cro.name);
5195   if (n_value != 0) {
5196     if (info->verbose && sym_name != nullptr)
5197       outs() << sym_name;
5198     else
5199       outs() << format("0x%" PRIx64, n_value);
5200     if (cro.name != 0)
5201       outs() << " + " << format("0x%" PRIx64, cro.name);
5202   } else
5203     outs() << format("0x%" PRIx64, cro.name);
5204   name = get_pointer_64(cro.name + n_value, xoffset, left, xS, info);
5205   if (name != nullptr)
5206     outs() << format(" %.*s", left, name);
5207   outs() << "\n";
5208 
5209   outs() << "              baseMethods ";
5210   sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, baseMethods),
5211                            S, info, n_value, cro.baseMethods);
5212   if (n_value != 0) {
5213     if (info->verbose && sym_name != nullptr)
5214       outs() << sym_name;
5215     else
5216       outs() << format("0x%" PRIx64, n_value);
5217     if (cro.baseMethods != 0)
5218       outs() << " + " << format("0x%" PRIx64, cro.baseMethods);
5219   } else
5220     outs() << format("0x%" PRIx64, cro.baseMethods);
5221   outs() << " (struct method_list_t *)\n";
5222   if (cro.baseMethods + n_value != 0)
5223     print_method_list64_t(cro.baseMethods + n_value, info, "");
5224 
5225   outs() << "            baseProtocols ";
5226   sym_name =
5227       get_symbol_64(offset + offsetof(struct class_ro64_t, baseProtocols), S,
5228                     info, n_value, cro.baseProtocols);
5229   if (n_value != 0) {
5230     if (info->verbose && sym_name != nullptr)
5231       outs() << sym_name;
5232     else
5233       outs() << format("0x%" PRIx64, n_value);
5234     if (cro.baseProtocols != 0)
5235       outs() << " + " << format("0x%" PRIx64, cro.baseProtocols);
5236   } else
5237     outs() << format("0x%" PRIx64, cro.baseProtocols);
5238   outs() << "\n";
5239   if (cro.baseProtocols + n_value != 0)
5240     print_protocol_list64_t(cro.baseProtocols + n_value, info);
5241 
5242   outs() << "                    ivars ";
5243   sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, ivars), S,
5244                            info, n_value, cro.ivars);
5245   if (n_value != 0) {
5246     if (info->verbose && sym_name != nullptr)
5247       outs() << sym_name;
5248     else
5249       outs() << format("0x%" PRIx64, n_value);
5250     if (cro.ivars != 0)
5251       outs() << " + " << format("0x%" PRIx64, cro.ivars);
5252   } else
5253     outs() << format("0x%" PRIx64, cro.ivars);
5254   outs() << "\n";
5255   if (cro.ivars + n_value != 0)
5256     print_ivar_list64_t(cro.ivars + n_value, info);
5257 
5258   outs() << "           weakIvarLayout ";
5259   sym_name =
5260       get_symbol_64(offset + offsetof(struct class_ro64_t, weakIvarLayout), S,
5261                     info, n_value, cro.weakIvarLayout);
5262   if (n_value != 0) {
5263     if (info->verbose && sym_name != nullptr)
5264       outs() << sym_name;
5265     else
5266       outs() << format("0x%" PRIx64, n_value);
5267     if (cro.weakIvarLayout != 0)
5268       outs() << " + " << format("0x%" PRIx64, cro.weakIvarLayout);
5269   } else
5270     outs() << format("0x%" PRIx64, cro.weakIvarLayout);
5271   outs() << "\n";
5272   print_layout_map64(cro.weakIvarLayout + n_value, info);
5273 
5274   outs() << "           baseProperties ";
5275   sym_name =
5276       get_symbol_64(offset + offsetof(struct class_ro64_t, baseProperties), S,
5277                     info, n_value, cro.baseProperties);
5278   if (n_value != 0) {
5279     if (info->verbose && sym_name != nullptr)
5280       outs() << sym_name;
5281     else
5282       outs() << format("0x%" PRIx64, n_value);
5283     if (cro.baseProperties != 0)
5284       outs() << " + " << format("0x%" PRIx64, cro.baseProperties);
5285   } else
5286     outs() << format("0x%" PRIx64, cro.baseProperties);
5287   outs() << "\n";
5288   if (cro.baseProperties + n_value != 0)
5289     print_objc_property_list64(cro.baseProperties + n_value, info);
5290 
5291   is_meta_class = (cro.flags & RO_META) != 0;
5292   return true;
5293 }
5294 
print_class_ro32_t(uint32_t p,struct DisassembleInfo * info,bool & is_meta_class)5295 static bool print_class_ro32_t(uint32_t p, struct DisassembleInfo *info,
5296                                bool &is_meta_class) {
5297   struct class_ro32_t cro;
5298   const char *r;
5299   uint32_t offset, xoffset, left;
5300   SectionRef S, xS;
5301   const char *name;
5302 
5303   r = get_pointer_32(p, offset, left, S, info);
5304   if (r == nullptr)
5305     return false;
5306   memset(&cro, '\0', sizeof(struct class_ro32_t));
5307   if (left < sizeof(struct class_ro32_t)) {
5308     memcpy(&cro, r, left);
5309     outs() << "   (class_ro_t entends past the end of the section)\n";
5310   } else
5311     memcpy(&cro, r, sizeof(struct class_ro32_t));
5312   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5313     swapStruct(cro);
5314   outs() << "                    flags " << format("0x%" PRIx32, cro.flags);
5315   if (cro.flags & RO_META)
5316     outs() << " RO_META";
5317   if (cro.flags & RO_ROOT)
5318     outs() << " RO_ROOT";
5319   if (cro.flags & RO_HAS_CXX_STRUCTORS)
5320     outs() << " RO_HAS_CXX_STRUCTORS";
5321   outs() << "\n";
5322   outs() << "            instanceStart " << cro.instanceStart << "\n";
5323   outs() << "             instanceSize " << cro.instanceSize << "\n";
5324   outs() << "               ivarLayout " << format("0x%" PRIx32, cro.ivarLayout)
5325          << "\n";
5326   print_layout_map32(cro.ivarLayout, info);
5327 
5328   outs() << "                     name " << format("0x%" PRIx32, cro.name);
5329   name = get_pointer_32(cro.name, xoffset, left, xS, info);
5330   if (name != nullptr)
5331     outs() << format(" %.*s", left, name);
5332   outs() << "\n";
5333 
5334   outs() << "              baseMethods "
5335          << format("0x%" PRIx32, cro.baseMethods)
5336          << " (struct method_list_t *)\n";
5337   if (cro.baseMethods != 0)
5338     print_method_list32_t(cro.baseMethods, info, "");
5339 
5340   outs() << "            baseProtocols "
5341          << format("0x%" PRIx32, cro.baseProtocols) << "\n";
5342   if (cro.baseProtocols != 0)
5343     print_protocol_list32_t(cro.baseProtocols, info);
5344   outs() << "                    ivars " << format("0x%" PRIx32, cro.ivars)
5345          << "\n";
5346   if (cro.ivars != 0)
5347     print_ivar_list32_t(cro.ivars, info);
5348   outs() << "           weakIvarLayout "
5349          << format("0x%" PRIx32, cro.weakIvarLayout) << "\n";
5350   print_layout_map32(cro.weakIvarLayout, info);
5351   outs() << "           baseProperties "
5352          << format("0x%" PRIx32, cro.baseProperties) << "\n";
5353   if (cro.baseProperties != 0)
5354     print_objc_property_list32(cro.baseProperties, info);
5355   is_meta_class = (cro.flags & RO_META) != 0;
5356   return true;
5357 }
5358 
print_class64_t(uint64_t p,struct DisassembleInfo * info)5359 static void print_class64_t(uint64_t p, struct DisassembleInfo *info) {
5360   struct class64_t c;
5361   const char *r;
5362   uint32_t offset, left;
5363   SectionRef S;
5364   const char *name;
5365   uint64_t isa_n_value, n_value;
5366 
5367   r = get_pointer_64(p, offset, left, S, info);
5368   if (r == nullptr || left < sizeof(struct class64_t))
5369     return;
5370   memcpy(&c, r, sizeof(struct class64_t));
5371   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5372     swapStruct(c);
5373 
5374   outs() << "           isa " << format("0x%" PRIx64, c.isa);
5375   name = get_symbol_64(offset + offsetof(struct class64_t, isa), S, info,
5376                        isa_n_value, c.isa);
5377   if (name != nullptr)
5378     outs() << " " << name;
5379   outs() << "\n";
5380 
5381   outs() << "    superclass " << format("0x%" PRIx64, c.superclass);
5382   name = get_symbol_64(offset + offsetof(struct class64_t, superclass), S, info,
5383                        n_value, c.superclass);
5384   if (name != nullptr)
5385     outs() << " " << name;
5386   else {
5387     name = get_dyld_bind_info_symbolname(S.getAddress() +
5388              offset + offsetof(struct class64_t, superclass), info);
5389     if (name != nullptr)
5390       outs() << " " << name;
5391   }
5392   outs() << "\n";
5393 
5394   outs() << "         cache " << format("0x%" PRIx64, c.cache);
5395   name = get_symbol_64(offset + offsetof(struct class64_t, cache), S, info,
5396                        n_value, c.cache);
5397   if (name != nullptr)
5398     outs() << " " << name;
5399   outs() << "\n";
5400 
5401   outs() << "        vtable " << format("0x%" PRIx64, c.vtable);
5402   name = get_symbol_64(offset + offsetof(struct class64_t, vtable), S, info,
5403                        n_value, c.vtable);
5404   if (name != nullptr)
5405     outs() << " " << name;
5406   outs() << "\n";
5407 
5408   name = get_symbol_64(offset + offsetof(struct class64_t, data), S, info,
5409                        n_value, c.data);
5410   outs() << "          data ";
5411   if (n_value != 0) {
5412     if (info->verbose && name != nullptr)
5413       outs() << name;
5414     else
5415       outs() << format("0x%" PRIx64, n_value);
5416     if (c.data != 0)
5417       outs() << " + " << format("0x%" PRIx64, c.data);
5418   } else
5419     outs() << format("0x%" PRIx64, c.data);
5420   outs() << " (struct class_ro_t *)";
5421 
5422   // This is a Swift class if some of the low bits of the pointer are set.
5423   if ((c.data + n_value) & 0x7)
5424     outs() << " Swift class";
5425   outs() << "\n";
5426   bool is_meta_class;
5427   if (!print_class_ro64_t((c.data + n_value) & ~0x7, info, is_meta_class))
5428     return;
5429 
5430   if (!is_meta_class &&
5431       c.isa + isa_n_value != p &&
5432       c.isa + isa_n_value != 0 &&
5433       info->depth < 100) {
5434       info->depth++;
5435       outs() << "Meta Class\n";
5436       print_class64_t(c.isa + isa_n_value, info);
5437   }
5438 }
5439 
print_class32_t(uint32_t p,struct DisassembleInfo * info)5440 static void print_class32_t(uint32_t p, struct DisassembleInfo *info) {
5441   struct class32_t c;
5442   const char *r;
5443   uint32_t offset, left;
5444   SectionRef S;
5445   const char *name;
5446 
5447   r = get_pointer_32(p, offset, left, S, info);
5448   if (r == nullptr)
5449     return;
5450   memset(&c, '\0', sizeof(struct class32_t));
5451   if (left < sizeof(struct class32_t)) {
5452     memcpy(&c, r, left);
5453     outs() << "   (class_t entends past the end of the section)\n";
5454   } else
5455     memcpy(&c, r, sizeof(struct class32_t));
5456   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5457     swapStruct(c);
5458 
5459   outs() << "           isa " << format("0x%" PRIx32, c.isa);
5460   name =
5461       get_symbol_32(offset + offsetof(struct class32_t, isa), S, info, c.isa);
5462   if (name != nullptr)
5463     outs() << " " << name;
5464   outs() << "\n";
5465 
5466   outs() << "    superclass " << format("0x%" PRIx32, c.superclass);
5467   name = get_symbol_32(offset + offsetof(struct class32_t, superclass), S, info,
5468                        c.superclass);
5469   if (name != nullptr)
5470     outs() << " " << name;
5471   outs() << "\n";
5472 
5473   outs() << "         cache " << format("0x%" PRIx32, c.cache);
5474   name = get_symbol_32(offset + offsetof(struct class32_t, cache), S, info,
5475                        c.cache);
5476   if (name != nullptr)
5477     outs() << " " << name;
5478   outs() << "\n";
5479 
5480   outs() << "        vtable " << format("0x%" PRIx32, c.vtable);
5481   name = get_symbol_32(offset + offsetof(struct class32_t, vtable), S, info,
5482                        c.vtable);
5483   if (name != nullptr)
5484     outs() << " " << name;
5485   outs() << "\n";
5486 
5487   name =
5488       get_symbol_32(offset + offsetof(struct class32_t, data), S, info, c.data);
5489   outs() << "          data " << format("0x%" PRIx32, c.data)
5490          << " (struct class_ro_t *)";
5491 
5492   // This is a Swift class if some of the low bits of the pointer are set.
5493   if (c.data & 0x3)
5494     outs() << " Swift class";
5495   outs() << "\n";
5496   bool is_meta_class;
5497   if (!print_class_ro32_t(c.data & ~0x3, info, is_meta_class))
5498     return;
5499 
5500   if (!is_meta_class) {
5501     outs() << "Meta Class\n";
5502     print_class32_t(c.isa, info);
5503   }
5504 }
5505 
print_objc_class_t(struct objc_class_t * objc_class,struct DisassembleInfo * info)5506 static void print_objc_class_t(struct objc_class_t *objc_class,
5507                                struct DisassembleInfo *info) {
5508   uint32_t offset, left, xleft;
5509   const char *name, *p, *ivar_list;
5510   SectionRef S;
5511   int32_t i;
5512   struct objc_ivar_list_t objc_ivar_list;
5513   struct objc_ivar_t ivar;
5514 
5515   outs() << "\t\t      isa " << format("0x%08" PRIx32, objc_class->isa);
5516   if (info->verbose && CLS_GETINFO(objc_class, CLS_META)) {
5517     name = get_pointer_32(objc_class->isa, offset, left, S, info, true);
5518     if (name != nullptr)
5519       outs() << format(" %.*s", left, name);
5520     else
5521       outs() << " (not in an __OBJC section)";
5522   }
5523   outs() << "\n";
5524 
5525   outs() << "\t      super_class "
5526          << format("0x%08" PRIx32, objc_class->super_class);
5527   if (info->verbose) {
5528     name = get_pointer_32(objc_class->super_class, offset, left, S, info, true);
5529     if (name != nullptr)
5530       outs() << format(" %.*s", left, name);
5531     else
5532       outs() << " (not in an __OBJC section)";
5533   }
5534   outs() << "\n";
5535 
5536   outs() << "\t\t     name " << format("0x%08" PRIx32, objc_class->name);
5537   if (info->verbose) {
5538     name = get_pointer_32(objc_class->name, offset, left, S, info, true);
5539     if (name != nullptr)
5540       outs() << format(" %.*s", left, name);
5541     else
5542       outs() << " (not in an __OBJC section)";
5543   }
5544   outs() << "\n";
5545 
5546   outs() << "\t\t  version " << format("0x%08" PRIx32, objc_class->version)
5547          << "\n";
5548 
5549   outs() << "\t\t     info " << format("0x%08" PRIx32, objc_class->info);
5550   if (info->verbose) {
5551     if (CLS_GETINFO(objc_class, CLS_CLASS))
5552       outs() << " CLS_CLASS";
5553     else if (CLS_GETINFO(objc_class, CLS_META))
5554       outs() << " CLS_META";
5555   }
5556   outs() << "\n";
5557 
5558   outs() << "\t    instance_size "
5559          << format("0x%08" PRIx32, objc_class->instance_size) << "\n";
5560 
5561   p = get_pointer_32(objc_class->ivars, offset, left, S, info, true);
5562   outs() << "\t\t    ivars " << format("0x%08" PRIx32, objc_class->ivars);
5563   if (p != nullptr) {
5564     if (left > sizeof(struct objc_ivar_list_t)) {
5565       outs() << "\n";
5566       memcpy(&objc_ivar_list, p, sizeof(struct objc_ivar_list_t));
5567     } else {
5568       outs() << " (entends past the end of the section)\n";
5569       memset(&objc_ivar_list, '\0', sizeof(struct objc_ivar_list_t));
5570       memcpy(&objc_ivar_list, p, left);
5571     }
5572     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5573       swapStruct(objc_ivar_list);
5574     outs() << "\t\t       ivar_count " << objc_ivar_list.ivar_count << "\n";
5575     ivar_list = p + sizeof(struct objc_ivar_list_t);
5576     for (i = 0; i < objc_ivar_list.ivar_count; i++) {
5577       if ((i + 1) * sizeof(struct objc_ivar_t) > left) {
5578         outs() << "\t\t remaining ivar's extend past the of the section\n";
5579         break;
5580       }
5581       memcpy(&ivar, ivar_list + i * sizeof(struct objc_ivar_t),
5582              sizeof(struct objc_ivar_t));
5583       if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5584         swapStruct(ivar);
5585 
5586       outs() << "\t\t\tivar_name " << format("0x%08" PRIx32, ivar.ivar_name);
5587       if (info->verbose) {
5588         name = get_pointer_32(ivar.ivar_name, offset, xleft, S, info, true);
5589         if (name != nullptr)
5590           outs() << format(" %.*s", xleft, name);
5591         else
5592           outs() << " (not in an __OBJC section)";
5593       }
5594       outs() << "\n";
5595 
5596       outs() << "\t\t\tivar_type " << format("0x%08" PRIx32, ivar.ivar_type);
5597       if (info->verbose) {
5598         name = get_pointer_32(ivar.ivar_type, offset, xleft, S, info, true);
5599         if (name != nullptr)
5600           outs() << format(" %.*s", xleft, name);
5601         else
5602           outs() << " (not in an __OBJC section)";
5603       }
5604       outs() << "\n";
5605 
5606       outs() << "\t\t      ivar_offset "
5607              << format("0x%08" PRIx32, ivar.ivar_offset) << "\n";
5608     }
5609   } else {
5610     outs() << " (not in an __OBJC section)\n";
5611   }
5612 
5613   outs() << "\t\t  methods " << format("0x%08" PRIx32, objc_class->methodLists);
5614   if (print_method_list(objc_class->methodLists, info))
5615     outs() << " (not in an __OBJC section)\n";
5616 
5617   outs() << "\t\t    cache " << format("0x%08" PRIx32, objc_class->cache)
5618          << "\n";
5619 
5620   outs() << "\t\tprotocols " << format("0x%08" PRIx32, objc_class->protocols);
5621   if (print_protocol_list(objc_class->protocols, 16, info))
5622     outs() << " (not in an __OBJC section)\n";
5623 }
5624 
print_objc_objc_category_t(struct objc_category_t * objc_category,struct DisassembleInfo * info)5625 static void print_objc_objc_category_t(struct objc_category_t *objc_category,
5626                                        struct DisassembleInfo *info) {
5627   uint32_t offset, left;
5628   const char *name;
5629   SectionRef S;
5630 
5631   outs() << "\t       category name "
5632          << format("0x%08" PRIx32, objc_category->category_name);
5633   if (info->verbose) {
5634     name = get_pointer_32(objc_category->category_name, offset, left, S, info,
5635                           true);
5636     if (name != nullptr)
5637       outs() << format(" %.*s", left, name);
5638     else
5639       outs() << " (not in an __OBJC section)";
5640   }
5641   outs() << "\n";
5642 
5643   outs() << "\t\t  class name "
5644          << format("0x%08" PRIx32, objc_category->class_name);
5645   if (info->verbose) {
5646     name =
5647         get_pointer_32(objc_category->class_name, offset, left, S, info, true);
5648     if (name != nullptr)
5649       outs() << format(" %.*s", left, name);
5650     else
5651       outs() << " (not in an __OBJC section)";
5652   }
5653   outs() << "\n";
5654 
5655   outs() << "\t    instance methods "
5656          << format("0x%08" PRIx32, objc_category->instance_methods);
5657   if (print_method_list(objc_category->instance_methods, info))
5658     outs() << " (not in an __OBJC section)\n";
5659 
5660   outs() << "\t       class methods "
5661          << format("0x%08" PRIx32, objc_category->class_methods);
5662   if (print_method_list(objc_category->class_methods, info))
5663     outs() << " (not in an __OBJC section)\n";
5664 }
5665 
print_category64_t(uint64_t p,struct DisassembleInfo * info)5666 static void print_category64_t(uint64_t p, struct DisassembleInfo *info) {
5667   struct category64_t c;
5668   const char *r;
5669   uint32_t offset, xoffset, left;
5670   SectionRef S, xS;
5671   const char *name, *sym_name;
5672   uint64_t n_value;
5673 
5674   r = get_pointer_64(p, offset, left, S, info);
5675   if (r == nullptr)
5676     return;
5677   memset(&c, '\0', sizeof(struct category64_t));
5678   if (left < sizeof(struct category64_t)) {
5679     memcpy(&c, r, left);
5680     outs() << "   (category_t entends past the end of the section)\n";
5681   } else
5682     memcpy(&c, r, sizeof(struct category64_t));
5683   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5684     swapStruct(c);
5685 
5686   outs() << "              name ";
5687   sym_name = get_symbol_64(offset + offsetof(struct category64_t, name), S,
5688                            info, n_value, c.name);
5689   if (n_value != 0) {
5690     if (info->verbose && sym_name != nullptr)
5691       outs() << sym_name;
5692     else
5693       outs() << format("0x%" PRIx64, n_value);
5694     if (c.name != 0)
5695       outs() << " + " << format("0x%" PRIx64, c.name);
5696   } else
5697     outs() << format("0x%" PRIx64, c.name);
5698   name = get_pointer_64(c.name + n_value, xoffset, left, xS, info);
5699   if (name != nullptr)
5700     outs() << format(" %.*s", left, name);
5701   outs() << "\n";
5702 
5703   outs() << "               cls ";
5704   sym_name = get_symbol_64(offset + offsetof(struct category64_t, cls), S, info,
5705                            n_value, c.cls);
5706   if (n_value != 0) {
5707     if (info->verbose && sym_name != nullptr)
5708       outs() << sym_name;
5709     else
5710       outs() << format("0x%" PRIx64, n_value);
5711     if (c.cls != 0)
5712       outs() << " + " << format("0x%" PRIx64, c.cls);
5713   } else
5714     outs() << format("0x%" PRIx64, c.cls);
5715   outs() << "\n";
5716   if (c.cls + n_value != 0)
5717     print_class64_t(c.cls + n_value, info);
5718 
5719   outs() << "   instanceMethods ";
5720   sym_name =
5721       get_symbol_64(offset + offsetof(struct category64_t, instanceMethods), S,
5722                     info, n_value, c.instanceMethods);
5723   if (n_value != 0) {
5724     if (info->verbose && sym_name != nullptr)
5725       outs() << sym_name;
5726     else
5727       outs() << format("0x%" PRIx64, n_value);
5728     if (c.instanceMethods != 0)
5729       outs() << " + " << format("0x%" PRIx64, c.instanceMethods);
5730   } else
5731     outs() << format("0x%" PRIx64, c.instanceMethods);
5732   outs() << "\n";
5733   if (c.instanceMethods + n_value != 0)
5734     print_method_list64_t(c.instanceMethods + n_value, info, "");
5735 
5736   outs() << "      classMethods ";
5737   sym_name = get_symbol_64(offset + offsetof(struct category64_t, classMethods),
5738                            S, info, n_value, c.classMethods);
5739   if (n_value != 0) {
5740     if (info->verbose && sym_name != nullptr)
5741       outs() << sym_name;
5742     else
5743       outs() << format("0x%" PRIx64, n_value);
5744     if (c.classMethods != 0)
5745       outs() << " + " << format("0x%" PRIx64, c.classMethods);
5746   } else
5747     outs() << format("0x%" PRIx64, c.classMethods);
5748   outs() << "\n";
5749   if (c.classMethods + n_value != 0)
5750     print_method_list64_t(c.classMethods + n_value, info, "");
5751 
5752   outs() << "         protocols ";
5753   sym_name = get_symbol_64(offset + offsetof(struct category64_t, protocols), S,
5754                            info, n_value, c.protocols);
5755   if (n_value != 0) {
5756     if (info->verbose && sym_name != nullptr)
5757       outs() << sym_name;
5758     else
5759       outs() << format("0x%" PRIx64, n_value);
5760     if (c.protocols != 0)
5761       outs() << " + " << format("0x%" PRIx64, c.protocols);
5762   } else
5763     outs() << format("0x%" PRIx64, c.protocols);
5764   outs() << "\n";
5765   if (c.protocols + n_value != 0)
5766     print_protocol_list64_t(c.protocols + n_value, info);
5767 
5768   outs() << "instanceProperties ";
5769   sym_name =
5770       get_symbol_64(offset + offsetof(struct category64_t, instanceProperties),
5771                     S, info, n_value, c.instanceProperties);
5772   if (n_value != 0) {
5773     if (info->verbose && sym_name != nullptr)
5774       outs() << sym_name;
5775     else
5776       outs() << format("0x%" PRIx64, n_value);
5777     if (c.instanceProperties != 0)
5778       outs() << " + " << format("0x%" PRIx64, c.instanceProperties);
5779   } else
5780     outs() << format("0x%" PRIx64, c.instanceProperties);
5781   outs() << "\n";
5782   if (c.instanceProperties + n_value != 0)
5783     print_objc_property_list64(c.instanceProperties + n_value, info);
5784 }
5785 
print_category32_t(uint32_t p,struct DisassembleInfo * info)5786 static void print_category32_t(uint32_t p, struct DisassembleInfo *info) {
5787   struct category32_t c;
5788   const char *r;
5789   uint32_t offset, left;
5790   SectionRef S, xS;
5791   const char *name;
5792 
5793   r = get_pointer_32(p, offset, left, S, info);
5794   if (r == nullptr)
5795     return;
5796   memset(&c, '\0', sizeof(struct category32_t));
5797   if (left < sizeof(struct category32_t)) {
5798     memcpy(&c, r, left);
5799     outs() << "   (category_t entends past the end of the section)\n";
5800   } else
5801     memcpy(&c, r, sizeof(struct category32_t));
5802   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5803     swapStruct(c);
5804 
5805   outs() << "              name " << format("0x%" PRIx32, c.name);
5806   name = get_symbol_32(offset + offsetof(struct category32_t, name), S, info,
5807                        c.name);
5808   if (name)
5809     outs() << " " << name;
5810   outs() << "\n";
5811 
5812   outs() << "               cls " << format("0x%" PRIx32, c.cls) << "\n";
5813   if (c.cls != 0)
5814     print_class32_t(c.cls, info);
5815   outs() << "   instanceMethods " << format("0x%" PRIx32, c.instanceMethods)
5816          << "\n";
5817   if (c.instanceMethods != 0)
5818     print_method_list32_t(c.instanceMethods, info, "");
5819   outs() << "      classMethods " << format("0x%" PRIx32, c.classMethods)
5820          << "\n";
5821   if (c.classMethods != 0)
5822     print_method_list32_t(c.classMethods, info, "");
5823   outs() << "         protocols " << format("0x%" PRIx32, c.protocols) << "\n";
5824   if (c.protocols != 0)
5825     print_protocol_list32_t(c.protocols, info);
5826   outs() << "instanceProperties " << format("0x%" PRIx32, c.instanceProperties)
5827          << "\n";
5828   if (c.instanceProperties != 0)
5829     print_objc_property_list32(c.instanceProperties, info);
5830 }
5831 
print_message_refs64(SectionRef S,struct DisassembleInfo * info)5832 static void print_message_refs64(SectionRef S, struct DisassembleInfo *info) {
5833   uint32_t i, left, offset, xoffset;
5834   uint64_t p, n_value;
5835   struct message_ref64 mr;
5836   const char *name, *sym_name;
5837   const char *r;
5838   SectionRef xS;
5839 
5840   if (S == SectionRef())
5841     return;
5842 
5843   StringRef SectName;
5844   Expected<StringRef> SecNameOrErr = S.getName();
5845   if (SecNameOrErr)
5846     SectName = *SecNameOrErr;
5847   else
5848     consumeError(SecNameOrErr.takeError());
5849 
5850   DataRefImpl Ref = S.getRawDataRefImpl();
5851   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5852   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5853   offset = 0;
5854   for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) {
5855     p = S.getAddress() + i;
5856     r = get_pointer_64(p, offset, left, S, info);
5857     if (r == nullptr)
5858       return;
5859     memset(&mr, '\0', sizeof(struct message_ref64));
5860     if (left < sizeof(struct message_ref64)) {
5861       memcpy(&mr, r, left);
5862       outs() << "   (message_ref entends past the end of the section)\n";
5863     } else
5864       memcpy(&mr, r, sizeof(struct message_ref64));
5865     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5866       swapStruct(mr);
5867 
5868     outs() << "  imp ";
5869     name = get_symbol_64(offset + offsetof(struct message_ref64, imp), S, info,
5870                          n_value, mr.imp);
5871     if (n_value != 0) {
5872       outs() << format("0x%" PRIx64, n_value) << " ";
5873       if (mr.imp != 0)
5874         outs() << "+ " << format("0x%" PRIx64, mr.imp) << " ";
5875     } else
5876       outs() << format("0x%" PRIx64, mr.imp) << " ";
5877     if (name != nullptr)
5878       outs() << " " << name;
5879     outs() << "\n";
5880 
5881     outs() << "  sel ";
5882     sym_name = get_symbol_64(offset + offsetof(struct message_ref64, sel), S,
5883                              info, n_value, mr.sel);
5884     if (n_value != 0) {
5885       if (info->verbose && sym_name != nullptr)
5886         outs() << sym_name;
5887       else
5888         outs() << format("0x%" PRIx64, n_value);
5889       if (mr.sel != 0)
5890         outs() << " + " << format("0x%" PRIx64, mr.sel);
5891     } else
5892       outs() << format("0x%" PRIx64, mr.sel);
5893     name = get_pointer_64(mr.sel + n_value, xoffset, left, xS, info);
5894     if (name != nullptr)
5895       outs() << format(" %.*s", left, name);
5896     outs() << "\n";
5897 
5898     offset += sizeof(struct message_ref64);
5899   }
5900 }
5901 
print_message_refs32(SectionRef S,struct DisassembleInfo * info)5902 static void print_message_refs32(SectionRef S, struct DisassembleInfo *info) {
5903   uint32_t i, left, offset, xoffset, p;
5904   struct message_ref32 mr;
5905   const char *name, *r;
5906   SectionRef xS;
5907 
5908   if (S == SectionRef())
5909     return;
5910 
5911   StringRef SectName;
5912   Expected<StringRef> SecNameOrErr = S.getName();
5913   if (SecNameOrErr)
5914     SectName = *SecNameOrErr;
5915   else
5916     consumeError(SecNameOrErr.takeError());
5917 
5918   DataRefImpl Ref = S.getRawDataRefImpl();
5919   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5920   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5921   offset = 0;
5922   for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) {
5923     p = S.getAddress() + i;
5924     r = get_pointer_32(p, offset, left, S, info);
5925     if (r == nullptr)
5926       return;
5927     memset(&mr, '\0', sizeof(struct message_ref32));
5928     if (left < sizeof(struct message_ref32)) {
5929       memcpy(&mr, r, left);
5930       outs() << "   (message_ref entends past the end of the section)\n";
5931     } else
5932       memcpy(&mr, r, sizeof(struct message_ref32));
5933     if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5934       swapStruct(mr);
5935 
5936     outs() << "  imp " << format("0x%" PRIx32, mr.imp);
5937     name = get_symbol_32(offset + offsetof(struct message_ref32, imp), S, info,
5938                          mr.imp);
5939     if (name != nullptr)
5940       outs() << " " << name;
5941     outs() << "\n";
5942 
5943     outs() << "  sel " << format("0x%" PRIx32, mr.sel);
5944     name = get_pointer_32(mr.sel, xoffset, left, xS, info);
5945     if (name != nullptr)
5946       outs() << " " << name;
5947     outs() << "\n";
5948 
5949     offset += sizeof(struct message_ref32);
5950   }
5951 }
5952 
print_image_info64(SectionRef S,struct DisassembleInfo * info)5953 static void print_image_info64(SectionRef S, struct DisassembleInfo *info) {
5954   uint32_t left, offset, swift_version;
5955   uint64_t p;
5956   struct objc_image_info64 o;
5957   const char *r;
5958 
5959   if (S == SectionRef())
5960     return;
5961 
5962   StringRef SectName;
5963   Expected<StringRef> SecNameOrErr = S.getName();
5964   if (SecNameOrErr)
5965     SectName = *SecNameOrErr;
5966   else
5967     consumeError(SecNameOrErr.takeError());
5968 
5969   DataRefImpl Ref = S.getRawDataRefImpl();
5970   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
5971   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
5972   p = S.getAddress();
5973   r = get_pointer_64(p, offset, left, S, info);
5974   if (r == nullptr)
5975     return;
5976   memset(&o, '\0', sizeof(struct objc_image_info64));
5977   if (left < sizeof(struct objc_image_info64)) {
5978     memcpy(&o, r, left);
5979     outs() << "   (objc_image_info entends past the end of the section)\n";
5980   } else
5981     memcpy(&o, r, sizeof(struct objc_image_info64));
5982   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
5983     swapStruct(o);
5984   outs() << "  version " << o.version << "\n";
5985   outs() << "    flags " << format("0x%" PRIx32, o.flags);
5986   if (o.flags & OBJC_IMAGE_IS_REPLACEMENT)
5987     outs() << " OBJC_IMAGE_IS_REPLACEMENT";
5988   if (o.flags & OBJC_IMAGE_SUPPORTS_GC)
5989     outs() << " OBJC_IMAGE_SUPPORTS_GC";
5990   if (o.flags & OBJC_IMAGE_IS_SIMULATED)
5991     outs() << " OBJC_IMAGE_IS_SIMULATED";
5992   if (o.flags & OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES)
5993     outs() << " OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES";
5994   swift_version = (o.flags >> 8) & 0xff;
5995   if (swift_version != 0) {
5996     if (swift_version == 1)
5997       outs() << " Swift 1.0";
5998     else if (swift_version == 2)
5999       outs() << " Swift 1.1";
6000     else if(swift_version == 3)
6001       outs() << " Swift 2.0";
6002     else if(swift_version == 4)
6003       outs() << " Swift 3.0";
6004     else if(swift_version == 5)
6005       outs() << " Swift 4.0";
6006     else if(swift_version == 6)
6007       outs() << " Swift 4.1/Swift 4.2";
6008     else if(swift_version == 7)
6009       outs() << " Swift 5 or later";
6010     else
6011       outs() << " unknown future Swift version (" << swift_version << ")";
6012   }
6013   outs() << "\n";
6014 }
6015 
print_image_info32(SectionRef S,struct DisassembleInfo * info)6016 static void print_image_info32(SectionRef S, struct DisassembleInfo *info) {
6017   uint32_t left, offset, swift_version, p;
6018   struct objc_image_info32 o;
6019   const char *r;
6020 
6021   if (S == SectionRef())
6022     return;
6023 
6024   StringRef SectName;
6025   Expected<StringRef> SecNameOrErr = S.getName();
6026   if (SecNameOrErr)
6027     SectName = *SecNameOrErr;
6028   else
6029     consumeError(SecNameOrErr.takeError());
6030 
6031   DataRefImpl Ref = S.getRawDataRefImpl();
6032   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
6033   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
6034   p = S.getAddress();
6035   r = get_pointer_32(p, offset, left, S, info);
6036   if (r == nullptr)
6037     return;
6038   memset(&o, '\0', sizeof(struct objc_image_info32));
6039   if (left < sizeof(struct objc_image_info32)) {
6040     memcpy(&o, r, left);
6041     outs() << "   (objc_image_info entends past the end of the section)\n";
6042   } else
6043     memcpy(&o, r, sizeof(struct objc_image_info32));
6044   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
6045     swapStruct(o);
6046   outs() << "  version " << o.version << "\n";
6047   outs() << "    flags " << format("0x%" PRIx32, o.flags);
6048   if (o.flags & OBJC_IMAGE_IS_REPLACEMENT)
6049     outs() << " OBJC_IMAGE_IS_REPLACEMENT";
6050   if (o.flags & OBJC_IMAGE_SUPPORTS_GC)
6051     outs() << " OBJC_IMAGE_SUPPORTS_GC";
6052   swift_version = (o.flags >> 8) & 0xff;
6053   if (swift_version != 0) {
6054     if (swift_version == 1)
6055       outs() << " Swift 1.0";
6056     else if (swift_version == 2)
6057       outs() << " Swift 1.1";
6058     else if(swift_version == 3)
6059       outs() << " Swift 2.0";
6060     else if(swift_version == 4)
6061       outs() << " Swift 3.0";
6062     else if(swift_version == 5)
6063       outs() << " Swift 4.0";
6064     else if(swift_version == 6)
6065       outs() << " Swift 4.1/Swift 4.2";
6066     else if(swift_version == 7)
6067       outs() << " Swift 5 or later";
6068     else
6069       outs() << " unknown future Swift version (" << swift_version << ")";
6070   }
6071   outs() << "\n";
6072 }
6073 
print_image_info(SectionRef S,struct DisassembleInfo * info)6074 static void print_image_info(SectionRef S, struct DisassembleInfo *info) {
6075   uint32_t left, offset, p;
6076   struct imageInfo_t o;
6077   const char *r;
6078 
6079   StringRef SectName;
6080   Expected<StringRef> SecNameOrErr = S.getName();
6081   if (SecNameOrErr)
6082     SectName = *SecNameOrErr;
6083   else
6084     consumeError(SecNameOrErr.takeError());
6085 
6086   DataRefImpl Ref = S.getRawDataRefImpl();
6087   StringRef SegName = info->O->getSectionFinalSegmentName(Ref);
6088   outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
6089   p = S.getAddress();
6090   r = get_pointer_32(p, offset, left, S, info);
6091   if (r == nullptr)
6092     return;
6093   memset(&o, '\0', sizeof(struct imageInfo_t));
6094   if (left < sizeof(struct imageInfo_t)) {
6095     memcpy(&o, r, left);
6096     outs() << " (imageInfo entends past the end of the section)\n";
6097   } else
6098     memcpy(&o, r, sizeof(struct imageInfo_t));
6099   if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
6100     swapStruct(o);
6101   outs() << "  version " << o.version << "\n";
6102   outs() << "    flags " << format("0x%" PRIx32, o.flags);
6103   if (o.flags & 0x1)
6104     outs() << "  F&C";
6105   if (o.flags & 0x2)
6106     outs() << " GC";
6107   if (o.flags & 0x4)
6108     outs() << " GC-only";
6109   else
6110     outs() << " RR";
6111   outs() << "\n";
6112 }
6113 
printObjc2_64bit_MetaData(MachOObjectFile * O,bool verbose)6114 static void printObjc2_64bit_MetaData(MachOObjectFile *O, bool verbose) {
6115   SymbolAddressMap AddrMap;
6116   if (verbose)
6117     CreateSymbolAddressMap(O, &AddrMap);
6118 
6119   std::vector<SectionRef> Sections;
6120   for (const SectionRef &Section : O->sections())
6121     Sections.push_back(Section);
6122 
6123   struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6124 
6125   SectionRef CL = get_section(O, "__OBJC2", "__class_list");
6126   if (CL == SectionRef())
6127     CL = get_section(O, "__DATA", "__objc_classlist");
6128   if (CL == SectionRef())
6129     CL = get_section(O, "__DATA_CONST", "__objc_classlist");
6130   if (CL == SectionRef())
6131     CL = get_section(O, "__DATA_DIRTY", "__objc_classlist");
6132   info.S = CL;
6133   walk_pointer_list_64("class", CL, O, &info, print_class64_t);
6134 
6135   SectionRef CR = get_section(O, "__OBJC2", "__class_refs");
6136   if (CR == SectionRef())
6137     CR = get_section(O, "__DATA", "__objc_classrefs");
6138   if (CR == SectionRef())
6139     CR = get_section(O, "__DATA_CONST", "__objc_classrefs");
6140   if (CR == SectionRef())
6141     CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs");
6142   info.S = CR;
6143   walk_pointer_list_64("class refs", CR, O, &info, nullptr);
6144 
6145   SectionRef SR = get_section(O, "__OBJC2", "__super_refs");
6146   if (SR == SectionRef())
6147     SR = get_section(O, "__DATA", "__objc_superrefs");
6148   if (SR == SectionRef())
6149     SR = get_section(O, "__DATA_CONST", "__objc_superrefs");
6150   if (SR == SectionRef())
6151     SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs");
6152   info.S = SR;
6153   walk_pointer_list_64("super refs", SR, O, &info, nullptr);
6154 
6155   SectionRef CA = get_section(O, "__OBJC2", "__category_list");
6156   if (CA == SectionRef())
6157     CA = get_section(O, "__DATA", "__objc_catlist");
6158   if (CA == SectionRef())
6159     CA = get_section(O, "__DATA_CONST", "__objc_catlist");
6160   if (CA == SectionRef())
6161     CA = get_section(O, "__DATA_DIRTY", "__objc_catlist");
6162   info.S = CA;
6163   walk_pointer_list_64("category", CA, O, &info, print_category64_t);
6164 
6165   SectionRef PL = get_section(O, "__OBJC2", "__protocol_list");
6166   if (PL == SectionRef())
6167     PL = get_section(O, "__DATA", "__objc_protolist");
6168   if (PL == SectionRef())
6169     PL = get_section(O, "__DATA_CONST", "__objc_protolist");
6170   if (PL == SectionRef())
6171     PL = get_section(O, "__DATA_DIRTY", "__objc_protolist");
6172   info.S = PL;
6173   walk_pointer_list_64("protocol", PL, O, &info, nullptr);
6174 
6175   SectionRef MR = get_section(O, "__OBJC2", "__message_refs");
6176   if (MR == SectionRef())
6177     MR = get_section(O, "__DATA", "__objc_msgrefs");
6178   if (MR == SectionRef())
6179     MR = get_section(O, "__DATA_CONST", "__objc_msgrefs");
6180   if (MR == SectionRef())
6181     MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs");
6182   info.S = MR;
6183   print_message_refs64(MR, &info);
6184 
6185   SectionRef II = get_section(O, "__OBJC2", "__image_info");
6186   if (II == SectionRef())
6187     II = get_section(O, "__DATA", "__objc_imageinfo");
6188   if (II == SectionRef())
6189     II = get_section(O, "__DATA_CONST", "__objc_imageinfo");
6190   if (II == SectionRef())
6191     II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo");
6192   info.S = II;
6193   print_image_info64(II, &info);
6194 }
6195 
printObjc2_32bit_MetaData(MachOObjectFile * O,bool verbose)6196 static void printObjc2_32bit_MetaData(MachOObjectFile *O, bool verbose) {
6197   SymbolAddressMap AddrMap;
6198   if (verbose)
6199     CreateSymbolAddressMap(O, &AddrMap);
6200 
6201   std::vector<SectionRef> Sections;
6202   for (const SectionRef &Section : O->sections())
6203     Sections.push_back(Section);
6204 
6205   struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6206 
6207   SectionRef CL = get_section(O, "__OBJC2", "__class_list");
6208   if (CL == SectionRef())
6209     CL = get_section(O, "__DATA", "__objc_classlist");
6210   if (CL == SectionRef())
6211     CL = get_section(O, "__DATA_CONST", "__objc_classlist");
6212   if (CL == SectionRef())
6213     CL = get_section(O, "__DATA_DIRTY", "__objc_classlist");
6214   info.S = CL;
6215   walk_pointer_list_32("class", CL, O, &info, print_class32_t);
6216 
6217   SectionRef CR = get_section(O, "__OBJC2", "__class_refs");
6218   if (CR == SectionRef())
6219     CR = get_section(O, "__DATA", "__objc_classrefs");
6220   if (CR == SectionRef())
6221     CR = get_section(O, "__DATA_CONST", "__objc_classrefs");
6222   if (CR == SectionRef())
6223     CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs");
6224   info.S = CR;
6225   walk_pointer_list_32("class refs", CR, O, &info, nullptr);
6226 
6227   SectionRef SR = get_section(O, "__OBJC2", "__super_refs");
6228   if (SR == SectionRef())
6229     SR = get_section(O, "__DATA", "__objc_superrefs");
6230   if (SR == SectionRef())
6231     SR = get_section(O, "__DATA_CONST", "__objc_superrefs");
6232   if (SR == SectionRef())
6233     SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs");
6234   info.S = SR;
6235   walk_pointer_list_32("super refs", SR, O, &info, nullptr);
6236 
6237   SectionRef CA = get_section(O, "__OBJC2", "__category_list");
6238   if (CA == SectionRef())
6239     CA = get_section(O, "__DATA", "__objc_catlist");
6240   if (CA == SectionRef())
6241     CA = get_section(O, "__DATA_CONST", "__objc_catlist");
6242   if (CA == SectionRef())
6243     CA = get_section(O, "__DATA_DIRTY", "__objc_catlist");
6244   info.S = CA;
6245   walk_pointer_list_32("category", CA, O, &info, print_category32_t);
6246 
6247   SectionRef PL = get_section(O, "__OBJC2", "__protocol_list");
6248   if (PL == SectionRef())
6249     PL = get_section(O, "__DATA", "__objc_protolist");
6250   if (PL == SectionRef())
6251     PL = get_section(O, "__DATA_CONST", "__objc_protolist");
6252   if (PL == SectionRef())
6253     PL = get_section(O, "__DATA_DIRTY", "__objc_protolist");
6254   info.S = PL;
6255   walk_pointer_list_32("protocol", PL, O, &info, nullptr);
6256 
6257   SectionRef MR = get_section(O, "__OBJC2", "__message_refs");
6258   if (MR == SectionRef())
6259     MR = get_section(O, "__DATA", "__objc_msgrefs");
6260   if (MR == SectionRef())
6261     MR = get_section(O, "__DATA_CONST", "__objc_msgrefs");
6262   if (MR == SectionRef())
6263     MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs");
6264   info.S = MR;
6265   print_message_refs32(MR, &info);
6266 
6267   SectionRef II = get_section(O, "__OBJC2", "__image_info");
6268   if (II == SectionRef())
6269     II = get_section(O, "__DATA", "__objc_imageinfo");
6270   if (II == SectionRef())
6271     II = get_section(O, "__DATA_CONST", "__objc_imageinfo");
6272   if (II == SectionRef())
6273     II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo");
6274   info.S = II;
6275   print_image_info32(II, &info);
6276 }
6277 
printObjc1_32bit_MetaData(MachOObjectFile * O,bool verbose)6278 static bool printObjc1_32bit_MetaData(MachOObjectFile *O, bool verbose) {
6279   uint32_t i, j, p, offset, xoffset, left, defs_left, def;
6280   const char *r, *name, *defs;
6281   struct objc_module_t module;
6282   SectionRef S, xS;
6283   struct objc_symtab_t symtab;
6284   struct objc_class_t objc_class;
6285   struct objc_category_t objc_category;
6286 
6287   outs() << "Objective-C segment\n";
6288   S = get_section(O, "__OBJC", "__module_info");
6289   if (S == SectionRef())
6290     return false;
6291 
6292   SymbolAddressMap AddrMap;
6293   if (verbose)
6294     CreateSymbolAddressMap(O, &AddrMap);
6295 
6296   std::vector<SectionRef> Sections;
6297   for (const SectionRef &Section : O->sections())
6298     Sections.push_back(Section);
6299 
6300   struct DisassembleInfo info(O, &AddrMap, &Sections, verbose);
6301 
6302   for (i = 0; i < S.getSize(); i += sizeof(struct objc_module_t)) {
6303     p = S.getAddress() + i;
6304     r = get_pointer_32(p, offset, left, S, &info, true);
6305     if (r == nullptr)
6306       return true;
6307     memset(&module, '\0', sizeof(struct objc_module_t));
6308     if (left < sizeof(struct objc_module_t)) {
6309       memcpy(&module, r, left);
6310       outs() << "   (module extends past end of __module_info section)\n";
6311     } else
6312       memcpy(&module, r, sizeof(struct objc_module_t));
6313     if (O->isLittleEndian() != sys::IsLittleEndianHost)
6314       swapStruct(module);
6315 
6316     outs() << "Module " << format("0x%" PRIx32, p) << "\n";
6317     outs() << "    version " << module.version << "\n";
6318     outs() << "       size " << module.size << "\n";
6319     outs() << "       name ";
6320     name = get_pointer_32(module.name, xoffset, left, xS, &info, true);
6321     if (name != nullptr)
6322       outs() << format("%.*s", left, name);
6323     else
6324       outs() << format("0x%08" PRIx32, module.name)
6325              << "(not in an __OBJC section)";
6326     outs() << "\n";
6327 
6328     r = get_pointer_32(module.symtab, xoffset, left, xS, &info, true);
6329     if (module.symtab == 0 || r == nullptr) {
6330       outs() << "     symtab " << format("0x%08" PRIx32, module.symtab)
6331              << " (not in an __OBJC section)\n";
6332       continue;
6333     }
6334     outs() << "     symtab " << format("0x%08" PRIx32, module.symtab) << "\n";
6335     memset(&symtab, '\0', sizeof(struct objc_symtab_t));
6336     defs_left = 0;
6337     defs = nullptr;
6338     if (left < sizeof(struct objc_symtab_t)) {
6339       memcpy(&symtab, r, left);
6340       outs() << "\tsymtab extends past end of an __OBJC section)\n";
6341     } else {
6342       memcpy(&symtab, r, sizeof(struct objc_symtab_t));
6343       if (left > sizeof(struct objc_symtab_t)) {
6344         defs_left = left - sizeof(struct objc_symtab_t);
6345         defs = r + sizeof(struct objc_symtab_t);
6346       }
6347     }
6348     if (O->isLittleEndian() != sys::IsLittleEndianHost)
6349       swapStruct(symtab);
6350 
6351     outs() << "\tsel_ref_cnt " << symtab.sel_ref_cnt << "\n";
6352     r = get_pointer_32(symtab.refs, xoffset, left, xS, &info, true);
6353     outs() << "\trefs " << format("0x%08" PRIx32, symtab.refs);
6354     if (r == nullptr)
6355       outs() << " (not in an __OBJC section)";
6356     outs() << "\n";
6357     outs() << "\tcls_def_cnt " << symtab.cls_def_cnt << "\n";
6358     outs() << "\tcat_def_cnt " << symtab.cat_def_cnt << "\n";
6359     if (symtab.cls_def_cnt > 0)
6360       outs() << "\tClass Definitions\n";
6361     for (j = 0; j < symtab.cls_def_cnt; j++) {
6362       if ((j + 1) * sizeof(uint32_t) > defs_left) {
6363         outs() << "\t(remaining class defs entries entends past the end of the "
6364                << "section)\n";
6365         break;
6366       }
6367       memcpy(&def, defs + j * sizeof(uint32_t), sizeof(uint32_t));
6368       if (O->isLittleEndian() != sys::IsLittleEndianHost)
6369         sys::swapByteOrder(def);
6370 
6371       r = get_pointer_32(def, xoffset, left, xS, &info, true);
6372       outs() << "\tdefs[" << j << "] " << format("0x%08" PRIx32, def);
6373       if (r != nullptr) {
6374         if (left > sizeof(struct objc_class_t)) {
6375           outs() << "\n";
6376           memcpy(&objc_class, r, sizeof(struct objc_class_t));
6377         } else {
6378           outs() << " (entends past the end of the section)\n";
6379           memset(&objc_class, '\0', sizeof(struct objc_class_t));
6380           memcpy(&objc_class, r, left);
6381         }
6382         if (O->isLittleEndian() != sys::IsLittleEndianHost)
6383           swapStruct(objc_class);
6384         print_objc_class_t(&objc_class, &info);
6385       } else {
6386         outs() << "(not in an __OBJC section)\n";
6387       }
6388 
6389       if (CLS_GETINFO(&objc_class, CLS_CLASS)) {
6390         outs() << "\tMeta Class";
6391         r = get_pointer_32(objc_class.isa, xoffset, left, xS, &info, true);
6392         if (r != nullptr) {
6393           if (left > sizeof(struct objc_class_t)) {
6394             outs() << "\n";
6395             memcpy(&objc_class, r, sizeof(struct objc_class_t));
6396           } else {
6397             outs() << " (entends past the end of the section)\n";
6398             memset(&objc_class, '\0', sizeof(struct objc_class_t));
6399             memcpy(&objc_class, r, left);
6400           }
6401           if (O->isLittleEndian() != sys::IsLittleEndianHost)
6402             swapStruct(objc_class);
6403           print_objc_class_t(&objc_class, &info);
6404         } else {
6405           outs() << "(not in an __OBJC section)\n";
6406         }
6407       }
6408     }
6409     if (symtab.cat_def_cnt > 0)
6410       outs() << "\tCategory Definitions\n";
6411     for (j = 0; j < symtab.cat_def_cnt; j++) {
6412       if ((j + symtab.cls_def_cnt + 1) * sizeof(uint32_t) > defs_left) {
6413         outs() << "\t(remaining category defs entries entends past the end of "
6414                << "the section)\n";
6415         break;
6416       }
6417       memcpy(&def, defs + (j + symtab.cls_def_cnt) * sizeof(uint32_t),
6418              sizeof(uint32_t));
6419       if (O->isLittleEndian() != sys::IsLittleEndianHost)
6420         sys::swapByteOrder(def);
6421 
6422       r = get_pointer_32(def, xoffset, left, xS, &info, true);
6423       outs() << "\tdefs[" << j + symtab.cls_def_cnt << "] "
6424              << format("0x%08" PRIx32, def);
6425       if (r != nullptr) {
6426         if (left > sizeof(struct objc_category_t)) {
6427           outs() << "\n";
6428           memcpy(&objc_category, r, sizeof(struct objc_category_t));
6429         } else {
6430           outs() << " (entends past the end of the section)\n";
6431           memset(&objc_category, '\0', sizeof(struct objc_category_t));
6432           memcpy(&objc_category, r, left);
6433         }
6434         if (O->isLittleEndian() != sys::IsLittleEndianHost)
6435           swapStruct(objc_category);
6436         print_objc_objc_category_t(&objc_category, &info);
6437       } else {
6438         outs() << "(not in an __OBJC section)\n";
6439       }
6440     }
6441   }
6442   const SectionRef II = get_section(O, "__OBJC", "__image_info");
6443   if (II != SectionRef())
6444     print_image_info(II, &info);
6445 
6446   return true;
6447 }
6448 
DumpProtocolSection(MachOObjectFile * O,const char * sect,uint32_t size,uint32_t addr)6449 static void DumpProtocolSection(MachOObjectFile *O, const char *sect,
6450                                 uint32_t size, uint32_t addr) {
6451   SymbolAddressMap AddrMap;
6452   CreateSymbolAddressMap(O, &AddrMap);
6453 
6454   std::vector<SectionRef> Sections;
6455   for (const SectionRef &Section : O->sections())
6456     Sections.push_back(Section);
6457 
6458   struct DisassembleInfo info(O, &AddrMap, &Sections, true);
6459 
6460   const char *p;
6461   struct objc_protocol_t protocol;
6462   uint32_t left, paddr;
6463   for (p = sect; p < sect + size; p += sizeof(struct objc_protocol_t)) {
6464     memset(&protocol, '\0', sizeof(struct objc_protocol_t));
6465     left = size - (p - sect);
6466     if (left < sizeof(struct objc_protocol_t)) {
6467       outs() << "Protocol extends past end of __protocol section\n";
6468       memcpy(&protocol, p, left);
6469     } else
6470       memcpy(&protocol, p, sizeof(struct objc_protocol_t));
6471     if (O->isLittleEndian() != sys::IsLittleEndianHost)
6472       swapStruct(protocol);
6473     paddr = addr + (p - sect);
6474     outs() << "Protocol " << format("0x%" PRIx32, paddr);
6475     if (print_protocol(paddr, 0, &info))
6476       outs() << "(not in an __OBJC section)\n";
6477   }
6478 }
6479 
6480 #ifdef HAVE_LIBXAR
swapStruct(struct xar_header & xar)6481 static inline void swapStruct(struct xar_header &xar) {
6482   sys::swapByteOrder(xar.magic);
6483   sys::swapByteOrder(xar.size);
6484   sys::swapByteOrder(xar.version);
6485   sys::swapByteOrder(xar.toc_length_compressed);
6486   sys::swapByteOrder(xar.toc_length_uncompressed);
6487   sys::swapByteOrder(xar.cksum_alg);
6488 }
6489 
PrintModeVerbose(uint32_t mode)6490 static void PrintModeVerbose(uint32_t mode) {
6491   switch(mode & S_IFMT){
6492   case S_IFDIR:
6493     outs() << "d";
6494     break;
6495   case S_IFCHR:
6496     outs() << "c";
6497     break;
6498   case S_IFBLK:
6499     outs() << "b";
6500     break;
6501   case S_IFREG:
6502     outs() << "-";
6503     break;
6504   case S_IFLNK:
6505     outs() << "l";
6506     break;
6507   case S_IFSOCK:
6508     outs() << "s";
6509     break;
6510   default:
6511     outs() << "?";
6512     break;
6513   }
6514 
6515   /* owner permissions */
6516   if(mode & S_IREAD)
6517     outs() << "r";
6518   else
6519     outs() << "-";
6520   if(mode & S_IWRITE)
6521     outs() << "w";
6522   else
6523     outs() << "-";
6524   if(mode & S_ISUID)
6525     outs() << "s";
6526   else if(mode & S_IEXEC)
6527     outs() << "x";
6528   else
6529     outs() << "-";
6530 
6531   /* group permissions */
6532   if(mode & (S_IREAD >> 3))
6533     outs() << "r";
6534   else
6535     outs() << "-";
6536   if(mode & (S_IWRITE >> 3))
6537     outs() << "w";
6538   else
6539     outs() << "-";
6540   if(mode & S_ISGID)
6541     outs() << "s";
6542   else if(mode & (S_IEXEC >> 3))
6543     outs() << "x";
6544   else
6545     outs() << "-";
6546 
6547   /* other permissions */
6548   if(mode & (S_IREAD >> 6))
6549     outs() << "r";
6550   else
6551     outs() << "-";
6552   if(mode & (S_IWRITE >> 6))
6553     outs() << "w";
6554   else
6555     outs() << "-";
6556   if(mode & S_ISVTX)
6557     outs() << "t";
6558   else if(mode & (S_IEXEC >> 6))
6559     outs() << "x";
6560   else
6561     outs() << "-";
6562 }
6563 
PrintXarFilesSummary(const char * XarFilename,xar_t xar)6564 static void PrintXarFilesSummary(const char *XarFilename, xar_t xar) {
6565   xar_file_t xf;
6566   const char *key, *type, *mode, *user, *group, *size, *mtime, *name, *m;
6567   char *endp;
6568   uint32_t mode_value;
6569 
6570   ScopedXarIter xi;
6571   if (!xi) {
6572     WithColor::error(errs(), "llvm-objdump")
6573         << "can't obtain an xar iterator for xar archive " << XarFilename
6574         << "\n";
6575     return;
6576   }
6577 
6578   // Go through the xar's files.
6579   for (xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)) {
6580     ScopedXarIter xp;
6581     if(!xp){
6582       WithColor::error(errs(), "llvm-objdump")
6583           << "can't obtain an xar iterator for xar archive " << XarFilename
6584           << "\n";
6585       return;
6586     }
6587     type = nullptr;
6588     mode = nullptr;
6589     user = nullptr;
6590     group = nullptr;
6591     size = nullptr;
6592     mtime = nullptr;
6593     name = nullptr;
6594     for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){
6595       const char *val = nullptr;
6596       xar_prop_get(xf, key, &val);
6597 #if 0 // Useful for debugging.
6598       outs() << "key: " << key << " value: " << val << "\n";
6599 #endif
6600       if(strcmp(key, "type") == 0)
6601         type = val;
6602       if(strcmp(key, "mode") == 0)
6603         mode = val;
6604       if(strcmp(key, "user") == 0)
6605         user = val;
6606       if(strcmp(key, "group") == 0)
6607         group = val;
6608       if(strcmp(key, "data/size") == 0)
6609         size = val;
6610       if(strcmp(key, "mtime") == 0)
6611         mtime = val;
6612       if(strcmp(key, "name") == 0)
6613         name = val;
6614     }
6615     if(mode != nullptr){
6616       mode_value = strtoul(mode, &endp, 8);
6617       if(*endp != '\0')
6618         outs() << "(mode: \"" << mode << "\" contains non-octal chars) ";
6619       if(strcmp(type, "file") == 0)
6620         mode_value |= S_IFREG;
6621       PrintModeVerbose(mode_value);
6622       outs() << " ";
6623     }
6624     if(user != nullptr)
6625       outs() << format("%10s/", user);
6626     if(group != nullptr)
6627       outs() << format("%-10s ", group);
6628     if(size != nullptr)
6629       outs() << format("%7s ", size);
6630     if(mtime != nullptr){
6631       for(m = mtime; *m != 'T' && *m != '\0'; m++)
6632         outs() << *m;
6633       if(*m == 'T')
6634         m++;
6635       outs() << " ";
6636       for( ; *m != 'Z' && *m != '\0'; m++)
6637         outs() << *m;
6638       outs() << " ";
6639     }
6640     if(name != nullptr)
6641       outs() << name;
6642     outs() << "\n";
6643   }
6644 }
6645 
DumpBitcodeSection(MachOObjectFile * O,const char * sect,uint32_t size,bool verbose,bool PrintXarHeader,bool PrintXarFileHeaders,std::string XarMemberName)6646 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect,
6647                                 uint32_t size, bool verbose,
6648                                 bool PrintXarHeader, bool PrintXarFileHeaders,
6649                                 std::string XarMemberName) {
6650   if(size < sizeof(struct xar_header)) {
6651     outs() << "size of (__LLVM,__bundle) section too small (smaller than size "
6652               "of struct xar_header)\n";
6653     return;
6654   }
6655   struct xar_header XarHeader;
6656   memcpy(&XarHeader, sect, sizeof(struct xar_header));
6657   if (sys::IsLittleEndianHost)
6658     swapStruct(XarHeader);
6659   if (PrintXarHeader) {
6660     if (!XarMemberName.empty())
6661       outs() << "In xar member " << XarMemberName << ": ";
6662     else
6663       outs() << "For (__LLVM,__bundle) section: ";
6664     outs() << "xar header\n";
6665     if (XarHeader.magic == XAR_HEADER_MAGIC)
6666       outs() << "                  magic XAR_HEADER_MAGIC\n";
6667     else
6668       outs() << "                  magic "
6669              << format_hex(XarHeader.magic, 10, true)
6670              << " (not XAR_HEADER_MAGIC)\n";
6671     outs() << "                   size " << XarHeader.size << "\n";
6672     outs() << "                version " << XarHeader.version << "\n";
6673     outs() << "  toc_length_compressed " << XarHeader.toc_length_compressed
6674            << "\n";
6675     outs() << "toc_length_uncompressed " << XarHeader.toc_length_uncompressed
6676            << "\n";
6677     outs() << "              cksum_alg ";
6678     switch (XarHeader.cksum_alg) {
6679       case XAR_CKSUM_NONE:
6680         outs() << "XAR_CKSUM_NONE\n";
6681         break;
6682       case XAR_CKSUM_SHA1:
6683         outs() << "XAR_CKSUM_SHA1\n";
6684         break;
6685       case XAR_CKSUM_MD5:
6686         outs() << "XAR_CKSUM_MD5\n";
6687         break;
6688 #ifdef XAR_CKSUM_SHA256
6689       case XAR_CKSUM_SHA256:
6690         outs() << "XAR_CKSUM_SHA256\n";
6691         break;
6692 #endif
6693 #ifdef XAR_CKSUM_SHA512
6694       case XAR_CKSUM_SHA512:
6695         outs() << "XAR_CKSUM_SHA512\n";
6696         break;
6697 #endif
6698       default:
6699         outs() << XarHeader.cksum_alg << "\n";
6700     }
6701   }
6702 
6703   SmallString<128> XarFilename;
6704   int FD;
6705   std::error_code XarEC =
6706       sys::fs::createTemporaryFile("llvm-objdump", "xar", FD, XarFilename);
6707   if (XarEC) {
6708     WithColor::error(errs(), "llvm-objdump") << XarEC.message() << "\n";
6709     return;
6710   }
6711   ToolOutputFile XarFile(XarFilename, FD);
6712   raw_fd_ostream &XarOut = XarFile.os();
6713   StringRef XarContents(sect, size);
6714   XarOut << XarContents;
6715   XarOut.close();
6716   if (XarOut.has_error())
6717     return;
6718 
6719   ScopedXarFile xar(XarFilename.c_str(), READ);
6720   if (!xar) {
6721     WithColor::error(errs(), "llvm-objdump")
6722         << "can't create temporary xar archive " << XarFilename << "\n";
6723     return;
6724   }
6725 
6726   SmallString<128> TocFilename;
6727   std::error_code TocEC =
6728       sys::fs::createTemporaryFile("llvm-objdump", "toc", TocFilename);
6729   if (TocEC) {
6730     WithColor::error(errs(), "llvm-objdump") << TocEC.message() << "\n";
6731     return;
6732   }
6733   xar_serialize(xar, TocFilename.c_str());
6734 
6735   if (PrintXarFileHeaders) {
6736     if (!XarMemberName.empty())
6737       outs() << "In xar member " << XarMemberName << ": ";
6738     else
6739       outs() << "For (__LLVM,__bundle) section: ";
6740     outs() << "xar archive files:\n";
6741     PrintXarFilesSummary(XarFilename.c_str(), xar);
6742   }
6743 
6744   ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr =
6745     MemoryBuffer::getFileOrSTDIN(TocFilename.c_str());
6746   if (std::error_code EC = FileOrErr.getError()) {
6747     WithColor::error(errs(), "llvm-objdump") << EC.message() << "\n";
6748     return;
6749   }
6750   std::unique_ptr<MemoryBuffer> &Buffer = FileOrErr.get();
6751 
6752   if (!XarMemberName.empty())
6753     outs() << "In xar member " << XarMemberName << ": ";
6754   else
6755     outs() << "For (__LLVM,__bundle) section: ";
6756   outs() << "xar table of contents:\n";
6757   outs() << Buffer->getBuffer() << "\n";
6758 
6759   // TODO: Go through the xar's files.
6760   ScopedXarIter xi;
6761   if(!xi){
6762     WithColor::error(errs(), "llvm-objdump")
6763         << "can't obtain an xar iterator for xar archive "
6764         << XarFilename.c_str() << "\n";
6765     return;
6766   }
6767   for(xar_file_t xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)){
6768     const char *key;
6769     const char *member_name, *member_type, *member_size_string;
6770     size_t member_size;
6771 
6772     ScopedXarIter xp;
6773     if(!xp){
6774       WithColor::error(errs(), "llvm-objdump")
6775           << "can't obtain an xar iterator for xar archive "
6776           << XarFilename.c_str() << "\n";
6777       return;
6778     }
6779     member_name = NULL;
6780     member_type = NULL;
6781     member_size_string = NULL;
6782     for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){
6783       const char *val = nullptr;
6784       xar_prop_get(xf, key, &val);
6785 #if 0 // Useful for debugging.
6786       outs() << "key: " << key << " value: " << val << "\n";
6787 #endif
6788       if (strcmp(key, "name") == 0)
6789         member_name = val;
6790       if (strcmp(key, "type") == 0)
6791         member_type = val;
6792       if (strcmp(key, "data/size") == 0)
6793         member_size_string = val;
6794     }
6795     /*
6796      * If we find a file with a name, date/size and type properties
6797      * and with the type being "file" see if that is a xar file.
6798      */
6799     if (member_name != NULL && member_type != NULL &&
6800         strcmp(member_type, "file") == 0 &&
6801         member_size_string != NULL){
6802       // Extract the file into a buffer.
6803       char *endptr;
6804       member_size = strtoul(member_size_string, &endptr, 10);
6805       if (*endptr == '\0' && member_size != 0) {
6806         char *buffer;
6807         if (xar_extract_tobuffersz(xar, xf, &buffer, &member_size) == 0) {
6808 #if 0 // Useful for debugging.
6809           outs() << "xar member: " << member_name << " extracted\n";
6810 #endif
6811           // Set the XarMemberName we want to see printed in the header.
6812           std::string OldXarMemberName;
6813           // If XarMemberName is already set this is nested. So
6814           // save the old name and create the nested name.
6815           if (!XarMemberName.empty()) {
6816             OldXarMemberName = XarMemberName;
6817             XarMemberName =
6818                 (Twine("[") + XarMemberName + "]" + member_name).str();
6819           } else {
6820             OldXarMemberName = "";
6821             XarMemberName = member_name;
6822           }
6823           // See if this is could be a xar file (nested).
6824           if (member_size >= sizeof(struct xar_header)) {
6825 #if 0 // Useful for debugging.
6826             outs() << "could be a xar file: " << member_name << "\n";
6827 #endif
6828             memcpy((char *)&XarHeader, buffer, sizeof(struct xar_header));
6829             if (sys::IsLittleEndianHost)
6830               swapStruct(XarHeader);
6831             if (XarHeader.magic == XAR_HEADER_MAGIC)
6832               DumpBitcodeSection(O, buffer, member_size, verbose,
6833                                  PrintXarHeader, PrintXarFileHeaders,
6834                                  XarMemberName);
6835           }
6836           XarMemberName = OldXarMemberName;
6837           delete buffer;
6838         }
6839       }
6840     }
6841   }
6842 }
6843 #endif // defined(HAVE_LIBXAR)
6844 
printObjcMetaData(MachOObjectFile * O,bool verbose)6845 static void printObjcMetaData(MachOObjectFile *O, bool verbose) {
6846   if (O->is64Bit())
6847     printObjc2_64bit_MetaData(O, verbose);
6848   else {
6849     MachO::mach_header H;
6850     H = O->getHeader();
6851     if (H.cputype == MachO::CPU_TYPE_ARM)
6852       printObjc2_32bit_MetaData(O, verbose);
6853     else {
6854       // This is the 32-bit non-arm cputype case.  Which is normally
6855       // the first Objective-C ABI.  But it may be the case of a
6856       // binary for the iOS simulator which is the second Objective-C
6857       // ABI.  In that case printObjc1_32bit_MetaData() will determine that
6858       // and return false.
6859       if (!printObjc1_32bit_MetaData(O, verbose))
6860         printObjc2_32bit_MetaData(O, verbose);
6861     }
6862   }
6863 }
6864 
6865 // GuessLiteralPointer returns a string which for the item in the Mach-O file
6866 // for the address passed in as ReferenceValue for printing as a comment with
6867 // the instruction and also returns the corresponding type of that item
6868 // indirectly through ReferenceType.
6869 //
6870 // If ReferenceValue is an address of literal cstring then a pointer to the
6871 // cstring is returned and ReferenceType is set to
6872 // LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr .
6873 //
6874 // If ReferenceValue is an address of an Objective-C CFString, Selector ref or
6875 // Class ref that name is returned and the ReferenceType is set accordingly.
6876 //
6877 // Lastly, literals which are Symbol address in a literal pool are looked for
6878 // and if found the symbol name is returned and ReferenceType is set to
6879 // LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr .
6880 //
6881 // If there is no item in the Mach-O file for the address passed in as
6882 // ReferenceValue nullptr is returned and ReferenceType is unchanged.
GuessLiteralPointer(uint64_t ReferenceValue,uint64_t ReferencePC,uint64_t * ReferenceType,struct DisassembleInfo * info)6883 static const char *GuessLiteralPointer(uint64_t ReferenceValue,
6884                                        uint64_t ReferencePC,
6885                                        uint64_t *ReferenceType,
6886                                        struct DisassembleInfo *info) {
6887   // First see if there is an external relocation entry at the ReferencePC.
6888   if (info->O->getHeader().filetype == MachO::MH_OBJECT) {
6889     uint64_t sect_addr = info->S.getAddress();
6890     uint64_t sect_offset = ReferencePC - sect_addr;
6891     bool reloc_found = false;
6892     DataRefImpl Rel;
6893     MachO::any_relocation_info RE;
6894     bool isExtern = false;
6895     SymbolRef Symbol;
6896     for (const RelocationRef &Reloc : info->S.relocations()) {
6897       uint64_t RelocOffset = Reloc.getOffset();
6898       if (RelocOffset == sect_offset) {
6899         Rel = Reloc.getRawDataRefImpl();
6900         RE = info->O->getRelocation(Rel);
6901         if (info->O->isRelocationScattered(RE))
6902           continue;
6903         isExtern = info->O->getPlainRelocationExternal(RE);
6904         if (isExtern) {
6905           symbol_iterator RelocSym = Reloc.getSymbol();
6906           Symbol = *RelocSym;
6907         }
6908         reloc_found = true;
6909         break;
6910       }
6911     }
6912     // If there is an external relocation entry for a symbol in a section
6913     // then used that symbol's value for the value of the reference.
6914     if (reloc_found && isExtern) {
6915       if (info->O->getAnyRelocationPCRel(RE)) {
6916         unsigned Type = info->O->getAnyRelocationType(RE);
6917         if (Type == MachO::X86_64_RELOC_SIGNED) {
6918           ReferenceValue = cantFail(Symbol.getValue());
6919         }
6920       }
6921     }
6922   }
6923 
6924   // Look for literals such as Objective-C CFStrings refs, Selector refs,
6925   // Message refs and Class refs.
6926   bool classref, selref, msgref, cfstring;
6927   uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref,
6928                                                selref, msgref, cfstring);
6929   if (classref && pointer_value == 0) {
6930     // Note the ReferenceValue is a pointer into the __objc_classrefs section.
6931     // And the pointer_value in that section is typically zero as it will be
6932     // set by dyld as part of the "bind information".
6933     const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info);
6934     if (name != nullptr) {
6935       *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
6936       const char *class_name = strrchr(name, '$');
6937       if (class_name != nullptr && class_name[1] == '_' &&
6938           class_name[2] != '\0') {
6939         info->class_name = class_name + 2;
6940         return name;
6941       }
6942     }
6943   }
6944 
6945   if (classref) {
6946     *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
6947     const char *name =
6948         get_objc2_64bit_class_name(pointer_value, ReferenceValue, info);
6949     if (name != nullptr)
6950       info->class_name = name;
6951     else
6952       name = "bad class ref";
6953     return name;
6954   }
6955 
6956   if (cfstring) {
6957     *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref;
6958     const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info);
6959     return name;
6960   }
6961 
6962   if (selref && pointer_value == 0)
6963     pointer_value = get_objc2_64bit_selref(ReferenceValue, info);
6964 
6965   if (pointer_value != 0)
6966     ReferenceValue = pointer_value;
6967 
6968   const char *name = GuessCstringPointer(ReferenceValue, info);
6969   if (name) {
6970     if (pointer_value != 0 && selref) {
6971       *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref;
6972       info->selector_name = name;
6973     } else if (pointer_value != 0 && msgref) {
6974       info->class_name = nullptr;
6975       *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref;
6976       info->selector_name = name;
6977     } else
6978       *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr;
6979     return name;
6980   }
6981 
6982   // Lastly look for an indirect symbol with this ReferenceValue which is in
6983   // a literal pool.  If found return that symbol name.
6984   name = GuessIndirectSymbol(ReferenceValue, info);
6985   if (name) {
6986     *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr;
6987     return name;
6988   }
6989 
6990   return nullptr;
6991 }
6992 
6993 // SymbolizerSymbolLookUp is the symbol lookup function passed when creating
6994 // the Symbolizer.  It looks up the ReferenceValue using the info passed via the
6995 // pointer to the struct DisassembleInfo that was passed when MCSymbolizer
6996 // is created and returns the symbol name that matches the ReferenceValue or
6997 // nullptr if none.  The ReferenceType is passed in for the IN type of
6998 // reference the instruction is making from the values in defined in the header
6999 // "llvm-c/Disassembler.h".  On return the ReferenceType can set to a specific
7000 // Out type and the ReferenceName will also be set which is added as a comment
7001 // to the disassembled instruction.
7002 //
7003 // If the symbol name is a C++ mangled name then the demangled name is
7004 // returned through ReferenceName and ReferenceType is set to
7005 // LLVMDisassembler_ReferenceType_DeMangled_Name .
7006 //
7007 // When this is called to get a symbol name for a branch target then the
7008 // ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then
7009 // SymbolValue will be looked for in the indirect symbol table to determine if
7010 // it is an address for a symbol stub.  If so then the symbol name for that
7011 // stub is returned indirectly through ReferenceName and then ReferenceType is
7012 // set to LLVMDisassembler_ReferenceType_Out_SymbolStub.
7013 //
7014 // When this is called with an value loaded via a PC relative load then
7015 // ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the
7016 // SymbolValue is checked to be an address of literal pointer, symbol pointer,
7017 // or an Objective-C meta data reference.  If so the output ReferenceType is
7018 // 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)7019 static const char *SymbolizerSymbolLookUp(void *DisInfo,
7020                                           uint64_t ReferenceValue,
7021                                           uint64_t *ReferenceType,
7022                                           uint64_t ReferencePC,
7023                                           const char **ReferenceName) {
7024   struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
7025   // If no verbose symbolic information is wanted then just return nullptr.
7026   if (!info->verbose) {
7027     *ReferenceName = nullptr;
7028     *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7029     return nullptr;
7030   }
7031 
7032   const char *SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
7033 
7034   if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) {
7035     *ReferenceName = GuessIndirectSymbol(ReferenceValue, info);
7036     if (*ReferenceName != nullptr) {
7037       method_reference(info, ReferenceType, ReferenceName);
7038       if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message)
7039         *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub;
7040     } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
7041       if (info->demangled_name != nullptr)
7042         free(info->demangled_name);
7043       int status;
7044       info->demangled_name =
7045           itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status);
7046       if (info->demangled_name != nullptr) {
7047         *ReferenceName = info->demangled_name;
7048         *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
7049       } else
7050         *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7051     } else
7052       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7053   } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) {
7054     *ReferenceName =
7055         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7056     if (*ReferenceName)
7057       method_reference(info, ReferenceType, ReferenceName);
7058     else
7059       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7060     // If this is arm64 and the reference is an adrp instruction save the
7061     // instruction, passed in ReferenceValue and the address of the instruction
7062     // for use later if we see and add immediate instruction.
7063   } else if (info->O->getArch() == Triple::aarch64 &&
7064              *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
7065     info->adrp_inst = ReferenceValue;
7066     info->adrp_addr = ReferencePC;
7067     SymbolName = nullptr;
7068     *ReferenceName = nullptr;
7069     *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7070     // If this is arm64 and reference is an add immediate instruction and we
7071     // have
7072     // seen an adrp instruction just before it and the adrp's Xd register
7073     // matches
7074     // this add's Xn register reconstruct the value being referenced and look to
7075     // see if it is a literal pointer.  Note the add immediate instruction is
7076     // passed in ReferenceValue.
7077   } else if (info->O->getArch() == Triple::aarch64 &&
7078              *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri &&
7079              ReferencePC - 4 == info->adrp_addr &&
7080              (info->adrp_inst & 0x9f000000) == 0x90000000 &&
7081              (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
7082     uint32_t addxri_inst;
7083     uint64_t adrp_imm, addxri_imm;
7084 
7085     adrp_imm =
7086         ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
7087     if (info->adrp_inst & 0x0200000)
7088       adrp_imm |= 0xfffffffffc000000LL;
7089 
7090     addxri_inst = ReferenceValue;
7091     addxri_imm = (addxri_inst >> 10) & 0xfff;
7092     if (((addxri_inst >> 22) & 0x3) == 1)
7093       addxri_imm <<= 12;
7094 
7095     ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
7096                      (adrp_imm << 12) + addxri_imm;
7097 
7098     *ReferenceName =
7099         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7100     if (*ReferenceName == nullptr)
7101       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7102     // If this is arm64 and the reference is a load register instruction and we
7103     // have seen an adrp instruction just before it and the adrp's Xd register
7104     // matches this add's Xn register reconstruct the value being referenced and
7105     // look to see if it is a literal pointer.  Note the load register
7106     // instruction is passed in ReferenceValue.
7107   } else if (info->O->getArch() == Triple::aarch64 &&
7108              *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui &&
7109              ReferencePC - 4 == info->adrp_addr &&
7110              (info->adrp_inst & 0x9f000000) == 0x90000000 &&
7111              (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
7112     uint32_t ldrxui_inst;
7113     uint64_t adrp_imm, ldrxui_imm;
7114 
7115     adrp_imm =
7116         ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
7117     if (info->adrp_inst & 0x0200000)
7118       adrp_imm |= 0xfffffffffc000000LL;
7119 
7120     ldrxui_inst = ReferenceValue;
7121     ldrxui_imm = (ldrxui_inst >> 10) & 0xfff;
7122 
7123     ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
7124                      (adrp_imm << 12) + (ldrxui_imm << 3);
7125 
7126     *ReferenceName =
7127         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7128     if (*ReferenceName == nullptr)
7129       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7130   }
7131   // If this arm64 and is an load register (PC-relative) instruction the
7132   // ReferenceValue is the PC plus the immediate value.
7133   else if (info->O->getArch() == Triple::aarch64 &&
7134            (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl ||
7135             *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) {
7136     *ReferenceName =
7137         GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
7138     if (*ReferenceName == nullptr)
7139       *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7140   } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
7141     if (info->demangled_name != nullptr)
7142       free(info->demangled_name);
7143     int status;
7144     info->demangled_name =
7145         itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status);
7146     if (info->demangled_name != nullptr) {
7147       *ReferenceName = info->demangled_name;
7148       *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
7149     }
7150   }
7151   else {
7152     *ReferenceName = nullptr;
7153     *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
7154   }
7155 
7156   return SymbolName;
7157 }
7158 
7159 /// Emits the comments that are stored in the CommentStream.
7160 /// 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)7161 static void emitComments(raw_svector_ostream &CommentStream,
7162                          SmallString<128> &CommentsToEmit,
7163                          formatted_raw_ostream &FormattedOS,
7164                          const MCAsmInfo &MAI) {
7165   // Flush the stream before taking its content.
7166   StringRef Comments = CommentsToEmit.str();
7167   // Get the default information for printing a comment.
7168   StringRef CommentBegin = MAI.getCommentString();
7169   unsigned CommentColumn = MAI.getCommentColumn();
7170   bool IsFirst = true;
7171   while (!Comments.empty()) {
7172     if (!IsFirst)
7173       FormattedOS << '\n';
7174     // Emit a line of comments.
7175     FormattedOS.PadToColumn(CommentColumn);
7176     size_t Position = Comments.find('\n');
7177     FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position);
7178     // Move after the newline character.
7179     Comments = Comments.substr(Position + 1);
7180     IsFirst = false;
7181   }
7182   FormattedOS.flush();
7183 
7184   // Tell the comment stream that the vector changed underneath it.
7185   CommentsToEmit.clear();
7186 }
7187 
DisassembleMachO(StringRef Filename,MachOObjectFile * MachOOF,StringRef DisSegName,StringRef DisSectName)7188 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
7189                              StringRef DisSegName, StringRef DisSectName) {
7190   const char *McpuDefault = nullptr;
7191   const Target *ThumbTarget = nullptr;
7192   const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget);
7193   if (!TheTarget) {
7194     // GetTarget prints out stuff.
7195     return;
7196   }
7197   std::string MachOMCPU;
7198   if (MCPU.empty() && McpuDefault)
7199     MachOMCPU = McpuDefault;
7200   else
7201     MachOMCPU = MCPU;
7202 
7203 #define CHECK_TARGET_INFO_CREATION(NAME)                                       \
7204   do {                                                                         \
7205     if (!NAME) {                                                               \
7206       WithColor::error(errs(), "llvm-objdump")                                 \
7207           << "couldn't initialize disassembler for target " << TripleName      \
7208           << '\n';                                                             \
7209       return;                                                                  \
7210     }                                                                          \
7211   } while (false)
7212 #define CHECK_THUMB_TARGET_INFO_CREATION(NAME)                                 \
7213   do {                                                                         \
7214     if (!NAME) {                                                               \
7215       WithColor::error(errs(), "llvm-objdump")                                 \
7216           << "couldn't initialize disassembler for target " << ThumbTripleName \
7217           << '\n';                                                             \
7218       return;                                                                  \
7219     }                                                                          \
7220   } while (false)
7221 
7222   std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
7223   CHECK_TARGET_INFO_CREATION(InstrInfo);
7224   std::unique_ptr<const MCInstrInfo> ThumbInstrInfo;
7225   if (ThumbTarget) {
7226     ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo());
7227     CHECK_THUMB_TARGET_INFO_CREATION(ThumbInstrInfo);
7228   }
7229 
7230   // Package up features to be passed to target/subtarget
7231   std::string FeaturesStr;
7232   if (!MAttrs.empty()) {
7233     SubtargetFeatures Features;
7234     for (unsigned i = 0; i != MAttrs.size(); ++i)
7235       Features.AddFeature(MAttrs[i]);
7236     FeaturesStr = Features.getString();
7237   }
7238 
7239   MCTargetOptions MCOptions;
7240   // Set up disassembler.
7241   std::unique_ptr<const MCRegisterInfo> MRI(
7242       TheTarget->createMCRegInfo(TripleName));
7243   CHECK_TARGET_INFO_CREATION(MRI);
7244   std::unique_ptr<const MCAsmInfo> AsmInfo(
7245       TheTarget->createMCAsmInfo(*MRI, TripleName, MCOptions));
7246   CHECK_TARGET_INFO_CREATION(AsmInfo);
7247   std::unique_ptr<const MCSubtargetInfo> STI(
7248       TheTarget->createMCSubtargetInfo(TripleName, MachOMCPU, FeaturesStr));
7249   CHECK_TARGET_INFO_CREATION(STI);
7250   MCContext Ctx(AsmInfo.get(), MRI.get(), nullptr);
7251   std::unique_ptr<MCDisassembler> DisAsm(
7252       TheTarget->createMCDisassembler(*STI, Ctx));
7253   CHECK_TARGET_INFO_CREATION(DisAsm);
7254   std::unique_ptr<MCSymbolizer> Symbolizer;
7255   struct DisassembleInfo SymbolizerInfo(nullptr, nullptr, nullptr, false);
7256   std::unique_ptr<MCRelocationInfo> RelInfo(
7257       TheTarget->createMCRelocationInfo(TripleName, Ctx));
7258   if (RelInfo) {
7259     Symbolizer.reset(TheTarget->createMCSymbolizer(
7260         TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
7261         &SymbolizerInfo, &Ctx, std::move(RelInfo)));
7262     DisAsm->setSymbolizer(std::move(Symbolizer));
7263   }
7264   int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
7265   std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
7266       Triple(TripleName), AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI));
7267   CHECK_TARGET_INFO_CREATION(IP);
7268   // Set the display preference for hex vs. decimal immediates.
7269   IP->setPrintImmHex(PrintImmHex);
7270   // Comment stream and backing vector.
7271   SmallString<128> CommentsToEmit;
7272   raw_svector_ostream CommentStream(CommentsToEmit);
7273   // FIXME: Setting the CommentStream in the InstPrinter is problematic in that
7274   // if it is done then arm64 comments for string literals don't get printed
7275   // and some constant get printed instead and not setting it causes intel
7276   // (32-bit and 64-bit) comments printed with different spacing before the
7277   // comment causing different diffs with the 'C' disassembler library API.
7278   // IP->setCommentStream(CommentStream);
7279 
7280   // Set up separate thumb disassembler if needed.
7281   std::unique_ptr<const MCRegisterInfo> ThumbMRI;
7282   std::unique_ptr<const MCAsmInfo> ThumbAsmInfo;
7283   std::unique_ptr<const MCSubtargetInfo> ThumbSTI;
7284   std::unique_ptr<MCDisassembler> ThumbDisAsm;
7285   std::unique_ptr<MCInstPrinter> ThumbIP;
7286   std::unique_ptr<MCContext> ThumbCtx;
7287   std::unique_ptr<MCSymbolizer> ThumbSymbolizer;
7288   struct DisassembleInfo ThumbSymbolizerInfo(nullptr, nullptr, nullptr, false);
7289   std::unique_ptr<MCRelocationInfo> ThumbRelInfo;
7290   if (ThumbTarget) {
7291     ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName));
7292     CHECK_THUMB_TARGET_INFO_CREATION(ThumbMRI);
7293     ThumbAsmInfo.reset(
7294         ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName, MCOptions));
7295     CHECK_THUMB_TARGET_INFO_CREATION(ThumbAsmInfo);
7296     ThumbSTI.reset(
7297         ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MachOMCPU,
7298                                            FeaturesStr));
7299     CHECK_THUMB_TARGET_INFO_CREATION(ThumbSTI);
7300     ThumbCtx.reset(new MCContext(ThumbAsmInfo.get(), ThumbMRI.get(), nullptr));
7301     ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx));
7302     CHECK_THUMB_TARGET_INFO_CREATION(ThumbDisAsm);
7303     MCContext *PtrThumbCtx = ThumbCtx.get();
7304     ThumbRelInfo.reset(
7305         ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx));
7306     if (ThumbRelInfo) {
7307       ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer(
7308           ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
7309           &ThumbSymbolizerInfo, PtrThumbCtx, std::move(ThumbRelInfo)));
7310       ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer));
7311     }
7312     int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect();
7313     ThumbIP.reset(ThumbTarget->createMCInstPrinter(
7314         Triple(ThumbTripleName), ThumbAsmPrinterVariant, *ThumbAsmInfo,
7315         *ThumbInstrInfo, *ThumbMRI));
7316     CHECK_THUMB_TARGET_INFO_CREATION(ThumbIP);
7317     // Set the display preference for hex vs. decimal immediates.
7318     ThumbIP->setPrintImmHex(PrintImmHex);
7319   }
7320 
7321 #undef CHECK_TARGET_INFO_CREATION
7322 #undef CHECK_THUMB_TARGET_INFO_CREATION
7323 
7324   MachO::mach_header Header = MachOOF->getHeader();
7325 
7326   // FIXME: Using the -cfg command line option, this code used to be able to
7327   // annotate relocations with the referenced symbol's name, and if this was
7328   // inside a __[cf]string section, the data it points to. This is now replaced
7329   // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
7330   std::vector<SectionRef> Sections;
7331   std::vector<SymbolRef> Symbols;
7332   SmallVector<uint64_t, 8> FoundFns;
7333   uint64_t BaseSegmentAddress = 0;
7334 
7335   getSectionsAndSymbols(MachOOF, Sections, Symbols, FoundFns,
7336                         BaseSegmentAddress);
7337 
7338   // Sort the symbols by address, just in case they didn't come in that way.
7339   llvm::sort(Symbols, SymbolSorter());
7340 
7341   // Build a data in code table that is sorted on by the address of each entry.
7342   uint64_t BaseAddress = 0;
7343   if (Header.filetype == MachO::MH_OBJECT)
7344     BaseAddress = Sections[0].getAddress();
7345   else
7346     BaseAddress = BaseSegmentAddress;
7347   DiceTable Dices;
7348   for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
7349        DI != DE; ++DI) {
7350     uint32_t Offset;
7351     DI->getOffset(Offset);
7352     Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
7353   }
7354   array_pod_sort(Dices.begin(), Dices.end());
7355 
7356   // Try to find debug info and set up the DIContext for it.
7357   std::unique_ptr<DIContext> diContext;
7358   std::unique_ptr<Binary> DSYMBinary;
7359   std::unique_ptr<MemoryBuffer> DSYMBuf;
7360   if (UseDbg) {
7361     ObjectFile *DbgObj = MachOOF;
7362 
7363     // A separate DSym file path was specified, parse it as a macho file,
7364     // get the sections and supply it to the section name parsing machinery.
7365     if (!DSYMFile.empty()) {
7366       std::string DSYMPath(DSYMFile);
7367 
7368       // If DSYMPath is a .dSYM directory, append the Mach-O file.
7369       if (llvm::sys::fs::is_directory(DSYMPath) &&
7370           llvm::sys::path::extension(DSYMPath) == ".dSYM") {
7371         SmallString<128> ShortName(llvm::sys::path::filename(DSYMPath));
7372         llvm::sys::path::replace_extension(ShortName, "");
7373         SmallString<1024> FullPath(DSYMPath);
7374         llvm::sys::path::append(FullPath, "Contents", "Resources", "DWARF",
7375                                 ShortName);
7376         DSYMPath = std::string(FullPath.str());
7377       }
7378 
7379       // Load the file.
7380       ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr =
7381           MemoryBuffer::getFileOrSTDIN(DSYMPath);
7382       if (std::error_code EC = BufOrErr.getError()) {
7383         reportError(errorCodeToError(EC), DSYMPath);
7384         return;
7385       }
7386 
7387       // We need to keep the file alive, because we're replacing DbgObj with it.
7388       DSYMBuf = std::move(BufOrErr.get());
7389 
7390       Expected<std::unique_ptr<Binary>> BinaryOrErr =
7391       createBinary(DSYMBuf.get()->getMemBufferRef());
7392       if (!BinaryOrErr) {
7393         reportError(BinaryOrErr.takeError(), DSYMPath);
7394         return;
7395       }
7396 
7397       // We need to keep the Binary alive with the buffer
7398       DSYMBinary = std::move(BinaryOrErr.get());
7399       if (ObjectFile *O = dyn_cast<ObjectFile>(DSYMBinary.get())) {
7400         // this is a Mach-O object file, use it
7401         if (MachOObjectFile *MachDSYM = dyn_cast<MachOObjectFile>(&*O)) {
7402           DbgObj = MachDSYM;
7403         }
7404         else {
7405           WithColor::error(errs(), "llvm-objdump")
7406             << DSYMPath << " is not a Mach-O file type.\n";
7407           return;
7408         }
7409       }
7410       else if (auto UB = dyn_cast<MachOUniversalBinary>(DSYMBinary.get())){
7411         // this is a Universal Binary, find a Mach-O for this architecture
7412         uint32_t CPUType, CPUSubType;
7413         const char *ArchFlag;
7414         if (MachOOF->is64Bit()) {
7415           const MachO::mach_header_64 H_64 = MachOOF->getHeader64();
7416           CPUType = H_64.cputype;
7417           CPUSubType = H_64.cpusubtype;
7418         } else {
7419           const MachO::mach_header H = MachOOF->getHeader();
7420           CPUType = H.cputype;
7421           CPUSubType = H.cpusubtype;
7422         }
7423         Triple T = MachOObjectFile::getArchTriple(CPUType, CPUSubType, nullptr,
7424                                                   &ArchFlag);
7425         Expected<std::unique_ptr<MachOObjectFile>> MachDSYM =
7426             UB->getMachOObjectForArch(ArchFlag);
7427         if (!MachDSYM) {
7428           reportError(MachDSYM.takeError(), DSYMPath);
7429           return;
7430         }
7431 
7432         // We need to keep the Binary alive with the buffer
7433         DbgObj = &*MachDSYM.get();
7434         DSYMBinary = std::move(*MachDSYM);
7435       }
7436       else {
7437         WithColor::error(errs(), "llvm-objdump")
7438           << DSYMPath << " is not a Mach-O or Universal file type.\n";
7439         return;
7440       }
7441     }
7442 
7443     // Setup the DIContext
7444     diContext = DWARFContext::create(*DbgObj);
7445   }
7446 
7447   if (FilterSections.empty())
7448     outs() << "(" << DisSegName << "," << DisSectName << ") section\n";
7449 
7450   for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
7451     Expected<StringRef> SecNameOrErr = Sections[SectIdx].getName();
7452     if (!SecNameOrErr) {
7453       consumeError(SecNameOrErr.takeError());
7454       continue;
7455     }
7456     if (*SecNameOrErr != DisSectName)
7457       continue;
7458 
7459     DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
7460 
7461     StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
7462     if (SegmentName != DisSegName)
7463       continue;
7464 
7465     StringRef BytesStr =
7466         unwrapOrError(Sections[SectIdx].getContents(), Filename);
7467     ArrayRef<uint8_t> Bytes = arrayRefFromStringRef(BytesStr);
7468     uint64_t SectAddress = Sections[SectIdx].getAddress();
7469 
7470     bool symbolTableWorked = false;
7471 
7472     // Create a map of symbol addresses to symbol names for use by
7473     // the SymbolizerSymbolLookUp() routine.
7474     SymbolAddressMap AddrMap;
7475     bool DisSymNameFound = false;
7476     for (const SymbolRef &Symbol : MachOOF->symbols()) {
7477       SymbolRef::Type ST =
7478           unwrapOrError(Symbol.getType(), MachOOF->getFileName());
7479       if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
7480           ST == SymbolRef::ST_Other) {
7481         uint64_t Address = cantFail(Symbol.getValue());
7482         StringRef SymName =
7483             unwrapOrError(Symbol.getName(), MachOOF->getFileName());
7484         AddrMap[Address] = SymName;
7485         if (!DisSymName.empty() && DisSymName == SymName)
7486           DisSymNameFound = true;
7487       }
7488     }
7489     if (!DisSymName.empty() && !DisSymNameFound) {
7490       outs() << "Can't find -dis-symname: " << DisSymName << "\n";
7491       return;
7492     }
7493     // Set up the block of info used by the Symbolizer call backs.
7494     SymbolizerInfo.verbose = !NoSymbolicOperands;
7495     SymbolizerInfo.O = MachOOF;
7496     SymbolizerInfo.S = Sections[SectIdx];
7497     SymbolizerInfo.AddrMap = &AddrMap;
7498     SymbolizerInfo.Sections = &Sections;
7499     // Same for the ThumbSymbolizer
7500     ThumbSymbolizerInfo.verbose = !NoSymbolicOperands;
7501     ThumbSymbolizerInfo.O = MachOOF;
7502     ThumbSymbolizerInfo.S = Sections[SectIdx];
7503     ThumbSymbolizerInfo.AddrMap = &AddrMap;
7504     ThumbSymbolizerInfo.Sections = &Sections;
7505 
7506     unsigned int Arch = MachOOF->getArch();
7507 
7508     // Skip all symbols if this is a stubs file.
7509     if (Bytes.empty())
7510       return;
7511 
7512     // If the section has symbols but no symbol at the start of the section
7513     // these are used to make sure the bytes before the first symbol are
7514     // disassembled.
7515     bool FirstSymbol = true;
7516     bool FirstSymbolAtSectionStart = true;
7517 
7518     // Disassemble symbol by symbol.
7519     for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
7520       StringRef SymName =
7521           unwrapOrError(Symbols[SymIdx].getName(), MachOOF->getFileName());
7522       SymbolRef::Type ST =
7523           unwrapOrError(Symbols[SymIdx].getType(), MachOOF->getFileName());
7524       if (ST != SymbolRef::ST_Function && ST != SymbolRef::ST_Data)
7525         continue;
7526 
7527       // Make sure the symbol is defined in this section.
7528       bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]);
7529       if (!containsSym) {
7530         if (!DisSymName.empty() && DisSymName == SymName) {
7531           outs() << "-dis-symname: " << DisSymName << " not in the section\n";
7532           return;
7533         }
7534         continue;
7535       }
7536       // The __mh_execute_header is special and we need to deal with that fact
7537       // this symbol is before the start of the (__TEXT,__text) section and at the
7538       // address of the start of the __TEXT segment.  This is because this symbol
7539       // is an N_SECT symbol in the (__TEXT,__text) but its address is before the
7540       // start of the section in a standard MH_EXECUTE filetype.
7541       if (!DisSymName.empty() && DisSymName == "__mh_execute_header") {
7542         outs() << "-dis-symname: __mh_execute_header not in any section\n";
7543         return;
7544       }
7545       // When this code is trying to disassemble a symbol at a time and in the
7546       // case there is only the __mh_execute_header symbol left as in a stripped
7547       // executable, we need to deal with this by ignoring this symbol so the
7548       // whole section is disassembled and this symbol is then not displayed.
7549       if (SymName == "__mh_execute_header" || SymName == "__mh_dylib_header" ||
7550           SymName == "__mh_bundle_header" || SymName == "__mh_object_header" ||
7551           SymName == "__mh_preload_header" || SymName == "__mh_dylinker_header")
7552         continue;
7553 
7554       // If we are only disassembling one symbol see if this is that symbol.
7555       if (!DisSymName.empty() && DisSymName != SymName)
7556         continue;
7557 
7558       // Start at the address of the symbol relative to the section's address.
7559       uint64_t SectSize = Sections[SectIdx].getSize();
7560       uint64_t Start = cantFail(Symbols[SymIdx].getValue());
7561       uint64_t SectionAddress = Sections[SectIdx].getAddress();
7562       Start -= SectionAddress;
7563 
7564       if (Start > SectSize) {
7565         outs() << "section data ends, " << SymName
7566                << " lies outside valid range\n";
7567         return;
7568       }
7569 
7570       // Stop disassembling either at the beginning of the next symbol or at
7571       // the end of the section.
7572       bool containsNextSym = false;
7573       uint64_t NextSym = 0;
7574       uint64_t NextSymIdx = SymIdx + 1;
7575       while (Symbols.size() > NextSymIdx) {
7576         SymbolRef::Type NextSymType = unwrapOrError(
7577             Symbols[NextSymIdx].getType(), MachOOF->getFileName());
7578         if (NextSymType == SymbolRef::ST_Function) {
7579           containsNextSym =
7580               Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]);
7581           NextSym = cantFail(Symbols[NextSymIdx].getValue());
7582           NextSym -= SectionAddress;
7583           break;
7584         }
7585         ++NextSymIdx;
7586       }
7587 
7588       uint64_t End = containsNextSym ? std::min(NextSym, SectSize) : SectSize;
7589       uint64_t Size;
7590 
7591       symbolTableWorked = true;
7592 
7593       DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl();
7594       uint32_t SymbolFlags = cantFail(MachOOF->getSymbolFlags(Symb));
7595       bool IsThumb = SymbolFlags & SymbolRef::SF_Thumb;
7596 
7597       // We only need the dedicated Thumb target if there's a real choice
7598       // (i.e. we're not targeting M-class) and the function is Thumb.
7599       bool UseThumbTarget = IsThumb && ThumbTarget;
7600 
7601       // If we are not specifying a symbol to start disassembly with and this
7602       // is the first symbol in the section but not at the start of the section
7603       // then move the disassembly index to the start of the section and
7604       // don't print the symbol name just yet.  This is so the bytes before the
7605       // first symbol are disassembled.
7606       uint64_t SymbolStart = Start;
7607       if (DisSymName.empty() && FirstSymbol && Start != 0) {
7608         FirstSymbolAtSectionStart = false;
7609         Start = 0;
7610       }
7611       else
7612         outs() << SymName << ":\n";
7613 
7614       DILineInfo lastLine;
7615       for (uint64_t Index = Start; Index < End; Index += Size) {
7616         MCInst Inst;
7617 
7618         // If this is the first symbol in the section and it was not at the
7619         // start of the section, see if we are at its Index now and if so print
7620         // the symbol name.
7621         if (FirstSymbol && !FirstSymbolAtSectionStart && Index == SymbolStart)
7622           outs() << SymName << ":\n";
7623 
7624         uint64_t PC = SectAddress + Index;
7625         if (!NoLeadingAddr) {
7626           if (FullLeadingAddr) {
7627             if (MachOOF->is64Bit())
7628               outs() << format("%016" PRIx64, PC);
7629             else
7630               outs() << format("%08" PRIx64, PC);
7631           } else {
7632             outs() << format("%8" PRIx64 ":", PC);
7633           }
7634         }
7635         if (!NoShowRawInsn || Arch == Triple::arm)
7636           outs() << "\t";
7637 
7638         if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, Size))
7639           continue;
7640 
7641         SmallVector<char, 64> AnnotationsBytes;
7642         raw_svector_ostream Annotations(AnnotationsBytes);
7643 
7644         bool gotInst;
7645         if (UseThumbTarget)
7646           gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index),
7647                                                 PC, Annotations);
7648         else
7649           gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC,
7650                                            Annotations);
7651         if (gotInst) {
7652           if (!NoShowRawInsn || Arch == Triple::arm) {
7653             dumpBytes(makeArrayRef(Bytes.data() + Index, Size), outs());
7654           }
7655           formatted_raw_ostream FormattedOS(outs());
7656           StringRef AnnotationsStr = Annotations.str();
7657           if (UseThumbTarget)
7658             ThumbIP->printInst(&Inst, PC, AnnotationsStr, *ThumbSTI,
7659                                FormattedOS);
7660           else
7661             IP->printInst(&Inst, PC, AnnotationsStr, *STI, FormattedOS);
7662           emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo);
7663 
7664           // Print debug info.
7665           if (diContext) {
7666             DILineInfo dli = diContext->getLineInfoForAddress({PC, SectIdx});
7667             // Print valid line info if it changed.
7668             if (dli != lastLine && dli.Line != 0)
7669               outs() << "\t## " << dli.FileName << ':' << dli.Line << ':'
7670                      << dli.Column;
7671             lastLine = dli;
7672           }
7673           outs() << "\n";
7674         } else {
7675           if (MachOOF->getArchTriple().isX86()) {
7676             outs() << format("\t.byte 0x%02x #bad opcode\n",
7677                              *(Bytes.data() + Index) & 0xff);
7678             Size = 1; // skip exactly one illegible byte and move on.
7679           } else if (Arch == Triple::aarch64 ||
7680                      (Arch == Triple::arm && !IsThumb)) {
7681             uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
7682                               (*(Bytes.data() + Index + 1) & 0xff) << 8 |
7683                               (*(Bytes.data() + Index + 2) & 0xff) << 16 |
7684                               (*(Bytes.data() + Index + 3) & 0xff) << 24;
7685             outs() << format("\t.long\t0x%08x\n", opcode);
7686             Size = 4;
7687           } else if (Arch == Triple::arm) {
7688             assert(IsThumb && "ARM mode should have been dealt with above");
7689             uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
7690                               (*(Bytes.data() + Index + 1) & 0xff) << 8;
7691             outs() << format("\t.short\t0x%04x\n", opcode);
7692             Size = 2;
7693           } else{
7694             WithColor::warning(errs(), "llvm-objdump")
7695                 << "invalid instruction encoding\n";
7696             if (Size == 0)
7697               Size = 1; // skip illegible bytes
7698           }
7699         }
7700       }
7701       // Now that we are done disassembled the first symbol set the bool that
7702       // were doing this to false.
7703       FirstSymbol = false;
7704     }
7705     if (!symbolTableWorked) {
7706       // Reading the symbol table didn't work, disassemble the whole section.
7707       uint64_t SectAddress = Sections[SectIdx].getAddress();
7708       uint64_t SectSize = Sections[SectIdx].getSize();
7709       uint64_t InstSize;
7710       for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
7711         MCInst Inst;
7712 
7713         uint64_t PC = SectAddress + Index;
7714 
7715         if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, InstSize))
7716           continue;
7717 
7718         SmallVector<char, 64> AnnotationsBytes;
7719         raw_svector_ostream Annotations(AnnotationsBytes);
7720         if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC,
7721                                    Annotations)) {
7722           if (!NoLeadingAddr) {
7723             if (FullLeadingAddr) {
7724               if (MachOOF->is64Bit())
7725                 outs() << format("%016" PRIx64, PC);
7726               else
7727                 outs() << format("%08" PRIx64, PC);
7728             } else {
7729               outs() << format("%8" PRIx64 ":", PC);
7730             }
7731           }
7732           if (!NoShowRawInsn || Arch == Triple::arm) {
7733             outs() << "\t";
7734             dumpBytes(makeArrayRef(Bytes.data() + Index, InstSize), outs());
7735           }
7736           StringRef AnnotationsStr = Annotations.str();
7737           IP->printInst(&Inst, PC, AnnotationsStr, *STI, outs());
7738           outs() << "\n";
7739         } else {
7740           if (MachOOF->getArchTriple().isX86()) {
7741             outs() << format("\t.byte 0x%02x #bad opcode\n",
7742                              *(Bytes.data() + Index) & 0xff);
7743             InstSize = 1; // skip exactly one illegible byte and move on.
7744           } else {
7745             WithColor::warning(errs(), "llvm-objdump")
7746                 << "invalid instruction encoding\n";
7747             if (InstSize == 0)
7748               InstSize = 1; // skip illegible bytes
7749           }
7750         }
7751       }
7752     }
7753     // The TripleName's need to be reset if we are called again for a different
7754     // architecture.
7755     TripleName = "";
7756     ThumbTripleName = "";
7757 
7758     if (SymbolizerInfo.demangled_name != nullptr)
7759       free(SymbolizerInfo.demangled_name);
7760     if (ThumbSymbolizerInfo.demangled_name != nullptr)
7761       free(ThumbSymbolizerInfo.demangled_name);
7762   }
7763 }
7764 
7765 //===----------------------------------------------------------------------===//
7766 // __compact_unwind section dumping
7767 //===----------------------------------------------------------------------===//
7768 
7769 namespace {
7770 
7771 template <typename T>
read(StringRef Contents,ptrdiff_t Offset)7772 static uint64_t read(StringRef Contents, ptrdiff_t Offset) {
7773   using llvm::support::little;
7774   using llvm::support::unaligned;
7775 
7776   if (Offset + sizeof(T) > Contents.size()) {
7777     outs() << "warning: attempt to read past end of buffer\n";
7778     return T();
7779   }
7780 
7781   uint64_t Val =
7782       support::endian::read<T, little, unaligned>(Contents.data() + Offset);
7783   return Val;
7784 }
7785 
7786 template <typename T>
readNext(StringRef Contents,ptrdiff_t & Offset)7787 static uint64_t readNext(StringRef Contents, ptrdiff_t &Offset) {
7788   T Val = read<T>(Contents, Offset);
7789   Offset += sizeof(T);
7790   return Val;
7791 }
7792 
7793 struct CompactUnwindEntry {
7794   uint32_t OffsetInSection;
7795 
7796   uint64_t FunctionAddr;
7797   uint32_t Length;
7798   uint32_t CompactEncoding;
7799   uint64_t PersonalityAddr;
7800   uint64_t LSDAAddr;
7801 
7802   RelocationRef FunctionReloc;
7803   RelocationRef PersonalityReloc;
7804   RelocationRef LSDAReloc;
7805 
CompactUnwindEntry__anon30756ae00b11::CompactUnwindEntry7806   CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64)
7807       : OffsetInSection(Offset) {
7808     if (Is64)
7809       read<uint64_t>(Contents, Offset);
7810     else
7811       read<uint32_t>(Contents, Offset);
7812   }
7813 
7814 private:
read__anon30756ae00b11::CompactUnwindEntry7815   template <typename UIntPtr> void read(StringRef Contents, ptrdiff_t Offset) {
7816     FunctionAddr = readNext<UIntPtr>(Contents, Offset);
7817     Length = readNext<uint32_t>(Contents, Offset);
7818     CompactEncoding = readNext<uint32_t>(Contents, Offset);
7819     PersonalityAddr = readNext<UIntPtr>(Contents, Offset);
7820     LSDAAddr = readNext<UIntPtr>(Contents, Offset);
7821   }
7822 };
7823 }
7824 
7825 /// Given a relocation from __compact_unwind, consisting of the RelocationRef
7826 /// and data being relocated, determine the best base Name and Addend to use for
7827 /// display purposes.
7828 ///
7829 /// 1. An Extern relocation will directly reference a symbol (and the data is
7830 ///    then already an addend), so use that.
7831 /// 2. Otherwise the data is an offset in the object file's layout; try to find
7832 //     a symbol before it in the same section, and use the offset from there.
7833 /// 3. Finally, if all that fails, fall back to an offset from the start of the
7834 ///    referenced section.
findUnwindRelocNameAddend(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const RelocationRef & Reloc,uint64_t Addr,StringRef & Name,uint64_t & Addend)7835 static void findUnwindRelocNameAddend(const MachOObjectFile *Obj,
7836                                       std::map<uint64_t, SymbolRef> &Symbols,
7837                                       const RelocationRef &Reloc, uint64_t Addr,
7838                                       StringRef &Name, uint64_t &Addend) {
7839   if (Reloc.getSymbol() != Obj->symbol_end()) {
7840     Name = unwrapOrError(Reloc.getSymbol()->getName(), Obj->getFileName());
7841     Addend = Addr;
7842     return;
7843   }
7844 
7845   auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl());
7846   SectionRef RelocSection = Obj->getAnyRelocationSection(RE);
7847 
7848   uint64_t SectionAddr = RelocSection.getAddress();
7849 
7850   auto Sym = Symbols.upper_bound(Addr);
7851   if (Sym == Symbols.begin()) {
7852     // The first symbol in the object is after this reference, the best we can
7853     // do is section-relative notation.
7854     if (Expected<StringRef> NameOrErr = RelocSection.getName())
7855       Name = *NameOrErr;
7856     else
7857       consumeError(NameOrErr.takeError());
7858 
7859     Addend = Addr - SectionAddr;
7860     return;
7861   }
7862 
7863   // Go back one so that SymbolAddress <= Addr.
7864   --Sym;
7865 
7866   section_iterator SymSection =
7867       unwrapOrError(Sym->second.getSection(), Obj->getFileName());
7868   if (RelocSection == *SymSection) {
7869     // There's a valid symbol in the same section before this reference.
7870     Name = unwrapOrError(Sym->second.getName(), Obj->getFileName());
7871     Addend = Addr - Sym->first;
7872     return;
7873   }
7874 
7875   // There is a symbol before this reference, but it's in a different
7876   // section. Probably not helpful to mention it, so use the section name.
7877   if (Expected<StringRef> NameOrErr = RelocSection.getName())
7878     Name = *NameOrErr;
7879   else
7880     consumeError(NameOrErr.takeError());
7881 
7882   Addend = Addr - SectionAddr;
7883 }
7884 
printUnwindRelocDest(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const RelocationRef & Reloc,uint64_t Addr)7885 static void printUnwindRelocDest(const MachOObjectFile *Obj,
7886                                  std::map<uint64_t, SymbolRef> &Symbols,
7887                                  const RelocationRef &Reloc, uint64_t Addr) {
7888   StringRef Name;
7889   uint64_t Addend;
7890 
7891   if (!Reloc.getObject())
7892     return;
7893 
7894   findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend);
7895 
7896   outs() << Name;
7897   if (Addend)
7898     outs() << " + " << format("0x%" PRIx64, Addend);
7899 }
7900 
7901 static void
printMachOCompactUnwindSection(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const SectionRef & CompactUnwind)7902 printMachOCompactUnwindSection(const MachOObjectFile *Obj,
7903                                std::map<uint64_t, SymbolRef> &Symbols,
7904                                const SectionRef &CompactUnwind) {
7905 
7906   if (!Obj->isLittleEndian()) {
7907     outs() << "Skipping big-endian __compact_unwind section\n";
7908     return;
7909   }
7910 
7911   bool Is64 = Obj->is64Bit();
7912   uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t);
7913   uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t);
7914 
7915   StringRef Contents =
7916       unwrapOrError(CompactUnwind.getContents(), Obj->getFileName());
7917   SmallVector<CompactUnwindEntry, 4> CompactUnwinds;
7918 
7919   // First populate the initial raw offsets, encodings and so on from the entry.
7920   for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) {
7921     CompactUnwindEntry Entry(Contents, Offset, Is64);
7922     CompactUnwinds.push_back(Entry);
7923   }
7924 
7925   // Next we need to look at the relocations to find out what objects are
7926   // actually being referred to.
7927   for (const RelocationRef &Reloc : CompactUnwind.relocations()) {
7928     uint64_t RelocAddress = Reloc.getOffset();
7929 
7930     uint32_t EntryIdx = RelocAddress / EntrySize;
7931     uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize;
7932     CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx];
7933 
7934     if (OffsetInEntry == 0)
7935       Entry.FunctionReloc = Reloc;
7936     else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t))
7937       Entry.PersonalityReloc = Reloc;
7938     else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t))
7939       Entry.LSDAReloc = Reloc;
7940     else {
7941       outs() << "Invalid relocation in __compact_unwind section\n";
7942       return;
7943     }
7944   }
7945 
7946   // Finally, we're ready to print the data we've gathered.
7947   outs() << "Contents of __compact_unwind section:\n";
7948   for (auto &Entry : CompactUnwinds) {
7949     outs() << "  Entry at offset "
7950            << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n";
7951 
7952     // 1. Start of the region this entry applies to.
7953     outs() << "    start:                " << format("0x%" PRIx64,
7954                                                      Entry.FunctionAddr) << ' ';
7955     printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr);
7956     outs() << '\n';
7957 
7958     // 2. Length of the region this entry applies to.
7959     outs() << "    length:               " << format("0x%" PRIx32, Entry.Length)
7960            << '\n';
7961     // 3. The 32-bit compact encoding.
7962     outs() << "    compact encoding:     "
7963            << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n';
7964 
7965     // 4. The personality function, if present.
7966     if (Entry.PersonalityReloc.getObject()) {
7967       outs() << "    personality function: "
7968              << format("0x%" PRIx64, Entry.PersonalityAddr) << ' ';
7969       printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc,
7970                            Entry.PersonalityAddr);
7971       outs() << '\n';
7972     }
7973 
7974     // 5. This entry's language-specific data area.
7975     if (Entry.LSDAReloc.getObject()) {
7976       outs() << "    LSDA:                 " << format("0x%" PRIx64,
7977                                                        Entry.LSDAAddr) << ' ';
7978       printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr);
7979       outs() << '\n';
7980     }
7981   }
7982 }
7983 
7984 //===----------------------------------------------------------------------===//
7985 // __unwind_info section dumping
7986 //===----------------------------------------------------------------------===//
7987 
printRegularSecondLevelUnwindPage(StringRef PageData)7988 static void printRegularSecondLevelUnwindPage(StringRef PageData) {
7989   ptrdiff_t Pos = 0;
7990   uint32_t Kind = readNext<uint32_t>(PageData, Pos);
7991   (void)Kind;
7992   assert(Kind == 2 && "kind for a regular 2nd level index should be 2");
7993 
7994   uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos);
7995   uint16_t NumEntries = readNext<uint16_t>(PageData, Pos);
7996 
7997   Pos = EntriesStart;
7998   for (unsigned i = 0; i < NumEntries; ++i) {
7999     uint32_t FunctionOffset = readNext<uint32_t>(PageData, Pos);
8000     uint32_t Encoding = readNext<uint32_t>(PageData, Pos);
8001 
8002     outs() << "      [" << i << "]: "
8003            << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8004            << ", "
8005            << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n';
8006   }
8007 }
8008 
printCompressedSecondLevelUnwindPage(StringRef PageData,uint32_t FunctionBase,const SmallVectorImpl<uint32_t> & CommonEncodings)8009 static void printCompressedSecondLevelUnwindPage(
8010     StringRef PageData, uint32_t FunctionBase,
8011     const SmallVectorImpl<uint32_t> &CommonEncodings) {
8012   ptrdiff_t Pos = 0;
8013   uint32_t Kind = readNext<uint32_t>(PageData, Pos);
8014   (void)Kind;
8015   assert(Kind == 3 && "kind for a compressed 2nd level index should be 3");
8016 
8017   uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos);
8018   uint16_t NumEntries = readNext<uint16_t>(PageData, Pos);
8019 
8020   uint16_t EncodingsStart = readNext<uint16_t>(PageData, Pos);
8021   readNext<uint16_t>(PageData, Pos);
8022   StringRef PageEncodings = PageData.substr(EncodingsStart, StringRef::npos);
8023 
8024   Pos = EntriesStart;
8025   for (unsigned i = 0; i < NumEntries; ++i) {
8026     uint32_t Entry = readNext<uint32_t>(PageData, Pos);
8027     uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff);
8028     uint32_t EncodingIdx = Entry >> 24;
8029 
8030     uint32_t Encoding;
8031     if (EncodingIdx < CommonEncodings.size())
8032       Encoding = CommonEncodings[EncodingIdx];
8033     else
8034       Encoding = read<uint32_t>(PageEncodings,
8035                                 sizeof(uint32_t) *
8036                                     (EncodingIdx - CommonEncodings.size()));
8037 
8038     outs() << "      [" << i << "]: "
8039            << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8040            << ", "
8041            << "encoding[" << EncodingIdx
8042            << "]=" << format("0x%08" PRIx32, Encoding) << '\n';
8043   }
8044 }
8045 
printMachOUnwindInfoSection(const MachOObjectFile * Obj,std::map<uint64_t,SymbolRef> & Symbols,const SectionRef & UnwindInfo)8046 static void printMachOUnwindInfoSection(const MachOObjectFile *Obj,
8047                                         std::map<uint64_t, SymbolRef> &Symbols,
8048                                         const SectionRef &UnwindInfo) {
8049 
8050   if (!Obj->isLittleEndian()) {
8051     outs() << "Skipping big-endian __unwind_info section\n";
8052     return;
8053   }
8054 
8055   outs() << "Contents of __unwind_info section:\n";
8056 
8057   StringRef Contents =
8058       unwrapOrError(UnwindInfo.getContents(), Obj->getFileName());
8059   ptrdiff_t Pos = 0;
8060 
8061   //===----------------------------------
8062   // Section header
8063   //===----------------------------------
8064 
8065   uint32_t Version = readNext<uint32_t>(Contents, Pos);
8066   outs() << "  Version:                                   "
8067          << format("0x%" PRIx32, Version) << '\n';
8068   if (Version != 1) {
8069     outs() << "    Skipping section with unknown version\n";
8070     return;
8071   }
8072 
8073   uint32_t CommonEncodingsStart = readNext<uint32_t>(Contents, Pos);
8074   outs() << "  Common encodings array section offset:     "
8075          << format("0x%" PRIx32, CommonEncodingsStart) << '\n';
8076   uint32_t NumCommonEncodings = readNext<uint32_t>(Contents, Pos);
8077   outs() << "  Number of common encodings in array:       "
8078          << format("0x%" PRIx32, NumCommonEncodings) << '\n';
8079 
8080   uint32_t PersonalitiesStart = readNext<uint32_t>(Contents, Pos);
8081   outs() << "  Personality function array section offset: "
8082          << format("0x%" PRIx32, PersonalitiesStart) << '\n';
8083   uint32_t NumPersonalities = readNext<uint32_t>(Contents, Pos);
8084   outs() << "  Number of personality functions in array:  "
8085          << format("0x%" PRIx32, NumPersonalities) << '\n';
8086 
8087   uint32_t IndicesStart = readNext<uint32_t>(Contents, Pos);
8088   outs() << "  Index array section offset:                "
8089          << format("0x%" PRIx32, IndicesStart) << '\n';
8090   uint32_t NumIndices = readNext<uint32_t>(Contents, Pos);
8091   outs() << "  Number of indices in array:                "
8092          << format("0x%" PRIx32, NumIndices) << '\n';
8093 
8094   //===----------------------------------
8095   // A shared list of common encodings
8096   //===----------------------------------
8097 
8098   // These occupy indices in the range [0, N] whenever an encoding is referenced
8099   // from a compressed 2nd level index table. In practice the linker only
8100   // creates ~128 of these, so that indices are available to embed encodings in
8101   // the 2nd level index.
8102 
8103   SmallVector<uint32_t, 64> CommonEncodings;
8104   outs() << "  Common encodings: (count = " << NumCommonEncodings << ")\n";
8105   Pos = CommonEncodingsStart;
8106   for (unsigned i = 0; i < NumCommonEncodings; ++i) {
8107     uint32_t Encoding = readNext<uint32_t>(Contents, Pos);
8108     CommonEncodings.push_back(Encoding);
8109 
8110     outs() << "    encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding)
8111            << '\n';
8112   }
8113 
8114   //===----------------------------------
8115   // Personality functions used in this executable
8116   //===----------------------------------
8117 
8118   // There should be only a handful of these (one per source language,
8119   // roughly). Particularly since they only get 2 bits in the compact encoding.
8120 
8121   outs() << "  Personality functions: (count = " << NumPersonalities << ")\n";
8122   Pos = PersonalitiesStart;
8123   for (unsigned i = 0; i < NumPersonalities; ++i) {
8124     uint32_t PersonalityFn = readNext<uint32_t>(Contents, Pos);
8125     outs() << "    personality[" << i + 1
8126            << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n';
8127   }
8128 
8129   //===----------------------------------
8130   // The level 1 index entries
8131   //===----------------------------------
8132 
8133   // These specify an approximate place to start searching for the more detailed
8134   // information, sorted by PC.
8135 
8136   struct IndexEntry {
8137     uint32_t FunctionOffset;
8138     uint32_t SecondLevelPageStart;
8139     uint32_t LSDAStart;
8140   };
8141 
8142   SmallVector<IndexEntry, 4> IndexEntries;
8143 
8144   outs() << "  Top level indices: (count = " << NumIndices << ")\n";
8145   Pos = IndicesStart;
8146   for (unsigned i = 0; i < NumIndices; ++i) {
8147     IndexEntry Entry;
8148 
8149     Entry.FunctionOffset = readNext<uint32_t>(Contents, Pos);
8150     Entry.SecondLevelPageStart = readNext<uint32_t>(Contents, Pos);
8151     Entry.LSDAStart = readNext<uint32_t>(Contents, Pos);
8152     IndexEntries.push_back(Entry);
8153 
8154     outs() << "    [" << i << "]: "
8155            << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset)
8156            << ", "
8157            << "2nd level page offset="
8158            << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", "
8159            << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n';
8160   }
8161 
8162   //===----------------------------------
8163   // Next come the LSDA tables
8164   //===----------------------------------
8165 
8166   // The LSDA layout is rather implicit: it's a contiguous array of entries from
8167   // the first top-level index's LSDAOffset to the last (sentinel).
8168 
8169   outs() << "  LSDA descriptors:\n";
8170   Pos = IndexEntries[0].LSDAStart;
8171   const uint32_t LSDASize = 2 * sizeof(uint32_t);
8172   int NumLSDAs =
8173       (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) / LSDASize;
8174 
8175   for (int i = 0; i < NumLSDAs; ++i) {
8176     uint32_t FunctionOffset = readNext<uint32_t>(Contents, Pos);
8177     uint32_t LSDAOffset = readNext<uint32_t>(Contents, Pos);
8178     outs() << "    [" << i << "]: "
8179            << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
8180            << ", "
8181            << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n';
8182   }
8183 
8184   //===----------------------------------
8185   // Finally, the 2nd level indices
8186   //===----------------------------------
8187 
8188   // Generally these are 4K in size, and have 2 possible forms:
8189   //   + Regular stores up to 511 entries with disparate encodings
8190   //   + Compressed stores up to 1021 entries if few enough compact encoding
8191   //     values are used.
8192   outs() << "  Second level indices:\n";
8193   for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) {
8194     // The final sentinel top-level index has no associated 2nd level page
8195     if (IndexEntries[i].SecondLevelPageStart == 0)
8196       break;
8197 
8198     outs() << "    Second level index[" << i << "]: "
8199            << "offset in section="
8200            << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart)
8201            << ", "
8202            << "base function offset="
8203            << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n';
8204 
8205     Pos = IndexEntries[i].SecondLevelPageStart;
8206     if (Pos + sizeof(uint32_t) > Contents.size()) {
8207       outs() << "warning: invalid offset for second level page: " << Pos << '\n';
8208       continue;
8209     }
8210 
8211     uint32_t Kind =
8212         *reinterpret_cast<const support::ulittle32_t *>(Contents.data() + Pos);
8213     if (Kind == 2)
8214       printRegularSecondLevelUnwindPage(Contents.substr(Pos, 4096));
8215     else if (Kind == 3)
8216       printCompressedSecondLevelUnwindPage(Contents.substr(Pos, 4096),
8217                                            IndexEntries[i].FunctionOffset,
8218                                            CommonEncodings);
8219     else
8220       outs() << "    Skipping 2nd level page with unknown kind " << Kind
8221              << '\n';
8222   }
8223 }
8224 
printMachOUnwindInfo(const MachOObjectFile * Obj)8225 void objdump::printMachOUnwindInfo(const MachOObjectFile *Obj) {
8226   std::map<uint64_t, SymbolRef> Symbols;
8227   for (const SymbolRef &SymRef : Obj->symbols()) {
8228     // Discard any undefined or absolute symbols. They're not going to take part
8229     // in the convenience lookup for unwind info and just take up resources.
8230     auto SectOrErr = SymRef.getSection();
8231     if (!SectOrErr) {
8232       // TODO: Actually report errors helpfully.
8233       consumeError(SectOrErr.takeError());
8234       continue;
8235     }
8236     section_iterator Section = *SectOrErr;
8237     if (Section == Obj->section_end())
8238       continue;
8239 
8240     uint64_t Addr = cantFail(SymRef.getValue());
8241     Symbols.insert(std::make_pair(Addr, SymRef));
8242   }
8243 
8244   for (const SectionRef &Section : Obj->sections()) {
8245     StringRef SectName;
8246     if (Expected<StringRef> NameOrErr = Section.getName())
8247       SectName = *NameOrErr;
8248     else
8249       consumeError(NameOrErr.takeError());
8250 
8251     if (SectName == "__compact_unwind")
8252       printMachOCompactUnwindSection(Obj, Symbols, Section);
8253     else if (SectName == "__unwind_info")
8254       printMachOUnwindInfoSection(Obj, Symbols, Section);
8255   }
8256 }
8257 
PrintMachHeader(uint32_t magic,uint32_t cputype,uint32_t cpusubtype,uint32_t filetype,uint32_t ncmds,uint32_t sizeofcmds,uint32_t flags,bool verbose)8258 static void PrintMachHeader(uint32_t magic, uint32_t cputype,
8259                             uint32_t cpusubtype, uint32_t filetype,
8260                             uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags,
8261                             bool verbose) {
8262   outs() << "Mach header\n";
8263   outs() << "      magic cputype cpusubtype  caps    filetype ncmds "
8264             "sizeofcmds      flags\n";
8265   if (verbose) {
8266     if (magic == MachO::MH_MAGIC)
8267       outs() << "   MH_MAGIC";
8268     else if (magic == MachO::MH_MAGIC_64)
8269       outs() << "MH_MAGIC_64";
8270     else
8271       outs() << format(" 0x%08" PRIx32, magic);
8272     switch (cputype) {
8273     case MachO::CPU_TYPE_I386:
8274       outs() << "    I386";
8275       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8276       case MachO::CPU_SUBTYPE_I386_ALL:
8277         outs() << "        ALL";
8278         break;
8279       default:
8280         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8281         break;
8282       }
8283       break;
8284     case MachO::CPU_TYPE_X86_64:
8285       outs() << "  X86_64";
8286       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8287       case MachO::CPU_SUBTYPE_X86_64_ALL:
8288         outs() << "        ALL";
8289         break;
8290       case MachO::CPU_SUBTYPE_X86_64_H:
8291         outs() << "    Haswell";
8292         break;
8293       default:
8294         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8295         break;
8296       }
8297       break;
8298     case MachO::CPU_TYPE_ARM:
8299       outs() << "     ARM";
8300       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8301       case MachO::CPU_SUBTYPE_ARM_ALL:
8302         outs() << "        ALL";
8303         break;
8304       case MachO::CPU_SUBTYPE_ARM_V4T:
8305         outs() << "        V4T";
8306         break;
8307       case MachO::CPU_SUBTYPE_ARM_V5TEJ:
8308         outs() << "      V5TEJ";
8309         break;
8310       case MachO::CPU_SUBTYPE_ARM_XSCALE:
8311         outs() << "     XSCALE";
8312         break;
8313       case MachO::CPU_SUBTYPE_ARM_V6:
8314         outs() << "         V6";
8315         break;
8316       case MachO::CPU_SUBTYPE_ARM_V6M:
8317         outs() << "        V6M";
8318         break;
8319       case MachO::CPU_SUBTYPE_ARM_V7:
8320         outs() << "         V7";
8321         break;
8322       case MachO::CPU_SUBTYPE_ARM_V7EM:
8323         outs() << "       V7EM";
8324         break;
8325       case MachO::CPU_SUBTYPE_ARM_V7K:
8326         outs() << "        V7K";
8327         break;
8328       case MachO::CPU_SUBTYPE_ARM_V7M:
8329         outs() << "        V7M";
8330         break;
8331       case MachO::CPU_SUBTYPE_ARM_V7S:
8332         outs() << "        V7S";
8333         break;
8334       default:
8335         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8336         break;
8337       }
8338       break;
8339     case MachO::CPU_TYPE_ARM64:
8340       outs() << "   ARM64";
8341       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8342       case MachO::CPU_SUBTYPE_ARM64_ALL:
8343         outs() << "        ALL";
8344         break;
8345       case MachO::CPU_SUBTYPE_ARM64_V8:
8346         outs() << "         V8";
8347         break;
8348       case MachO::CPU_SUBTYPE_ARM64E:
8349         outs() << "          E";
8350         break;
8351       default:
8352         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8353         break;
8354       }
8355       break;
8356     case MachO::CPU_TYPE_ARM64_32:
8357       outs() << " ARM64_32";
8358       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8359       case MachO::CPU_SUBTYPE_ARM64_32_V8:
8360         outs() << "        V8";
8361         break;
8362       default:
8363         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8364         break;
8365       }
8366       break;
8367     case MachO::CPU_TYPE_POWERPC:
8368       outs() << "     PPC";
8369       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8370       case MachO::CPU_SUBTYPE_POWERPC_ALL:
8371         outs() << "        ALL";
8372         break;
8373       default:
8374         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8375         break;
8376       }
8377       break;
8378     case MachO::CPU_TYPE_POWERPC64:
8379       outs() << "   PPC64";
8380       switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
8381       case MachO::CPU_SUBTYPE_POWERPC_ALL:
8382         outs() << "        ALL";
8383         break;
8384       default:
8385         outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8386         break;
8387       }
8388       break;
8389     default:
8390       outs() << format(" %7d", cputype);
8391       outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8392       break;
8393     }
8394     if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) {
8395       outs() << " LIB64";
8396     } else {
8397       outs() << format("  0x%02" PRIx32,
8398                        (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
8399     }
8400     switch (filetype) {
8401     case MachO::MH_OBJECT:
8402       outs() << "      OBJECT";
8403       break;
8404     case MachO::MH_EXECUTE:
8405       outs() << "     EXECUTE";
8406       break;
8407     case MachO::MH_FVMLIB:
8408       outs() << "      FVMLIB";
8409       break;
8410     case MachO::MH_CORE:
8411       outs() << "        CORE";
8412       break;
8413     case MachO::MH_PRELOAD:
8414       outs() << "     PRELOAD";
8415       break;
8416     case MachO::MH_DYLIB:
8417       outs() << "       DYLIB";
8418       break;
8419     case MachO::MH_DYLIB_STUB:
8420       outs() << "  DYLIB_STUB";
8421       break;
8422     case MachO::MH_DYLINKER:
8423       outs() << "    DYLINKER";
8424       break;
8425     case MachO::MH_BUNDLE:
8426       outs() << "      BUNDLE";
8427       break;
8428     case MachO::MH_DSYM:
8429       outs() << "        DSYM";
8430       break;
8431     case MachO::MH_KEXT_BUNDLE:
8432       outs() << "  KEXTBUNDLE";
8433       break;
8434     default:
8435       outs() << format("  %10u", filetype);
8436       break;
8437     }
8438     outs() << format(" %5u", ncmds);
8439     outs() << format(" %10u", sizeofcmds);
8440     uint32_t f = flags;
8441     if (f & MachO::MH_NOUNDEFS) {
8442       outs() << "   NOUNDEFS";
8443       f &= ~MachO::MH_NOUNDEFS;
8444     }
8445     if (f & MachO::MH_INCRLINK) {
8446       outs() << " INCRLINK";
8447       f &= ~MachO::MH_INCRLINK;
8448     }
8449     if (f & MachO::MH_DYLDLINK) {
8450       outs() << " DYLDLINK";
8451       f &= ~MachO::MH_DYLDLINK;
8452     }
8453     if (f & MachO::MH_BINDATLOAD) {
8454       outs() << " BINDATLOAD";
8455       f &= ~MachO::MH_BINDATLOAD;
8456     }
8457     if (f & MachO::MH_PREBOUND) {
8458       outs() << " PREBOUND";
8459       f &= ~MachO::MH_PREBOUND;
8460     }
8461     if (f & MachO::MH_SPLIT_SEGS) {
8462       outs() << " SPLIT_SEGS";
8463       f &= ~MachO::MH_SPLIT_SEGS;
8464     }
8465     if (f & MachO::MH_LAZY_INIT) {
8466       outs() << " LAZY_INIT";
8467       f &= ~MachO::MH_LAZY_INIT;
8468     }
8469     if (f & MachO::MH_TWOLEVEL) {
8470       outs() << " TWOLEVEL";
8471       f &= ~MachO::MH_TWOLEVEL;
8472     }
8473     if (f & MachO::MH_FORCE_FLAT) {
8474       outs() << " FORCE_FLAT";
8475       f &= ~MachO::MH_FORCE_FLAT;
8476     }
8477     if (f & MachO::MH_NOMULTIDEFS) {
8478       outs() << " NOMULTIDEFS";
8479       f &= ~MachO::MH_NOMULTIDEFS;
8480     }
8481     if (f & MachO::MH_NOFIXPREBINDING) {
8482       outs() << " NOFIXPREBINDING";
8483       f &= ~MachO::MH_NOFIXPREBINDING;
8484     }
8485     if (f & MachO::MH_PREBINDABLE) {
8486       outs() << " PREBINDABLE";
8487       f &= ~MachO::MH_PREBINDABLE;
8488     }
8489     if (f & MachO::MH_ALLMODSBOUND) {
8490       outs() << " ALLMODSBOUND";
8491       f &= ~MachO::MH_ALLMODSBOUND;
8492     }
8493     if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) {
8494       outs() << " SUBSECTIONS_VIA_SYMBOLS";
8495       f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
8496     }
8497     if (f & MachO::MH_CANONICAL) {
8498       outs() << " CANONICAL";
8499       f &= ~MachO::MH_CANONICAL;
8500     }
8501     if (f & MachO::MH_WEAK_DEFINES) {
8502       outs() << " WEAK_DEFINES";
8503       f &= ~MachO::MH_WEAK_DEFINES;
8504     }
8505     if (f & MachO::MH_BINDS_TO_WEAK) {
8506       outs() << " BINDS_TO_WEAK";
8507       f &= ~MachO::MH_BINDS_TO_WEAK;
8508     }
8509     if (f & MachO::MH_ALLOW_STACK_EXECUTION) {
8510       outs() << " ALLOW_STACK_EXECUTION";
8511       f &= ~MachO::MH_ALLOW_STACK_EXECUTION;
8512     }
8513     if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) {
8514       outs() << " DEAD_STRIPPABLE_DYLIB";
8515       f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB;
8516     }
8517     if (f & MachO::MH_PIE) {
8518       outs() << " PIE";
8519       f &= ~MachO::MH_PIE;
8520     }
8521     if (f & MachO::MH_NO_REEXPORTED_DYLIBS) {
8522       outs() << " NO_REEXPORTED_DYLIBS";
8523       f &= ~MachO::MH_NO_REEXPORTED_DYLIBS;
8524     }
8525     if (f & MachO::MH_HAS_TLV_DESCRIPTORS) {
8526       outs() << " MH_HAS_TLV_DESCRIPTORS";
8527       f &= ~MachO::MH_HAS_TLV_DESCRIPTORS;
8528     }
8529     if (f & MachO::MH_NO_HEAP_EXECUTION) {
8530       outs() << " MH_NO_HEAP_EXECUTION";
8531       f &= ~MachO::MH_NO_HEAP_EXECUTION;
8532     }
8533     if (f & MachO::MH_APP_EXTENSION_SAFE) {
8534       outs() << " APP_EXTENSION_SAFE";
8535       f &= ~MachO::MH_APP_EXTENSION_SAFE;
8536     }
8537     if (f & MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO) {
8538       outs() << " NLIST_OUTOFSYNC_WITH_DYLDINFO";
8539       f &= ~MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO;
8540     }
8541     if (f != 0 || flags == 0)
8542       outs() << format(" 0x%08" PRIx32, f);
8543   } else {
8544     outs() << format(" 0x%08" PRIx32, magic);
8545     outs() << format(" %7d", cputype);
8546     outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
8547     outs() << format("  0x%02" PRIx32,
8548                      (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
8549     outs() << format("  %10u", filetype);
8550     outs() << format(" %5u", ncmds);
8551     outs() << format(" %10u", sizeofcmds);
8552     outs() << format(" 0x%08" PRIx32, flags);
8553   }
8554   outs() << "\n";
8555 }
8556 
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)8557 static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize,
8558                                 StringRef SegName, uint64_t vmaddr,
8559                                 uint64_t vmsize, uint64_t fileoff,
8560                                 uint64_t filesize, uint32_t maxprot,
8561                                 uint32_t initprot, uint32_t nsects,
8562                                 uint32_t flags, uint32_t object_size,
8563                                 bool verbose) {
8564   uint64_t expected_cmdsize;
8565   if (cmd == MachO::LC_SEGMENT) {
8566     outs() << "      cmd LC_SEGMENT\n";
8567     expected_cmdsize = nsects;
8568     expected_cmdsize *= sizeof(struct MachO::section);
8569     expected_cmdsize += sizeof(struct MachO::segment_command);
8570   } else {
8571     outs() << "      cmd LC_SEGMENT_64\n";
8572     expected_cmdsize = nsects;
8573     expected_cmdsize *= sizeof(struct MachO::section_64);
8574     expected_cmdsize += sizeof(struct MachO::segment_command_64);
8575   }
8576   outs() << "  cmdsize " << cmdsize;
8577   if (cmdsize != expected_cmdsize)
8578     outs() << " Inconsistent size\n";
8579   else
8580     outs() << "\n";
8581   outs() << "  segname " << SegName << "\n";
8582   if (cmd == MachO::LC_SEGMENT_64) {
8583     outs() << "   vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n";
8584     outs() << "   vmsize " << format("0x%016" PRIx64, vmsize) << "\n";
8585   } else {
8586     outs() << "   vmaddr " << format("0x%08" PRIx64, vmaddr) << "\n";
8587     outs() << "   vmsize " << format("0x%08" PRIx64, vmsize) << "\n";
8588   }
8589   outs() << "  fileoff " << fileoff;
8590   if (fileoff > object_size)
8591     outs() << " (past end of file)\n";
8592   else
8593     outs() << "\n";
8594   outs() << " filesize " << filesize;
8595   if (fileoff + filesize > object_size)
8596     outs() << " (past end of file)\n";
8597   else
8598     outs() << "\n";
8599   if (verbose) {
8600     if ((maxprot &
8601          ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
8602            MachO::VM_PROT_EXECUTE)) != 0)
8603       outs() << "  maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n";
8604     else {
8605       outs() << "  maxprot ";
8606       outs() << ((maxprot & MachO::VM_PROT_READ) ? "r" : "-");
8607       outs() << ((maxprot & MachO::VM_PROT_WRITE) ? "w" : "-");
8608       outs() << ((maxprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n");
8609     }
8610     if ((initprot &
8611          ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
8612            MachO::VM_PROT_EXECUTE)) != 0)
8613       outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n";
8614     else {
8615       outs() << " initprot ";
8616       outs() << ((initprot & MachO::VM_PROT_READ) ? "r" : "-");
8617       outs() << ((initprot & MachO::VM_PROT_WRITE) ? "w" : "-");
8618       outs() << ((initprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n");
8619     }
8620   } else {
8621     outs() << "  maxprot " << format("0x%08" PRIx32, maxprot) << "\n";
8622     outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n";
8623   }
8624   outs() << "   nsects " << nsects << "\n";
8625   if (verbose) {
8626     outs() << "    flags";
8627     if (flags == 0)
8628       outs() << " (none)\n";
8629     else {
8630       if (flags & MachO::SG_HIGHVM) {
8631         outs() << " HIGHVM";
8632         flags &= ~MachO::SG_HIGHVM;
8633       }
8634       if (flags & MachO::SG_FVMLIB) {
8635         outs() << " FVMLIB";
8636         flags &= ~MachO::SG_FVMLIB;
8637       }
8638       if (flags & MachO::SG_NORELOC) {
8639         outs() << " NORELOC";
8640         flags &= ~MachO::SG_NORELOC;
8641       }
8642       if (flags & MachO::SG_PROTECTED_VERSION_1) {
8643         outs() << " PROTECTED_VERSION_1";
8644         flags &= ~MachO::SG_PROTECTED_VERSION_1;
8645       }
8646       if (flags)
8647         outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n";
8648       else
8649         outs() << "\n";
8650     }
8651   } else {
8652     outs() << "    flags " << format("0x%" PRIx32, flags) << "\n";
8653   }
8654 }
8655 
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)8656 static void PrintSection(const char *sectname, const char *segname,
8657                          uint64_t addr, uint64_t size, uint32_t offset,
8658                          uint32_t align, uint32_t reloff, uint32_t nreloc,
8659                          uint32_t flags, uint32_t reserved1, uint32_t reserved2,
8660                          uint32_t cmd, const char *sg_segname,
8661                          uint32_t filetype, uint32_t object_size,
8662                          bool verbose) {
8663   outs() << "Section\n";
8664   outs() << "  sectname " << format("%.16s\n", sectname);
8665   outs() << "   segname " << format("%.16s", segname);
8666   if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0)
8667     outs() << " (does not match segment)\n";
8668   else
8669     outs() << "\n";
8670   if (cmd == MachO::LC_SEGMENT_64) {
8671     outs() << "      addr " << format("0x%016" PRIx64, addr) << "\n";
8672     outs() << "      size " << format("0x%016" PRIx64, size);
8673   } else {
8674     outs() << "      addr " << format("0x%08" PRIx64, addr) << "\n";
8675     outs() << "      size " << format("0x%08" PRIx64, size);
8676   }
8677   if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size)
8678     outs() << " (past end of file)\n";
8679   else
8680     outs() << "\n";
8681   outs() << "    offset " << offset;
8682   if (offset > object_size)
8683     outs() << " (past end of file)\n";
8684   else
8685     outs() << "\n";
8686   uint32_t align_shifted = 1 << align;
8687   outs() << "     align 2^" << align << " (" << align_shifted << ")\n";
8688   outs() << "    reloff " << reloff;
8689   if (reloff > object_size)
8690     outs() << " (past end of file)\n";
8691   else
8692     outs() << "\n";
8693   outs() << "    nreloc " << nreloc;
8694   if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size)
8695     outs() << " (past end of file)\n";
8696   else
8697     outs() << "\n";
8698   uint32_t section_type = flags & MachO::SECTION_TYPE;
8699   if (verbose) {
8700     outs() << "      type";
8701     if (section_type == MachO::S_REGULAR)
8702       outs() << " S_REGULAR\n";
8703     else if (section_type == MachO::S_ZEROFILL)
8704       outs() << " S_ZEROFILL\n";
8705     else if (section_type == MachO::S_CSTRING_LITERALS)
8706       outs() << " S_CSTRING_LITERALS\n";
8707     else if (section_type == MachO::S_4BYTE_LITERALS)
8708       outs() << " S_4BYTE_LITERALS\n";
8709     else if (section_type == MachO::S_8BYTE_LITERALS)
8710       outs() << " S_8BYTE_LITERALS\n";
8711     else if (section_type == MachO::S_16BYTE_LITERALS)
8712       outs() << " S_16BYTE_LITERALS\n";
8713     else if (section_type == MachO::S_LITERAL_POINTERS)
8714       outs() << " S_LITERAL_POINTERS\n";
8715     else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS)
8716       outs() << " S_NON_LAZY_SYMBOL_POINTERS\n";
8717     else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS)
8718       outs() << " S_LAZY_SYMBOL_POINTERS\n";
8719     else if (section_type == MachO::S_SYMBOL_STUBS)
8720       outs() << " S_SYMBOL_STUBS\n";
8721     else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS)
8722       outs() << " S_MOD_INIT_FUNC_POINTERS\n";
8723     else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS)
8724       outs() << " S_MOD_TERM_FUNC_POINTERS\n";
8725     else if (section_type == MachO::S_COALESCED)
8726       outs() << " S_COALESCED\n";
8727     else if (section_type == MachO::S_INTERPOSING)
8728       outs() << " S_INTERPOSING\n";
8729     else if (section_type == MachO::S_DTRACE_DOF)
8730       outs() << " S_DTRACE_DOF\n";
8731     else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS)
8732       outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n";
8733     else if (section_type == MachO::S_THREAD_LOCAL_REGULAR)
8734       outs() << " S_THREAD_LOCAL_REGULAR\n";
8735     else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL)
8736       outs() << " S_THREAD_LOCAL_ZEROFILL\n";
8737     else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES)
8738       outs() << " S_THREAD_LOCAL_VARIABLES\n";
8739     else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
8740       outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n";
8741     else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS)
8742       outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n";
8743     else
8744       outs() << format("0x%08" PRIx32, section_type) << "\n";
8745     outs() << "attributes";
8746     uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES;
8747     if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS)
8748       outs() << " PURE_INSTRUCTIONS";
8749     if (section_attributes & MachO::S_ATTR_NO_TOC)
8750       outs() << " NO_TOC";
8751     if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS)
8752       outs() << " STRIP_STATIC_SYMS";
8753     if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP)
8754       outs() << " NO_DEAD_STRIP";
8755     if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT)
8756       outs() << " LIVE_SUPPORT";
8757     if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE)
8758       outs() << " SELF_MODIFYING_CODE";
8759     if (section_attributes & MachO::S_ATTR_DEBUG)
8760       outs() << " DEBUG";
8761     if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS)
8762       outs() << " SOME_INSTRUCTIONS";
8763     if (section_attributes & MachO::S_ATTR_EXT_RELOC)
8764       outs() << " EXT_RELOC";
8765     if (section_attributes & MachO::S_ATTR_LOC_RELOC)
8766       outs() << " LOC_RELOC";
8767     if (section_attributes == 0)
8768       outs() << " (none)";
8769     outs() << "\n";
8770   } else
8771     outs() << "     flags " << format("0x%08" PRIx32, flags) << "\n";
8772   outs() << " reserved1 " << reserved1;
8773   if (section_type == MachO::S_SYMBOL_STUBS ||
8774       section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
8775       section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
8776       section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
8777       section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
8778     outs() << " (index into indirect symbol table)\n";
8779   else
8780     outs() << "\n";
8781   outs() << " reserved2 " << reserved2;
8782   if (section_type == MachO::S_SYMBOL_STUBS)
8783     outs() << " (size of stubs)\n";
8784   else
8785     outs() << "\n";
8786 }
8787 
PrintSymtabLoadCommand(MachO::symtab_command st,bool Is64Bit,uint32_t object_size)8788 static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit,
8789                                    uint32_t object_size) {
8790   outs() << "     cmd LC_SYMTAB\n";
8791   outs() << " cmdsize " << st.cmdsize;
8792   if (st.cmdsize != sizeof(struct MachO::symtab_command))
8793     outs() << " Incorrect size\n";
8794   else
8795     outs() << "\n";
8796   outs() << "  symoff " << st.symoff;
8797   if (st.symoff > object_size)
8798     outs() << " (past end of file)\n";
8799   else
8800     outs() << "\n";
8801   outs() << "   nsyms " << st.nsyms;
8802   uint64_t big_size;
8803   if (Is64Bit) {
8804     big_size = st.nsyms;
8805     big_size *= sizeof(struct MachO::nlist_64);
8806     big_size += st.symoff;
8807     if (big_size > object_size)
8808       outs() << " (past end of file)\n";
8809     else
8810       outs() << "\n";
8811   } else {
8812     big_size = st.nsyms;
8813     big_size *= sizeof(struct MachO::nlist);
8814     big_size += st.symoff;
8815     if (big_size > object_size)
8816       outs() << " (past end of file)\n";
8817     else
8818       outs() << "\n";
8819   }
8820   outs() << "  stroff " << st.stroff;
8821   if (st.stroff > object_size)
8822     outs() << " (past end of file)\n";
8823   else
8824     outs() << "\n";
8825   outs() << " strsize " << st.strsize;
8826   big_size = st.stroff;
8827   big_size += st.strsize;
8828   if (big_size > object_size)
8829     outs() << " (past end of file)\n";
8830   else
8831     outs() << "\n";
8832 }
8833 
PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,uint32_t nsyms,uint32_t object_size,bool Is64Bit)8834 static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,
8835                                      uint32_t nsyms, uint32_t object_size,
8836                                      bool Is64Bit) {
8837   outs() << "            cmd LC_DYSYMTAB\n";
8838   outs() << "        cmdsize " << dyst.cmdsize;
8839   if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command))
8840     outs() << " Incorrect size\n";
8841   else
8842     outs() << "\n";
8843   outs() << "      ilocalsym " << dyst.ilocalsym;
8844   if (dyst.ilocalsym > nsyms)
8845     outs() << " (greater than the number of symbols)\n";
8846   else
8847     outs() << "\n";
8848   outs() << "      nlocalsym " << dyst.nlocalsym;
8849   uint64_t big_size;
8850   big_size = dyst.ilocalsym;
8851   big_size += dyst.nlocalsym;
8852   if (big_size > nsyms)
8853     outs() << " (past the end of the symbol table)\n";
8854   else
8855     outs() << "\n";
8856   outs() << "     iextdefsym " << dyst.iextdefsym;
8857   if (dyst.iextdefsym > nsyms)
8858     outs() << " (greater than the number of symbols)\n";
8859   else
8860     outs() << "\n";
8861   outs() << "     nextdefsym " << dyst.nextdefsym;
8862   big_size = dyst.iextdefsym;
8863   big_size += dyst.nextdefsym;
8864   if (big_size > nsyms)
8865     outs() << " (past the end of the symbol table)\n";
8866   else
8867     outs() << "\n";
8868   outs() << "      iundefsym " << dyst.iundefsym;
8869   if (dyst.iundefsym > nsyms)
8870     outs() << " (greater than the number of symbols)\n";
8871   else
8872     outs() << "\n";
8873   outs() << "      nundefsym " << dyst.nundefsym;
8874   big_size = dyst.iundefsym;
8875   big_size += dyst.nundefsym;
8876   if (big_size > nsyms)
8877     outs() << " (past the end of the symbol table)\n";
8878   else
8879     outs() << "\n";
8880   outs() << "         tocoff " << dyst.tocoff;
8881   if (dyst.tocoff > object_size)
8882     outs() << " (past end of file)\n";
8883   else
8884     outs() << "\n";
8885   outs() << "           ntoc " << dyst.ntoc;
8886   big_size = dyst.ntoc;
8887   big_size *= sizeof(struct MachO::dylib_table_of_contents);
8888   big_size += dyst.tocoff;
8889   if (big_size > object_size)
8890     outs() << " (past end of file)\n";
8891   else
8892     outs() << "\n";
8893   outs() << "      modtaboff " << dyst.modtaboff;
8894   if (dyst.modtaboff > object_size)
8895     outs() << " (past end of file)\n";
8896   else
8897     outs() << "\n";
8898   outs() << "        nmodtab " << dyst.nmodtab;
8899   uint64_t modtabend;
8900   if (Is64Bit) {
8901     modtabend = dyst.nmodtab;
8902     modtabend *= sizeof(struct MachO::dylib_module_64);
8903     modtabend += dyst.modtaboff;
8904   } else {
8905     modtabend = dyst.nmodtab;
8906     modtabend *= sizeof(struct MachO::dylib_module);
8907     modtabend += dyst.modtaboff;
8908   }
8909   if (modtabend > object_size)
8910     outs() << " (past end of file)\n";
8911   else
8912     outs() << "\n";
8913   outs() << "   extrefsymoff " << dyst.extrefsymoff;
8914   if (dyst.extrefsymoff > object_size)
8915     outs() << " (past end of file)\n";
8916   else
8917     outs() << "\n";
8918   outs() << "    nextrefsyms " << dyst.nextrefsyms;
8919   big_size = dyst.nextrefsyms;
8920   big_size *= sizeof(struct MachO::dylib_reference);
8921   big_size += dyst.extrefsymoff;
8922   if (big_size > object_size)
8923     outs() << " (past end of file)\n";
8924   else
8925     outs() << "\n";
8926   outs() << " indirectsymoff " << dyst.indirectsymoff;
8927   if (dyst.indirectsymoff > object_size)
8928     outs() << " (past end of file)\n";
8929   else
8930     outs() << "\n";
8931   outs() << "  nindirectsyms " << dyst.nindirectsyms;
8932   big_size = dyst.nindirectsyms;
8933   big_size *= sizeof(uint32_t);
8934   big_size += dyst.indirectsymoff;
8935   if (big_size > object_size)
8936     outs() << " (past end of file)\n";
8937   else
8938     outs() << "\n";
8939   outs() << "      extreloff " << dyst.extreloff;
8940   if (dyst.extreloff > object_size)
8941     outs() << " (past end of file)\n";
8942   else
8943     outs() << "\n";
8944   outs() << "        nextrel " << dyst.nextrel;
8945   big_size = dyst.nextrel;
8946   big_size *= sizeof(struct MachO::relocation_info);
8947   big_size += dyst.extreloff;
8948   if (big_size > object_size)
8949     outs() << " (past end of file)\n";
8950   else
8951     outs() << "\n";
8952   outs() << "      locreloff " << dyst.locreloff;
8953   if (dyst.locreloff > object_size)
8954     outs() << " (past end of file)\n";
8955   else
8956     outs() << "\n";
8957   outs() << "        nlocrel " << dyst.nlocrel;
8958   big_size = dyst.nlocrel;
8959   big_size *= sizeof(struct MachO::relocation_info);
8960   big_size += dyst.locreloff;
8961   if (big_size > object_size)
8962     outs() << " (past end of file)\n";
8963   else
8964     outs() << "\n";
8965 }
8966 
PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,uint32_t object_size)8967 static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,
8968                                      uint32_t object_size) {
8969   if (dc.cmd == MachO::LC_DYLD_INFO)
8970     outs() << "            cmd LC_DYLD_INFO\n";
8971   else
8972     outs() << "            cmd LC_DYLD_INFO_ONLY\n";
8973   outs() << "        cmdsize " << dc.cmdsize;
8974   if (dc.cmdsize != sizeof(struct MachO::dyld_info_command))
8975     outs() << " Incorrect size\n";
8976   else
8977     outs() << "\n";
8978   outs() << "     rebase_off " << dc.rebase_off;
8979   if (dc.rebase_off > object_size)
8980     outs() << " (past end of file)\n";
8981   else
8982     outs() << "\n";
8983   outs() << "    rebase_size " << dc.rebase_size;
8984   uint64_t big_size;
8985   big_size = dc.rebase_off;
8986   big_size += dc.rebase_size;
8987   if (big_size > object_size)
8988     outs() << " (past end of file)\n";
8989   else
8990     outs() << "\n";
8991   outs() << "       bind_off " << dc.bind_off;
8992   if (dc.bind_off > object_size)
8993     outs() << " (past end of file)\n";
8994   else
8995     outs() << "\n";
8996   outs() << "      bind_size " << dc.bind_size;
8997   big_size = dc.bind_off;
8998   big_size += dc.bind_size;
8999   if (big_size > object_size)
9000     outs() << " (past end of file)\n";
9001   else
9002     outs() << "\n";
9003   outs() << "  weak_bind_off " << dc.weak_bind_off;
9004   if (dc.weak_bind_off > object_size)
9005     outs() << " (past end of file)\n";
9006   else
9007     outs() << "\n";
9008   outs() << " weak_bind_size " << dc.weak_bind_size;
9009   big_size = dc.weak_bind_off;
9010   big_size += dc.weak_bind_size;
9011   if (big_size > object_size)
9012     outs() << " (past end of file)\n";
9013   else
9014     outs() << "\n";
9015   outs() << "  lazy_bind_off " << dc.lazy_bind_off;
9016   if (dc.lazy_bind_off > object_size)
9017     outs() << " (past end of file)\n";
9018   else
9019     outs() << "\n";
9020   outs() << " lazy_bind_size " << dc.lazy_bind_size;
9021   big_size = dc.lazy_bind_off;
9022   big_size += dc.lazy_bind_size;
9023   if (big_size > object_size)
9024     outs() << " (past end of file)\n";
9025   else
9026     outs() << "\n";
9027   outs() << "     export_off " << dc.export_off;
9028   if (dc.export_off > object_size)
9029     outs() << " (past end of file)\n";
9030   else
9031     outs() << "\n";
9032   outs() << "    export_size " << dc.export_size;
9033   big_size = dc.export_off;
9034   big_size += dc.export_size;
9035   if (big_size > object_size)
9036     outs() << " (past end of file)\n";
9037   else
9038     outs() << "\n";
9039 }
9040 
PrintDyldLoadCommand(MachO::dylinker_command dyld,const char * Ptr)9041 static void PrintDyldLoadCommand(MachO::dylinker_command dyld,
9042                                  const char *Ptr) {
9043   if (dyld.cmd == MachO::LC_ID_DYLINKER)
9044     outs() << "          cmd LC_ID_DYLINKER\n";
9045   else if (dyld.cmd == MachO::LC_LOAD_DYLINKER)
9046     outs() << "          cmd LC_LOAD_DYLINKER\n";
9047   else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT)
9048     outs() << "          cmd LC_DYLD_ENVIRONMENT\n";
9049   else
9050     outs() << "          cmd ?(" << dyld.cmd << ")\n";
9051   outs() << "      cmdsize " << dyld.cmdsize;
9052   if (dyld.cmdsize < sizeof(struct MachO::dylinker_command))
9053     outs() << " Incorrect size\n";
9054   else
9055     outs() << "\n";
9056   if (dyld.name >= dyld.cmdsize)
9057     outs() << "         name ?(bad offset " << dyld.name << ")\n";
9058   else {
9059     const char *P = (const char *)(Ptr) + dyld.name;
9060     outs() << "         name " << P << " (offset " << dyld.name << ")\n";
9061   }
9062 }
9063 
PrintUuidLoadCommand(MachO::uuid_command uuid)9064 static void PrintUuidLoadCommand(MachO::uuid_command uuid) {
9065   outs() << "     cmd LC_UUID\n";
9066   outs() << " cmdsize " << uuid.cmdsize;
9067   if (uuid.cmdsize != sizeof(struct MachO::uuid_command))
9068     outs() << " Incorrect size\n";
9069   else
9070     outs() << "\n";
9071   outs() << "    uuid ";
9072   for (int i = 0; i < 16; ++i) {
9073     outs() << format("%02" PRIX32, uuid.uuid[i]);
9074     if (i == 3 || i == 5 || i == 7 || i == 9)
9075       outs() << "-";
9076   }
9077   outs() << "\n";
9078 }
9079 
PrintRpathLoadCommand(MachO::rpath_command rpath,const char * Ptr)9080 static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) {
9081   outs() << "          cmd LC_RPATH\n";
9082   outs() << "      cmdsize " << rpath.cmdsize;
9083   if (rpath.cmdsize < sizeof(struct MachO::rpath_command))
9084     outs() << " Incorrect size\n";
9085   else
9086     outs() << "\n";
9087   if (rpath.path >= rpath.cmdsize)
9088     outs() << "         path ?(bad offset " << rpath.path << ")\n";
9089   else {
9090     const char *P = (const char *)(Ptr) + rpath.path;
9091     outs() << "         path " << P << " (offset " << rpath.path << ")\n";
9092   }
9093 }
9094 
PrintVersionMinLoadCommand(MachO::version_min_command vd)9095 static void PrintVersionMinLoadCommand(MachO::version_min_command vd) {
9096   StringRef LoadCmdName;
9097   switch (vd.cmd) {
9098   case MachO::LC_VERSION_MIN_MACOSX:
9099     LoadCmdName = "LC_VERSION_MIN_MACOSX";
9100     break;
9101   case MachO::LC_VERSION_MIN_IPHONEOS:
9102     LoadCmdName = "LC_VERSION_MIN_IPHONEOS";
9103     break;
9104   case MachO::LC_VERSION_MIN_TVOS:
9105     LoadCmdName = "LC_VERSION_MIN_TVOS";
9106     break;
9107   case MachO::LC_VERSION_MIN_WATCHOS:
9108     LoadCmdName = "LC_VERSION_MIN_WATCHOS";
9109     break;
9110   default:
9111     llvm_unreachable("Unknown version min load command");
9112   }
9113 
9114   outs() << "      cmd " << LoadCmdName << '\n';
9115   outs() << "  cmdsize " << vd.cmdsize;
9116   if (vd.cmdsize != sizeof(struct MachO::version_min_command))
9117     outs() << " Incorrect size\n";
9118   else
9119     outs() << "\n";
9120   outs() << "  version "
9121          << MachOObjectFile::getVersionMinMajor(vd, false) << "."
9122          << MachOObjectFile::getVersionMinMinor(vd, false);
9123   uint32_t Update = MachOObjectFile::getVersionMinUpdate(vd, false);
9124   if (Update != 0)
9125     outs() << "." << Update;
9126   outs() << "\n";
9127   if (vd.sdk == 0)
9128     outs() << "      sdk n/a";
9129   else {
9130     outs() << "      sdk "
9131            << MachOObjectFile::getVersionMinMajor(vd, true) << "."
9132            << MachOObjectFile::getVersionMinMinor(vd, true);
9133   }
9134   Update = MachOObjectFile::getVersionMinUpdate(vd, true);
9135   if (Update != 0)
9136     outs() << "." << Update;
9137   outs() << "\n";
9138 }
9139 
PrintNoteLoadCommand(MachO::note_command Nt)9140 static void PrintNoteLoadCommand(MachO::note_command Nt) {
9141   outs() << "       cmd LC_NOTE\n";
9142   outs() << "   cmdsize " << Nt.cmdsize;
9143   if (Nt.cmdsize != sizeof(struct MachO::note_command))
9144     outs() << " Incorrect size\n";
9145   else
9146     outs() << "\n";
9147   const char *d = Nt.data_owner;
9148   outs() << "data_owner " << format("%.16s\n", d);
9149   outs() << "    offset " << Nt.offset << "\n";
9150   outs() << "      size " << Nt.size << "\n";
9151 }
9152 
PrintBuildToolVersion(MachO::build_tool_version bv)9153 static void PrintBuildToolVersion(MachO::build_tool_version bv) {
9154   outs() << "      tool " << MachOObjectFile::getBuildTool(bv.tool) << "\n";
9155   outs() << "   version " << MachOObjectFile::getVersionString(bv.version)
9156          << "\n";
9157 }
9158 
PrintBuildVersionLoadCommand(const MachOObjectFile * obj,MachO::build_version_command bd)9159 static void PrintBuildVersionLoadCommand(const MachOObjectFile *obj,
9160                                          MachO::build_version_command bd) {
9161   outs() << "       cmd LC_BUILD_VERSION\n";
9162   outs() << "   cmdsize " << bd.cmdsize;
9163   if (bd.cmdsize !=
9164       sizeof(struct MachO::build_version_command) +
9165           bd.ntools * sizeof(struct MachO::build_tool_version))
9166     outs() << " Incorrect size\n";
9167   else
9168     outs() << "\n";
9169   outs() << "  platform " << MachOObjectFile::getBuildPlatform(bd.platform)
9170          << "\n";
9171   if (bd.sdk)
9172     outs() << "       sdk " << MachOObjectFile::getVersionString(bd.sdk)
9173            << "\n";
9174   else
9175     outs() << "       sdk n/a\n";
9176   outs() << "     minos " << MachOObjectFile::getVersionString(bd.minos)
9177          << "\n";
9178   outs() << "    ntools " << bd.ntools << "\n";
9179   for (unsigned i = 0; i < bd.ntools; ++i) {
9180     MachO::build_tool_version bv = obj->getBuildToolVersion(i);
9181     PrintBuildToolVersion(bv);
9182   }
9183 }
9184 
PrintSourceVersionCommand(MachO::source_version_command sd)9185 static void PrintSourceVersionCommand(MachO::source_version_command sd) {
9186   outs() << "      cmd LC_SOURCE_VERSION\n";
9187   outs() << "  cmdsize " << sd.cmdsize;
9188   if (sd.cmdsize != sizeof(struct MachO::source_version_command))
9189     outs() << " Incorrect size\n";
9190   else
9191     outs() << "\n";
9192   uint64_t a = (sd.version >> 40) & 0xffffff;
9193   uint64_t b = (sd.version >> 30) & 0x3ff;
9194   uint64_t c = (sd.version >> 20) & 0x3ff;
9195   uint64_t d = (sd.version >> 10) & 0x3ff;
9196   uint64_t e = sd.version & 0x3ff;
9197   outs() << "  version " << a << "." << b;
9198   if (e != 0)
9199     outs() << "." << c << "." << d << "." << e;
9200   else if (d != 0)
9201     outs() << "." << c << "." << d;
9202   else if (c != 0)
9203     outs() << "." << c;
9204   outs() << "\n";
9205 }
9206 
PrintEntryPointCommand(MachO::entry_point_command ep)9207 static void PrintEntryPointCommand(MachO::entry_point_command ep) {
9208   outs() << "       cmd LC_MAIN\n";
9209   outs() << "   cmdsize " << ep.cmdsize;
9210   if (ep.cmdsize != sizeof(struct MachO::entry_point_command))
9211     outs() << " Incorrect size\n";
9212   else
9213     outs() << "\n";
9214   outs() << "  entryoff " << ep.entryoff << "\n";
9215   outs() << " stacksize " << ep.stacksize << "\n";
9216 }
9217 
PrintEncryptionInfoCommand(MachO::encryption_info_command ec,uint32_t object_size)9218 static void PrintEncryptionInfoCommand(MachO::encryption_info_command ec,
9219                                        uint32_t object_size) {
9220   outs() << "          cmd LC_ENCRYPTION_INFO\n";
9221   outs() << "      cmdsize " << ec.cmdsize;
9222   if (ec.cmdsize != sizeof(struct MachO::encryption_info_command))
9223     outs() << " Incorrect size\n";
9224   else
9225     outs() << "\n";
9226   outs() << "     cryptoff " << ec.cryptoff;
9227   if (ec.cryptoff > object_size)
9228     outs() << " (past end of file)\n";
9229   else
9230     outs() << "\n";
9231   outs() << "    cryptsize " << ec.cryptsize;
9232   if (ec.cryptsize > object_size)
9233     outs() << " (past end of file)\n";
9234   else
9235     outs() << "\n";
9236   outs() << "      cryptid " << ec.cryptid << "\n";
9237 }
9238 
PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,uint32_t object_size)9239 static void PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,
9240                                          uint32_t object_size) {
9241   outs() << "          cmd LC_ENCRYPTION_INFO_64\n";
9242   outs() << "      cmdsize " << ec.cmdsize;
9243   if (ec.cmdsize != sizeof(struct MachO::encryption_info_command_64))
9244     outs() << " Incorrect size\n";
9245   else
9246     outs() << "\n";
9247   outs() << "     cryptoff " << ec.cryptoff;
9248   if (ec.cryptoff > object_size)
9249     outs() << " (past end of file)\n";
9250   else
9251     outs() << "\n";
9252   outs() << "    cryptsize " << ec.cryptsize;
9253   if (ec.cryptsize > object_size)
9254     outs() << " (past end of file)\n";
9255   else
9256     outs() << "\n";
9257   outs() << "      cryptid " << ec.cryptid << "\n";
9258   outs() << "          pad " << ec.pad << "\n";
9259 }
9260 
PrintLinkerOptionCommand(MachO::linker_option_command lo,const char * Ptr)9261 static void PrintLinkerOptionCommand(MachO::linker_option_command lo,
9262                                      const char *Ptr) {
9263   outs() << "     cmd LC_LINKER_OPTION\n";
9264   outs() << " cmdsize " << lo.cmdsize;
9265   if (lo.cmdsize < sizeof(struct MachO::linker_option_command))
9266     outs() << " Incorrect size\n";
9267   else
9268     outs() << "\n";
9269   outs() << "   count " << lo.count << "\n";
9270   const char *string = Ptr + sizeof(struct MachO::linker_option_command);
9271   uint32_t left = lo.cmdsize - sizeof(struct MachO::linker_option_command);
9272   uint32_t i = 0;
9273   while (left > 0) {
9274     while (*string == '\0' && left > 0) {
9275       string++;
9276       left--;
9277     }
9278     if (left > 0) {
9279       i++;
9280       outs() << "  string #" << i << " " << format("%.*s\n", left, string);
9281       uint32_t NullPos = StringRef(string, left).find('\0');
9282       uint32_t len = std::min(NullPos, left) + 1;
9283       string += len;
9284       left -= len;
9285     }
9286   }
9287   if (lo.count != i)
9288     outs() << "   count " << lo.count << " does not match number of strings "
9289            << i << "\n";
9290 }
9291 
PrintSubFrameworkCommand(MachO::sub_framework_command sub,const char * Ptr)9292 static void PrintSubFrameworkCommand(MachO::sub_framework_command sub,
9293                                      const char *Ptr) {
9294   outs() << "          cmd LC_SUB_FRAMEWORK\n";
9295   outs() << "      cmdsize " << sub.cmdsize;
9296   if (sub.cmdsize < sizeof(struct MachO::sub_framework_command))
9297     outs() << " Incorrect size\n";
9298   else
9299     outs() << "\n";
9300   if (sub.umbrella < sub.cmdsize) {
9301     const char *P = Ptr + sub.umbrella;
9302     outs() << "     umbrella " << P << " (offset " << sub.umbrella << ")\n";
9303   } else {
9304     outs() << "     umbrella ?(bad offset " << sub.umbrella << ")\n";
9305   }
9306 }
9307 
PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,const char * Ptr)9308 static void PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,
9309                                     const char *Ptr) {
9310   outs() << "          cmd LC_SUB_UMBRELLA\n";
9311   outs() << "      cmdsize " << sub.cmdsize;
9312   if (sub.cmdsize < sizeof(struct MachO::sub_umbrella_command))
9313     outs() << " Incorrect size\n";
9314   else
9315     outs() << "\n";
9316   if (sub.sub_umbrella < sub.cmdsize) {
9317     const char *P = Ptr + sub.sub_umbrella;
9318     outs() << " sub_umbrella " << P << " (offset " << sub.sub_umbrella << ")\n";
9319   } else {
9320     outs() << " sub_umbrella ?(bad offset " << sub.sub_umbrella << ")\n";
9321   }
9322 }
9323 
PrintSubLibraryCommand(MachO::sub_library_command sub,const char * Ptr)9324 static void PrintSubLibraryCommand(MachO::sub_library_command sub,
9325                                    const char *Ptr) {
9326   outs() << "          cmd LC_SUB_LIBRARY\n";
9327   outs() << "      cmdsize " << sub.cmdsize;
9328   if (sub.cmdsize < sizeof(struct MachO::sub_library_command))
9329     outs() << " Incorrect size\n";
9330   else
9331     outs() << "\n";
9332   if (sub.sub_library < sub.cmdsize) {
9333     const char *P = Ptr + sub.sub_library;
9334     outs() << "  sub_library " << P << " (offset " << sub.sub_library << ")\n";
9335   } else {
9336     outs() << "  sub_library ?(bad offset " << sub.sub_library << ")\n";
9337   }
9338 }
9339 
PrintSubClientCommand(MachO::sub_client_command sub,const char * Ptr)9340 static void PrintSubClientCommand(MachO::sub_client_command sub,
9341                                   const char *Ptr) {
9342   outs() << "          cmd LC_SUB_CLIENT\n";
9343   outs() << "      cmdsize " << sub.cmdsize;
9344   if (sub.cmdsize < sizeof(struct MachO::sub_client_command))
9345     outs() << " Incorrect size\n";
9346   else
9347     outs() << "\n";
9348   if (sub.client < sub.cmdsize) {
9349     const char *P = Ptr + sub.client;
9350     outs() << "       client " << P << " (offset " << sub.client << ")\n";
9351   } else {
9352     outs() << "       client ?(bad offset " << sub.client << ")\n";
9353   }
9354 }
9355 
PrintRoutinesCommand(MachO::routines_command r)9356 static void PrintRoutinesCommand(MachO::routines_command r) {
9357   outs() << "          cmd LC_ROUTINES\n";
9358   outs() << "      cmdsize " << r.cmdsize;
9359   if (r.cmdsize != sizeof(struct MachO::routines_command))
9360     outs() << " Incorrect size\n";
9361   else
9362     outs() << "\n";
9363   outs() << " init_address " << format("0x%08" PRIx32, r.init_address) << "\n";
9364   outs() << "  init_module " << r.init_module << "\n";
9365   outs() << "    reserved1 " << r.reserved1 << "\n";
9366   outs() << "    reserved2 " << r.reserved2 << "\n";
9367   outs() << "    reserved3 " << r.reserved3 << "\n";
9368   outs() << "    reserved4 " << r.reserved4 << "\n";
9369   outs() << "    reserved5 " << r.reserved5 << "\n";
9370   outs() << "    reserved6 " << r.reserved6 << "\n";
9371 }
9372 
PrintRoutinesCommand64(MachO::routines_command_64 r)9373 static void PrintRoutinesCommand64(MachO::routines_command_64 r) {
9374   outs() << "          cmd LC_ROUTINES_64\n";
9375   outs() << "      cmdsize " << r.cmdsize;
9376   if (r.cmdsize != sizeof(struct MachO::routines_command_64))
9377     outs() << " Incorrect size\n";
9378   else
9379     outs() << "\n";
9380   outs() << " init_address " << format("0x%016" PRIx64, r.init_address) << "\n";
9381   outs() << "  init_module " << r.init_module << "\n";
9382   outs() << "    reserved1 " << r.reserved1 << "\n";
9383   outs() << "    reserved2 " << r.reserved2 << "\n";
9384   outs() << "    reserved3 " << r.reserved3 << "\n";
9385   outs() << "    reserved4 " << r.reserved4 << "\n";
9386   outs() << "    reserved5 " << r.reserved5 << "\n";
9387   outs() << "    reserved6 " << r.reserved6 << "\n";
9388 }
9389 
Print_x86_thread_state32_t(MachO::x86_thread_state32_t & cpu32)9390 static void Print_x86_thread_state32_t(MachO::x86_thread_state32_t &cpu32) {
9391   outs() << "\t    eax " << format("0x%08" PRIx32, cpu32.eax);
9392   outs() << " ebx    " << format("0x%08" PRIx32, cpu32.ebx);
9393   outs() << " ecx " << format("0x%08" PRIx32, cpu32.ecx);
9394   outs() << " edx " << format("0x%08" PRIx32, cpu32.edx) << "\n";
9395   outs() << "\t    edi " << format("0x%08" PRIx32, cpu32.edi);
9396   outs() << " esi    " << format("0x%08" PRIx32, cpu32.esi);
9397   outs() << " ebp " << format("0x%08" PRIx32, cpu32.ebp);
9398   outs() << " esp " << format("0x%08" PRIx32, cpu32.esp) << "\n";
9399   outs() << "\t    ss  " << format("0x%08" PRIx32, cpu32.ss);
9400   outs() << " eflags " << format("0x%08" PRIx32, cpu32.eflags);
9401   outs() << " eip " << format("0x%08" PRIx32, cpu32.eip);
9402   outs() << " cs  " << format("0x%08" PRIx32, cpu32.cs) << "\n";
9403   outs() << "\t    ds  " << format("0x%08" PRIx32, cpu32.ds);
9404   outs() << " es     " << format("0x%08" PRIx32, cpu32.es);
9405   outs() << " fs  " << format("0x%08" PRIx32, cpu32.fs);
9406   outs() << " gs  " << format("0x%08" PRIx32, cpu32.gs) << "\n";
9407 }
9408 
Print_x86_thread_state64_t(MachO::x86_thread_state64_t & cpu64)9409 static void Print_x86_thread_state64_t(MachO::x86_thread_state64_t &cpu64) {
9410   outs() << "   rax  " << format("0x%016" PRIx64, cpu64.rax);
9411   outs() << " rbx " << format("0x%016" PRIx64, cpu64.rbx);
9412   outs() << " rcx  " << format("0x%016" PRIx64, cpu64.rcx) << "\n";
9413   outs() << "   rdx  " << format("0x%016" PRIx64, cpu64.rdx);
9414   outs() << " rdi " << format("0x%016" PRIx64, cpu64.rdi);
9415   outs() << " rsi  " << format("0x%016" PRIx64, cpu64.rsi) << "\n";
9416   outs() << "   rbp  " << format("0x%016" PRIx64, cpu64.rbp);
9417   outs() << " rsp " << format("0x%016" PRIx64, cpu64.rsp);
9418   outs() << " r8   " << format("0x%016" PRIx64, cpu64.r8) << "\n";
9419   outs() << "    r9  " << format("0x%016" PRIx64, cpu64.r9);
9420   outs() << " r10 " << format("0x%016" PRIx64, cpu64.r10);
9421   outs() << " r11  " << format("0x%016" PRIx64, cpu64.r11) << "\n";
9422   outs() << "   r12  " << format("0x%016" PRIx64, cpu64.r12);
9423   outs() << " r13 " << format("0x%016" PRIx64, cpu64.r13);
9424   outs() << " r14  " << format("0x%016" PRIx64, cpu64.r14) << "\n";
9425   outs() << "   r15  " << format("0x%016" PRIx64, cpu64.r15);
9426   outs() << " rip " << format("0x%016" PRIx64, cpu64.rip) << "\n";
9427   outs() << "rflags  " << format("0x%016" PRIx64, cpu64.rflags);
9428   outs() << " cs  " << format("0x%016" PRIx64, cpu64.cs);
9429   outs() << " fs   " << format("0x%016" PRIx64, cpu64.fs) << "\n";
9430   outs() << "    gs  " << format("0x%016" PRIx64, cpu64.gs) << "\n";
9431 }
9432 
Print_mmst_reg(MachO::mmst_reg_t & r)9433 static void Print_mmst_reg(MachO::mmst_reg_t &r) {
9434   uint32_t f;
9435   outs() << "\t      mmst_reg  ";
9436   for (f = 0; f < 10; f++)
9437     outs() << format("%02" PRIx32, (r.mmst_reg[f] & 0xff)) << " ";
9438   outs() << "\n";
9439   outs() << "\t      mmst_rsrv ";
9440   for (f = 0; f < 6; f++)
9441     outs() << format("%02" PRIx32, (r.mmst_rsrv[f] & 0xff)) << " ";
9442   outs() << "\n";
9443 }
9444 
Print_xmm_reg(MachO::xmm_reg_t & r)9445 static void Print_xmm_reg(MachO::xmm_reg_t &r) {
9446   uint32_t f;
9447   outs() << "\t      xmm_reg ";
9448   for (f = 0; f < 16; f++)
9449     outs() << format("%02" PRIx32, (r.xmm_reg[f] & 0xff)) << " ";
9450   outs() << "\n";
9451 }
9452 
Print_x86_float_state_t(MachO::x86_float_state64_t & fpu)9453 static void Print_x86_float_state_t(MachO::x86_float_state64_t &fpu) {
9454   outs() << "\t    fpu_reserved[0] " << fpu.fpu_reserved[0];
9455   outs() << " fpu_reserved[1] " << fpu.fpu_reserved[1] << "\n";
9456   outs() << "\t    control: invalid " << fpu.fpu_fcw.invalid;
9457   outs() << " denorm " << fpu.fpu_fcw.denorm;
9458   outs() << " zdiv " << fpu.fpu_fcw.zdiv;
9459   outs() << " ovrfl " << fpu.fpu_fcw.ovrfl;
9460   outs() << " undfl " << fpu.fpu_fcw.undfl;
9461   outs() << " precis " << fpu.fpu_fcw.precis << "\n";
9462   outs() << "\t\t     pc ";
9463   if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_24B)
9464     outs() << "FP_PREC_24B ";
9465   else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_53B)
9466     outs() << "FP_PREC_53B ";
9467   else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_64B)
9468     outs() << "FP_PREC_64B ";
9469   else
9470     outs() << fpu.fpu_fcw.pc << " ";
9471   outs() << "rc ";
9472   if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_NEAR)
9473     outs() << "FP_RND_NEAR ";
9474   else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_DOWN)
9475     outs() << "FP_RND_DOWN ";
9476   else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_UP)
9477     outs() << "FP_RND_UP ";
9478   else if (fpu.fpu_fcw.rc == MachO::x86_FP_CHOP)
9479     outs() << "FP_CHOP ";
9480   outs() << "\n";
9481   outs() << "\t    status: invalid " << fpu.fpu_fsw.invalid;
9482   outs() << " denorm " << fpu.fpu_fsw.denorm;
9483   outs() << " zdiv " << fpu.fpu_fsw.zdiv;
9484   outs() << " ovrfl " << fpu.fpu_fsw.ovrfl;
9485   outs() << " undfl " << fpu.fpu_fsw.undfl;
9486   outs() << " precis " << fpu.fpu_fsw.precis;
9487   outs() << " stkflt " << fpu.fpu_fsw.stkflt << "\n";
9488   outs() << "\t            errsumm " << fpu.fpu_fsw.errsumm;
9489   outs() << " c0 " << fpu.fpu_fsw.c0;
9490   outs() << " c1 " << fpu.fpu_fsw.c1;
9491   outs() << " c2 " << fpu.fpu_fsw.c2;
9492   outs() << " tos " << fpu.fpu_fsw.tos;
9493   outs() << " c3 " << fpu.fpu_fsw.c3;
9494   outs() << " busy " << fpu.fpu_fsw.busy << "\n";
9495   outs() << "\t    fpu_ftw " << format("0x%02" PRIx32, fpu.fpu_ftw);
9496   outs() << " fpu_rsrv1 " << format("0x%02" PRIx32, fpu.fpu_rsrv1);
9497   outs() << " fpu_fop " << format("0x%04" PRIx32, fpu.fpu_fop);
9498   outs() << " fpu_ip " << format("0x%08" PRIx32, fpu.fpu_ip) << "\n";
9499   outs() << "\t    fpu_cs " << format("0x%04" PRIx32, fpu.fpu_cs);
9500   outs() << " fpu_rsrv2 " << format("0x%04" PRIx32, fpu.fpu_rsrv2);
9501   outs() << " fpu_dp " << format("0x%08" PRIx32, fpu.fpu_dp);
9502   outs() << " fpu_ds " << format("0x%04" PRIx32, fpu.fpu_ds) << "\n";
9503   outs() << "\t    fpu_rsrv3 " << format("0x%04" PRIx32, fpu.fpu_rsrv3);
9504   outs() << " fpu_mxcsr " << format("0x%08" PRIx32, fpu.fpu_mxcsr);
9505   outs() << " fpu_mxcsrmask " << format("0x%08" PRIx32, fpu.fpu_mxcsrmask);
9506   outs() << "\n";
9507   outs() << "\t    fpu_stmm0:\n";
9508   Print_mmst_reg(fpu.fpu_stmm0);
9509   outs() << "\t    fpu_stmm1:\n";
9510   Print_mmst_reg(fpu.fpu_stmm1);
9511   outs() << "\t    fpu_stmm2:\n";
9512   Print_mmst_reg(fpu.fpu_stmm2);
9513   outs() << "\t    fpu_stmm3:\n";
9514   Print_mmst_reg(fpu.fpu_stmm3);
9515   outs() << "\t    fpu_stmm4:\n";
9516   Print_mmst_reg(fpu.fpu_stmm4);
9517   outs() << "\t    fpu_stmm5:\n";
9518   Print_mmst_reg(fpu.fpu_stmm5);
9519   outs() << "\t    fpu_stmm6:\n";
9520   Print_mmst_reg(fpu.fpu_stmm6);
9521   outs() << "\t    fpu_stmm7:\n";
9522   Print_mmst_reg(fpu.fpu_stmm7);
9523   outs() << "\t    fpu_xmm0:\n";
9524   Print_xmm_reg(fpu.fpu_xmm0);
9525   outs() << "\t    fpu_xmm1:\n";
9526   Print_xmm_reg(fpu.fpu_xmm1);
9527   outs() << "\t    fpu_xmm2:\n";
9528   Print_xmm_reg(fpu.fpu_xmm2);
9529   outs() << "\t    fpu_xmm3:\n";
9530   Print_xmm_reg(fpu.fpu_xmm3);
9531   outs() << "\t    fpu_xmm4:\n";
9532   Print_xmm_reg(fpu.fpu_xmm4);
9533   outs() << "\t    fpu_xmm5:\n";
9534   Print_xmm_reg(fpu.fpu_xmm5);
9535   outs() << "\t    fpu_xmm6:\n";
9536   Print_xmm_reg(fpu.fpu_xmm6);
9537   outs() << "\t    fpu_xmm7:\n";
9538   Print_xmm_reg(fpu.fpu_xmm7);
9539   outs() << "\t    fpu_xmm8:\n";
9540   Print_xmm_reg(fpu.fpu_xmm8);
9541   outs() << "\t    fpu_xmm9:\n";
9542   Print_xmm_reg(fpu.fpu_xmm9);
9543   outs() << "\t    fpu_xmm10:\n";
9544   Print_xmm_reg(fpu.fpu_xmm10);
9545   outs() << "\t    fpu_xmm11:\n";
9546   Print_xmm_reg(fpu.fpu_xmm11);
9547   outs() << "\t    fpu_xmm12:\n";
9548   Print_xmm_reg(fpu.fpu_xmm12);
9549   outs() << "\t    fpu_xmm13:\n";
9550   Print_xmm_reg(fpu.fpu_xmm13);
9551   outs() << "\t    fpu_xmm14:\n";
9552   Print_xmm_reg(fpu.fpu_xmm14);
9553   outs() << "\t    fpu_xmm15:\n";
9554   Print_xmm_reg(fpu.fpu_xmm15);
9555   outs() << "\t    fpu_rsrv4:\n";
9556   for (uint32_t f = 0; f < 6; f++) {
9557     outs() << "\t            ";
9558     for (uint32_t g = 0; g < 16; g++)
9559       outs() << format("%02" PRIx32, fpu.fpu_rsrv4[f * g]) << " ";
9560     outs() << "\n";
9561   }
9562   outs() << "\t    fpu_reserved1 " << format("0x%08" PRIx32, fpu.fpu_reserved1);
9563   outs() << "\n";
9564 }
9565 
Print_x86_exception_state_t(MachO::x86_exception_state64_t & exc64)9566 static void Print_x86_exception_state_t(MachO::x86_exception_state64_t &exc64) {
9567   outs() << "\t    trapno " << format("0x%08" PRIx32, exc64.trapno);
9568   outs() << " err " << format("0x%08" PRIx32, exc64.err);
9569   outs() << " faultvaddr " << format("0x%016" PRIx64, exc64.faultvaddr) << "\n";
9570 }
9571 
Print_arm_thread_state32_t(MachO::arm_thread_state32_t & cpu32)9572 static void Print_arm_thread_state32_t(MachO::arm_thread_state32_t &cpu32) {
9573   outs() << "\t    r0  " << format("0x%08" PRIx32, cpu32.r[0]);
9574   outs() << " r1     "   << format("0x%08" PRIx32, cpu32.r[1]);
9575   outs() << " r2  "      << format("0x%08" PRIx32, cpu32.r[2]);
9576   outs() << " r3  "      << format("0x%08" PRIx32, cpu32.r[3]) << "\n";
9577   outs() << "\t    r4  " << format("0x%08" PRIx32, cpu32.r[4]);
9578   outs() << " r5     "   << format("0x%08" PRIx32, cpu32.r[5]);
9579   outs() << " r6  "      << format("0x%08" PRIx32, cpu32.r[6]);
9580   outs() << " r7  "      << format("0x%08" PRIx32, cpu32.r[7]) << "\n";
9581   outs() << "\t    r8  " << format("0x%08" PRIx32, cpu32.r[8]);
9582   outs() << " r9     "   << format("0x%08" PRIx32, cpu32.r[9]);
9583   outs() << " r10 "      << format("0x%08" PRIx32, cpu32.r[10]);
9584   outs() << " r11 "      << format("0x%08" PRIx32, cpu32.r[11]) << "\n";
9585   outs() << "\t    r12 " << format("0x%08" PRIx32, cpu32.r[12]);
9586   outs() << " sp     "   << format("0x%08" PRIx32, cpu32.sp);
9587   outs() << " lr  "      << format("0x%08" PRIx32, cpu32.lr);
9588   outs() << " pc  "      << format("0x%08" PRIx32, cpu32.pc) << "\n";
9589   outs() << "\t   cpsr " << format("0x%08" PRIx32, cpu32.cpsr) << "\n";
9590 }
9591 
Print_arm_thread_state64_t(MachO::arm_thread_state64_t & cpu64)9592 static void Print_arm_thread_state64_t(MachO::arm_thread_state64_t &cpu64) {
9593   outs() << "\t    x0  " << format("0x%016" PRIx64, cpu64.x[0]);
9594   outs() << " x1  "      << format("0x%016" PRIx64, cpu64.x[1]);
9595   outs() << " x2  "      << format("0x%016" PRIx64, cpu64.x[2]) << "\n";
9596   outs() << "\t    x3  " << format("0x%016" PRIx64, cpu64.x[3]);
9597   outs() << " x4  "      << format("0x%016" PRIx64, cpu64.x[4]);
9598   outs() << " x5  "      << format("0x%016" PRIx64, cpu64.x[5]) << "\n";
9599   outs() << "\t    x6  " << format("0x%016" PRIx64, cpu64.x[6]);
9600   outs() << " x7  "      << format("0x%016" PRIx64, cpu64.x[7]);
9601   outs() << " x8  "      << format("0x%016" PRIx64, cpu64.x[8]) << "\n";
9602   outs() << "\t    x9  " << format("0x%016" PRIx64, cpu64.x[9]);
9603   outs() << " x10 "      << format("0x%016" PRIx64, cpu64.x[10]);
9604   outs() << " x11 "      << format("0x%016" PRIx64, cpu64.x[11]) << "\n";
9605   outs() << "\t    x12 " << format("0x%016" PRIx64, cpu64.x[12]);
9606   outs() << " x13 "      << format("0x%016" PRIx64, cpu64.x[13]);
9607   outs() << " x14 "      << format("0x%016" PRIx64, cpu64.x[14]) << "\n";
9608   outs() << "\t    x15 " << format("0x%016" PRIx64, cpu64.x[15]);
9609   outs() << " x16 "      << format("0x%016" PRIx64, cpu64.x[16]);
9610   outs() << " x17 "      << format("0x%016" PRIx64, cpu64.x[17]) << "\n";
9611   outs() << "\t    x18 " << format("0x%016" PRIx64, cpu64.x[18]);
9612   outs() << " x19 "      << format("0x%016" PRIx64, cpu64.x[19]);
9613   outs() << " x20 "      << format("0x%016" PRIx64, cpu64.x[20]) << "\n";
9614   outs() << "\t    x21 " << format("0x%016" PRIx64, cpu64.x[21]);
9615   outs() << " x22 "      << format("0x%016" PRIx64, cpu64.x[22]);
9616   outs() << " x23 "      << format("0x%016" PRIx64, cpu64.x[23]) << "\n";
9617   outs() << "\t    x24 " << format("0x%016" PRIx64, cpu64.x[24]);
9618   outs() << " x25 "      << format("0x%016" PRIx64, cpu64.x[25]);
9619   outs() << " x26 "      << format("0x%016" PRIx64, cpu64.x[26]) << "\n";
9620   outs() << "\t    x27 " << format("0x%016" PRIx64, cpu64.x[27]);
9621   outs() << " x28 "      << format("0x%016" PRIx64, cpu64.x[28]);
9622   outs() << "  fp "      << format("0x%016" PRIx64, cpu64.fp) << "\n";
9623   outs() << "\t     lr " << format("0x%016" PRIx64, cpu64.lr);
9624   outs() << " sp  "      << format("0x%016" PRIx64, cpu64.sp);
9625   outs() << "  pc "      << format("0x%016" PRIx64, cpu64.pc) << "\n";
9626   outs() << "\t   cpsr " << format("0x%08"  PRIx32, cpu64.cpsr) << "\n";
9627 }
9628 
PrintThreadCommand(MachO::thread_command t,const char * Ptr,bool isLittleEndian,uint32_t cputype)9629 static void PrintThreadCommand(MachO::thread_command t, const char *Ptr,
9630                                bool isLittleEndian, uint32_t cputype) {
9631   if (t.cmd == MachO::LC_THREAD)
9632     outs() << "        cmd LC_THREAD\n";
9633   else if (t.cmd == MachO::LC_UNIXTHREAD)
9634     outs() << "        cmd LC_UNIXTHREAD\n";
9635   else
9636     outs() << "        cmd " << t.cmd << " (unknown)\n";
9637   outs() << "    cmdsize " << t.cmdsize;
9638   if (t.cmdsize < sizeof(struct MachO::thread_command) + 2 * sizeof(uint32_t))
9639     outs() << " Incorrect size\n";
9640   else
9641     outs() << "\n";
9642 
9643   const char *begin = Ptr + sizeof(struct MachO::thread_command);
9644   const char *end = Ptr + t.cmdsize;
9645   uint32_t flavor, count, left;
9646   if (cputype == MachO::CPU_TYPE_I386) {
9647     while (begin < end) {
9648       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9649         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9650         begin += sizeof(uint32_t);
9651       } else {
9652         flavor = 0;
9653         begin = end;
9654       }
9655       if (isLittleEndian != sys::IsLittleEndianHost)
9656         sys::swapByteOrder(flavor);
9657       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9658         memcpy((char *)&count, begin, sizeof(uint32_t));
9659         begin += sizeof(uint32_t);
9660       } else {
9661         count = 0;
9662         begin = end;
9663       }
9664       if (isLittleEndian != sys::IsLittleEndianHost)
9665         sys::swapByteOrder(count);
9666       if (flavor == MachO::x86_THREAD_STATE32) {
9667         outs() << "     flavor i386_THREAD_STATE\n";
9668         if (count == MachO::x86_THREAD_STATE32_COUNT)
9669           outs() << "      count i386_THREAD_STATE_COUNT\n";
9670         else
9671           outs() << "      count " << count
9672                  << " (not x86_THREAD_STATE32_COUNT)\n";
9673         MachO::x86_thread_state32_t cpu32;
9674         left = end - begin;
9675         if (left >= sizeof(MachO::x86_thread_state32_t)) {
9676           memcpy(&cpu32, begin, sizeof(MachO::x86_thread_state32_t));
9677           begin += sizeof(MachO::x86_thread_state32_t);
9678         } else {
9679           memset(&cpu32, '\0', sizeof(MachO::x86_thread_state32_t));
9680           memcpy(&cpu32, begin, left);
9681           begin += left;
9682         }
9683         if (isLittleEndian != sys::IsLittleEndianHost)
9684           swapStruct(cpu32);
9685         Print_x86_thread_state32_t(cpu32);
9686       } else if (flavor == MachO::x86_THREAD_STATE) {
9687         outs() << "     flavor x86_THREAD_STATE\n";
9688         if (count == MachO::x86_THREAD_STATE_COUNT)
9689           outs() << "      count x86_THREAD_STATE_COUNT\n";
9690         else
9691           outs() << "      count " << count
9692                  << " (not x86_THREAD_STATE_COUNT)\n";
9693         struct MachO::x86_thread_state_t ts;
9694         left = end - begin;
9695         if (left >= sizeof(MachO::x86_thread_state_t)) {
9696           memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
9697           begin += sizeof(MachO::x86_thread_state_t);
9698         } else {
9699           memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
9700           memcpy(&ts, begin, left);
9701           begin += left;
9702         }
9703         if (isLittleEndian != sys::IsLittleEndianHost)
9704           swapStruct(ts);
9705         if (ts.tsh.flavor == MachO::x86_THREAD_STATE32) {
9706           outs() << "\t    tsh.flavor x86_THREAD_STATE32 ";
9707           if (ts.tsh.count == MachO::x86_THREAD_STATE32_COUNT)
9708             outs() << "tsh.count x86_THREAD_STATE32_COUNT\n";
9709           else
9710             outs() << "tsh.count " << ts.tsh.count
9711                    << " (not x86_THREAD_STATE32_COUNT\n";
9712           Print_x86_thread_state32_t(ts.uts.ts32);
9713         } else {
9714           outs() << "\t    tsh.flavor " << ts.tsh.flavor << "  tsh.count "
9715                  << ts.tsh.count << "\n";
9716         }
9717       } else {
9718         outs() << "     flavor " << flavor << " (unknown)\n";
9719         outs() << "      count " << count << "\n";
9720         outs() << "      state (unknown)\n";
9721         begin += count * sizeof(uint32_t);
9722       }
9723     }
9724   } else if (cputype == MachO::CPU_TYPE_X86_64) {
9725     while (begin < end) {
9726       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9727         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9728         begin += sizeof(uint32_t);
9729       } else {
9730         flavor = 0;
9731         begin = end;
9732       }
9733       if (isLittleEndian != sys::IsLittleEndianHost)
9734         sys::swapByteOrder(flavor);
9735       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9736         memcpy((char *)&count, begin, sizeof(uint32_t));
9737         begin += sizeof(uint32_t);
9738       } else {
9739         count = 0;
9740         begin = end;
9741       }
9742       if (isLittleEndian != sys::IsLittleEndianHost)
9743         sys::swapByteOrder(count);
9744       if (flavor == MachO::x86_THREAD_STATE64) {
9745         outs() << "     flavor x86_THREAD_STATE64\n";
9746         if (count == MachO::x86_THREAD_STATE64_COUNT)
9747           outs() << "      count x86_THREAD_STATE64_COUNT\n";
9748         else
9749           outs() << "      count " << count
9750                  << " (not x86_THREAD_STATE64_COUNT)\n";
9751         MachO::x86_thread_state64_t cpu64;
9752         left = end - begin;
9753         if (left >= sizeof(MachO::x86_thread_state64_t)) {
9754           memcpy(&cpu64, begin, sizeof(MachO::x86_thread_state64_t));
9755           begin += sizeof(MachO::x86_thread_state64_t);
9756         } else {
9757           memset(&cpu64, '\0', sizeof(MachO::x86_thread_state64_t));
9758           memcpy(&cpu64, begin, left);
9759           begin += left;
9760         }
9761         if (isLittleEndian != sys::IsLittleEndianHost)
9762           swapStruct(cpu64);
9763         Print_x86_thread_state64_t(cpu64);
9764       } else if (flavor == MachO::x86_THREAD_STATE) {
9765         outs() << "     flavor x86_THREAD_STATE\n";
9766         if (count == MachO::x86_THREAD_STATE_COUNT)
9767           outs() << "      count x86_THREAD_STATE_COUNT\n";
9768         else
9769           outs() << "      count " << count
9770                  << " (not x86_THREAD_STATE_COUNT)\n";
9771         struct MachO::x86_thread_state_t ts;
9772         left = end - begin;
9773         if (left >= sizeof(MachO::x86_thread_state_t)) {
9774           memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
9775           begin += sizeof(MachO::x86_thread_state_t);
9776         } else {
9777           memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
9778           memcpy(&ts, begin, left);
9779           begin += left;
9780         }
9781         if (isLittleEndian != sys::IsLittleEndianHost)
9782           swapStruct(ts);
9783         if (ts.tsh.flavor == MachO::x86_THREAD_STATE64) {
9784           outs() << "\t    tsh.flavor x86_THREAD_STATE64 ";
9785           if (ts.tsh.count == MachO::x86_THREAD_STATE64_COUNT)
9786             outs() << "tsh.count x86_THREAD_STATE64_COUNT\n";
9787           else
9788             outs() << "tsh.count " << ts.tsh.count
9789                    << " (not x86_THREAD_STATE64_COUNT\n";
9790           Print_x86_thread_state64_t(ts.uts.ts64);
9791         } else {
9792           outs() << "\t    tsh.flavor " << ts.tsh.flavor << "  tsh.count "
9793                  << ts.tsh.count << "\n";
9794         }
9795       } else if (flavor == MachO::x86_FLOAT_STATE) {
9796         outs() << "     flavor x86_FLOAT_STATE\n";
9797         if (count == MachO::x86_FLOAT_STATE_COUNT)
9798           outs() << "      count x86_FLOAT_STATE_COUNT\n";
9799         else
9800           outs() << "      count " << count << " (not x86_FLOAT_STATE_COUNT)\n";
9801         struct MachO::x86_float_state_t fs;
9802         left = end - begin;
9803         if (left >= sizeof(MachO::x86_float_state_t)) {
9804           memcpy(&fs, begin, sizeof(MachO::x86_float_state_t));
9805           begin += sizeof(MachO::x86_float_state_t);
9806         } else {
9807           memset(&fs, '\0', sizeof(MachO::x86_float_state_t));
9808           memcpy(&fs, begin, left);
9809           begin += left;
9810         }
9811         if (isLittleEndian != sys::IsLittleEndianHost)
9812           swapStruct(fs);
9813         if (fs.fsh.flavor == MachO::x86_FLOAT_STATE64) {
9814           outs() << "\t    fsh.flavor x86_FLOAT_STATE64 ";
9815           if (fs.fsh.count == MachO::x86_FLOAT_STATE64_COUNT)
9816             outs() << "fsh.count x86_FLOAT_STATE64_COUNT\n";
9817           else
9818             outs() << "fsh.count " << fs.fsh.count
9819                    << " (not x86_FLOAT_STATE64_COUNT\n";
9820           Print_x86_float_state_t(fs.ufs.fs64);
9821         } else {
9822           outs() << "\t    fsh.flavor " << fs.fsh.flavor << "  fsh.count "
9823                  << fs.fsh.count << "\n";
9824         }
9825       } else if (flavor == MachO::x86_EXCEPTION_STATE) {
9826         outs() << "     flavor x86_EXCEPTION_STATE\n";
9827         if (count == MachO::x86_EXCEPTION_STATE_COUNT)
9828           outs() << "      count x86_EXCEPTION_STATE_COUNT\n";
9829         else
9830           outs() << "      count " << count
9831                  << " (not x86_EXCEPTION_STATE_COUNT)\n";
9832         struct MachO::x86_exception_state_t es;
9833         left = end - begin;
9834         if (left >= sizeof(MachO::x86_exception_state_t)) {
9835           memcpy(&es, begin, sizeof(MachO::x86_exception_state_t));
9836           begin += sizeof(MachO::x86_exception_state_t);
9837         } else {
9838           memset(&es, '\0', sizeof(MachO::x86_exception_state_t));
9839           memcpy(&es, begin, left);
9840           begin += left;
9841         }
9842         if (isLittleEndian != sys::IsLittleEndianHost)
9843           swapStruct(es);
9844         if (es.esh.flavor == MachO::x86_EXCEPTION_STATE64) {
9845           outs() << "\t    esh.flavor x86_EXCEPTION_STATE64\n";
9846           if (es.esh.count == MachO::x86_EXCEPTION_STATE64_COUNT)
9847             outs() << "\t    esh.count x86_EXCEPTION_STATE64_COUNT\n";
9848           else
9849             outs() << "\t    esh.count " << es.esh.count
9850                    << " (not x86_EXCEPTION_STATE64_COUNT\n";
9851           Print_x86_exception_state_t(es.ues.es64);
9852         } else {
9853           outs() << "\t    esh.flavor " << es.esh.flavor << "  esh.count "
9854                  << es.esh.count << "\n";
9855         }
9856       } else if (flavor == MachO::x86_EXCEPTION_STATE64) {
9857         outs() << "     flavor x86_EXCEPTION_STATE64\n";
9858         if (count == MachO::x86_EXCEPTION_STATE64_COUNT)
9859           outs() << "      count x86_EXCEPTION_STATE64_COUNT\n";
9860         else
9861           outs() << "      count " << count
9862                  << " (not x86_EXCEPTION_STATE64_COUNT)\n";
9863         struct MachO::x86_exception_state64_t es64;
9864         left = end - begin;
9865         if (left >= sizeof(MachO::x86_exception_state64_t)) {
9866           memcpy(&es64, begin, sizeof(MachO::x86_exception_state64_t));
9867           begin += sizeof(MachO::x86_exception_state64_t);
9868         } else {
9869           memset(&es64, '\0', sizeof(MachO::x86_exception_state64_t));
9870           memcpy(&es64, begin, left);
9871           begin += left;
9872         }
9873         if (isLittleEndian != sys::IsLittleEndianHost)
9874           swapStruct(es64);
9875         Print_x86_exception_state_t(es64);
9876       } else {
9877         outs() << "     flavor " << flavor << " (unknown)\n";
9878         outs() << "      count " << count << "\n";
9879         outs() << "      state (unknown)\n";
9880         begin += count * sizeof(uint32_t);
9881       }
9882     }
9883   } else if (cputype == MachO::CPU_TYPE_ARM) {
9884     while (begin < end) {
9885       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9886         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9887         begin += sizeof(uint32_t);
9888       } else {
9889         flavor = 0;
9890         begin = end;
9891       }
9892       if (isLittleEndian != sys::IsLittleEndianHost)
9893         sys::swapByteOrder(flavor);
9894       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9895         memcpy((char *)&count, begin, sizeof(uint32_t));
9896         begin += sizeof(uint32_t);
9897       } else {
9898         count = 0;
9899         begin = end;
9900       }
9901       if (isLittleEndian != sys::IsLittleEndianHost)
9902         sys::swapByteOrder(count);
9903       if (flavor == MachO::ARM_THREAD_STATE) {
9904         outs() << "     flavor ARM_THREAD_STATE\n";
9905         if (count == MachO::ARM_THREAD_STATE_COUNT)
9906           outs() << "      count ARM_THREAD_STATE_COUNT\n";
9907         else
9908           outs() << "      count " << count
9909                  << " (not ARM_THREAD_STATE_COUNT)\n";
9910         MachO::arm_thread_state32_t cpu32;
9911         left = end - begin;
9912         if (left >= sizeof(MachO::arm_thread_state32_t)) {
9913           memcpy(&cpu32, begin, sizeof(MachO::arm_thread_state32_t));
9914           begin += sizeof(MachO::arm_thread_state32_t);
9915         } else {
9916           memset(&cpu32, '\0', sizeof(MachO::arm_thread_state32_t));
9917           memcpy(&cpu32, begin, left);
9918           begin += left;
9919         }
9920         if (isLittleEndian != sys::IsLittleEndianHost)
9921           swapStruct(cpu32);
9922         Print_arm_thread_state32_t(cpu32);
9923       } else {
9924         outs() << "     flavor " << flavor << " (unknown)\n";
9925         outs() << "      count " << count << "\n";
9926         outs() << "      state (unknown)\n";
9927         begin += count * sizeof(uint32_t);
9928       }
9929     }
9930   } else if (cputype == MachO::CPU_TYPE_ARM64 ||
9931              cputype == MachO::CPU_TYPE_ARM64_32) {
9932     while (begin < end) {
9933       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9934         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9935         begin += sizeof(uint32_t);
9936       } else {
9937         flavor = 0;
9938         begin = end;
9939       }
9940       if (isLittleEndian != sys::IsLittleEndianHost)
9941         sys::swapByteOrder(flavor);
9942       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9943         memcpy((char *)&count, begin, sizeof(uint32_t));
9944         begin += sizeof(uint32_t);
9945       } else {
9946         count = 0;
9947         begin = end;
9948       }
9949       if (isLittleEndian != sys::IsLittleEndianHost)
9950         sys::swapByteOrder(count);
9951       if (flavor == MachO::ARM_THREAD_STATE64) {
9952         outs() << "     flavor ARM_THREAD_STATE64\n";
9953         if (count == MachO::ARM_THREAD_STATE64_COUNT)
9954           outs() << "      count ARM_THREAD_STATE64_COUNT\n";
9955         else
9956           outs() << "      count " << count
9957                  << " (not ARM_THREAD_STATE64_COUNT)\n";
9958         MachO::arm_thread_state64_t cpu64;
9959         left = end - begin;
9960         if (left >= sizeof(MachO::arm_thread_state64_t)) {
9961           memcpy(&cpu64, begin, sizeof(MachO::arm_thread_state64_t));
9962           begin += sizeof(MachO::arm_thread_state64_t);
9963         } else {
9964           memset(&cpu64, '\0', sizeof(MachO::arm_thread_state64_t));
9965           memcpy(&cpu64, begin, left);
9966           begin += left;
9967         }
9968         if (isLittleEndian != sys::IsLittleEndianHost)
9969           swapStruct(cpu64);
9970         Print_arm_thread_state64_t(cpu64);
9971       } else {
9972         outs() << "     flavor " << flavor << " (unknown)\n";
9973         outs() << "      count " << count << "\n";
9974         outs() << "      state (unknown)\n";
9975         begin += count * sizeof(uint32_t);
9976       }
9977     }
9978   } else {
9979     while (begin < end) {
9980       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9981         memcpy((char *)&flavor, begin, sizeof(uint32_t));
9982         begin += sizeof(uint32_t);
9983       } else {
9984         flavor = 0;
9985         begin = end;
9986       }
9987       if (isLittleEndian != sys::IsLittleEndianHost)
9988         sys::swapByteOrder(flavor);
9989       if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
9990         memcpy((char *)&count, begin, sizeof(uint32_t));
9991         begin += sizeof(uint32_t);
9992       } else {
9993         count = 0;
9994         begin = end;
9995       }
9996       if (isLittleEndian != sys::IsLittleEndianHost)
9997         sys::swapByteOrder(count);
9998       outs() << "     flavor " << flavor << "\n";
9999       outs() << "      count " << count << "\n";
10000       outs() << "      state (Unknown cputype/cpusubtype)\n";
10001       begin += count * sizeof(uint32_t);
10002     }
10003   }
10004 }
10005 
PrintDylibCommand(MachO::dylib_command dl,const char * Ptr)10006 static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) {
10007   if (dl.cmd == MachO::LC_ID_DYLIB)
10008     outs() << "          cmd LC_ID_DYLIB\n";
10009   else if (dl.cmd == MachO::LC_LOAD_DYLIB)
10010     outs() << "          cmd LC_LOAD_DYLIB\n";
10011   else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB)
10012     outs() << "          cmd LC_LOAD_WEAK_DYLIB\n";
10013   else if (dl.cmd == MachO::LC_REEXPORT_DYLIB)
10014     outs() << "          cmd LC_REEXPORT_DYLIB\n";
10015   else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB)
10016     outs() << "          cmd LC_LAZY_LOAD_DYLIB\n";
10017   else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
10018     outs() << "          cmd LC_LOAD_UPWARD_DYLIB\n";
10019   else
10020     outs() << "          cmd " << dl.cmd << " (unknown)\n";
10021   outs() << "      cmdsize " << dl.cmdsize;
10022   if (dl.cmdsize < sizeof(struct MachO::dylib_command))
10023     outs() << " Incorrect size\n";
10024   else
10025     outs() << "\n";
10026   if (dl.dylib.name < dl.cmdsize) {
10027     const char *P = (const char *)(Ptr) + dl.dylib.name;
10028     outs() << "         name " << P << " (offset " << dl.dylib.name << ")\n";
10029   } else {
10030     outs() << "         name ?(bad offset " << dl.dylib.name << ")\n";
10031   }
10032   outs() << "   time stamp " << dl.dylib.timestamp << " ";
10033   time_t t = dl.dylib.timestamp;
10034   outs() << ctime(&t);
10035   outs() << "      current version ";
10036   if (dl.dylib.current_version == 0xffffffff)
10037     outs() << "n/a\n";
10038   else
10039     outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "."
10040            << ((dl.dylib.current_version >> 8) & 0xff) << "."
10041            << (dl.dylib.current_version & 0xff) << "\n";
10042   outs() << "compatibility version ";
10043   if (dl.dylib.compatibility_version == 0xffffffff)
10044     outs() << "n/a\n";
10045   else
10046     outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
10047            << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
10048            << (dl.dylib.compatibility_version & 0xff) << "\n";
10049 }
10050 
PrintLinkEditDataCommand(MachO::linkedit_data_command ld,uint32_t object_size)10051 static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld,
10052                                      uint32_t object_size) {
10053   if (ld.cmd == MachO::LC_CODE_SIGNATURE)
10054     outs() << "      cmd LC_CODE_SIGNATURE\n";
10055   else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO)
10056     outs() << "      cmd LC_SEGMENT_SPLIT_INFO\n";
10057   else if (ld.cmd == MachO::LC_FUNCTION_STARTS)
10058     outs() << "      cmd LC_FUNCTION_STARTS\n";
10059   else if (ld.cmd == MachO::LC_DATA_IN_CODE)
10060     outs() << "      cmd LC_DATA_IN_CODE\n";
10061   else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS)
10062     outs() << "      cmd LC_DYLIB_CODE_SIGN_DRS\n";
10063   else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT)
10064     outs() << "      cmd LC_LINKER_OPTIMIZATION_HINT\n";
10065   else
10066     outs() << "      cmd " << ld.cmd << " (?)\n";
10067   outs() << "  cmdsize " << ld.cmdsize;
10068   if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command))
10069     outs() << " Incorrect size\n";
10070   else
10071     outs() << "\n";
10072   outs() << "  dataoff " << ld.dataoff;
10073   if (ld.dataoff > object_size)
10074     outs() << " (past end of file)\n";
10075   else
10076     outs() << "\n";
10077   outs() << " datasize " << ld.datasize;
10078   uint64_t big_size = ld.dataoff;
10079   big_size += ld.datasize;
10080   if (big_size > object_size)
10081     outs() << " (past end of file)\n";
10082   else
10083     outs() << "\n";
10084 }
10085 
PrintLoadCommands(const MachOObjectFile * Obj,uint32_t filetype,uint32_t cputype,bool verbose)10086 static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t filetype,
10087                               uint32_t cputype, bool verbose) {
10088   StringRef Buf = Obj->getData();
10089   unsigned Index = 0;
10090   for (const auto &Command : Obj->load_commands()) {
10091     outs() << "Load command " << Index++ << "\n";
10092     if (Command.C.cmd == MachO::LC_SEGMENT) {
10093       MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command);
10094       const char *sg_segname = SLC.segname;
10095       PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr,
10096                           SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot,
10097                           SLC.initprot, SLC.nsects, SLC.flags, Buf.size(),
10098                           verbose);
10099       for (unsigned j = 0; j < SLC.nsects; j++) {
10100         MachO::section S = Obj->getSection(Command, j);
10101         PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align,
10102                      S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2,
10103                      SLC.cmd, sg_segname, filetype, Buf.size(), verbose);
10104       }
10105     } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
10106       MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command);
10107       const char *sg_segname = SLC_64.segname;
10108       PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname,
10109                           SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff,
10110                           SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot,
10111                           SLC_64.nsects, SLC_64.flags, Buf.size(), verbose);
10112       for (unsigned j = 0; j < SLC_64.nsects; j++) {
10113         MachO::section_64 S_64 = Obj->getSection64(Command, j);
10114         PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size,
10115                      S_64.offset, S_64.align, S_64.reloff, S_64.nreloc,
10116                      S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd,
10117                      sg_segname, filetype, Buf.size(), verbose);
10118       }
10119     } else if (Command.C.cmd == MachO::LC_SYMTAB) {
10120       MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
10121       PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size());
10122     } else if (Command.C.cmd == MachO::LC_DYSYMTAB) {
10123       MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand();
10124       MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
10125       PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(),
10126                                Obj->is64Bit());
10127     } else if (Command.C.cmd == MachO::LC_DYLD_INFO ||
10128                Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) {
10129       MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command);
10130       PrintDyldInfoLoadCommand(DyldInfo, Buf.size());
10131     } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER ||
10132                Command.C.cmd == MachO::LC_ID_DYLINKER ||
10133                Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) {
10134       MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command);
10135       PrintDyldLoadCommand(Dyld, Command.Ptr);
10136     } else if (Command.C.cmd == MachO::LC_UUID) {
10137       MachO::uuid_command Uuid = Obj->getUuidCommand(Command);
10138       PrintUuidLoadCommand(Uuid);
10139     } else if (Command.C.cmd == MachO::LC_RPATH) {
10140       MachO::rpath_command Rpath = Obj->getRpathCommand(Command);
10141       PrintRpathLoadCommand(Rpath, Command.Ptr);
10142     } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX ||
10143                Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS ||
10144                Command.C.cmd == MachO::LC_VERSION_MIN_TVOS ||
10145                Command.C.cmd == MachO::LC_VERSION_MIN_WATCHOS) {
10146       MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command);
10147       PrintVersionMinLoadCommand(Vd);
10148     } else if (Command.C.cmd == MachO::LC_NOTE) {
10149       MachO::note_command Nt = Obj->getNoteLoadCommand(Command);
10150       PrintNoteLoadCommand(Nt);
10151     } else if (Command.C.cmd == MachO::LC_BUILD_VERSION) {
10152       MachO::build_version_command Bv =
10153           Obj->getBuildVersionLoadCommand(Command);
10154       PrintBuildVersionLoadCommand(Obj, Bv);
10155     } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) {
10156       MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command);
10157       PrintSourceVersionCommand(Sd);
10158     } else if (Command.C.cmd == MachO::LC_MAIN) {
10159       MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command);
10160       PrintEntryPointCommand(Ep);
10161     } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO) {
10162       MachO::encryption_info_command Ei =
10163           Obj->getEncryptionInfoCommand(Command);
10164       PrintEncryptionInfoCommand(Ei, Buf.size());
10165     } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO_64) {
10166       MachO::encryption_info_command_64 Ei =
10167           Obj->getEncryptionInfoCommand64(Command);
10168       PrintEncryptionInfoCommand64(Ei, Buf.size());
10169     } else if (Command.C.cmd == MachO::LC_LINKER_OPTION) {
10170       MachO::linker_option_command Lo =
10171           Obj->getLinkerOptionLoadCommand(Command);
10172       PrintLinkerOptionCommand(Lo, Command.Ptr);
10173     } else if (Command.C.cmd == MachO::LC_SUB_FRAMEWORK) {
10174       MachO::sub_framework_command Sf = Obj->getSubFrameworkCommand(Command);
10175       PrintSubFrameworkCommand(Sf, Command.Ptr);
10176     } else if (Command.C.cmd == MachO::LC_SUB_UMBRELLA) {
10177       MachO::sub_umbrella_command Sf = Obj->getSubUmbrellaCommand(Command);
10178       PrintSubUmbrellaCommand(Sf, Command.Ptr);
10179     } else if (Command.C.cmd == MachO::LC_SUB_LIBRARY) {
10180       MachO::sub_library_command Sl = Obj->getSubLibraryCommand(Command);
10181       PrintSubLibraryCommand(Sl, Command.Ptr);
10182     } else if (Command.C.cmd == MachO::LC_SUB_CLIENT) {
10183       MachO::sub_client_command Sc = Obj->getSubClientCommand(Command);
10184       PrintSubClientCommand(Sc, Command.Ptr);
10185     } else if (Command.C.cmd == MachO::LC_ROUTINES) {
10186       MachO::routines_command Rc = Obj->getRoutinesCommand(Command);
10187       PrintRoutinesCommand(Rc);
10188     } else if (Command.C.cmd == MachO::LC_ROUTINES_64) {
10189       MachO::routines_command_64 Rc = Obj->getRoutinesCommand64(Command);
10190       PrintRoutinesCommand64(Rc);
10191     } else if (Command.C.cmd == MachO::LC_THREAD ||
10192                Command.C.cmd == MachO::LC_UNIXTHREAD) {
10193       MachO::thread_command Tc = Obj->getThreadCommand(Command);
10194       PrintThreadCommand(Tc, Command.Ptr, Obj->isLittleEndian(), cputype);
10195     } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB ||
10196                Command.C.cmd == MachO::LC_ID_DYLIB ||
10197                Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
10198                Command.C.cmd == MachO::LC_REEXPORT_DYLIB ||
10199                Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
10200                Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
10201       MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command);
10202       PrintDylibCommand(Dl, Command.Ptr);
10203     } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE ||
10204                Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO ||
10205                Command.C.cmd == MachO::LC_FUNCTION_STARTS ||
10206                Command.C.cmd == MachO::LC_DATA_IN_CODE ||
10207                Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS ||
10208                Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) {
10209       MachO::linkedit_data_command Ld =
10210           Obj->getLinkeditDataLoadCommand(Command);
10211       PrintLinkEditDataCommand(Ld, Buf.size());
10212     } else {
10213       outs() << "      cmd ?(" << format("0x%08" PRIx32, Command.C.cmd)
10214              << ")\n";
10215       outs() << "  cmdsize " << Command.C.cmdsize << "\n";
10216       // TODO: get and print the raw bytes of the load command.
10217     }
10218     // TODO: print all the other kinds of load commands.
10219   }
10220 }
10221 
PrintMachHeader(const MachOObjectFile * Obj,bool verbose)10222 static void PrintMachHeader(const MachOObjectFile *Obj, bool verbose) {
10223   if (Obj->is64Bit()) {
10224     MachO::mach_header_64 H_64;
10225     H_64 = Obj->getHeader64();
10226     PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype,
10227                     H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose);
10228   } else {
10229     MachO::mach_header H;
10230     H = Obj->getHeader();
10231     PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds,
10232                     H.sizeofcmds, H.flags, verbose);
10233   }
10234 }
10235 
printMachOFileHeader(const object::ObjectFile * Obj)10236 void objdump::printMachOFileHeader(const object::ObjectFile *Obj) {
10237   const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
10238   PrintMachHeader(file, !NonVerbose);
10239 }
10240 
printMachOLoadCommands(const object::ObjectFile * Obj)10241 void objdump::printMachOLoadCommands(const object::ObjectFile *Obj) {
10242   const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
10243   uint32_t filetype = 0;
10244   uint32_t cputype = 0;
10245   if (file->is64Bit()) {
10246     MachO::mach_header_64 H_64;
10247     H_64 = file->getHeader64();
10248     filetype = H_64.filetype;
10249     cputype = H_64.cputype;
10250   } else {
10251     MachO::mach_header H;
10252     H = file->getHeader();
10253     filetype = H.filetype;
10254     cputype = H.cputype;
10255   }
10256   PrintLoadCommands(file, filetype, cputype, !NonVerbose);
10257 }
10258 
10259 //===----------------------------------------------------------------------===//
10260 // export trie dumping
10261 //===----------------------------------------------------------------------===//
10262 
printMachOExportsTrie(const object::MachOObjectFile * Obj)10263 static void printMachOExportsTrie(const object::MachOObjectFile *Obj) {
10264   uint64_t BaseSegmentAddress = 0;
10265   for (const auto &Command : Obj->load_commands()) {
10266     if (Command.C.cmd == MachO::LC_SEGMENT) {
10267       MachO::segment_command Seg = Obj->getSegmentLoadCommand(Command);
10268       if (Seg.fileoff == 0 && Seg.filesize != 0) {
10269         BaseSegmentAddress = Seg.vmaddr;
10270         break;
10271       }
10272     } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
10273       MachO::segment_command_64 Seg = Obj->getSegment64LoadCommand(Command);
10274       if (Seg.fileoff == 0 && Seg.filesize != 0) {
10275         BaseSegmentAddress = Seg.vmaddr;
10276         break;
10277       }
10278     }
10279   }
10280   Error Err = Error::success();
10281   for (const object::ExportEntry &Entry : Obj->exports(Err)) {
10282     uint64_t Flags = Entry.flags();
10283     bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT);
10284     bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION);
10285     bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
10286                         MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL);
10287     bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
10288                 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE);
10289     bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER);
10290     if (ReExport)
10291       outs() << "[re-export] ";
10292     else
10293       outs() << format("0x%08llX  ",
10294                        Entry.address() + BaseSegmentAddress);
10295     outs() << Entry.name();
10296     if (WeakDef || ThreadLocal || Resolver || Abs) {
10297       bool NeedsComma = false;
10298       outs() << " [";
10299       if (WeakDef) {
10300         outs() << "weak_def";
10301         NeedsComma = true;
10302       }
10303       if (ThreadLocal) {
10304         if (NeedsComma)
10305           outs() << ", ";
10306         outs() << "per-thread";
10307         NeedsComma = true;
10308       }
10309       if (Abs) {
10310         if (NeedsComma)
10311           outs() << ", ";
10312         outs() << "absolute";
10313         NeedsComma = true;
10314       }
10315       if (Resolver) {
10316         if (NeedsComma)
10317           outs() << ", ";
10318         outs() << format("resolver=0x%08llX", Entry.other());
10319         NeedsComma = true;
10320       }
10321       outs() << "]";
10322     }
10323     if (ReExport) {
10324       StringRef DylibName = "unknown";
10325       int Ordinal = Entry.other() - 1;
10326       Obj->getLibraryShortNameByIndex(Ordinal, DylibName);
10327       if (Entry.otherName().empty())
10328         outs() << " (from " << DylibName << ")";
10329       else
10330         outs() << " (" << Entry.otherName() << " from " << DylibName << ")";
10331     }
10332     outs() << "\n";
10333   }
10334   if (Err)
10335     reportError(std::move(Err), Obj->getFileName());
10336 }
10337 
10338 //===----------------------------------------------------------------------===//
10339 // rebase table dumping
10340 //===----------------------------------------------------------------------===//
10341 
printMachORebaseTable(object::MachOObjectFile * Obj)10342 static void printMachORebaseTable(object::MachOObjectFile *Obj) {
10343   outs() << "segment  section            address     type\n";
10344   Error Err = Error::success();
10345   for (const object::MachORebaseEntry &Entry : Obj->rebaseTable(Err)) {
10346     StringRef SegmentName = Entry.segmentName();
10347     StringRef SectionName = Entry.sectionName();
10348     uint64_t Address = Entry.address();
10349 
10350     // Table lines look like: __DATA  __nl_symbol_ptr  0x0000F00C  pointer
10351     outs() << format("%-8s %-18s 0x%08" PRIX64 "  %s\n",
10352                      SegmentName.str().c_str(), SectionName.str().c_str(),
10353                      Address, Entry.typeName().str().c_str());
10354   }
10355   if (Err)
10356     reportError(std::move(Err), Obj->getFileName());
10357 }
10358 
ordinalName(const object::MachOObjectFile * Obj,int Ordinal)10359 static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) {
10360   StringRef DylibName;
10361   switch (Ordinal) {
10362   case MachO::BIND_SPECIAL_DYLIB_SELF:
10363     return "this-image";
10364   case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE:
10365     return "main-executable";
10366   case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP:
10367     return "flat-namespace";
10368   default:
10369     if (Ordinal > 0) {
10370       std::error_code EC =
10371           Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName);
10372       if (EC)
10373         return "<<bad library ordinal>>";
10374       return DylibName;
10375     }
10376   }
10377   return "<<unknown special ordinal>>";
10378 }
10379 
10380 //===----------------------------------------------------------------------===//
10381 // bind table dumping
10382 //===----------------------------------------------------------------------===//
10383 
printMachOBindTable(object::MachOObjectFile * Obj)10384 static void printMachOBindTable(object::MachOObjectFile *Obj) {
10385   // Build table of sections so names can used in final output.
10386   outs() << "segment  section            address    type       "
10387             "addend dylib            symbol\n";
10388   Error Err = Error::success();
10389   for (const object::MachOBindEntry &Entry : Obj->bindTable(Err)) {
10390     StringRef SegmentName = Entry.segmentName();
10391     StringRef SectionName = Entry.sectionName();
10392     uint64_t Address = Entry.address();
10393 
10394     // Table lines look like:
10395     //  __DATA  __got  0x00012010    pointer   0 libSystem ___stack_chk_guard
10396     StringRef Attr;
10397     if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT)
10398       Attr = " (weak_import)";
10399     outs() << left_justify(SegmentName, 8) << " "
10400            << left_justify(SectionName, 18) << " "
10401            << format_hex(Address, 10, true) << " "
10402            << left_justify(Entry.typeName(), 8) << " "
10403            << format_decimal(Entry.addend(), 8) << " "
10404            << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
10405            << Entry.symbolName() << Attr << "\n";
10406   }
10407   if (Err)
10408     reportError(std::move(Err), Obj->getFileName());
10409 }
10410 
10411 //===----------------------------------------------------------------------===//
10412 // lazy bind table dumping
10413 //===----------------------------------------------------------------------===//
10414 
printMachOLazyBindTable(object::MachOObjectFile * Obj)10415 static void printMachOLazyBindTable(object::MachOObjectFile *Obj) {
10416   outs() << "segment  section            address     "
10417             "dylib            symbol\n";
10418   Error Err = Error::success();
10419   for (const object::MachOBindEntry &Entry : Obj->lazyBindTable(Err)) {
10420     StringRef SegmentName = Entry.segmentName();
10421     StringRef SectionName = Entry.sectionName();
10422     uint64_t Address = Entry.address();
10423 
10424     // Table lines look like:
10425     //  __DATA  __got  0x00012010 libSystem ___stack_chk_guard
10426     outs() << left_justify(SegmentName, 8) << " "
10427            << left_justify(SectionName, 18) << " "
10428            << format_hex(Address, 10, true) << " "
10429            << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
10430            << Entry.symbolName() << "\n";
10431   }
10432   if (Err)
10433     reportError(std::move(Err), Obj->getFileName());
10434 }
10435 
10436 //===----------------------------------------------------------------------===//
10437 // weak bind table dumping
10438 //===----------------------------------------------------------------------===//
10439 
printMachOWeakBindTable(object::MachOObjectFile * Obj)10440 static void printMachOWeakBindTable(object::MachOObjectFile *Obj) {
10441   outs() << "segment  section            address     "
10442             "type       addend   symbol\n";
10443   Error Err = Error::success();
10444   for (const object::MachOBindEntry &Entry : Obj->weakBindTable(Err)) {
10445     // Strong symbols don't have a location to update.
10446     if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) {
10447       outs() << "                                        strong              "
10448              << Entry.symbolName() << "\n";
10449       continue;
10450     }
10451     StringRef SegmentName = Entry.segmentName();
10452     StringRef SectionName = Entry.sectionName();
10453     uint64_t Address = Entry.address();
10454 
10455     // Table lines look like:
10456     // __DATA  __data  0x00001000  pointer    0   _foo
10457     outs() << left_justify(SegmentName, 8) << " "
10458            << left_justify(SectionName, 18) << " "
10459            << format_hex(Address, 10, true) << " "
10460            << left_justify(Entry.typeName(), 8) << " "
10461            << format_decimal(Entry.addend(), 8) << "   " << Entry.symbolName()
10462            << "\n";
10463   }
10464   if (Err)
10465     reportError(std::move(Err), Obj->getFileName());
10466 }
10467 
10468 // get_dyld_bind_info_symbolname() is used for disassembly and passed an
10469 // address, ReferenceValue, in the Mach-O file and looks in the dyld bind
10470 // information for that address. If the address is found its binding symbol
10471 // name is returned.  If not nullptr is returned.
get_dyld_bind_info_symbolname(uint64_t ReferenceValue,struct DisassembleInfo * info)10472 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
10473                                                  struct DisassembleInfo *info) {
10474   if (info->bindtable == nullptr) {
10475     info->bindtable = std::make_unique<SymbolAddressMap>();
10476     Error Err = Error::success();
10477     for (const object::MachOBindEntry &Entry : info->O->bindTable(Err)) {
10478       uint64_t Address = Entry.address();
10479       StringRef name = Entry.symbolName();
10480       if (!name.empty())
10481         (*info->bindtable)[Address] = name;
10482     }
10483     if (Err)
10484       reportError(std::move(Err), info->O->getFileName());
10485   }
10486   auto name = info->bindtable->lookup(ReferenceValue);
10487   return !name.empty() ? name.data() : nullptr;
10488 }
10489 
printLazyBindTable(ObjectFile * o)10490 void objdump::printLazyBindTable(ObjectFile *o) {
10491   outs() << "Lazy bind table:\n";
10492   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10493     printMachOLazyBindTable(MachO);
10494   else
10495     WithColor::error()
10496         << "This operation is only currently supported "
10497            "for Mach-O executable files.\n";
10498 }
10499 
printWeakBindTable(ObjectFile * o)10500 void objdump::printWeakBindTable(ObjectFile *o) {
10501   outs() << "Weak bind table:\n";
10502   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10503     printMachOWeakBindTable(MachO);
10504   else
10505     WithColor::error()
10506         << "This operation is only currently supported "
10507            "for Mach-O executable files.\n";
10508 }
10509 
printExportsTrie(const ObjectFile * o)10510 void objdump::printExportsTrie(const ObjectFile *o) {
10511   outs() << "Exports trie:\n";
10512   if (const MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10513     printMachOExportsTrie(MachO);
10514   else
10515     WithColor::error()
10516         << "This operation is only currently supported "
10517            "for Mach-O executable files.\n";
10518 }
10519 
printRebaseTable(ObjectFile * o)10520 void objdump::printRebaseTable(ObjectFile *o) {
10521   outs() << "Rebase table:\n";
10522   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10523     printMachORebaseTable(MachO);
10524   else
10525     WithColor::error()
10526         << "This operation is only currently supported "
10527            "for Mach-O executable files.\n";
10528 }
10529 
printBindTable(ObjectFile * o)10530 void objdump::printBindTable(ObjectFile *o) {
10531   outs() << "Bind table:\n";
10532   if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o))
10533     printMachOBindTable(MachO);
10534   else
10535     WithColor::error()
10536         << "This operation is only currently supported "
10537            "for Mach-O executable files.\n";
10538 }
10539