blob: 3c8b9bce4c92b0ee328a6324ba02e3b9eacb8c56 [file] [log] [blame]
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001//===-- MachODump.cpp - Object file dumping utility for llvm --------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file implements the MachO-specific dumper for llvm-objdump.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm-objdump.h"
Kevin Enderby98c9acc2014-09-16 18:00:57 +000015#include "llvm-c/Disassembler.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000016#include "llvm/ADT/STLExtras.h"
Ahmed Bougachaaa790682013-05-24 01:07:04 +000017#include "llvm/ADT/StringExtras.h"
Chandler Carruth4d88a1c2012-12-04 10:44:52 +000018#include "llvm/ADT/Triple.h"
Kevin Enderby04bf6932014-10-28 23:39:46 +000019#include "llvm/Config/config.h"
Zachary Turner82af9432015-01-30 18:07:45 +000020#include "llvm/DebugInfo/DWARF/DIContext.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000021#include "llvm/MC/MCAsmInfo.h"
Lang Hamesa1bc0f52014-04-15 04:40:56 +000022#include "llvm/MC/MCContext.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000023#include "llvm/MC/MCDisassembler.h"
24#include "llvm/MC/MCInst.h"
25#include "llvm/MC/MCInstPrinter.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000026#include "llvm/MC/MCInstrDesc.h"
27#include "llvm/MC/MCInstrInfo.h"
Jim Grosbachfd93a592012-03-05 19:33:20 +000028#include "llvm/MC/MCRegisterInfo.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000029#include "llvm/MC/MCSubtargetInfo.h"
Chandler Carruth4d88a1c2012-12-04 10:44:52 +000030#include "llvm/Object/MachO.h"
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +000031#include "llvm/Object/MachOUniversal.h"
Rafael Espindola9b709252013-04-13 01:45:40 +000032#include "llvm/Support/Casting.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000033#include "llvm/Support/CommandLine.h"
34#include "llvm/Support/Debug.h"
Tim Northover4bd286a2014-08-01 13:07:19 +000035#include "llvm/Support/Endian.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000036#include "llvm/Support/Format.h"
Chandler Carruthd9903882015-01-14 11:23:27 +000037#include "llvm/Support/FormattedStream.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000038#include "llvm/Support/GraphWriter.h"
Kevin Enderby9a509442015-01-27 21:28:24 +000039#include "llvm/Support/LEB128.h"
Chandler Carruth4d88a1c2012-12-04 10:44:52 +000040#include "llvm/Support/MachO.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000041#include "llvm/Support/MemoryBuffer.h"
42#include "llvm/Support/TargetRegistry.h"
43#include "llvm/Support/TargetSelect.h"
44#include "llvm/Support/raw_ostream.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000045#include <algorithm>
46#include <cstring>
Rafael Espindolaa6e9c3e2014-06-12 17:38:55 +000047#include <system_error>
Kevin Enderby04bf6932014-10-28 23:39:46 +000048
49#if HAVE_CXXABI_H
50#include <cxxabi.h>
51#endif
52
Benjamin Kramer43a772e2011-09-19 17:56:04 +000053using namespace llvm;
54using namespace object;
55
56static cl::opt<bool>
Kevin Enderbyb28ed012014-10-29 21:28:24 +000057 UseDbg("g",
58 cl::desc("Print line information from debug info if available"));
Benjamin Kramer699128e2011-09-21 01:13:19 +000059
Kevin Enderbyb28ed012014-10-29 21:28:24 +000060static cl::opt<std::string> DSYMFile("dsym",
61 cl::desc("Use .dSYM file for debug info"));
Benjamin Kramer699128e2011-09-21 01:13:19 +000062
Kevin Enderbyb28ed012014-10-29 21:28:24 +000063static cl::opt<bool> FullLeadingAddr("full-leading-addr",
64 cl::desc("Print full leading address"));
Kevin Enderbybf246f52014-09-24 23:08:22 +000065
Kevin Enderbyab5e6c92015-03-17 21:07:39 +000066static cl::opt<bool> NoLeadingAddr("no-leading-addr",
67 cl::desc("Print no leading address"));
68
Kevin Enderbybf246f52014-09-24 23:08:22 +000069static cl::opt<bool>
70 PrintImmHex("print-imm-hex",
71 cl::desc("Use hex format for immediate values"));
72
Kevin Enderby13023a12015-01-15 23:19:11 +000073cl::opt<bool> llvm::UniversalHeaders("universal-headers",
74 cl::desc("Print Mach-O universal headers "
75 "(requires -macho)"));
76
Kevin Enderby131d1772015-01-09 19:22:37 +000077cl::opt<bool>
Kevin Enderby13023a12015-01-15 23:19:11 +000078 llvm::ArchiveHeaders("archive-headers",
79 cl::desc("Print archive headers for Mach-O archives "
80 "(requires -macho)"));
Kevin Enderby131d1772015-01-09 19:22:37 +000081
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +000082cl::opt<bool>
83 llvm::IndirectSymbols("indirect-symbols",
84 cl::desc("Print indirect symbol table for Mach-O "
85 "objects (requires -macho)"));
86
Kevin Enderby69fe98d2015-01-23 18:52:17 +000087cl::opt<bool>
88 llvm::DataInCode("data-in-code",
89 cl::desc("Print the data in code table for Mach-O objects "
90 "(requires -macho)"));
91
Kevin Enderby9a509442015-01-27 21:28:24 +000092cl::opt<bool>
93 llvm::LinkOptHints("link-opt-hints",
94 cl::desc("Print the linker optimization hints for "
95 "Mach-O objects (requires -macho)"));
96
Kevin Enderbyf6d25852015-01-31 00:37:11 +000097cl::list<std::string>
98 llvm::DumpSections("section",
99 cl::desc("Prints the specified segment,section for "
100 "Mach-O objects (requires -macho)"));
101
Kevin Enderbycd66be52015-03-11 22:06:32 +0000102cl::opt<bool>
103 llvm::InfoPlist("info-plist",
104 cl::desc("Print the info plist section as strings for "
105 "Mach-O objects (requires -macho)"));
106
Kevin Enderbyf0640752015-03-13 17:56:32 +0000107cl::opt<bool>
Kevin Enderbybc847fa2015-03-16 20:08:09 +0000108 llvm::DylibsUsed("dylibs-used",
109 cl::desc("Print the shared libraries used for linked "
110 "Mach-O files (requires -macho)"));
111
112cl::opt<bool>
113 llvm::DylibId("dylib-id",
114 cl::desc("Print the shared library's id for the dylib Mach-O "
115 "file (requires -macho)"));
116
117cl::opt<bool>
Kevin Enderbyf0640752015-03-13 17:56:32 +0000118 llvm::NonVerbose("non-verbose",
119 cl::desc("Print the info for Mach-O objects in "
120 "non-verbose or numeric form (requires -macho)"));
121
Kevin Enderby6a221752015-03-17 17:10:57 +0000122cl::opt<std::string> llvm::DisSymName(
123 "dis-symname",
124 cl::desc("disassemble just this symbol's instructions (requires -macho"));
125
Kevin Enderby8e29ec92015-03-17 22:26:11 +0000126static cl::opt<bool> NoSymbolicOperands(
127 "no-symbolic-operands",
128 cl::desc("do not symbolic operands when disassembling (requires -macho)"));
129
130
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +0000131static cl::list<std::string>
132 ArchFlags("arch", cl::desc("architecture(s) from a Mach-O file to dump"),
133 cl::ZeroOrMore);
134bool ArchAll = false;
135
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000136static std::string ThumbTripleName;
137
138static const Target *GetTarget(const MachOObjectFile *MachOObj,
139 const char **McpuDefault,
140 const Target **ThumbTarget) {
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000141 // Figure out the target triple.
Cameron Zwarich88cc16a2012-02-03 06:35:22 +0000142 if (TripleName.empty()) {
143 llvm::Triple TT("unknown-unknown-unknown");
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000144 llvm::Triple ThumbTriple = Triple();
145 TT = MachOObj->getArch(McpuDefault, &ThumbTriple);
Cameron Zwarich88cc16a2012-02-03 06:35:22 +0000146 TripleName = TT.str();
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000147 ThumbTripleName = ThumbTriple.str();
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000148 }
149
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000150 // Get the target specific parser.
151 std::string Error;
152 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000153 if (TheTarget && ThumbTripleName.empty())
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000154 return TheTarget;
155
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000156 *ThumbTarget = TargetRegistry::lookupTarget(ThumbTripleName, Error);
157 if (*ThumbTarget)
158 return TheTarget;
159
160 errs() << "llvm-objdump: error: unable to get target for '";
161 if (!TheTarget)
162 errs() << TripleName;
163 else
164 errs() << ThumbTripleName;
165 errs() << "', see --version and --triple.\n";
Craig Toppere6cb63e2014-04-25 04:24:47 +0000166 return nullptr;
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000167}
168
Owen Andersond9243c42011-10-17 21:37:35 +0000169struct SymbolSorter {
170 bool operator()(const SymbolRef &A, const SymbolRef &B) {
171 SymbolRef::Type AType, BType;
172 A.getType(AType);
173 B.getType(BType);
174
175 uint64_t AAddr, BAddr;
176 if (AType != SymbolRef::ST_Function)
177 AAddr = 0;
178 else
179 A.getAddress(AAddr);
180 if (BType != SymbolRef::ST_Function)
181 BAddr = 0;
182 else
183 B.getAddress(BAddr);
184 return AAddr < BAddr;
185 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000186};
187
Kevin Enderby273ae012013-06-06 17:20:50 +0000188// Types for the storted data in code table that is built before disassembly
189// and the predicate function to sort them.
190typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
191typedef std::vector<DiceTableEntry> DiceTable;
192typedef DiceTable::iterator dice_table_iterator;
193
Kevin Enderby930fdc72014-11-06 19:00:13 +0000194// This is used to search for a data in code table entry for the PC being
195// disassembled. The j parameter has the PC in j.first. A single data in code
196// table entry can cover many bytes for each of its Kind's. So if the offset,
197// aka the i.first value, of the data in code table entry plus its Length
198// covers the PC being searched for this will return true. If not it will
199// return false.
David Majnemerea9b8ee2014-11-04 08:41:48 +0000200static bool compareDiceTableEntries(const DiceTableEntry &i,
201 const DiceTableEntry &j) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000202 uint16_t Length;
203 i.second.getLength(Length);
204
205 return j.first >= i.first && j.first < i.first + Length;
Kevin Enderby273ae012013-06-06 17:20:50 +0000206}
207
Kevin Enderby930fdc72014-11-06 19:00:13 +0000208static uint64_t DumpDataInCode(const char *bytes, uint64_t Length,
209 unsigned short Kind) {
210 uint32_t Value, Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000211
212 switch (Kind) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000213 default:
Charles Davis8bdfafd2013-09-01 04:28:48 +0000214 case MachO::DICE_KIND_DATA:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000215 if (Length >= 4) {
216 if (!NoShowRawInsn)
217 DumpBytes(StringRef(bytes, 4));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000218 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
Kevin Enderby273ae012013-06-06 17:20:50 +0000219 outs() << "\t.long " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000220 Size = 4;
221 } else if (Length >= 2) {
222 if (!NoShowRawInsn)
223 DumpBytes(StringRef(bytes, 2));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000224 Value = bytes[1] << 8 | bytes[0];
Kevin Enderby273ae012013-06-06 17:20:50 +0000225 outs() << "\t.short " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000226 Size = 2;
227 } else {
228 if (!NoShowRawInsn)
229 DumpBytes(StringRef(bytes, 2));
Kevin Enderby273ae012013-06-06 17:20:50 +0000230 Value = bytes[0];
231 outs() << "\t.byte " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000232 Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000233 }
Kevin Enderby930fdc72014-11-06 19:00:13 +0000234 if (Kind == MachO::DICE_KIND_DATA)
235 outs() << "\t@ KIND_DATA\n";
236 else
237 outs() << "\t@ data in code kind = " << Kind << "\n";
Kevin Enderby273ae012013-06-06 17:20:50 +0000238 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000239 case MachO::DICE_KIND_JUMP_TABLE8:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000240 if (!NoShowRawInsn)
241 DumpBytes(StringRef(bytes, 1));
Kevin Enderby273ae012013-06-06 17:20:50 +0000242 Value = bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000243 outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n";
244 Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000245 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000246 case MachO::DICE_KIND_JUMP_TABLE16:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000247 if (!NoShowRawInsn)
248 DumpBytes(StringRef(bytes, 2));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000249 Value = bytes[1] << 8 | bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000250 outs() << "\t.short " << format("%5u", Value & 0xffff)
251 << "\t@ KIND_JUMP_TABLE16\n";
252 Size = 2;
Kevin Enderby273ae012013-06-06 17:20:50 +0000253 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000254 case MachO::DICE_KIND_JUMP_TABLE32:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000255 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
256 if (!NoShowRawInsn)
257 DumpBytes(StringRef(bytes, 4));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000258 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000259 outs() << "\t.long " << Value;
260 if (Kind == MachO::DICE_KIND_JUMP_TABLE32)
261 outs() << "\t@ KIND_JUMP_TABLE32\n";
262 else
263 outs() << "\t@ KIND_ABS_JUMP_TABLE32\n";
264 Size = 4;
Kevin Enderby273ae012013-06-06 17:20:50 +0000265 break;
266 }
Kevin Enderby930fdc72014-11-06 19:00:13 +0000267 return Size;
Kevin Enderby273ae012013-06-06 17:20:50 +0000268}
269
Alexey Samsonov464d2e42014-03-17 07:28:19 +0000270static void getSectionsAndSymbols(const MachO::mach_header Header,
271 MachOObjectFile *MachOObj,
272 std::vector<SectionRef> &Sections,
273 std::vector<SymbolRef> &Symbols,
274 SmallVectorImpl<uint64_t> &FoundFns,
275 uint64_t &BaseSegmentAddress) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +0000276 for (const SymbolRef &Symbol : MachOObj->symbols()) {
277 StringRef SymName;
278 Symbol.getName(SymName);
279 if (!SymName.startswith("ltmp"))
280 Symbols.push_back(Symbol);
281 }
Owen Andersond9243c42011-10-17 21:37:35 +0000282
Alexey Samsonov48803e52014-03-13 14:37:36 +0000283 for (const SectionRef &Section : MachOObj->sections()) {
Owen Andersond9243c42011-10-17 21:37:35 +0000284 StringRef SectName;
Alexey Samsonov48803e52014-03-13 14:37:36 +0000285 Section.getName(SectName);
286 Sections.push_back(Section);
Owen Andersond9243c42011-10-17 21:37:35 +0000287 }
288
Rafael Espindola56f976f2013-04-18 18:08:55 +0000289 MachOObjectFile::LoadCommandInfo Command =
Alexey Samsonov48803e52014-03-13 14:37:36 +0000290 MachOObj->getFirstLoadCommandInfo();
Kevin Enderby273ae012013-06-06 17:20:50 +0000291 bool BaseSegmentAddressSet = false;
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000292 for (unsigned i = 0;; ++i) {
Charles Davis8bdfafd2013-09-01 04:28:48 +0000293 if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) {
Benjamin Kramer699128e2011-09-21 01:13:19 +0000294 // We found a function starts segment, parse the addresses for later
295 // consumption.
Charles Davis8bdfafd2013-09-01 04:28:48 +0000296 MachO::linkedit_data_command LLC =
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000297 MachOObj->getLinkeditDataLoadCommand(Command);
Benjamin Kramer699128e2011-09-21 01:13:19 +0000298
Charles Davis8bdfafd2013-09-01 04:28:48 +0000299 MachOObj->ReadULEB128s(LLC.dataoff, FoundFns);
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000300 } else if (Command.C.cmd == MachO::LC_SEGMENT) {
301 MachO::segment_command SLC = MachOObj->getSegmentLoadCommand(Command);
Charles Davis8bdfafd2013-09-01 04:28:48 +0000302 StringRef SegName = SLC.segname;
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000303 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
Kevin Enderby273ae012013-06-06 17:20:50 +0000304 BaseSegmentAddressSet = true;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000305 BaseSegmentAddress = SLC.vmaddr;
Kevin Enderby273ae012013-06-06 17:20:50 +0000306 }
307 }
Rafael Espindolafeef8c22013-04-19 11:36:47 +0000308
Charles Davis8bdfafd2013-09-01 04:28:48 +0000309 if (i == Header.ncmds - 1)
Rafael Espindolafeef8c22013-04-19 11:36:47 +0000310 break;
311 else
312 Command = MachOObj->getNextLoadCommandInfo(Command);
Benjamin Kramer8a529dc2011-09-21 22:16:43 +0000313 }
Benjamin Kramer699128e2011-09-21 01:13:19 +0000314}
315
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +0000316static void PrintIndirectSymbolTable(MachOObjectFile *O, bool verbose,
317 uint32_t n, uint32_t count,
318 uint32_t stride, uint64_t addr) {
319 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
320 uint32_t nindirectsyms = Dysymtab.nindirectsyms;
321 if (n > nindirectsyms)
322 outs() << " (entries start past the end of the indirect symbol "
323 "table) (reserved1 field greater than the table size)";
324 else if (n + count > nindirectsyms)
325 outs() << " (entries extends past the end of the indirect symbol "
326 "table)";
327 outs() << "\n";
328 uint32_t cputype = O->getHeader().cputype;
329 if (cputype & MachO::CPU_ARCH_ABI64)
330 outs() << "address index";
331 else
332 outs() << "address index";
333 if (verbose)
334 outs() << " name\n";
335 else
336 outs() << "\n";
337 for (uint32_t j = 0; j < count && n + j < nindirectsyms; j++) {
338 if (cputype & MachO::CPU_ARCH_ABI64)
339 outs() << format("0x%016" PRIx64, addr + j * stride) << " ";
340 else
341 outs() << format("0x%08" PRIx32, addr + j * stride) << " ";
342 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
343 uint32_t indirect_symbol = O->getIndirectSymbolTableEntry(Dysymtab, n + j);
344 if (indirect_symbol == MachO::INDIRECT_SYMBOL_LOCAL) {
345 outs() << "LOCAL\n";
346 continue;
347 }
348 if (indirect_symbol ==
349 (MachO::INDIRECT_SYMBOL_LOCAL | MachO::INDIRECT_SYMBOL_ABS)) {
350 outs() << "LOCAL ABSOLUTE\n";
351 continue;
352 }
353 if (indirect_symbol == MachO::INDIRECT_SYMBOL_ABS) {
354 outs() << "ABSOLUTE\n";
355 continue;
356 }
357 outs() << format("%5u ", indirect_symbol);
Kevin Enderbyf0640752015-03-13 17:56:32 +0000358 if (verbose) {
359 MachO::symtab_command Symtab = O->getSymtabLoadCommand();
360 if (indirect_symbol < Symtab.nsyms) {
361 symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol);
362 SymbolRef Symbol = *Sym;
363 StringRef SymName;
364 Symbol.getName(SymName);
365 outs() << SymName;
366 } else {
367 outs() << "?";
368 }
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +0000369 }
370 outs() << "\n";
371 }
372}
373
374static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) {
375 uint32_t LoadCommandCount = O->getHeader().ncmds;
376 MachOObjectFile::LoadCommandInfo Load = O->getFirstLoadCommandInfo();
377 for (unsigned I = 0;; ++I) {
378 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
379 MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load);
380 for (unsigned J = 0; J < Seg.nsects; ++J) {
381 MachO::section_64 Sec = O->getSection64(Load, J);
382 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
383 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
384 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
385 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
386 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
387 section_type == MachO::S_SYMBOL_STUBS) {
388 uint32_t stride;
389 if (section_type == MachO::S_SYMBOL_STUBS)
390 stride = Sec.reserved2;
391 else
392 stride = 8;
393 if (stride == 0) {
394 outs() << "Can't print indirect symbols for (" << Sec.segname << ","
395 << Sec.sectname << ") "
396 << "(size of stubs in reserved2 field is zero)\n";
397 continue;
398 }
399 uint32_t count = Sec.size / stride;
400 outs() << "Indirect symbols for (" << Sec.segname << ","
401 << Sec.sectname << ") " << count << " entries";
402 uint32_t n = Sec.reserved1;
403 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
404 }
405 }
406 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
407 MachO::segment_command Seg = O->getSegmentLoadCommand(Load);
408 for (unsigned J = 0; J < Seg.nsects; ++J) {
409 MachO::section Sec = O->getSection(Load, J);
410 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
411 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
412 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
413 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
414 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
415 section_type == MachO::S_SYMBOL_STUBS) {
416 uint32_t stride;
417 if (section_type == MachO::S_SYMBOL_STUBS)
418 stride = Sec.reserved2;
419 else
420 stride = 4;
421 if (stride == 0) {
422 outs() << "Can't print indirect symbols for (" << Sec.segname << ","
423 << Sec.sectname << ") "
424 << "(size of stubs in reserved2 field is zero)\n";
425 continue;
426 }
427 uint32_t count = Sec.size / stride;
428 outs() << "Indirect symbols for (" << Sec.segname << ","
429 << Sec.sectname << ") " << count << " entries";
430 uint32_t n = Sec.reserved1;
431 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
432 }
433 }
434 }
435 if (I == LoadCommandCount - 1)
436 break;
437 else
438 Load = O->getNextLoadCommandInfo(Load);
439 }
440}
441
Kevin Enderby69fe98d2015-01-23 18:52:17 +0000442static void PrintDataInCodeTable(MachOObjectFile *O, bool verbose) {
443 MachO::linkedit_data_command DIC = O->getDataInCodeLoadCommand();
444 uint32_t nentries = DIC.datasize / sizeof(struct MachO::data_in_code_entry);
445 outs() << "Data in code table (" << nentries << " entries)\n";
446 outs() << "offset length kind\n";
447 for (dice_iterator DI = O->begin_dices(), DE = O->end_dices(); DI != DE;
448 ++DI) {
449 uint32_t Offset;
450 DI->getOffset(Offset);
451 outs() << format("0x%08" PRIx32, Offset) << " ";
452 uint16_t Length;
453 DI->getLength(Length);
454 outs() << format("%6u", Length) << " ";
455 uint16_t Kind;
456 DI->getKind(Kind);
457 if (verbose) {
458 switch (Kind) {
459 case MachO::DICE_KIND_DATA:
460 outs() << "DATA";
461 break;
462 case MachO::DICE_KIND_JUMP_TABLE8:
463 outs() << "JUMP_TABLE8";
464 break;
465 case MachO::DICE_KIND_JUMP_TABLE16:
466 outs() << "JUMP_TABLE16";
467 break;
468 case MachO::DICE_KIND_JUMP_TABLE32:
469 outs() << "JUMP_TABLE32";
470 break;
471 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
472 outs() << "ABS_JUMP_TABLE32";
473 break;
474 default:
475 outs() << format("0x%04" PRIx32, Kind);
476 break;
477 }
478 } else
479 outs() << format("0x%04" PRIx32, Kind);
480 outs() << "\n";
481 }
482}
483
Kevin Enderby9a509442015-01-27 21:28:24 +0000484static void PrintLinkOptHints(MachOObjectFile *O) {
485 MachO::linkedit_data_command LohLC = O->getLinkOptHintsLoadCommand();
486 const char *loh = O->getData().substr(LohLC.dataoff, 1).data();
487 uint32_t nloh = LohLC.datasize;
488 outs() << "Linker optimiztion hints (" << nloh << " total bytes)\n";
489 for (uint32_t i = 0; i < nloh;) {
490 unsigned n;
491 uint64_t identifier = decodeULEB128((const uint8_t *)(loh + i), &n);
492 i += n;
493 outs() << " identifier " << identifier << " ";
494 if (i >= nloh)
495 return;
496 switch (identifier) {
497 case 1:
498 outs() << "AdrpAdrp\n";
499 break;
500 case 2:
501 outs() << "AdrpLdr\n";
502 break;
503 case 3:
504 outs() << "AdrpAddLdr\n";
505 break;
506 case 4:
507 outs() << "AdrpLdrGotLdr\n";
508 break;
509 case 5:
510 outs() << "AdrpAddStr\n";
511 break;
512 case 6:
513 outs() << "AdrpLdrGotStr\n";
514 break;
515 case 7:
516 outs() << "AdrpAdd\n";
517 break;
518 case 8:
519 outs() << "AdrpLdrGot\n";
520 break;
521 default:
522 outs() << "Unknown identifier value\n";
523 break;
524 }
525 uint64_t narguments = decodeULEB128((const uint8_t *)(loh + i), &n);
526 i += n;
527 outs() << " narguments " << narguments << "\n";
528 if (i >= nloh)
529 return;
530
531 for (uint32_t j = 0; j < narguments; j++) {
532 uint64_t value = decodeULEB128((const uint8_t *)(loh + i), &n);
533 i += n;
534 outs() << "\tvalue " << format("0x%" PRIx64, value) << "\n";
535 if (i >= nloh)
536 return;
537 }
538 }
539}
540
Kevin Enderbybc847fa2015-03-16 20:08:09 +0000541static void PrintDylibs(MachOObjectFile *O, bool JustId) {
542 uint32_t LoadCommandCount = O->getHeader().ncmds;
543 MachOObjectFile::LoadCommandInfo Load = O->getFirstLoadCommandInfo();
544 for (unsigned I = 0;; ++I) {
545 if ((JustId && Load.C.cmd == MachO::LC_ID_DYLIB) ||
546 (!JustId && (Load.C.cmd == MachO::LC_ID_DYLIB ||
547 Load.C.cmd == MachO::LC_LOAD_DYLIB ||
548 Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
549 Load.C.cmd == MachO::LC_REEXPORT_DYLIB ||
550 Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
551 Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB))) {
552 MachO::dylib_command dl = O->getDylibIDLoadCommand(Load);
553 if (dl.dylib.name < dl.cmdsize) {
554 const char *p = (const char *)(Load.Ptr) + dl.dylib.name;
555 if (JustId)
556 outs() << p << "\n";
557 else {
558 outs() << "\t" << p;
559 outs() << " (compatibility version "
560 << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
561 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
562 << (dl.dylib.compatibility_version & 0xff) << ",";
563 outs() << " current version "
564 << ((dl.dylib.current_version >> 16) & 0xffff) << "."
565 << ((dl.dylib.current_version >> 8) & 0xff) << "."
566 << (dl.dylib.current_version & 0xff) << ")\n";
567 }
568 } else {
569 outs() << "\tBad offset (" << dl.dylib.name << ") for name of ";
570 if (Load.C.cmd == MachO::LC_ID_DYLIB)
571 outs() << "LC_ID_DYLIB ";
572 else if (Load.C.cmd == MachO::LC_LOAD_DYLIB)
573 outs() << "LC_LOAD_DYLIB ";
574 else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB)
575 outs() << "LC_LOAD_WEAK_DYLIB ";
576 else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB)
577 outs() << "LC_LAZY_LOAD_DYLIB ";
578 else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB)
579 outs() << "LC_REEXPORT_DYLIB ";
580 else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
581 outs() << "LC_LOAD_UPWARD_DYLIB ";
582 else
583 outs() << "LC_??? ";
584 outs() << "command " << I << "\n";
585 }
586 }
587 if (I == LoadCommandCount - 1)
588 break;
589 else
590 Load = O->getNextLoadCommandInfo(Load);
591 }
592}
593
Kevin Enderbyf6d25852015-01-31 00:37:11 +0000594typedef DenseMap<uint64_t, StringRef> SymbolAddressMap;
595
596static void CreateSymbolAddressMap(MachOObjectFile *O,
597 SymbolAddressMap *AddrMap) {
598 // Create a map of symbol addresses to symbol names.
599 for (const SymbolRef &Symbol : O->symbols()) {
600 SymbolRef::Type ST;
601 Symbol.getType(ST);
602 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
603 ST == SymbolRef::ST_Other) {
604 uint64_t Address;
605 Symbol.getAddress(Address);
606 StringRef SymName;
607 Symbol.getName(SymName);
608 (*AddrMap)[Address] = SymName;
609 }
610 }
611}
612
613// GuessSymbolName is passed the address of what might be a symbol and a
614// pointer to the SymbolAddressMap. It returns the name of a symbol
615// with that address or nullptr if no symbol is found with that address.
616static const char *GuessSymbolName(uint64_t value, SymbolAddressMap *AddrMap) {
617 const char *SymbolName = nullptr;
618 // A DenseMap can't lookup up some values.
619 if (value != 0xffffffffffffffffULL && value != 0xfffffffffffffffeULL) {
620 StringRef name = AddrMap->lookup(value);
621 if (!name.empty())
622 SymbolName = name.data();
623 }
624 return SymbolName;
625}
626
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000627static void DumpCstringChar(const char c) {
628 char p[2];
629 p[0] = c;
630 p[1] = '\0';
631 outs().write_escaped(p);
632}
633
Kevin Enderby10ba0412015-02-04 21:38:42 +0000634static void DumpCstringSection(MachOObjectFile *O, const char *sect,
635 uint32_t sect_size, uint64_t sect_addr,
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000636 bool print_addresses) {
Kevin Enderby10ba0412015-02-04 21:38:42 +0000637 for (uint32_t i = 0; i < sect_size; i++) {
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000638 if (print_addresses) {
639 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000640 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000641 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000642 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000643 }
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000644 for (; i < sect_size && sect[i] != '\0'; i++)
645 DumpCstringChar(sect[i]);
Kevin Enderby10ba0412015-02-04 21:38:42 +0000646 if (i < sect_size && sect[i] == '\0')
647 outs() << "\n";
648 }
649}
650
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000651static void DumpLiteral4(uint32_t l, float f) {
652 outs() << format("0x%08" PRIx32, l);
653 if ((l & 0x7f800000) != 0x7f800000)
654 outs() << format(" (%.16e)\n", f);
655 else {
656 if (l == 0x7f800000)
657 outs() << " (+Infinity)\n";
658 else if (l == 0xff800000)
659 outs() << " (-Infinity)\n";
660 else if ((l & 0x00400000) == 0x00400000)
661 outs() << " (non-signaling Not-a-Number)\n";
662 else
663 outs() << " (signaling Not-a-Number)\n";
664 }
665}
666
667static void DumpLiteral4Section(MachOObjectFile *O, const char *sect,
668 uint32_t sect_size, uint64_t sect_addr,
669 bool print_addresses) {
670 for (uint32_t i = 0; i < sect_size; i += sizeof(float)) {
671 if (print_addresses) {
672 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000673 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000674 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000675 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000676 }
677 float f;
678 memcpy(&f, sect + i, sizeof(float));
679 if (O->isLittleEndian() != sys::IsLittleEndianHost)
680 sys::swapByteOrder(f);
681 uint32_t l;
682 memcpy(&l, sect + i, sizeof(uint32_t));
683 if (O->isLittleEndian() != sys::IsLittleEndianHost)
684 sys::swapByteOrder(l);
685 DumpLiteral4(l, f);
686 }
687}
688
689static void DumpLiteral8(MachOObjectFile *O, uint32_t l0, uint32_t l1,
690 double d) {
691 outs() << format("0x%08" PRIx32, l0) << " " << format("0x%08" PRIx32, l1);
692 uint32_t Hi, Lo;
693 if (O->isLittleEndian()) {
694 Hi = l1;
695 Lo = l0;
696 } else {
697 Hi = l0;
698 Lo = l1;
699 }
700 // Hi is the high word, so this is equivalent to if(isfinite(d))
701 if ((Hi & 0x7ff00000) != 0x7ff00000)
702 outs() << format(" (%.16e)\n", d);
703 else {
704 if (Hi == 0x7ff00000 && Lo == 0)
705 outs() << " (+Infinity)\n";
706 else if (Hi == 0xfff00000 && Lo == 0)
707 outs() << " (-Infinity)\n";
708 else if ((Hi & 0x00080000) == 0x00080000)
709 outs() << " (non-signaling Not-a-Number)\n";
710 else
711 outs() << " (signaling Not-a-Number)\n";
712 }
713}
714
715static void DumpLiteral8Section(MachOObjectFile *O, const char *sect,
716 uint32_t sect_size, uint64_t sect_addr,
717 bool print_addresses) {
718 for (uint32_t i = 0; i < sect_size; i += sizeof(double)) {
719 if (print_addresses) {
720 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000721 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000722 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000723 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000724 }
725 double d;
726 memcpy(&d, sect + i, sizeof(double));
727 if (O->isLittleEndian() != sys::IsLittleEndianHost)
728 sys::swapByteOrder(d);
729 uint32_t l0, l1;
730 memcpy(&l0, sect + i, sizeof(uint32_t));
731 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
732 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
733 sys::swapByteOrder(l0);
734 sys::swapByteOrder(l1);
735 }
736 DumpLiteral8(O, l0, l1, d);
737 }
738}
739
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000740static void DumpLiteral16(uint32_t l0, uint32_t l1, uint32_t l2, uint32_t l3) {
741 outs() << format("0x%08" PRIx32, l0) << " ";
742 outs() << format("0x%08" PRIx32, l1) << " ";
743 outs() << format("0x%08" PRIx32, l2) << " ";
744 outs() << format("0x%08" PRIx32, l3) << "\n";
745}
746
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000747static void DumpLiteral16Section(MachOObjectFile *O, const char *sect,
748 uint32_t sect_size, uint64_t sect_addr,
749 bool print_addresses) {
750 for (uint32_t i = 0; i < sect_size; i += 16) {
751 if (print_addresses) {
752 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000753 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000754 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000755 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000756 }
757 uint32_t l0, l1, l2, l3;
758 memcpy(&l0, sect + i, sizeof(uint32_t));
759 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
760 memcpy(&l2, sect + i + 2 * sizeof(uint32_t), sizeof(uint32_t));
761 memcpy(&l3, sect + i + 3 * sizeof(uint32_t), sizeof(uint32_t));
762 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
763 sys::swapByteOrder(l0);
764 sys::swapByteOrder(l1);
765 sys::swapByteOrder(l2);
766 sys::swapByteOrder(l3);
767 }
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000768 DumpLiteral16(l0, l1, l2, l3);
769 }
770}
771
772static void DumpLiteralPointerSection(MachOObjectFile *O,
773 const SectionRef &Section,
774 const char *sect, uint32_t sect_size,
775 uint64_t sect_addr,
776 bool print_addresses) {
777 // Collect the literal sections in this Mach-O file.
778 std::vector<SectionRef> LiteralSections;
779 for (const SectionRef &Section : O->sections()) {
780 DataRefImpl Ref = Section.getRawDataRefImpl();
781 uint32_t section_type;
782 if (O->is64Bit()) {
783 const MachO::section_64 Sec = O->getSection64(Ref);
784 section_type = Sec.flags & MachO::SECTION_TYPE;
785 } else {
786 const MachO::section Sec = O->getSection(Ref);
787 section_type = Sec.flags & MachO::SECTION_TYPE;
788 }
789 if (section_type == MachO::S_CSTRING_LITERALS ||
790 section_type == MachO::S_4BYTE_LITERALS ||
791 section_type == MachO::S_8BYTE_LITERALS ||
792 section_type == MachO::S_16BYTE_LITERALS)
793 LiteralSections.push_back(Section);
794 }
795
796 // Set the size of the literal pointer.
797 uint32_t lp_size = O->is64Bit() ? 8 : 4;
798
799 // Collect the external relocation symbols for the the literal pointers.
800 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
801 for (const RelocationRef &Reloc : Section.relocations()) {
802 DataRefImpl Rel;
803 MachO::any_relocation_info RE;
804 bool isExtern = false;
805 Rel = Reloc.getRawDataRefImpl();
806 RE = O->getRelocation(Rel);
807 isExtern = O->getPlainRelocationExternal(RE);
808 if (isExtern) {
809 uint64_t RelocOffset;
810 Reloc.getOffset(RelocOffset);
811 symbol_iterator RelocSym = Reloc.getSymbol();
812 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
813 }
814 }
815 array_pod_sort(Relocs.begin(), Relocs.end());
816
817 // Dump each literal pointer.
818 for (uint32_t i = 0; i < sect_size; i += lp_size) {
819 if (print_addresses) {
820 if (O->is64Bit())
821 outs() << format("%016" PRIx64, sect_addr + i) << " ";
822 else
823 outs() << format("%08" PRIx64, sect_addr + i) << " ";
824 }
825 uint64_t lp;
826 if (O->is64Bit()) {
827 memcpy(&lp, sect + i, sizeof(uint64_t));
828 if (O->isLittleEndian() != sys::IsLittleEndianHost)
829 sys::swapByteOrder(lp);
830 } else {
831 uint32_t li;
832 memcpy(&li, sect + i, sizeof(uint32_t));
833 if (O->isLittleEndian() != sys::IsLittleEndianHost)
834 sys::swapByteOrder(li);
835 lp = li;
836 }
837
838 // First look for an external relocation entry for this literal pointer.
839 bool reloc_found = false;
840 for (unsigned j = 0, e = Relocs.size(); j != e; ++j) {
841 if (Relocs[i].first == i) {
842 symbol_iterator RelocSym = Relocs[j].second;
843 StringRef SymName;
844 RelocSym->getName(SymName);
845 outs() << "external relocation entry for symbol:" << SymName << "\n";
846 reloc_found = true;
847 }
848 }
849 if (reloc_found == true)
850 continue;
851
852 // For local references see what the section the literal pointer points to.
853 bool found = false;
854 for (unsigned SectIdx = 0; SectIdx != LiteralSections.size(); SectIdx++) {
855 uint64_t SectAddress = LiteralSections[SectIdx].getAddress();
856 uint64_t SectSize = LiteralSections[SectIdx].getSize();
857 if (lp >= SectAddress && lp < SectAddress + SectSize) {
858 found = true;
859
860 StringRef SectName;
861 LiteralSections[SectIdx].getName(SectName);
862 DataRefImpl Ref = LiteralSections[SectIdx].getRawDataRefImpl();
863 StringRef SegmentName = O->getSectionFinalSegmentName(Ref);
864 outs() << SegmentName << ":" << SectName << ":";
865
866 uint32_t section_type;
867 if (O->is64Bit()) {
868 const MachO::section_64 Sec = O->getSection64(Ref);
869 section_type = Sec.flags & MachO::SECTION_TYPE;
870 } else {
871 const MachO::section Sec = O->getSection(Ref);
872 section_type = Sec.flags & MachO::SECTION_TYPE;
873 }
874
875 StringRef BytesStr;
876 LiteralSections[SectIdx].getContents(BytesStr);
877 const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
878
879 switch (section_type) {
880 case MachO::S_CSTRING_LITERALS:
881 for (uint64_t i = lp - SectAddress;
882 i < SectSize && Contents[i] != '\0'; i++) {
883 DumpCstringChar(Contents[i]);
884 }
885 outs() << "\n";
886 break;
887 case MachO::S_4BYTE_LITERALS:
888 float f;
889 memcpy(&f, Contents + (lp - SectAddress), sizeof(float));
890 uint32_t l;
891 memcpy(&l, Contents + (lp - SectAddress), sizeof(uint32_t));
892 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
893 sys::swapByteOrder(f);
894 sys::swapByteOrder(l);
895 }
896 DumpLiteral4(l, f);
897 break;
898 case MachO::S_8BYTE_LITERALS: {
899 double d;
900 memcpy(&d, Contents + (lp - SectAddress), sizeof(double));
901 uint32_t l0, l1;
902 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
903 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
904 sizeof(uint32_t));
905 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
906 sys::swapByteOrder(f);
907 sys::swapByteOrder(l0);
908 sys::swapByteOrder(l1);
909 }
910 DumpLiteral8(O, l0, l1, d);
911 break;
912 }
913 case MachO::S_16BYTE_LITERALS: {
914 uint32_t l0, l1, l2, l3;
915 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
916 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
917 sizeof(uint32_t));
918 memcpy(&l2, Contents + (lp - SectAddress) + 2 * sizeof(uint32_t),
919 sizeof(uint32_t));
920 memcpy(&l3, Contents + (lp - SectAddress) + 3 * sizeof(uint32_t),
921 sizeof(uint32_t));
922 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
923 sys::swapByteOrder(l0);
924 sys::swapByteOrder(l1);
925 sys::swapByteOrder(l2);
926 sys::swapByteOrder(l3);
927 }
928 DumpLiteral16(l0, l1, l2, l3);
929 break;
930 }
931 }
932 }
933 }
934 if (found == false)
935 outs() << format("0x%" PRIx64, lp) << " (not in a literal section)\n";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000936 }
937}
938
Kevin Enderbyf6d25852015-01-31 00:37:11 +0000939static void DumpInitTermPointerSection(MachOObjectFile *O, const char *sect,
940 uint32_t sect_size, uint64_t sect_addr,
941 SymbolAddressMap *AddrMap,
942 bool verbose) {
943 uint32_t stride;
944 if (O->is64Bit())
945 stride = sizeof(uint64_t);
946 else
947 stride = sizeof(uint32_t);
948 for (uint32_t i = 0; i < sect_size; i += stride) {
949 const char *SymbolName = nullptr;
950 if (O->is64Bit()) {
951 outs() << format("0x%016" PRIx64, sect_addr + i * stride) << " ";
952 uint64_t pointer_value;
953 memcpy(&pointer_value, sect + i, stride);
954 if (O->isLittleEndian() != sys::IsLittleEndianHost)
955 sys::swapByteOrder(pointer_value);
956 outs() << format("0x%016" PRIx64, pointer_value);
957 if (verbose)
958 SymbolName = GuessSymbolName(pointer_value, AddrMap);
959 } else {
960 outs() << format("0x%08" PRIx64, sect_addr + i * stride) << " ";
961 uint32_t pointer_value;
962 memcpy(&pointer_value, sect + i, stride);
963 if (O->isLittleEndian() != sys::IsLittleEndianHost)
964 sys::swapByteOrder(pointer_value);
965 outs() << format("0x%08" PRIx32, pointer_value);
966 if (verbose)
967 SymbolName = GuessSymbolName(pointer_value, AddrMap);
968 }
969 if (SymbolName)
970 outs() << " " << SymbolName;
971 outs() << "\n";
972 }
973}
974
975static void DumpRawSectionContents(MachOObjectFile *O, const char *sect,
976 uint32_t size, uint64_t addr) {
977 uint32_t cputype = O->getHeader().cputype;
978 if (cputype == MachO::CPU_TYPE_I386 || cputype == MachO::CPU_TYPE_X86_64) {
979 uint32_t j;
980 for (uint32_t i = 0; i < size; i += j, addr += j) {
981 if (O->is64Bit())
982 outs() << format("%016" PRIx64, addr) << "\t";
983 else
Kevin Enderbyf0640752015-03-13 17:56:32 +0000984 outs() << format("%08" PRIx64, addr) << "\t";
Kevin Enderbyf6d25852015-01-31 00:37:11 +0000985 for (j = 0; j < 16 && i + j < size; j++) {
986 uint8_t byte_word = *(sect + i + j);
987 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
988 }
989 outs() << "\n";
990 }
991 } else {
992 uint32_t j;
993 for (uint32_t i = 0; i < size; i += j, addr += j) {
994 if (O->is64Bit())
995 outs() << format("%016" PRIx64, addr) << "\t";
996 else
997 outs() << format("%08" PRIx64, sect) << "\t";
998 for (j = 0; j < 4 * sizeof(int32_t) && i + j < size;
999 j += sizeof(int32_t)) {
1000 if (i + j + sizeof(int32_t) < size) {
1001 uint32_t long_word;
1002 memcpy(&long_word, sect + i + j, sizeof(int32_t));
1003 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1004 sys::swapByteOrder(long_word);
1005 outs() << format("%08" PRIx32, long_word) << " ";
1006 } else {
1007 for (uint32_t k = 0; i + j + k < size; k++) {
1008 uint8_t byte_word = *(sect + i + j);
1009 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
1010 }
1011 }
1012 }
1013 outs() << "\n";
1014 }
1015 }
1016}
1017
Kevin Enderby95df54c2015-02-04 01:01:38 +00001018static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
1019 StringRef DisSegName, StringRef DisSectName);
1020
1021static void DumpSectionContents(StringRef Filename, MachOObjectFile *O,
1022 bool verbose) {
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001023 SymbolAddressMap AddrMap;
1024 if (verbose)
1025 CreateSymbolAddressMap(O, &AddrMap);
1026
1027 for (unsigned i = 0; i < DumpSections.size(); ++i) {
1028 StringRef DumpSection = DumpSections[i];
1029 std::pair<StringRef, StringRef> DumpSegSectName;
1030 DumpSegSectName = DumpSection.split(',');
1031 StringRef DumpSegName, DumpSectName;
1032 if (DumpSegSectName.second.size()) {
1033 DumpSegName = DumpSegSectName.first;
1034 DumpSectName = DumpSegSectName.second;
1035 } else {
1036 DumpSegName = "";
1037 DumpSectName = DumpSegSectName.first;
1038 }
1039 for (const SectionRef &Section : O->sections()) {
1040 StringRef SectName;
1041 Section.getName(SectName);
1042 DataRefImpl Ref = Section.getRawDataRefImpl();
1043 StringRef SegName = O->getSectionFinalSegmentName(Ref);
1044 if ((DumpSegName.empty() || SegName == DumpSegName) &&
1045 (SectName == DumpSectName)) {
1046 outs() << "Contents of (" << SegName << "," << SectName
1047 << ") section\n";
Kevin Enderby95df54c2015-02-04 01:01:38 +00001048 uint32_t section_flags;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001049 if (O->is64Bit()) {
1050 const MachO::section_64 Sec = O->getSection64(Ref);
Kevin Enderby95df54c2015-02-04 01:01:38 +00001051 section_flags = Sec.flags;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001052
1053 } else {
1054 const MachO::section Sec = O->getSection(Ref);
Kevin Enderby95df54c2015-02-04 01:01:38 +00001055 section_flags = Sec.flags;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001056 }
Kevin Enderby95df54c2015-02-04 01:01:38 +00001057 uint32_t section_type = section_flags & MachO::SECTION_TYPE;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001058
1059 StringRef BytesStr;
1060 Section.getContents(BytesStr);
1061 const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1062 uint32_t sect_size = BytesStr.size();
1063 uint64_t sect_addr = Section.getAddress();
1064
1065 if (verbose) {
Kevin Enderby95df54c2015-02-04 01:01:38 +00001066 if ((section_flags & MachO::S_ATTR_PURE_INSTRUCTIONS) ||
1067 (section_flags & MachO::S_ATTR_SOME_INSTRUCTIONS)) {
1068 DisassembleMachO(Filename, O, SegName, SectName);
1069 continue;
1070 }
Kevin Enderbycd66be52015-03-11 22:06:32 +00001071 if (SegName == "__TEXT" && SectName == "__info_plist") {
1072 outs() << sect;
1073 continue;
1074 }
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001075 switch (section_type) {
1076 case MachO::S_REGULAR:
1077 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1078 break;
1079 case MachO::S_ZEROFILL:
1080 outs() << "zerofill section and has no contents in the file\n";
1081 break;
Kevin Enderby10ba0412015-02-04 21:38:42 +00001082 case MachO::S_CSTRING_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001083 DumpCstringSection(O, sect, sect_size, sect_addr, !NoLeadingAddr);
Kevin Enderby10ba0412015-02-04 21:38:42 +00001084 break;
Kevin Enderby74b43cb2015-02-06 23:25:38 +00001085 case MachO::S_4BYTE_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001086 DumpLiteral4Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
Kevin Enderby74b43cb2015-02-06 23:25:38 +00001087 break;
1088 case MachO::S_8BYTE_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001089 DumpLiteral8Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
Kevin Enderby74b43cb2015-02-06 23:25:38 +00001090 break;
1091 case MachO::S_16BYTE_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001092 DumpLiteral16Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
1093 break;
Kevin Enderby578fe5a2015-02-17 21:35:48 +00001094 case MachO::S_LITERAL_POINTERS:
1095 DumpLiteralPointerSection(O, Section, sect, sect_size, sect_addr,
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001096 !NoLeadingAddr);
Kevin Enderby578fe5a2015-02-17 21:35:48 +00001097 break;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001098 case MachO::S_MOD_INIT_FUNC_POINTERS:
1099 case MachO::S_MOD_TERM_FUNC_POINTERS:
1100 DumpInitTermPointerSection(O, sect, sect_size, sect_addr, &AddrMap,
1101 verbose);
1102 break;
1103 default:
1104 outs() << "Unknown section type ("
1105 << format("0x%08" PRIx32, section_type) << ")\n";
1106 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1107 break;
1108 }
1109 } else {
1110 if (section_type == MachO::S_ZEROFILL)
1111 outs() << "zerofill section and has no contents in the file\n";
1112 else
1113 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1114 }
1115 }
1116 }
1117 }
1118}
1119
Kevin Enderbycd66be52015-03-11 22:06:32 +00001120static void DumpInfoPlistSectionContents(StringRef Filename,
1121 MachOObjectFile *O) {
1122 for (const SectionRef &Section : O->sections()) {
1123 StringRef SectName;
1124 Section.getName(SectName);
1125 DataRefImpl Ref = Section.getRawDataRefImpl();
1126 StringRef SegName = O->getSectionFinalSegmentName(Ref);
1127 if (SegName == "__TEXT" && SectName == "__info_plist") {
1128 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
1129 StringRef BytesStr;
1130 Section.getContents(BytesStr);
1131 const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1132 outs() << sect;
1133 return;
1134 }
1135 }
1136}
1137
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001138// checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file
1139// and if it is and there is a list of architecture flags is specified then
1140// check to make sure this Mach-O file is one of those architectures or all
1141// architectures were specified. If not then an error is generated and this
1142// routine returns false. Else it returns true.
1143static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) {
1144 if (isa<MachOObjectFile>(O) && !ArchAll && ArchFlags.size() != 0) {
1145 MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(O);
1146 bool ArchFound = false;
1147 MachO::mach_header H;
1148 MachO::mach_header_64 H_64;
1149 Triple T;
1150 if (MachO->is64Bit()) {
1151 H_64 = MachO->MachOObjectFile::getHeader64();
1152 T = MachOObjectFile::getArch(H_64.cputype, H_64.cpusubtype);
1153 } else {
1154 H = MachO->MachOObjectFile::getHeader();
1155 T = MachOObjectFile::getArch(H.cputype, H.cpusubtype);
1156 }
1157 unsigned i;
1158 for (i = 0; i < ArchFlags.size(); ++i) {
1159 if (ArchFlags[i] == T.getArchName())
1160 ArchFound = true;
1161 break;
1162 }
1163 if (!ArchFound) {
1164 errs() << "llvm-objdump: file: " + Filename + " does not contain "
1165 << "architecture: " + ArchFlags[i] + "\n";
1166 return false;
1167 }
1168 }
1169 return true;
1170}
1171
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001172// ProcessMachO() is passed a single opened Mach-O file, which may be an
1173// archive member and or in a slice of a universal file. It prints the
1174// the file name and header info and then processes it according to the
1175// command line options.
1176static void ProcessMachO(StringRef Filename, MachOObjectFile *MachOOF,
1177 StringRef ArchiveMemberName = StringRef(),
1178 StringRef ArchitectureName = StringRef()) {
Kevin Enderby131d1772015-01-09 19:22:37 +00001179 // If we are doing some processing here on the Mach-O file print the header
1180 // info. And don't print it otherwise like in the case of printing the
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +00001181 // UniversalHeaders or ArchiveHeaders.
Kevin Enderby131d1772015-01-09 19:22:37 +00001182 if (Disassemble || PrivateHeaders || ExportsTrie || Rebase || Bind ||
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001183 LazyBind || WeakBind || IndirectSymbols || DataInCode || LinkOptHints ||
Kevin Enderbybc847fa2015-03-16 20:08:09 +00001184 DylibsUsed || DylibId || DumpSections.size() != 0) {
Kevin Enderby131d1772015-01-09 19:22:37 +00001185 outs() << Filename;
1186 if (!ArchiveMemberName.empty())
1187 outs() << '(' << ArchiveMemberName << ')';
1188 if (!ArchitectureName.empty())
1189 outs() << " (architecture " << ArchitectureName << ")";
1190 outs() << ":\n";
1191 }
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001192
1193 if (Disassemble)
Kevin Enderby95df54c2015-02-04 01:01:38 +00001194 DisassembleMachO(Filename, MachOOF, "__TEXT", "__text");
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +00001195 if (IndirectSymbols)
Kevin Enderbyf0640752015-03-13 17:56:32 +00001196 PrintIndirectSymbols(MachOOF, !NonVerbose);
Kevin Enderby69fe98d2015-01-23 18:52:17 +00001197 if (DataInCode)
Kevin Enderbyf0640752015-03-13 17:56:32 +00001198 PrintDataInCodeTable(MachOOF, !NonVerbose);
Kevin Enderby9a509442015-01-27 21:28:24 +00001199 if (LinkOptHints)
1200 PrintLinkOptHints(MachOOF);
Kevin Enderby98da6132015-01-20 21:47:46 +00001201 if (Relocations)
1202 PrintRelocations(MachOOF);
1203 if (SectionHeaders)
1204 PrintSectionHeaders(MachOOF);
1205 if (SectionContents)
1206 PrintSectionContents(MachOOF);
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001207 if (DumpSections.size() != 0)
Kevin Enderbyf0640752015-03-13 17:56:32 +00001208 DumpSectionContents(Filename, MachOOF, !NonVerbose);
Kevin Enderbycd66be52015-03-11 22:06:32 +00001209 if (InfoPlist)
1210 DumpInfoPlistSectionContents(Filename, MachOOF);
Kevin Enderbybc847fa2015-03-16 20:08:09 +00001211 if (DylibsUsed)
1212 PrintDylibs(MachOOF, false);
1213 if (DylibId)
1214 PrintDylibs(MachOOF, true);
Kevin Enderby98da6132015-01-20 21:47:46 +00001215 if (SymbolTable)
1216 PrintSymbolTable(MachOOF);
1217 if (UnwindInfo)
1218 printMachOUnwindInfo(MachOOF);
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001219 if (PrivateHeaders)
1220 printMachOFileHeader(MachOOF);
1221 if (ExportsTrie)
1222 printExportsTrie(MachOOF);
1223 if (Rebase)
1224 printRebaseTable(MachOOF);
1225 if (Bind)
1226 printBindTable(MachOOF);
1227 if (LazyBind)
1228 printLazyBindTable(MachOOF);
1229 if (WeakBind)
1230 printWeakBindTable(MachOOF);
1231}
1232
Kevin Enderby131d1772015-01-09 19:22:37 +00001233// printUnknownCPUType() helps print_fat_headers for unknown CPU's.
1234static void printUnknownCPUType(uint32_t cputype, uint32_t cpusubtype) {
1235 outs() << " cputype (" << cputype << ")\n";
1236 outs() << " cpusubtype (" << cpusubtype << ")\n";
1237}
1238
1239// printCPUType() helps print_fat_headers by printing the cputype and
1240// pusubtype (symbolically for the one's it knows about).
1241static void printCPUType(uint32_t cputype, uint32_t cpusubtype) {
1242 switch (cputype) {
1243 case MachO::CPU_TYPE_I386:
1244 switch (cpusubtype) {
1245 case MachO::CPU_SUBTYPE_I386_ALL:
1246 outs() << " cputype CPU_TYPE_I386\n";
1247 outs() << " cpusubtype CPU_SUBTYPE_I386_ALL\n";
1248 break;
1249 default:
1250 printUnknownCPUType(cputype, cpusubtype);
1251 break;
1252 }
1253 break;
1254 case MachO::CPU_TYPE_X86_64:
1255 switch (cpusubtype) {
1256 case MachO::CPU_SUBTYPE_X86_64_ALL:
1257 outs() << " cputype CPU_TYPE_X86_64\n";
1258 outs() << " cpusubtype CPU_SUBTYPE_X86_64_ALL\n";
1259 break;
1260 case MachO::CPU_SUBTYPE_X86_64_H:
1261 outs() << " cputype CPU_TYPE_X86_64\n";
1262 outs() << " cpusubtype CPU_SUBTYPE_X86_64_H\n";
1263 break;
1264 default:
1265 printUnknownCPUType(cputype, cpusubtype);
1266 break;
1267 }
1268 break;
1269 case MachO::CPU_TYPE_ARM:
1270 switch (cpusubtype) {
1271 case MachO::CPU_SUBTYPE_ARM_ALL:
1272 outs() << " cputype CPU_TYPE_ARM\n";
1273 outs() << " cpusubtype CPU_SUBTYPE_ARM_ALL\n";
1274 break;
1275 case MachO::CPU_SUBTYPE_ARM_V4T:
1276 outs() << " cputype CPU_TYPE_ARM\n";
1277 outs() << " cpusubtype CPU_SUBTYPE_ARM_V4T\n";
1278 break;
1279 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
1280 outs() << " cputype CPU_TYPE_ARM\n";
1281 outs() << " cpusubtype CPU_SUBTYPE_ARM_V5TEJ\n";
1282 break;
1283 case MachO::CPU_SUBTYPE_ARM_XSCALE:
1284 outs() << " cputype CPU_TYPE_ARM\n";
1285 outs() << " cpusubtype CPU_SUBTYPE_ARM_XSCALE\n";
1286 break;
1287 case MachO::CPU_SUBTYPE_ARM_V6:
1288 outs() << " cputype CPU_TYPE_ARM\n";
1289 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6\n";
1290 break;
1291 case MachO::CPU_SUBTYPE_ARM_V6M:
1292 outs() << " cputype CPU_TYPE_ARM\n";
1293 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6M\n";
1294 break;
1295 case MachO::CPU_SUBTYPE_ARM_V7:
1296 outs() << " cputype CPU_TYPE_ARM\n";
1297 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7\n";
1298 break;
1299 case MachO::CPU_SUBTYPE_ARM_V7EM:
1300 outs() << " cputype CPU_TYPE_ARM\n";
1301 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7EM\n";
1302 break;
1303 case MachO::CPU_SUBTYPE_ARM_V7K:
1304 outs() << " cputype CPU_TYPE_ARM\n";
1305 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7K\n";
1306 break;
1307 case MachO::CPU_SUBTYPE_ARM_V7M:
1308 outs() << " cputype CPU_TYPE_ARM\n";
1309 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7M\n";
1310 break;
1311 case MachO::CPU_SUBTYPE_ARM_V7S:
1312 outs() << " cputype CPU_TYPE_ARM\n";
1313 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7S\n";
1314 break;
1315 default:
1316 printUnknownCPUType(cputype, cpusubtype);
1317 break;
1318 }
1319 break;
1320 case MachO::CPU_TYPE_ARM64:
1321 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
1322 case MachO::CPU_SUBTYPE_ARM64_ALL:
1323 outs() << " cputype CPU_TYPE_ARM64\n";
1324 outs() << " cpusubtype CPU_SUBTYPE_ARM64_ALL\n";
1325 break;
1326 default:
1327 printUnknownCPUType(cputype, cpusubtype);
1328 break;
1329 }
1330 break;
1331 default:
1332 printUnknownCPUType(cputype, cpusubtype);
1333 break;
1334 }
1335}
1336
1337static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB,
1338 bool verbose) {
1339 outs() << "Fat headers\n";
1340 if (verbose)
1341 outs() << "fat_magic FAT_MAGIC\n";
1342 else
1343 outs() << "fat_magic " << format("0x%" PRIx32, MachO::FAT_MAGIC) << "\n";
1344
1345 uint32_t nfat_arch = UB->getNumberOfObjects();
1346 StringRef Buf = UB->getData();
1347 uint64_t size = Buf.size();
1348 uint64_t big_size = sizeof(struct MachO::fat_header) +
1349 nfat_arch * sizeof(struct MachO::fat_arch);
1350 outs() << "nfat_arch " << UB->getNumberOfObjects();
1351 if (nfat_arch == 0)
1352 outs() << " (malformed, contains zero architecture types)\n";
1353 else if (big_size > size)
1354 outs() << " (malformed, architectures past end of file)\n";
1355 else
1356 outs() << "\n";
1357
1358 for (uint32_t i = 0; i < nfat_arch; ++i) {
1359 MachOUniversalBinary::ObjectForArch OFA(UB, i);
1360 uint32_t cputype = OFA.getCPUType();
1361 uint32_t cpusubtype = OFA.getCPUSubType();
1362 outs() << "architecture ";
1363 for (uint32_t j = 0; i != 0 && j <= i - 1; j++) {
1364 MachOUniversalBinary::ObjectForArch other_OFA(UB, j);
1365 uint32_t other_cputype = other_OFA.getCPUType();
1366 uint32_t other_cpusubtype = other_OFA.getCPUSubType();
Kevin Enderby0512bd72015-01-09 21:55:03 +00001367 if (cputype != 0 && cpusubtype != 0 && cputype == other_cputype &&
Kevin Enderby131d1772015-01-09 19:22:37 +00001368 (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) ==
Kevin Enderby0512bd72015-01-09 21:55:03 +00001369 (other_cpusubtype & ~MachO::CPU_SUBTYPE_MASK)) {
Kevin Enderby131d1772015-01-09 19:22:37 +00001370 outs() << "(illegal duplicate architecture) ";
1371 break;
Kevin Enderby0512bd72015-01-09 21:55:03 +00001372 }
Kevin Enderby131d1772015-01-09 19:22:37 +00001373 }
1374 if (verbose) {
1375 outs() << OFA.getArchTypeName() << "\n";
1376 printCPUType(cputype, cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
1377 } else {
1378 outs() << i << "\n";
1379 outs() << " cputype " << cputype << "\n";
1380 outs() << " cpusubtype " << (cpusubtype & ~MachO::CPU_SUBTYPE_MASK)
1381 << "\n";
1382 }
1383 if (verbose &&
1384 (cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64)
1385 outs() << " capabilities CPU_SUBTYPE_LIB64\n";
1386 else
1387 outs() << " capabilities "
1388 << format("0x%" PRIx32,
1389 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24) << "\n";
1390 outs() << " offset " << OFA.getOffset();
1391 if (OFA.getOffset() > size)
1392 outs() << " (past end of file)";
1393 if (OFA.getOffset() % (1 << OFA.getAlign()) != 0)
1394 outs() << " (not aligned on it's alignment (2^" << OFA.getAlign() << ")";
1395 outs() << "\n";
1396 outs() << " size " << OFA.getSize();
1397 big_size = OFA.getOffset() + OFA.getSize();
1398 if (big_size > size)
1399 outs() << " (past end of file)";
1400 outs() << "\n";
1401 outs() << " align 2^" << OFA.getAlign() << " (" << (1 << OFA.getAlign())
1402 << ")\n";
1403 }
1404}
1405
Kevin Enderby13023a12015-01-15 23:19:11 +00001406static void printArchiveChild(Archive::Child &C, bool verbose,
1407 bool print_offset) {
1408 if (print_offset)
1409 outs() << C.getChildOffset() << "\t";
1410 sys::fs::perms Mode = C.getAccessMode();
1411 if (verbose) {
1412 // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG.
1413 // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG.
1414 outs() << "-";
1415 if (Mode & sys::fs::owner_read)
1416 outs() << "r";
1417 else
1418 outs() << "-";
1419 if (Mode & sys::fs::owner_write)
1420 outs() << "w";
1421 else
1422 outs() << "-";
1423 if (Mode & sys::fs::owner_exe)
1424 outs() << "x";
1425 else
1426 outs() << "-";
1427 if (Mode & sys::fs::group_read)
1428 outs() << "r";
1429 else
1430 outs() << "-";
1431 if (Mode & sys::fs::group_write)
1432 outs() << "w";
1433 else
1434 outs() << "-";
1435 if (Mode & sys::fs::group_exe)
1436 outs() << "x";
1437 else
1438 outs() << "-";
1439 if (Mode & sys::fs::others_read)
1440 outs() << "r";
1441 else
1442 outs() << "-";
1443 if (Mode & sys::fs::others_write)
1444 outs() << "w";
1445 else
1446 outs() << "-";
1447 if (Mode & sys::fs::others_exe)
1448 outs() << "x";
1449 else
1450 outs() << "-";
1451 } else {
1452 outs() << format("0%o ", Mode);
1453 }
1454
1455 unsigned UID = C.getUID();
1456 outs() << format("%3d/", UID);
1457 unsigned GID = C.getGID();
1458 outs() << format("%-3d ", GID);
Kevin Enderbyc12718932015-01-16 22:10:36 +00001459 uint64_t Size = C.getRawSize();
Kevin Enderby479ee612015-01-23 21:02:44 +00001460 outs() << format("%5" PRId64, Size) << " ";
Kevin Enderby13023a12015-01-15 23:19:11 +00001461
1462 StringRef RawLastModified = C.getRawLastModified();
1463 if (verbose) {
1464 unsigned Seconds;
1465 if (RawLastModified.getAsInteger(10, Seconds))
1466 outs() << "(date: \"%s\" contains non-decimal chars) " << RawLastModified;
1467 else {
1468 // Since cime(3) returns a 26 character string of the form:
1469 // "Sun Sep 16 01:03:52 1973\n\0"
1470 // just print 24 characters.
1471 time_t t = Seconds;
1472 outs() << format("%.24s ", ctime(&t));
1473 }
1474 } else {
1475 outs() << RawLastModified << " ";
1476 }
1477
1478 if (verbose) {
1479 ErrorOr<StringRef> NameOrErr = C.getName();
1480 if (NameOrErr.getError()) {
1481 StringRef RawName = C.getRawName();
1482 outs() << RawName << "\n";
1483 } else {
1484 StringRef Name = NameOrErr.get();
1485 outs() << Name << "\n";
1486 }
1487 } else {
1488 StringRef RawName = C.getRawName();
1489 outs() << RawName << "\n";
1490 }
1491}
1492
1493static void printArchiveHeaders(Archive *A, bool verbose, bool print_offset) {
1494 if (A->hasSymbolTable()) {
1495 Archive::child_iterator S = A->getSymbolTableChild();
1496 Archive::Child C = *S;
1497 printArchiveChild(C, verbose, print_offset);
1498 }
1499 for (Archive::child_iterator I = A->child_begin(), E = A->child_end(); I != E;
1500 ++I) {
1501 Archive::Child C = *I;
1502 printArchiveChild(C, verbose, print_offset);
1503 }
1504}
1505
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001506// ParseInputMachO() parses the named Mach-O file in Filename and handles the
1507// -arch flags selecting just those slices as specified by them and also parses
1508// archive files. Then for each individual Mach-O file ProcessMachO() is
1509// called to process the file based on the command line options.
1510void llvm::ParseInputMachO(StringRef Filename) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001511 // Check for -arch all and verifiy the -arch flags are valid.
1512 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
1513 if (ArchFlags[i] == "all") {
1514 ArchAll = true;
1515 } else {
1516 if (!MachOObjectFile::isValidArch(ArchFlags[i])) {
1517 errs() << "llvm-objdump: Unknown architecture named '" + ArchFlags[i] +
1518 "'for the -arch option\n";
1519 return;
1520 }
1521 }
1522 }
1523
1524 // Attempt to open the binary.
1525 ErrorOr<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename);
1526 if (std::error_code EC = BinaryOrErr.getError()) {
1527 errs() << "llvm-objdump: '" << Filename << "': " << EC.message() << ".\n";
Rafael Espindolade882cd2014-12-03 23:29:34 +00001528 return;
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001529 }
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001530 Binary &Bin = *BinaryOrErr.get().getBinary();
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001531
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001532 if (Archive *A = dyn_cast<Archive>(&Bin)) {
1533 outs() << "Archive : " << Filename << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001534 if (ArchiveHeaders)
1535 printArchiveHeaders(A, true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001536 for (Archive::child_iterator I = A->child_begin(), E = A->child_end();
1537 I != E; ++I) {
1538 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = I->getAsBinary();
1539 if (ChildOrErr.getError())
1540 continue;
1541 if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
1542 if (!checkMachOAndArchFlags(O, Filename))
1543 return;
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001544 ProcessMachO(Filename, O, O->getFileName());
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001545 }
1546 }
1547 return;
1548 }
Kevin Enderby131d1772015-01-09 19:22:37 +00001549 if (UniversalHeaders) {
1550 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin))
Kevin Enderbyf0640752015-03-13 17:56:32 +00001551 printMachOUniversalHeaders(UB, !NonVerbose);
Kevin Enderby131d1772015-01-09 19:22:37 +00001552 }
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001553 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) {
1554 // If we have a list of architecture flags specified dump only those.
1555 if (!ArchAll && ArchFlags.size() != 0) {
1556 // Look for a slice in the universal binary that matches each ArchFlag.
1557 bool ArchFound;
1558 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
1559 ArchFound = false;
1560 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
1561 E = UB->end_objects();
1562 I != E; ++I) {
1563 if (ArchFlags[i] == I->getArchTypeName()) {
1564 ArchFound = true;
1565 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr =
1566 I->getAsObjectFile();
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001567 std::string ArchitectureName = "";
1568 if (ArchFlags.size() > 1)
1569 ArchitectureName = I->getArchTypeName();
1570 if (ObjOrErr) {
1571 ObjectFile &O = *ObjOrErr.get();
1572 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001573 ProcessMachO(Filename, MachOOF, "", ArchitectureName);
Rafael Espindola0bfe8282014-12-09 21:05:36 +00001574 } else if (ErrorOr<std::unique_ptr<Archive>> AOrErr =
1575 I->getAsArchive()) {
1576 std::unique_ptr<Archive> &A = *AOrErr;
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001577 outs() << "Archive : " << Filename;
1578 if (!ArchitectureName.empty())
1579 outs() << " (architecture " << ArchitectureName << ")";
1580 outs() << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001581 if (ArchiveHeaders)
1582 printArchiveHeaders(A.get(), true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001583 for (Archive::child_iterator AI = A->child_begin(),
1584 AE = A->child_end();
1585 AI != AE; ++AI) {
1586 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
1587 if (ChildOrErr.getError())
1588 continue;
1589 if (MachOObjectFile *O =
1590 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001591 ProcessMachO(Filename, O, O->getFileName(), ArchitectureName);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001592 }
1593 }
1594 }
1595 }
1596 if (!ArchFound) {
1597 errs() << "llvm-objdump: file: " + Filename + " does not contain "
1598 << "architecture: " + ArchFlags[i] + "\n";
1599 return;
1600 }
1601 }
1602 return;
1603 }
1604 // No architecture flags were specified so if this contains a slice that
1605 // matches the host architecture dump only that.
1606 if (!ArchAll) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001607 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
1608 E = UB->end_objects();
1609 I != E; ++I) {
Kevin Enderby0512bd72015-01-09 21:55:03 +00001610 if (MachOObjectFile::getHostArch().getArchName() ==
1611 I->getArchTypeName()) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001612 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001613 std::string ArchiveName;
1614 ArchiveName.clear();
1615 if (ObjOrErr) {
1616 ObjectFile &O = *ObjOrErr.get();
1617 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001618 ProcessMachO(Filename, MachOOF);
Rafael Espindola0bfe8282014-12-09 21:05:36 +00001619 } else if (ErrorOr<std::unique_ptr<Archive>> AOrErr =
1620 I->getAsArchive()) {
1621 std::unique_ptr<Archive> &A = *AOrErr;
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001622 outs() << "Archive : " << Filename << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001623 if (ArchiveHeaders)
1624 printArchiveHeaders(A.get(), true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001625 for (Archive::child_iterator AI = A->child_begin(),
1626 AE = A->child_end();
1627 AI != AE; ++AI) {
1628 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
1629 if (ChildOrErr.getError())
1630 continue;
1631 if (MachOObjectFile *O =
1632 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001633 ProcessMachO(Filename, O, O->getFileName());
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001634 }
1635 }
1636 return;
1637 }
1638 }
1639 }
1640 // Either all architectures have been specified or none have been specified
1641 // and this does not contain the host architecture so dump all the slices.
1642 bool moreThanOneArch = UB->getNumberOfObjects() > 1;
1643 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
1644 E = UB->end_objects();
1645 I != E; ++I) {
1646 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001647 std::string ArchitectureName = "";
1648 if (moreThanOneArch)
1649 ArchitectureName = I->getArchTypeName();
1650 if (ObjOrErr) {
1651 ObjectFile &Obj = *ObjOrErr.get();
1652 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001653 ProcessMachO(Filename, MachOOF, "", ArchitectureName);
Rafael Espindola0bfe8282014-12-09 21:05:36 +00001654 } else if (ErrorOr<std::unique_ptr<Archive>> AOrErr = I->getAsArchive()) {
1655 std::unique_ptr<Archive> &A = *AOrErr;
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001656 outs() << "Archive : " << Filename;
1657 if (!ArchitectureName.empty())
1658 outs() << " (architecture " << ArchitectureName << ")";
1659 outs() << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001660 if (ArchiveHeaders)
1661 printArchiveHeaders(A.get(), true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001662 for (Archive::child_iterator AI = A->child_begin(), AE = A->child_end();
1663 AI != AE; ++AI) {
1664 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
1665 if (ChildOrErr.getError())
1666 continue;
1667 if (MachOObjectFile *O =
1668 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
1669 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001670 ProcessMachO(Filename, MachOOF, MachOOF->getFileName(),
1671 ArchitectureName);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001672 }
1673 }
1674 }
1675 }
1676 return;
1677 }
1678 if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) {
1679 if (!checkMachOAndArchFlags(O, Filename))
1680 return;
1681 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O)) {
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001682 ProcessMachO(Filename, MachOOF);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001683 } else
1684 errs() << "llvm-objdump: '" << Filename << "': "
1685 << "Object is not a Mach-O file type.\n";
1686 } else
1687 errs() << "llvm-objdump: '" << Filename << "': "
1688 << "Unrecognized file type.\n";
Rafael Espindola9b709252013-04-13 01:45:40 +00001689}
1690
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001691typedef std::pair<uint64_t, const char *> BindInfoEntry;
1692typedef std::vector<BindInfoEntry> BindTable;
1693typedef BindTable::iterator bind_table_iterator;
Kevin Enderbybf246f52014-09-24 23:08:22 +00001694
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001695// The block of info used by the Symbolizer call backs.
1696struct DisassembleInfo {
1697 bool verbose;
1698 MachOObjectFile *O;
1699 SectionRef S;
Kevin Enderbybf246f52014-09-24 23:08:22 +00001700 SymbolAddressMap *AddrMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001701 std::vector<SectionRef> *Sections;
1702 const char *class_name;
1703 const char *selector_name;
1704 char *method;
Kevin Enderby04bf6932014-10-28 23:39:46 +00001705 char *demangled_name;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001706 uint64_t adrp_addr;
1707 uint32_t adrp_inst;
Kevin Enderby078be602014-10-23 19:53:12 +00001708 BindTable *bindtable;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001709};
1710
1711// SymbolizerGetOpInfo() is the operand information call back function.
1712// This is called to get the symbolic information for operand(s) of an
1713// instruction when it is being done. This routine does this from
1714// the relocation information, symbol table, etc. That block of information
1715// is a pointer to the struct DisassembleInfo that was passed when the
1716// disassembler context was created and passed to back to here when
1717// called back by the disassembler for instruction operands that could have
1718// relocation information. The address of the instruction containing operand is
1719// at the Pc parameter. The immediate value the operand has is passed in
1720// op_info->Value and is at Offset past the start of the instruction and has a
1721// byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the
1722// LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol
1723// names and addends of the symbolic expression to add for the operand. The
1724// value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic
1725// information is returned then this function returns 1 else it returns 0.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00001726static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset,
1727 uint64_t Size, int TagType, void *TagBuf) {
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001728 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
1729 struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001730 uint64_t value = op_info->Value;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001731
1732 // Make sure all fields returned are zero if we don't set them.
1733 memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1));
1734 op_info->Value = value;
1735
1736 // If the TagType is not the value 1 which it code knows about or if no
1737 // verbose symbolic information is wanted then just return 0, indicating no
1738 // information is being returned.
1739 if (TagType != 1 || info->verbose == false)
1740 return 0;
1741
1742 unsigned int Arch = info->O->getArch();
1743 if (Arch == Triple::x86) {
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001744 if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
1745 return 0;
1746 // First search the section's relocation entries (if any) for an entry
1747 // for this section offset.
1748 uint32_t sect_addr = info->S.getAddress();
1749 uint32_t sect_offset = (Pc + Offset) - sect_addr;
1750 bool reloc_found = false;
1751 DataRefImpl Rel;
1752 MachO::any_relocation_info RE;
1753 bool isExtern = false;
1754 SymbolRef Symbol;
1755 bool r_scattered = false;
1756 uint32_t r_value, pair_r_value, r_type;
1757 for (const RelocationRef &Reloc : info->S.relocations()) {
1758 uint64_t RelocOffset;
1759 Reloc.getOffset(RelocOffset);
1760 if (RelocOffset == sect_offset) {
1761 Rel = Reloc.getRawDataRefImpl();
1762 RE = info->O->getRelocation(Rel);
Kevin Enderby3eb73e12014-11-11 19:16:45 +00001763 r_type = info->O->getAnyRelocationType(RE);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001764 r_scattered = info->O->isRelocationScattered(RE);
1765 if (r_scattered) {
1766 r_value = info->O->getScatteredRelocationValue(RE);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001767 if (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
1768 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) {
1769 DataRefImpl RelNext = Rel;
1770 info->O->moveRelocationNext(RelNext);
1771 MachO::any_relocation_info RENext;
1772 RENext = info->O->getRelocation(RelNext);
1773 if (info->O->isRelocationScattered(RENext))
Kevin Enderby930fdc72014-11-06 19:00:13 +00001774 pair_r_value = info->O->getScatteredRelocationValue(RENext);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001775 else
1776 return 0;
1777 }
1778 } else {
1779 isExtern = info->O->getPlainRelocationExternal(RE);
1780 if (isExtern) {
1781 symbol_iterator RelocSym = Reloc.getSymbol();
1782 Symbol = *RelocSym;
1783 }
1784 }
1785 reloc_found = true;
1786 break;
1787 }
1788 }
1789 if (reloc_found && isExtern) {
1790 StringRef SymName;
1791 Symbol.getName(SymName);
1792 const char *name = SymName.data();
1793 op_info->AddSymbol.Present = 1;
1794 op_info->AddSymbol.Name = name;
1795 // For i386 extern relocation entries the value in the instruction is
1796 // the offset from the symbol, and value is already set in op_info->Value.
1797 return 1;
1798 }
1799 if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
1800 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) {
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001801 const char *add = GuessSymbolName(r_value, info->AddrMap);
1802 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001803 uint32_t offset = value - (r_value - pair_r_value);
1804 op_info->AddSymbol.Present = 1;
1805 if (add != nullptr)
1806 op_info->AddSymbol.Name = add;
1807 else
1808 op_info->AddSymbol.Value = r_value;
1809 op_info->SubtractSymbol.Present = 1;
1810 if (sub != nullptr)
1811 op_info->SubtractSymbol.Name = sub;
1812 else
1813 op_info->SubtractSymbol.Value = pair_r_value;
1814 op_info->Value = offset;
1815 return 1;
1816 }
1817 // TODO:
1818 // Second search the external relocation entries of a fully linked image
1819 // (if any) for an entry that matches this segment offset.
1820 // uint32_t seg_offset = (Pc + Offset);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001821 return 0;
1822 } else if (Arch == Triple::x86_64) {
1823 if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
1824 return 0;
1825 // First search the section's relocation entries (if any) for an entry
1826 // for this section offset.
Rafael Espindola80291272014-10-08 15:28:58 +00001827 uint64_t sect_addr = info->S.getAddress();
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001828 uint64_t sect_offset = (Pc + Offset) - sect_addr;
1829 bool reloc_found = false;
1830 DataRefImpl Rel;
1831 MachO::any_relocation_info RE;
1832 bool isExtern = false;
1833 SymbolRef Symbol;
1834 for (const RelocationRef &Reloc : info->S.relocations()) {
1835 uint64_t RelocOffset;
1836 Reloc.getOffset(RelocOffset);
1837 if (RelocOffset == sect_offset) {
1838 Rel = Reloc.getRawDataRefImpl();
1839 RE = info->O->getRelocation(Rel);
1840 // NOTE: Scattered relocations don't exist on x86_64.
1841 isExtern = info->O->getPlainRelocationExternal(RE);
1842 if (isExtern) {
1843 symbol_iterator RelocSym = Reloc.getSymbol();
1844 Symbol = *RelocSym;
1845 }
1846 reloc_found = true;
1847 break;
1848 }
1849 }
1850 if (reloc_found && isExtern) {
1851 // The Value passed in will be adjusted by the Pc if the instruction
1852 // adds the Pc. But for x86_64 external relocation entries the Value
1853 // is the offset from the external symbol.
1854 if (info->O->getAnyRelocationPCRel(RE))
1855 op_info->Value -= Pc + Offset + Size;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001856 StringRef SymName;
1857 Symbol.getName(SymName);
1858 const char *name = SymName.data();
1859 unsigned Type = info->O->getAnyRelocationType(RE);
1860 if (Type == MachO::X86_64_RELOC_SUBTRACTOR) {
1861 DataRefImpl RelNext = Rel;
1862 info->O->moveRelocationNext(RelNext);
1863 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
1864 unsigned TypeNext = info->O->getAnyRelocationType(RENext);
1865 bool isExternNext = info->O->getPlainRelocationExternal(RENext);
1866 unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext);
1867 if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) {
1868 op_info->SubtractSymbol.Present = 1;
1869 op_info->SubtractSymbol.Name = name;
1870 symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum);
1871 Symbol = *RelocSymNext;
1872 StringRef SymNameNext;
1873 Symbol.getName(SymNameNext);
1874 name = SymNameNext.data();
1875 }
1876 }
1877 // TODO: add the VariantKinds to op_info->VariantKind for relocation types
1878 // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT.
1879 op_info->AddSymbol.Present = 1;
1880 op_info->AddSymbol.Name = name;
1881 return 1;
1882 }
1883 // TODO:
1884 // Second search the external relocation entries of a fully linked image
1885 // (if any) for an entry that matches this segment offset.
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001886 // uint64_t seg_offset = (Pc + Offset);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001887 return 0;
1888 } else if (Arch == Triple::arm) {
Kevin Enderby930fdc72014-11-06 19:00:13 +00001889 if (Offset != 0 || (Size != 4 && Size != 2))
1890 return 0;
1891 // First search the section's relocation entries (if any) for an entry
1892 // for this section offset.
1893 uint32_t sect_addr = info->S.getAddress();
1894 uint32_t sect_offset = (Pc + Offset) - sect_addr;
1895 bool reloc_found = false;
1896 DataRefImpl Rel;
1897 MachO::any_relocation_info RE;
1898 bool isExtern = false;
1899 SymbolRef Symbol;
1900 bool r_scattered = false;
1901 uint32_t r_value, pair_r_value, r_type, r_length, other_half;
1902 for (const RelocationRef &Reloc : info->S.relocations()) {
1903 uint64_t RelocOffset;
1904 Reloc.getOffset(RelocOffset);
1905 if (RelocOffset == sect_offset) {
1906 Rel = Reloc.getRawDataRefImpl();
1907 RE = info->O->getRelocation(Rel);
1908 r_length = info->O->getAnyRelocationLength(RE);
1909 r_scattered = info->O->isRelocationScattered(RE);
1910 if (r_scattered) {
1911 r_value = info->O->getScatteredRelocationValue(RE);
1912 r_type = info->O->getScatteredRelocationType(RE);
1913 } else {
1914 r_type = info->O->getAnyRelocationType(RE);
1915 isExtern = info->O->getPlainRelocationExternal(RE);
1916 if (isExtern) {
1917 symbol_iterator RelocSym = Reloc.getSymbol();
1918 Symbol = *RelocSym;
1919 }
1920 }
1921 if (r_type == MachO::ARM_RELOC_HALF ||
1922 r_type == MachO::ARM_RELOC_SECTDIFF ||
1923 r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
1924 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1925 DataRefImpl RelNext = Rel;
1926 info->O->moveRelocationNext(RelNext);
1927 MachO::any_relocation_info RENext;
1928 RENext = info->O->getRelocation(RelNext);
1929 other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff;
1930 if (info->O->isRelocationScattered(RENext))
1931 pair_r_value = info->O->getScatteredRelocationValue(RENext);
1932 }
1933 reloc_found = true;
1934 break;
1935 }
1936 }
1937 if (reloc_found && isExtern) {
1938 StringRef SymName;
1939 Symbol.getName(SymName);
1940 const char *name = SymName.data();
1941 op_info->AddSymbol.Present = 1;
1942 op_info->AddSymbol.Name = name;
Sylvestre Ledru648cced2015-02-05 17:00:23 +00001943 switch (r_type) {
1944 case MachO::ARM_RELOC_HALF:
1945 if ((r_length & 0x1) == 1) {
1946 op_info->Value = value << 16 | other_half;
1947 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
1948 } else {
1949 op_info->Value = other_half << 16 | value;
1950 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
Sylvestre Ledrufe0c7ad2015-02-05 16:35:44 +00001951 }
Sylvestre Ledru648cced2015-02-05 17:00:23 +00001952 break;
1953 default:
1954 break;
Kevin Enderby930fdc72014-11-06 19:00:13 +00001955 }
1956 return 1;
1957 }
1958 // If we have a branch that is not an external relocation entry then
1959 // return 0 so the code in tryAddingSymbolicOperand() can use the
1960 // SymbolLookUp call back with the branch target address to look up the
1961 // symbol and possiblity add an annotation for a symbol stub.
1962 if (reloc_found && isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 ||
1963 r_type == MachO::ARM_THUMB_RELOC_BR22))
1964 return 0;
1965
1966 uint32_t offset = 0;
1967 if (reloc_found) {
1968 if (r_type == MachO::ARM_RELOC_HALF ||
1969 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1970 if ((r_length & 0x1) == 1)
1971 value = value << 16 | other_half;
1972 else
1973 value = other_half << 16 | value;
1974 }
1975 if (r_scattered && (r_type != MachO::ARM_RELOC_HALF &&
1976 r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) {
1977 offset = value - r_value;
1978 value = r_value;
1979 }
1980 }
1981
1982 if (reloc_found && r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1983 if ((r_length & 0x1) == 1)
1984 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
1985 else
1986 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001987 const char *add = GuessSymbolName(r_value, info->AddrMap);
1988 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
Kevin Enderby930fdc72014-11-06 19:00:13 +00001989 int32_t offset = value - (r_value - pair_r_value);
1990 op_info->AddSymbol.Present = 1;
1991 if (add != nullptr)
1992 op_info->AddSymbol.Name = add;
1993 else
1994 op_info->AddSymbol.Value = r_value;
1995 op_info->SubtractSymbol.Present = 1;
1996 if (sub != nullptr)
1997 op_info->SubtractSymbol.Name = sub;
1998 else
1999 op_info->SubtractSymbol.Value = pair_r_value;
2000 op_info->Value = offset;
2001 return 1;
2002 }
2003
2004 if (reloc_found == false)
2005 return 0;
2006
2007 op_info->AddSymbol.Present = 1;
2008 op_info->Value = offset;
2009 if (reloc_found) {
2010 if (r_type == MachO::ARM_RELOC_HALF) {
2011 if ((r_length & 0x1) == 1)
2012 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2013 else
2014 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2015 }
2016 }
Kevin Enderbyf6d25852015-01-31 00:37:11 +00002017 const char *add = GuessSymbolName(value, info->AddrMap);
Kevin Enderby930fdc72014-11-06 19:00:13 +00002018 if (add != nullptr) {
2019 op_info->AddSymbol.Name = add;
2020 return 1;
2021 }
2022 op_info->AddSymbol.Value = value;
2023 return 1;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002024 } else if (Arch == Triple::aarch64) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002025 if (Offset != 0 || Size != 4)
2026 return 0;
2027 // First search the section's relocation entries (if any) for an entry
2028 // for this section offset.
2029 uint64_t sect_addr = info->S.getAddress();
2030 uint64_t sect_offset = (Pc + Offset) - sect_addr;
2031 bool reloc_found = false;
2032 DataRefImpl Rel;
2033 MachO::any_relocation_info RE;
2034 bool isExtern = false;
2035 SymbolRef Symbol;
2036 uint32_t r_type = 0;
2037 for (const RelocationRef &Reloc : info->S.relocations()) {
2038 uint64_t RelocOffset;
2039 Reloc.getOffset(RelocOffset);
2040 if (RelocOffset == sect_offset) {
2041 Rel = Reloc.getRawDataRefImpl();
2042 RE = info->O->getRelocation(Rel);
2043 r_type = info->O->getAnyRelocationType(RE);
2044 if (r_type == MachO::ARM64_RELOC_ADDEND) {
2045 DataRefImpl RelNext = Rel;
2046 info->O->moveRelocationNext(RelNext);
2047 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2048 if (value == 0) {
2049 value = info->O->getPlainRelocationSymbolNum(RENext);
2050 op_info->Value = value;
2051 }
2052 }
2053 // NOTE: Scattered relocations don't exist on arm64.
2054 isExtern = info->O->getPlainRelocationExternal(RE);
2055 if (isExtern) {
2056 symbol_iterator RelocSym = Reloc.getSymbol();
2057 Symbol = *RelocSym;
2058 }
2059 reloc_found = true;
2060 break;
2061 }
2062 }
2063 if (reloc_found && isExtern) {
2064 StringRef SymName;
2065 Symbol.getName(SymName);
2066 const char *name = SymName.data();
2067 op_info->AddSymbol.Present = 1;
2068 op_info->AddSymbol.Name = name;
2069
2070 switch (r_type) {
2071 case MachO::ARM64_RELOC_PAGE21:
2072 /* @page */
2073 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE;
2074 break;
2075 case MachO::ARM64_RELOC_PAGEOFF12:
2076 /* @pageoff */
2077 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF;
2078 break;
2079 case MachO::ARM64_RELOC_GOT_LOAD_PAGE21:
2080 /* @gotpage */
2081 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE;
2082 break;
2083 case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12:
2084 /* @gotpageoff */
2085 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF;
2086 break;
2087 case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21:
2088 /* @tvlppage is not implemented in llvm-mc */
2089 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP;
2090 break;
2091 case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12:
2092 /* @tvlppageoff is not implemented in llvm-mc */
2093 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF;
2094 break;
2095 default:
2096 case MachO::ARM64_RELOC_BRANCH26:
2097 op_info->VariantKind = LLVMDisassembler_VariantKind_None;
2098 break;
2099 }
2100 return 1;
2101 }
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002102 return 0;
2103 } else {
2104 return 0;
2105 }
2106}
2107
Kevin Enderbybf246f52014-09-24 23:08:22 +00002108// GuessCstringPointer is passed the address of what might be a pointer to a
2109// literal string in a cstring section. If that address is in a cstring section
2110// it returns a pointer to that string. Else it returns nullptr.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002111static const char *GuessCstringPointer(uint64_t ReferenceValue,
2112 struct DisassembleInfo *info) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002113 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
2114 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
2115 for (unsigned I = 0;; ++I) {
2116 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
2117 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
2118 for (unsigned J = 0; J < Seg.nsects; ++J) {
2119 MachO::section_64 Sec = info->O->getSection64(Load, J);
2120 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2121 if (section_type == MachO::S_CSTRING_LITERALS &&
2122 ReferenceValue >= Sec.addr &&
2123 ReferenceValue < Sec.addr + Sec.size) {
2124 uint64_t sect_offset = ReferenceValue - Sec.addr;
2125 uint64_t object_offset = Sec.offset + sect_offset;
2126 StringRef MachOContents = info->O->getData();
2127 uint64_t object_size = MachOContents.size();
2128 const char *object_addr = (const char *)MachOContents.data();
2129 if (object_offset < object_size) {
2130 const char *name = object_addr + object_offset;
2131 return name;
2132 } else {
2133 return nullptr;
2134 }
2135 }
2136 }
2137 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
2138 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
2139 for (unsigned J = 0; J < Seg.nsects; ++J) {
2140 MachO::section Sec = info->O->getSection(Load, J);
2141 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2142 if (section_type == MachO::S_CSTRING_LITERALS &&
2143 ReferenceValue >= Sec.addr &&
2144 ReferenceValue < Sec.addr + Sec.size) {
2145 uint64_t sect_offset = ReferenceValue - Sec.addr;
2146 uint64_t object_offset = Sec.offset + sect_offset;
2147 StringRef MachOContents = info->O->getData();
2148 uint64_t object_size = MachOContents.size();
2149 const char *object_addr = (const char *)MachOContents.data();
2150 if (object_offset < object_size) {
2151 const char *name = object_addr + object_offset;
2152 return name;
2153 } else {
2154 return nullptr;
2155 }
2156 }
2157 }
2158 }
2159 if (I == LoadCommandCount - 1)
2160 break;
2161 else
2162 Load = info->O->getNextLoadCommandInfo(Load);
2163 }
2164 return nullptr;
2165}
2166
Kevin Enderby85974882014-09-26 22:20:44 +00002167// GuessIndirectSymbol returns the name of the indirect symbol for the
2168// ReferenceValue passed in or nullptr. This is used when ReferenceValue maybe
2169// an address of a symbol stub or a lazy or non-lazy pointer to associate the
2170// symbol name being referenced by the stub or pointer.
2171static const char *GuessIndirectSymbol(uint64_t ReferenceValue,
2172 struct DisassembleInfo *info) {
2173 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
2174 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
2175 MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand();
2176 MachO::symtab_command Symtab = info->O->getSymtabLoadCommand();
2177 for (unsigned I = 0;; ++I) {
2178 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
2179 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
2180 for (unsigned J = 0; J < Seg.nsects; ++J) {
2181 MachO::section_64 Sec = info->O->getSection64(Load, J);
2182 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2183 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
2184 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
2185 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
2186 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
2187 section_type == MachO::S_SYMBOL_STUBS) &&
2188 ReferenceValue >= Sec.addr &&
2189 ReferenceValue < Sec.addr + Sec.size) {
2190 uint32_t stride;
2191 if (section_type == MachO::S_SYMBOL_STUBS)
2192 stride = Sec.reserved2;
2193 else
2194 stride = 8;
2195 if (stride == 0)
2196 return nullptr;
2197 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
2198 if (index < Dysymtab.nindirectsyms) {
2199 uint32_t indirect_symbol =
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002200 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
Kevin Enderby85974882014-09-26 22:20:44 +00002201 if (indirect_symbol < Symtab.nsyms) {
2202 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
2203 SymbolRef Symbol = *Sym;
2204 StringRef SymName;
2205 Symbol.getName(SymName);
2206 const char *name = SymName.data();
2207 return name;
2208 }
2209 }
2210 }
2211 }
2212 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
2213 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
2214 for (unsigned J = 0; J < Seg.nsects; ++J) {
2215 MachO::section Sec = info->O->getSection(Load, J);
2216 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2217 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
2218 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
2219 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
2220 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
2221 section_type == MachO::S_SYMBOL_STUBS) &&
2222 ReferenceValue >= Sec.addr &&
2223 ReferenceValue < Sec.addr + Sec.size) {
2224 uint32_t stride;
2225 if (section_type == MachO::S_SYMBOL_STUBS)
2226 stride = Sec.reserved2;
2227 else
2228 stride = 4;
2229 if (stride == 0)
2230 return nullptr;
2231 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
2232 if (index < Dysymtab.nindirectsyms) {
2233 uint32_t indirect_symbol =
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002234 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
Kevin Enderby85974882014-09-26 22:20:44 +00002235 if (indirect_symbol < Symtab.nsyms) {
2236 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
2237 SymbolRef Symbol = *Sym;
2238 StringRef SymName;
2239 Symbol.getName(SymName);
2240 const char *name = SymName.data();
2241 return name;
2242 }
2243 }
2244 }
2245 }
2246 }
2247 if (I == LoadCommandCount - 1)
2248 break;
2249 else
2250 Load = info->O->getNextLoadCommandInfo(Load);
2251 }
2252 return nullptr;
2253}
2254
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002255// method_reference() is called passing it the ReferenceName that might be
2256// a reference it to an Objective-C method call. If so then it allocates and
2257// assembles a method call string with the values last seen and saved in
2258// the DisassembleInfo's class_name and selector_name fields. This is saved
2259// into the method field of the info and any previous string is free'ed.
2260// Then the class_name field in the info is set to nullptr. The method call
2261// string is set into ReferenceName and ReferenceType is set to
2262// LLVMDisassembler_ReferenceType_Out_Objc_Message. If this not a method call
2263// then both ReferenceType and ReferenceName are left unchanged.
2264static void method_reference(struct DisassembleInfo *info,
2265 uint64_t *ReferenceType,
2266 const char **ReferenceName) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002267 unsigned int Arch = info->O->getArch();
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002268 if (*ReferenceName != nullptr) {
2269 if (strcmp(*ReferenceName, "_objc_msgSend") == 0) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002270 if (info->selector_name != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002271 if (info->method != nullptr)
2272 free(info->method);
2273 if (info->class_name != nullptr) {
2274 info->method = (char *)malloc(5 + strlen(info->class_name) +
2275 strlen(info->selector_name));
2276 if (info->method != nullptr) {
2277 strcpy(info->method, "+[");
2278 strcat(info->method, info->class_name);
2279 strcat(info->method, " ");
2280 strcat(info->method, info->selector_name);
2281 strcat(info->method, "]");
2282 *ReferenceName = info->method;
2283 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
2284 }
2285 } else {
2286 info->method = (char *)malloc(9 + strlen(info->selector_name));
2287 if (info->method != nullptr) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002288 if (Arch == Triple::x86_64)
2289 strcpy(info->method, "-[%rdi ");
2290 else if (Arch == Triple::aarch64)
2291 strcpy(info->method, "-[x0 ");
2292 else
2293 strcpy(info->method, "-[r? ");
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002294 strcat(info->method, info->selector_name);
2295 strcat(info->method, "]");
2296 *ReferenceName = info->method;
2297 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
2298 }
2299 }
2300 info->class_name = nullptr;
2301 }
2302 } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002303 if (info->selector_name != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002304 if (info->method != nullptr)
2305 free(info->method);
2306 info->method = (char *)malloc(17 + strlen(info->selector_name));
2307 if (info->method != nullptr) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002308 if (Arch == Triple::x86_64)
2309 strcpy(info->method, "-[[%rdi super] ");
2310 else if (Arch == Triple::aarch64)
2311 strcpy(info->method, "-[[x0 super] ");
2312 else
2313 strcpy(info->method, "-[[r? super] ");
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002314 strcat(info->method, info->selector_name);
2315 strcat(info->method, "]");
2316 *ReferenceName = info->method;
2317 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
2318 }
2319 info->class_name = nullptr;
2320 }
2321 }
2322 }
2323}
2324
2325// GuessPointerPointer() is passed the address of what might be a pointer to
2326// a reference to an Objective-C class, selector, message ref or cfstring.
2327// If so the value of the pointer is returned and one of the booleans are set
2328// to true. If not zero is returned and all the booleans are set to false.
2329static uint64_t GuessPointerPointer(uint64_t ReferenceValue,
2330 struct DisassembleInfo *info,
2331 bool &classref, bool &selref, bool &msgref,
2332 bool &cfstring) {
2333 classref = false;
2334 selref = false;
2335 msgref = false;
2336 cfstring = false;
2337 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
2338 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
2339 for (unsigned I = 0;; ++I) {
2340 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
2341 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
2342 for (unsigned J = 0; J < Seg.nsects; ++J) {
2343 MachO::section_64 Sec = info->O->getSection64(Load, J);
2344 if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 ||
2345 strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
2346 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 ||
2347 strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 ||
2348 strncmp(Sec.sectname, "__cfstring", 16) == 0) &&
2349 ReferenceValue >= Sec.addr &&
2350 ReferenceValue < Sec.addr + Sec.size) {
2351 uint64_t sect_offset = ReferenceValue - Sec.addr;
2352 uint64_t object_offset = Sec.offset + sect_offset;
2353 StringRef MachOContents = info->O->getData();
2354 uint64_t object_size = MachOContents.size();
2355 const char *object_addr = (const char *)MachOContents.data();
2356 if (object_offset < object_size) {
2357 uint64_t pointer_value;
2358 memcpy(&pointer_value, object_addr + object_offset,
2359 sizeof(uint64_t));
2360 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2361 sys::swapByteOrder(pointer_value);
2362 if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0)
2363 selref = true;
2364 else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
2365 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0)
2366 classref = true;
2367 else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 &&
2368 ReferenceValue + 8 < Sec.addr + Sec.size) {
2369 msgref = true;
2370 memcpy(&pointer_value, object_addr + object_offset + 8,
2371 sizeof(uint64_t));
2372 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2373 sys::swapByteOrder(pointer_value);
2374 } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0)
2375 cfstring = true;
2376 return pointer_value;
2377 } else {
2378 return 0;
2379 }
2380 }
2381 }
2382 }
2383 // TODO: Look for LC_SEGMENT for 32-bit Mach-O files.
2384 if (I == LoadCommandCount - 1)
2385 break;
2386 else
2387 Load = info->O->getNextLoadCommandInfo(Load);
2388 }
2389 return 0;
2390}
2391
2392// get_pointer_64 returns a pointer to the bytes in the object file at the
2393// Address from a section in the Mach-O file. And indirectly returns the
2394// offset into the section, number of bytes left in the section past the offset
2395// and which section is was being referenced. If the Address is not in a
2396// section nullptr is returned.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002397static const char *get_pointer_64(uint64_t Address, uint32_t &offset,
2398 uint32_t &left, SectionRef &S,
2399 DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002400 offset = 0;
2401 left = 0;
2402 S = SectionRef();
2403 for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) {
2404 uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress();
2405 uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize();
2406 if (Address >= SectAddress && Address < SectAddress + SectSize) {
2407 S = (*(info->Sections))[SectIdx];
2408 offset = Address - SectAddress;
2409 left = SectSize - offset;
2410 StringRef SectContents;
2411 ((*(info->Sections))[SectIdx]).getContents(SectContents);
2412 return SectContents.data() + offset;
2413 }
2414 }
2415 return nullptr;
2416}
2417
2418// get_symbol_64() returns the name of a symbol (or nullptr) and the address of
2419// the symbol indirectly through n_value. Based on the relocation information
2420// for the specified section offset in the specified section reference.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002421static const char *get_symbol_64(uint32_t sect_offset, SectionRef S,
2422 DisassembleInfo *info, uint64_t &n_value) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002423 n_value = 0;
2424 if (info->verbose == false)
2425 return nullptr;
2426
2427 // See if there is an external relocation entry at the sect_offset.
2428 bool reloc_found = false;
2429 DataRefImpl Rel;
2430 MachO::any_relocation_info RE;
2431 bool isExtern = false;
2432 SymbolRef Symbol;
2433 for (const RelocationRef &Reloc : S.relocations()) {
2434 uint64_t RelocOffset;
2435 Reloc.getOffset(RelocOffset);
2436 if (RelocOffset == sect_offset) {
2437 Rel = Reloc.getRawDataRefImpl();
2438 RE = info->O->getRelocation(Rel);
2439 if (info->O->isRelocationScattered(RE))
2440 continue;
2441 isExtern = info->O->getPlainRelocationExternal(RE);
2442 if (isExtern) {
2443 symbol_iterator RelocSym = Reloc.getSymbol();
2444 Symbol = *RelocSym;
2445 }
2446 reloc_found = true;
2447 break;
2448 }
2449 }
2450 // If there is an external relocation entry for a symbol in this section
2451 // at this section_offset then use that symbol's value for the n_value
2452 // and return its name.
2453 const char *SymbolName = nullptr;
2454 if (reloc_found && isExtern) {
2455 Symbol.getAddress(n_value);
2456 StringRef name;
2457 Symbol.getName(name);
2458 if (!name.empty()) {
2459 SymbolName = name.data();
2460 return SymbolName;
2461 }
2462 }
2463
2464 // TODO: For fully linked images, look through the external relocation
2465 // entries off the dynamic symtab command. For these the r_offset is from the
2466 // start of the first writeable segment in the Mach-O file. So the offset
2467 // to this section from that segment is passed to this routine by the caller,
2468 // as the database_offset. Which is the difference of the section's starting
2469 // address and the first writable segment.
2470 //
2471 // NOTE: need add passing the database_offset to this routine.
2472
2473 // TODO: We did not find an external relocation entry so look up the
2474 // ReferenceValue as an address of a symbol and if found return that symbol's
2475 // name.
2476 //
2477 // NOTE: need add passing the ReferenceValue to this routine. Then that code
2478 // would simply be this:
Kevin Enderbyf6d25852015-01-31 00:37:11 +00002479 // SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002480
2481 return SymbolName;
2482}
2483
2484// These are structs in the Objective-C meta data and read to produce the
2485// comments for disassembly. While these are part of the ABI they are no
2486// public defintions. So the are here not in include/llvm/Support/MachO.h .
2487
2488// The cfstring object in a 64-bit Mach-O file.
2489struct cfstring64_t {
2490 uint64_t isa; // class64_t * (64-bit pointer)
2491 uint64_t flags; // flag bits
2492 uint64_t characters; // char * (64-bit pointer)
2493 uint64_t length; // number of non-NULL characters in above
2494};
2495
2496// The class object in a 64-bit Mach-O file.
2497struct class64_t {
2498 uint64_t isa; // class64_t * (64-bit pointer)
2499 uint64_t superclass; // class64_t * (64-bit pointer)
2500 uint64_t cache; // Cache (64-bit pointer)
2501 uint64_t vtable; // IMP * (64-bit pointer)
2502 uint64_t data; // class_ro64_t * (64-bit pointer)
2503};
2504
2505struct class_ro64_t {
2506 uint32_t flags;
2507 uint32_t instanceStart;
2508 uint32_t instanceSize;
2509 uint32_t reserved;
2510 uint64_t ivarLayout; // const uint8_t * (64-bit pointer)
2511 uint64_t name; // const char * (64-bit pointer)
2512 uint64_t baseMethods; // const method_list_t * (64-bit pointer)
2513 uint64_t baseProtocols; // const protocol_list_t * (64-bit pointer)
2514 uint64_t ivars; // const ivar_list_t * (64-bit pointer)
2515 uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer)
2516 uint64_t baseProperties; // const struct objc_property_list (64-bit pointer)
2517};
2518
2519inline void swapStruct(struct cfstring64_t &cfs) {
2520 sys::swapByteOrder(cfs.isa);
2521 sys::swapByteOrder(cfs.flags);
2522 sys::swapByteOrder(cfs.characters);
2523 sys::swapByteOrder(cfs.length);
2524}
2525
2526inline void swapStruct(struct class64_t &c) {
2527 sys::swapByteOrder(c.isa);
2528 sys::swapByteOrder(c.superclass);
2529 sys::swapByteOrder(c.cache);
2530 sys::swapByteOrder(c.vtable);
2531 sys::swapByteOrder(c.data);
2532}
2533
2534inline void swapStruct(struct class_ro64_t &cro) {
2535 sys::swapByteOrder(cro.flags);
2536 sys::swapByteOrder(cro.instanceStart);
2537 sys::swapByteOrder(cro.instanceSize);
2538 sys::swapByteOrder(cro.reserved);
2539 sys::swapByteOrder(cro.ivarLayout);
2540 sys::swapByteOrder(cro.name);
2541 sys::swapByteOrder(cro.baseMethods);
2542 sys::swapByteOrder(cro.baseProtocols);
2543 sys::swapByteOrder(cro.ivars);
2544 sys::swapByteOrder(cro.weakIvarLayout);
2545 sys::swapByteOrder(cro.baseProperties);
2546}
2547
2548static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
2549 struct DisassembleInfo *info);
2550
2551// get_objc2_64bit_class_name() is used for disassembly and is passed a pointer
2552// to an Objective-C class and returns the class name. It is also passed the
2553// address of the pointer, so when the pointer is zero as it can be in an .o
2554// file, that is used to look for an external relocation entry with a symbol
2555// name.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002556static const char *get_objc2_64bit_class_name(uint64_t pointer_value,
2557 uint64_t ReferenceValue,
2558 struct DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002559 const char *r;
2560 uint32_t offset, left;
2561 SectionRef S;
2562
2563 // The pointer_value can be 0 in an object file and have a relocation
2564 // entry for the class symbol at the ReferenceValue (the address of the
2565 // pointer).
2566 if (pointer_value == 0) {
2567 r = get_pointer_64(ReferenceValue, offset, left, S, info);
2568 if (r == nullptr || left < sizeof(uint64_t))
2569 return nullptr;
2570 uint64_t n_value;
2571 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
2572 if (symbol_name == nullptr)
2573 return nullptr;
Hans Wennborgdb53e302014-10-23 21:59:17 +00002574 const char *class_name = strrchr(symbol_name, '$');
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002575 if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0')
2576 return class_name + 2;
2577 else
2578 return nullptr;
2579 }
2580
2581 // The case were the pointer_value is non-zero and points to a class defined
2582 // in this Mach-O file.
2583 r = get_pointer_64(pointer_value, offset, left, S, info);
2584 if (r == nullptr || left < sizeof(struct class64_t))
2585 return nullptr;
2586 struct class64_t c;
2587 memcpy(&c, r, sizeof(struct class64_t));
2588 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2589 swapStruct(c);
2590 if (c.data == 0)
2591 return nullptr;
2592 r = get_pointer_64(c.data, offset, left, S, info);
2593 if (r == nullptr || left < sizeof(struct class_ro64_t))
2594 return nullptr;
2595 struct class_ro64_t cro;
2596 memcpy(&cro, r, sizeof(struct class_ro64_t));
2597 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2598 swapStruct(cro);
2599 if (cro.name == 0)
2600 return nullptr;
2601 const char *name = get_pointer_64(cro.name, offset, left, S, info);
2602 return name;
2603}
2604
2605// get_objc2_64bit_cfstring_name is used for disassembly and is passed a
2606// pointer to a cfstring and returns its name or nullptr.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002607static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,
2608 struct DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002609 const char *r, *name;
2610 uint32_t offset, left;
2611 SectionRef S;
2612 struct cfstring64_t cfs;
2613 uint64_t cfs_characters;
2614
2615 r = get_pointer_64(ReferenceValue, offset, left, S, info);
2616 if (r == nullptr || left < sizeof(struct cfstring64_t))
2617 return nullptr;
2618 memcpy(&cfs, r, sizeof(struct cfstring64_t));
2619 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2620 swapStruct(cfs);
2621 if (cfs.characters == 0) {
2622 uint64_t n_value;
2623 const char *symbol_name = get_symbol_64(
2624 offset + offsetof(struct cfstring64_t, characters), S, info, n_value);
2625 if (symbol_name == nullptr)
2626 return nullptr;
2627 cfs_characters = n_value;
2628 } else
2629 cfs_characters = cfs.characters;
2630 name = get_pointer_64(cfs_characters, offset, left, S, info);
2631
2632 return name;
2633}
2634
2635// get_objc2_64bit_selref() is used for disassembly and is passed a the address
2636// of a pointer to an Objective-C selector reference when the pointer value is
2637// zero as in a .o file and is likely to have a external relocation entry with
2638// who's symbol's n_value is the real pointer to the selector name. If that is
2639// the case the real pointer to the selector name is returned else 0 is
2640// returned
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002641static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue,
2642 struct DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002643 uint32_t offset, left;
2644 SectionRef S;
2645
2646 const char *r = get_pointer_64(ReferenceValue, offset, left, S, info);
2647 if (r == nullptr || left < sizeof(uint64_t))
2648 return 0;
2649 uint64_t n_value;
2650 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
2651 if (symbol_name == nullptr)
2652 return 0;
2653 return n_value;
2654}
2655
Kevin Enderbybf246f52014-09-24 23:08:22 +00002656// GuessLiteralPointer returns a string which for the item in the Mach-O file
2657// for the address passed in as ReferenceValue for printing as a comment with
2658// the instruction and also returns the corresponding type of that item
2659// indirectly through ReferenceType.
2660//
2661// If ReferenceValue is an address of literal cstring then a pointer to the
2662// cstring is returned and ReferenceType is set to
2663// LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr .
2664//
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002665// If ReferenceValue is an address of an Objective-C CFString, Selector ref or
2666// Class ref that name is returned and the ReferenceType is set accordingly.
2667//
2668// Lastly, literals which are Symbol address in a literal pool are looked for
2669// and if found the symbol name is returned and ReferenceType is set to
2670// LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr .
2671//
2672// If there is no item in the Mach-O file for the address passed in as
2673// ReferenceValue nullptr is returned and ReferenceType is unchanged.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002674static const char *GuessLiteralPointer(uint64_t ReferenceValue,
2675 uint64_t ReferencePC,
2676 uint64_t *ReferenceType,
2677 struct DisassembleInfo *info) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002678 // First see if there is an external relocation entry at the ReferencePC.
Rafael Espindola80291272014-10-08 15:28:58 +00002679 uint64_t sect_addr = info->S.getAddress();
Kevin Enderbybf246f52014-09-24 23:08:22 +00002680 uint64_t sect_offset = ReferencePC - sect_addr;
2681 bool reloc_found = false;
2682 DataRefImpl Rel;
2683 MachO::any_relocation_info RE;
2684 bool isExtern = false;
2685 SymbolRef Symbol;
2686 for (const RelocationRef &Reloc : info->S.relocations()) {
2687 uint64_t RelocOffset;
2688 Reloc.getOffset(RelocOffset);
2689 if (RelocOffset == sect_offset) {
2690 Rel = Reloc.getRawDataRefImpl();
2691 RE = info->O->getRelocation(Rel);
2692 if (info->O->isRelocationScattered(RE))
2693 continue;
2694 isExtern = info->O->getPlainRelocationExternal(RE);
2695 if (isExtern) {
2696 symbol_iterator RelocSym = Reloc.getSymbol();
2697 Symbol = *RelocSym;
2698 }
2699 reloc_found = true;
2700 break;
2701 }
2702 }
2703 // If there is an external relocation entry for a symbol in a section
2704 // then used that symbol's value for the value of the reference.
2705 if (reloc_found && isExtern) {
2706 if (info->O->getAnyRelocationPCRel(RE)) {
2707 unsigned Type = info->O->getAnyRelocationType(RE);
2708 if (Type == MachO::X86_64_RELOC_SIGNED) {
2709 Symbol.getAddress(ReferenceValue);
2710 }
2711 }
2712 }
2713
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002714 // Look for literals such as Objective-C CFStrings refs, Selector refs,
2715 // Message refs and Class refs.
2716 bool classref, selref, msgref, cfstring;
2717 uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref,
2718 selref, msgref, cfstring);
2719 if (classref == true && pointer_value == 0) {
2720 // Note the ReferenceValue is a pointer into the __objc_classrefs section.
2721 // And the pointer_value in that section is typically zero as it will be
2722 // set by dyld as part of the "bind information".
2723 const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info);
2724 if (name != nullptr) {
2725 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
Hans Wennborgdb53e302014-10-23 21:59:17 +00002726 const char *class_name = strrchr(name, '$');
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002727 if (class_name != nullptr && class_name[1] == '_' &&
2728 class_name[2] != '\0') {
2729 info->class_name = class_name + 2;
2730 return name;
2731 }
2732 }
2733 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00002734
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002735 if (classref == true) {
2736 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
2737 const char *name =
2738 get_objc2_64bit_class_name(pointer_value, ReferenceValue, info);
2739 if (name != nullptr)
2740 info->class_name = name;
2741 else
2742 name = "bad class ref";
Kevin Enderbybf246f52014-09-24 23:08:22 +00002743 return name;
2744 }
2745
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002746 if (cfstring == true) {
2747 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref;
2748 const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info);
2749 return name;
2750 }
2751
2752 if (selref == true && pointer_value == 0)
2753 pointer_value = get_objc2_64bit_selref(ReferenceValue, info);
2754
2755 if (pointer_value != 0)
2756 ReferenceValue = pointer_value;
2757
2758 const char *name = GuessCstringPointer(ReferenceValue, info);
2759 if (name) {
2760 if (pointer_value != 0 && selref == true) {
2761 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref;
2762 info->selector_name = name;
2763 } else if (pointer_value != 0 && msgref == true) {
2764 info->class_name = nullptr;
2765 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref;
2766 info->selector_name = name;
2767 } else
2768 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr;
2769 return name;
2770 }
2771
2772 // Lastly look for an indirect symbol with this ReferenceValue which is in
2773 // a literal pool. If found return that symbol name.
2774 name = GuessIndirectSymbol(ReferenceValue, info);
2775 if (name) {
2776 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr;
2777 return name;
2778 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00002779
2780 return nullptr;
2781}
2782
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002783// SymbolizerSymbolLookUp is the symbol lookup function passed when creating
Kevin Enderbybf246f52014-09-24 23:08:22 +00002784// the Symbolizer. It looks up the ReferenceValue using the info passed via the
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002785// pointer to the struct DisassembleInfo that was passed when MCSymbolizer
2786// is created and returns the symbol name that matches the ReferenceValue or
2787// nullptr if none. The ReferenceType is passed in for the IN type of
2788// reference the instruction is making from the values in defined in the header
2789// "llvm-c/Disassembler.h". On return the ReferenceType can set to a specific
2790// Out type and the ReferenceName will also be set which is added as a comment
2791// to the disassembled instruction.
2792//
Kevin Enderby04bf6932014-10-28 23:39:46 +00002793#if HAVE_CXXABI_H
2794// If the symbol name is a C++ mangled name then the demangled name is
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002795// returned through ReferenceName and ReferenceType is set to
2796// LLVMDisassembler_ReferenceType_DeMangled_Name .
Kevin Enderby04bf6932014-10-28 23:39:46 +00002797#endif
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002798//
2799// When this is called to get a symbol name for a branch target then the
2800// ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then
2801// SymbolValue will be looked for in the indirect symbol table to determine if
2802// it is an address for a symbol stub. If so then the symbol name for that
2803// stub is returned indirectly through ReferenceName and then ReferenceType is
2804// set to LLVMDisassembler_ReferenceType_Out_SymbolStub.
2805//
Kevin Enderbybf246f52014-09-24 23:08:22 +00002806// When this is called with an value loaded via a PC relative load then
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002807// ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the
2808// SymbolValue is checked to be an address of literal pointer, symbol pointer,
2809// or an Objective-C meta data reference. If so the output ReferenceType is
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002810// set to correspond to that as well as setting the ReferenceName.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002811static const char *SymbolizerSymbolLookUp(void *DisInfo,
2812 uint64_t ReferenceValue,
2813 uint64_t *ReferenceType,
2814 uint64_t ReferencePC,
2815 const char **ReferenceName) {
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002816 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
Kevin Enderbybf246f52014-09-24 23:08:22 +00002817 // If no verbose symbolic information is wanted then just return nullptr.
2818 if (info->verbose == false) {
2819 *ReferenceName = nullptr;
2820 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002821 return nullptr;
2822 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00002823
Kevin Enderbyf6d25852015-01-31 00:37:11 +00002824 const char *SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
Kevin Enderbybf246f52014-09-24 23:08:22 +00002825
Kevin Enderby85974882014-09-26 22:20:44 +00002826 if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) {
2827 *ReferenceName = GuessIndirectSymbol(ReferenceValue, info);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002828 if (*ReferenceName != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002829 method_reference(info, ReferenceType, ReferenceName);
2830 if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message)
2831 *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub;
2832 } else
Kevin Enderby04bf6932014-10-28 23:39:46 +00002833#if HAVE_CXXABI_H
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002834 if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
Kevin Enderby04bf6932014-10-28 23:39:46 +00002835 if (info->demangled_name != nullptr)
2836 free(info->demangled_name);
2837 int status;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002838 info->demangled_name =
2839 abi::__cxa_demangle(SymbolName + 1, nullptr, nullptr, &status);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002840 if (info->demangled_name != nullptr) {
2841 *ReferenceName = info->demangled_name;
2842 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
2843 } else
2844 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2845 } else
2846#endif
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002847 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2848 } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) {
2849 *ReferenceName =
2850 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
Kevin Enderby85974882014-09-26 22:20:44 +00002851 if (*ReferenceName)
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002852 method_reference(info, ReferenceType, ReferenceName);
Kevin Enderby85974882014-09-26 22:20:44 +00002853 else
2854 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002855 // If this is arm64 and the reference is an adrp instruction save the
2856 // instruction, passed in ReferenceValue and the address of the instruction
2857 // for use later if we see and add immediate instruction.
2858 } else if (info->O->getArch() == Triple::aarch64 &&
2859 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
2860 info->adrp_inst = ReferenceValue;
2861 info->adrp_addr = ReferencePC;
2862 SymbolName = nullptr;
2863 *ReferenceName = nullptr;
2864 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2865 // If this is arm64 and reference is an add immediate instruction and we
2866 // have
2867 // seen an adrp instruction just before it and the adrp's Xd register
2868 // matches
2869 // this add's Xn register reconstruct the value being referenced and look to
2870 // see if it is a literal pointer. Note the add immediate instruction is
2871 // passed in ReferenceValue.
2872 } else if (info->O->getArch() == Triple::aarch64 &&
2873 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri &&
2874 ReferencePC - 4 == info->adrp_addr &&
2875 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
2876 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
2877 uint32_t addxri_inst;
2878 uint64_t adrp_imm, addxri_imm;
2879
2880 adrp_imm =
2881 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
2882 if (info->adrp_inst & 0x0200000)
2883 adrp_imm |= 0xfffffffffc000000LL;
2884
2885 addxri_inst = ReferenceValue;
2886 addxri_imm = (addxri_inst >> 10) & 0xfff;
2887 if (((addxri_inst >> 22) & 0x3) == 1)
2888 addxri_imm <<= 12;
2889
2890 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
2891 (adrp_imm << 12) + addxri_imm;
2892
2893 *ReferenceName =
2894 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
2895 if (*ReferenceName == nullptr)
2896 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2897 // If this is arm64 and the reference is a load register instruction and we
2898 // have seen an adrp instruction just before it and the adrp's Xd register
2899 // matches this add's Xn register reconstruct the value being referenced and
2900 // look to see if it is a literal pointer. Note the load register
2901 // instruction is passed in ReferenceValue.
2902 } else if (info->O->getArch() == Triple::aarch64 &&
2903 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui &&
2904 ReferencePC - 4 == info->adrp_addr &&
2905 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
2906 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
2907 uint32_t ldrxui_inst;
2908 uint64_t adrp_imm, ldrxui_imm;
2909
2910 adrp_imm =
2911 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
2912 if (info->adrp_inst & 0x0200000)
2913 adrp_imm |= 0xfffffffffc000000LL;
2914
2915 ldrxui_inst = ReferenceValue;
2916 ldrxui_imm = (ldrxui_inst >> 10) & 0xfff;
2917
2918 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
2919 (adrp_imm << 12) + (ldrxui_imm << 3);
2920
2921 *ReferenceName =
2922 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
2923 if (*ReferenceName == nullptr)
2924 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2925 }
2926 // If this arm64 and is an load register (PC-relative) instruction the
2927 // ReferenceValue is the PC plus the immediate value.
2928 else if (info->O->getArch() == Triple::aarch64 &&
2929 (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl ||
2930 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) {
2931 *ReferenceName =
2932 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
2933 if (*ReferenceName == nullptr)
2934 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderby85974882014-09-26 22:20:44 +00002935 }
Kevin Enderby04bf6932014-10-28 23:39:46 +00002936#if HAVE_CXXABI_H
2937 else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
2938 if (info->demangled_name != nullptr)
2939 free(info->demangled_name);
2940 int status;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002941 info->demangled_name =
2942 abi::__cxa_demangle(SymbolName + 1, nullptr, nullptr, &status);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002943 if (info->demangled_name != nullptr) {
2944 *ReferenceName = info->demangled_name;
2945 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
2946 }
2947 }
2948#endif
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002949 else {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002950 *ReferenceName = nullptr;
2951 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2952 }
2953
2954 return SymbolName;
2955}
2956
Kevin Enderbybf246f52014-09-24 23:08:22 +00002957/// \brief Emits the comments that are stored in the CommentStream.
2958/// Each comment in the CommentStream must end with a newline.
2959static void emitComments(raw_svector_ostream &CommentStream,
2960 SmallString<128> &CommentsToEmit,
2961 formatted_raw_ostream &FormattedOS,
2962 const MCAsmInfo &MAI) {
2963 // Flush the stream before taking its content.
2964 CommentStream.flush();
2965 StringRef Comments = CommentsToEmit.str();
2966 // Get the default information for printing a comment.
2967 const char *CommentBegin = MAI.getCommentString();
2968 unsigned CommentColumn = MAI.getCommentColumn();
2969 bool IsFirst = true;
2970 while (!Comments.empty()) {
2971 if (!IsFirst)
2972 FormattedOS << '\n';
2973 // Emit a line of comments.
2974 FormattedOS.PadToColumn(CommentColumn);
2975 size_t Position = Comments.find('\n');
2976 FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position);
2977 // Move after the newline character.
2978 Comments = Comments.substr(Position + 1);
2979 IsFirst = false;
2980 }
2981 FormattedOS.flush();
2982
2983 // Tell the comment stream that the vector changed underneath it.
2984 CommentsToEmit.clear();
2985 CommentStream.resync();
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002986}
2987
Kevin Enderby95df54c2015-02-04 01:01:38 +00002988static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
2989 StringRef DisSegName, StringRef DisSectName) {
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002990 const char *McpuDefault = nullptr;
2991 const Target *ThumbTarget = nullptr;
2992 const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget);
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002993 if (!TheTarget) {
2994 // GetTarget prints out stuff.
2995 return;
2996 }
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002997 if (MCPU.empty() && McpuDefault)
2998 MCPU = McpuDefault;
2999
Ahmed Charles56440fd2014-03-06 05:51:42 +00003000 std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003001 std::unique_ptr<const MCInstrInfo> ThumbInstrInfo;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003002 if (ThumbTarget)
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003003 ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo());
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003004
Kevin Enderbyc9595622014-08-06 23:24:41 +00003005 // Package up features to be passed to target/subtarget
3006 std::string FeaturesStr;
3007 if (MAttrs.size()) {
3008 SubtargetFeatures Features;
3009 for (unsigned i = 0; i != MAttrs.size(); ++i)
3010 Features.AddFeature(MAttrs[i]);
3011 FeaturesStr = Features.getString();
3012 }
3013
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003014 // Set up disassembler.
Ahmed Charles56440fd2014-03-06 05:51:42 +00003015 std::unique_ptr<const MCRegisterInfo> MRI(
3016 TheTarget->createMCRegInfo(TripleName));
3017 std::unique_ptr<const MCAsmInfo> AsmInfo(
Rafael Espindola227144c2013-05-13 01:16:13 +00003018 TheTarget->createMCAsmInfo(*MRI, TripleName));
Ahmed Charles56440fd2014-03-06 05:51:42 +00003019 std::unique_ptr<const MCSubtargetInfo> STI(
Kevin Enderbyc9595622014-08-06 23:24:41 +00003020 TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
Craig Toppere6cb63e2014-04-25 04:24:47 +00003021 MCContext Ctx(AsmInfo.get(), MRI.get(), nullptr);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003022 std::unique_ptr<MCDisassembler> DisAsm(
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003023 TheTarget->createMCDisassembler(*STI, Ctx));
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003024 std::unique_ptr<MCSymbolizer> Symbolizer;
3025 struct DisassembleInfo SymbolizerInfo;
3026 std::unique_ptr<MCRelocationInfo> RelInfo(
3027 TheTarget->createMCRelocationInfo(TripleName, Ctx));
3028 if (RelInfo) {
3029 Symbolizer.reset(TheTarget->createMCSymbolizer(
3030 TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
David Blaikie186db432015-01-18 20:45:48 +00003031 &SymbolizerInfo, &Ctx, std::move(RelInfo)));
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003032 DisAsm->setSymbolizer(std::move(Symbolizer));
3033 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003034 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
Ahmed Charles56440fd2014-03-06 05:51:42 +00003035 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
3036 AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI, *STI));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003037 // Set the display preference for hex vs. decimal immediates.
3038 IP->setPrintImmHex(PrintImmHex);
3039 // Comment stream and backing vector.
3040 SmallString<128> CommentsToEmit;
3041 raw_svector_ostream CommentStream(CommentsToEmit);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00003042 // FIXME: Setting the CommentStream in the InstPrinter is problematic in that
3043 // if it is done then arm64 comments for string literals don't get printed
3044 // and some constant get printed instead and not setting it causes intel
3045 // (32-bit and 64-bit) comments printed with different spacing before the
3046 // comment causing different diffs with the 'C' disassembler library API.
3047 // IP->setCommentStream(CommentStream);
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003048
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003049 if (!AsmInfo || !STI || !DisAsm || !IP) {
Michael J. Spencerc1363cf2011-10-07 19:25:47 +00003050 errs() << "error: couldn't initialize disassembler for target "
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003051 << TripleName << '\n';
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003052 return;
3053 }
3054
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003055 // Set up thumb disassembler.
3056 std::unique_ptr<const MCRegisterInfo> ThumbMRI;
3057 std::unique_ptr<const MCAsmInfo> ThumbAsmInfo;
3058 std::unique_ptr<const MCSubtargetInfo> ThumbSTI;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003059 std::unique_ptr<MCDisassembler> ThumbDisAsm;
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003060 std::unique_ptr<MCInstPrinter> ThumbIP;
3061 std::unique_ptr<MCContext> ThumbCtx;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003062 std::unique_ptr<MCSymbolizer> ThumbSymbolizer;
3063 struct DisassembleInfo ThumbSymbolizerInfo;
3064 std::unique_ptr<MCRelocationInfo> ThumbRelInfo;
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003065 if (ThumbTarget) {
3066 ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName));
3067 ThumbAsmInfo.reset(
3068 ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName));
3069 ThumbSTI.reset(
3070 ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MCPU, FeaturesStr));
3071 ThumbCtx.reset(new MCContext(ThumbAsmInfo.get(), ThumbMRI.get(), nullptr));
3072 ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx));
Kevin Enderby930fdc72014-11-06 19:00:13 +00003073 MCContext *PtrThumbCtx = ThumbCtx.get();
3074 ThumbRelInfo.reset(
3075 ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx));
3076 if (ThumbRelInfo) {
3077 ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer(
3078 ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
David Blaikie186db432015-01-18 20:45:48 +00003079 &ThumbSymbolizerInfo, PtrThumbCtx, std::move(ThumbRelInfo)));
Kevin Enderby930fdc72014-11-06 19:00:13 +00003080 ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer));
3081 }
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003082 int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect();
3083 ThumbIP.reset(ThumbTarget->createMCInstPrinter(
3084 ThumbAsmPrinterVariant, *ThumbAsmInfo, *ThumbInstrInfo, *ThumbMRI,
3085 *ThumbSTI));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003086 // Set the display preference for hex vs. decimal immediates.
3087 ThumbIP->setPrintImmHex(PrintImmHex);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003088 }
3089
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003090 if (ThumbTarget && (!ThumbAsmInfo || !ThumbSTI || !ThumbDisAsm || !ThumbIP)) {
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003091 errs() << "error: couldn't initialize disassembler for target "
3092 << ThumbTripleName << '\n';
3093 return;
3094 }
3095
Charles Davis8bdfafd2013-09-01 04:28:48 +00003096 MachO::mach_header Header = MachOOF->getHeader();
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003097
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003098 // FIXME: Using the -cfg command line option, this code used to be able to
3099 // annotate relocations with the referenced symbol's name, and if this was
3100 // inside a __[cf]string section, the data it points to. This is now replaced
3101 // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
Owen Andersond9243c42011-10-17 21:37:35 +00003102 std::vector<SectionRef> Sections;
3103 std::vector<SymbolRef> Symbols;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003104 SmallVector<uint64_t, 8> FoundFns;
Kevin Enderby273ae012013-06-06 17:20:50 +00003105 uint64_t BaseSegmentAddress;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003106
Kevin Enderby273ae012013-06-06 17:20:50 +00003107 getSectionsAndSymbols(Header, MachOOF, Sections, Symbols, FoundFns,
3108 BaseSegmentAddress);
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003109
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003110 // Sort the symbols by address, just in case they didn't come in that way.
Owen Andersond9243c42011-10-17 21:37:35 +00003111 std::sort(Symbols.begin(), Symbols.end(), SymbolSorter());
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003112
Kevin Enderby273ae012013-06-06 17:20:50 +00003113 // Build a data in code table that is sorted on by the address of each entry.
3114 uint64_t BaseAddress = 0;
Charles Davis8bdfafd2013-09-01 04:28:48 +00003115 if (Header.filetype == MachO::MH_OBJECT)
Rafael Espindola80291272014-10-08 15:28:58 +00003116 BaseAddress = Sections[0].getAddress();
Kevin Enderby273ae012013-06-06 17:20:50 +00003117 else
3118 BaseAddress = BaseSegmentAddress;
3119 DiceTable Dices;
Kevin Enderby273ae012013-06-06 17:20:50 +00003120 for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
Rafael Espindola5e812af2014-01-30 02:49:50 +00003121 DI != DE; ++DI) {
Kevin Enderby273ae012013-06-06 17:20:50 +00003122 uint32_t Offset;
3123 DI->getOffset(Offset);
3124 Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
3125 }
3126 array_pod_sort(Dices.begin(), Dices.end());
3127
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003128#ifndef NDEBUG
3129 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
3130#else
3131 raw_ostream &DebugOut = nulls();
3132#endif
3133
Ahmed Charles56440fd2014-03-06 05:51:42 +00003134 std::unique_ptr<DIContext> diContext;
Rafael Espindola9b709252013-04-13 01:45:40 +00003135 ObjectFile *DbgObj = MachOOF;
Benjamin Kramer699128e2011-09-21 01:13:19 +00003136 // Try to find debug info and set up the DIContext for it.
3137 if (UseDbg) {
Benjamin Kramer699128e2011-09-21 01:13:19 +00003138 // A separate DSym file path was specified, parse it as a macho file,
3139 // get the sections and supply it to the section name parsing machinery.
3140 if (!DSYMFile.empty()) {
Rafael Espindola48af1c22014-08-19 18:44:46 +00003141 ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr =
Rafael Espindolaadf21f22014-07-06 17:43:13 +00003142 MemoryBuffer::getFileOrSTDIN(DSYMFile);
Rafael Espindola48af1c22014-08-19 18:44:46 +00003143 if (std::error_code EC = BufOrErr.getError()) {
Rafael Espindolaadf21f22014-07-06 17:43:13 +00003144 errs() << "llvm-objdump: " << Filename << ": " << EC.message() << '\n';
Benjamin Kramer699128e2011-09-21 01:13:19 +00003145 return;
3146 }
Rafael Espindola48af1c22014-08-19 18:44:46 +00003147 DbgObj =
3148 ObjectFile::createMachOObjectFile(BufOrErr.get()->getMemBufferRef())
3149 .get()
3150 .release();
Benjamin Kramer699128e2011-09-21 01:13:19 +00003151 }
3152
Eric Christopher7370b552012-11-12 21:40:38 +00003153 // Setup the DIContext
Rafael Espindolaa04bb5b2014-07-31 20:19:36 +00003154 diContext.reset(DIContext::getDWARFContext(*DbgObj));
Benjamin Kramer699128e2011-09-21 01:13:19 +00003155 }
3156
Kevin Enderby95df54c2015-02-04 01:01:38 +00003157 if (DumpSections.size() == 0)
3158 outs() << "(" << DisSegName << "," << DisSectName << ") section\n";
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00003159
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003160 for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
Owen Andersond9243c42011-10-17 21:37:35 +00003161 StringRef SectName;
Kevin Enderby95df54c2015-02-04 01:01:38 +00003162 if (Sections[SectIdx].getName(SectName) || SectName != DisSectName)
3163 continue;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003164
Rafael Espindolaa9f810b2012-12-21 03:47:03 +00003165 DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003166
Rafael Espindolab0f76a42013-04-05 15:15:22 +00003167 StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
Kevin Enderby95df54c2015-02-04 01:01:38 +00003168 if (SegmentName != DisSegName)
Rafael Espindolaa9f810b2012-12-21 03:47:03 +00003169 continue;
3170
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003171 StringRef BytesStr;
3172 Sections[SectIdx].getContents(BytesStr);
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003173 ArrayRef<uint8_t> Bytes(reinterpret_cast<const uint8_t *>(BytesStr.data()),
3174 BytesStr.size());
Rafael Espindola80291272014-10-08 15:28:58 +00003175 uint64_t SectAddress = Sections[SectIdx].getAddress();
Rafael Espindolabd604f22014-11-07 00:52:15 +00003176
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003177 bool symbolTableWorked = false;
3178
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003179 // Parse relocations.
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00003180 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
3181 for (const RelocationRef &Reloc : Sections[SectIdx].relocations()) {
Rafael Espindola80291272014-10-08 15:28:58 +00003182 uint64_t RelocOffset;
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00003183 Reloc.getOffset(RelocOffset);
Rafael Espindola80291272014-10-08 15:28:58 +00003184 uint64_t SectionAddress = Sections[SectIdx].getAddress();
Owen Andersond9243c42011-10-17 21:37:35 +00003185 RelocOffset -= SectionAddress;
3186
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00003187 symbol_iterator RelocSym = Reloc.getSymbol();
Owen Andersond9243c42011-10-17 21:37:35 +00003188
Rafael Espindola806f0062013-06-05 01:33:53 +00003189 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003190 }
3191 array_pod_sort(Relocs.begin(), Relocs.end());
3192
Kevin Enderbybf246f52014-09-24 23:08:22 +00003193 // Create a map of symbol addresses to symbol names for use by
3194 // the SymbolizerSymbolLookUp() routine.
3195 SymbolAddressMap AddrMap;
Kevin Enderby6a221752015-03-17 17:10:57 +00003196 bool DisSymNameFound = false;
Kevin Enderbybf246f52014-09-24 23:08:22 +00003197 for (const SymbolRef &Symbol : MachOOF->symbols()) {
3198 SymbolRef::Type ST;
3199 Symbol.getType(ST);
3200 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
3201 ST == SymbolRef::ST_Other) {
3202 uint64_t Address;
3203 Symbol.getAddress(Address);
3204 StringRef SymName;
3205 Symbol.getName(SymName);
3206 AddrMap[Address] = SymName;
Kevin Enderby6a221752015-03-17 17:10:57 +00003207 if (!DisSymName.empty() && DisSymName == SymName)
3208 DisSymNameFound = true;
Kevin Enderbybf246f52014-09-24 23:08:22 +00003209 }
3210 }
Kevin Enderby6a221752015-03-17 17:10:57 +00003211 if (!DisSymName.empty() && DisSymNameFound == false) {
3212 outs() << "Can't find -dis-symname: " << DisSymName << "\n";
3213 return;
3214 }
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003215 // Set up the block of info used by the Symbolizer call backs.
Kevin Enderby8e29ec92015-03-17 22:26:11 +00003216 SymbolizerInfo.verbose = !NoSymbolicOperands;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003217 SymbolizerInfo.O = MachOOF;
3218 SymbolizerInfo.S = Sections[SectIdx];
Kevin Enderbybf246f52014-09-24 23:08:22 +00003219 SymbolizerInfo.AddrMap = &AddrMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003220 SymbolizerInfo.Sections = &Sections;
3221 SymbolizerInfo.class_name = nullptr;
3222 SymbolizerInfo.selector_name = nullptr;
3223 SymbolizerInfo.method = nullptr;
Kevin Enderby04bf6932014-10-28 23:39:46 +00003224 SymbolizerInfo.demangled_name = nullptr;
Kevin Enderby078be602014-10-23 19:53:12 +00003225 SymbolizerInfo.bindtable = nullptr;
Kevin Enderby10738222014-11-19 20:20:16 +00003226 SymbolizerInfo.adrp_addr = 0;
3227 SymbolizerInfo.adrp_inst = 0;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003228 // Same for the ThumbSymbolizer
Kevin Enderby8e29ec92015-03-17 22:26:11 +00003229 ThumbSymbolizerInfo.verbose = !NoSymbolicOperands;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003230 ThumbSymbolizerInfo.O = MachOOF;
3231 ThumbSymbolizerInfo.S = Sections[SectIdx];
3232 ThumbSymbolizerInfo.AddrMap = &AddrMap;
3233 ThumbSymbolizerInfo.Sections = &Sections;
3234 ThumbSymbolizerInfo.class_name = nullptr;
3235 ThumbSymbolizerInfo.selector_name = nullptr;
3236 ThumbSymbolizerInfo.method = nullptr;
3237 ThumbSymbolizerInfo.demangled_name = nullptr;
3238 ThumbSymbolizerInfo.bindtable = nullptr;
Kevin Enderby10738222014-11-19 20:20:16 +00003239 ThumbSymbolizerInfo.adrp_addr = 0;
3240 ThumbSymbolizerInfo.adrp_inst = 0;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003241
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003242 // Disassemble symbol by symbol.
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003243 for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
Owen Andersond9243c42011-10-17 21:37:35 +00003244 StringRef SymName;
3245 Symbols[SymIdx].getName(SymName);
3246
3247 SymbolRef::Type ST;
3248 Symbols[SymIdx].getType(ST);
3249 if (ST != SymbolRef::ST_Function)
3250 continue;
3251
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003252 // Make sure the symbol is defined in this section.
Rafael Espindola80291272014-10-08 15:28:58 +00003253 bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]);
Owen Andersond9243c42011-10-17 21:37:35 +00003254 if (!containsSym)
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003255 continue;
3256
Kevin Enderby6a221752015-03-17 17:10:57 +00003257 // If we are only disassembling one symbol see if this is that symbol.
3258 if (!DisSymName.empty() && DisSymName != SymName)
3259 continue;
3260
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003261 // Start at the address of the symbol relative to the section's address.
Owen Andersond9243c42011-10-17 21:37:35 +00003262 uint64_t Start = 0;
Rafael Espindola80291272014-10-08 15:28:58 +00003263 uint64_t SectionAddress = Sections[SectIdx].getAddress();
Danil Malyshevcbe72fc2011-11-29 17:40:10 +00003264 Symbols[SymIdx].getAddress(Start);
Cameron Zwarich54478a52012-02-03 05:42:17 +00003265 Start -= SectionAddress;
Owen Andersond9243c42011-10-17 21:37:35 +00003266
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003267 // Stop disassembling either at the beginning of the next symbol or at
3268 // the end of the section.
Kevin Enderbyedd58722012-05-15 18:57:14 +00003269 bool containsNextSym = false;
Owen Andersond9243c42011-10-17 21:37:35 +00003270 uint64_t NextSym = 0;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003271 uint64_t NextSymIdx = SymIdx + 1;
Owen Andersond9243c42011-10-17 21:37:35 +00003272 while (Symbols.size() > NextSymIdx) {
3273 SymbolRef::Type NextSymType;
3274 Symbols[NextSymIdx].getType(NextSymType);
3275 if (NextSymType == SymbolRef::ST_Function) {
Rafael Espindola80291272014-10-08 15:28:58 +00003276 containsNextSym =
3277 Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]);
Danil Malyshevcbe72fc2011-11-29 17:40:10 +00003278 Symbols[NextSymIdx].getAddress(NextSym);
Cameron Zwarich54478a52012-02-03 05:42:17 +00003279 NextSym -= SectionAddress;
Owen Andersond9243c42011-10-17 21:37:35 +00003280 break;
3281 }
3282 ++NextSymIdx;
3283 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003284
Rafael Espindola80291272014-10-08 15:28:58 +00003285 uint64_t SectSize = Sections[SectIdx].getSize();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003286 uint64_t End = containsNextSym ? NextSym : SectSize;
Owen Andersond9243c42011-10-17 21:37:35 +00003287 uint64_t Size;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003288
3289 symbolTableWorked = true;
Rafael Espindolabd604f22014-11-07 00:52:15 +00003290
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003291 DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl();
3292 bool isThumb =
3293 (MachOOF->getSymbolFlags(Symb) & SymbolRef::SF_Thumb) && ThumbTarget;
3294
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003295 outs() << SymName << ":\n";
3296 DILineInfo lastLine;
3297 for (uint64_t Index = Start; Index < End; Index += Size) {
3298 MCInst Inst;
Owen Andersond9243c42011-10-17 21:37:35 +00003299
Kevin Enderbybf246f52014-09-24 23:08:22 +00003300 uint64_t PC = SectAddress + Index;
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00003301 if (!NoLeadingAddr) {
3302 if (FullLeadingAddr) {
3303 if (MachOOF->is64Bit())
3304 outs() << format("%016" PRIx64, PC);
3305 else
3306 outs() << format("%08" PRIx64, PC);
3307 } else {
3308 outs() << format("%8" PRIx64 ":", PC);
3309 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00003310 }
3311 if (!NoShowRawInsn)
3312 outs() << "\t";
Kevin Enderby273ae012013-06-06 17:20:50 +00003313
3314 // Check the data in code table here to see if this is data not an
3315 // instruction to be disassembled.
3316 DiceTable Dice;
Kevin Enderbybf246f52014-09-24 23:08:22 +00003317 Dice.push_back(std::make_pair(PC, DiceRef()));
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003318 dice_table_iterator DTI =
3319 std::search(Dices.begin(), Dices.end(), Dice.begin(), Dice.end(),
3320 compareDiceTableEntries);
3321 if (DTI != Dices.end()) {
Kevin Enderby273ae012013-06-06 17:20:50 +00003322 uint16_t Length;
3323 DTI->second.getLength(Length);
Kevin Enderby273ae012013-06-06 17:20:50 +00003324 uint16_t Kind;
3325 DTI->second.getKind(Kind);
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003326 Size = DumpDataInCode(reinterpret_cast<const char *>(Bytes.data()) +
3327 Index,
3328 Length, Kind);
Kevin Enderby930fdc72014-11-06 19:00:13 +00003329 if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) &&
3330 (PC == (DTI->first + Length - 1)) && (Length & 1))
3331 Size++;
Kevin Enderby273ae012013-06-06 17:20:50 +00003332 continue;
3333 }
3334
Kevin Enderbybf246f52014-09-24 23:08:22 +00003335 SmallVector<char, 64> AnnotationsBytes;
3336 raw_svector_ostream Annotations(AnnotationsBytes);
3337
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003338 bool gotInst;
3339 if (isThumb)
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003340 gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index),
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003341 PC, DebugOut, Annotations);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003342 else
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003343 gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC,
Kevin Enderbybf246f52014-09-24 23:08:22 +00003344 DebugOut, Annotations);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003345 if (gotInst) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00003346 if (!NoShowRawInsn) {
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003347 DumpBytes(StringRef(
3348 reinterpret_cast<const char *>(Bytes.data()) + Index, Size));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003349 }
3350 formatted_raw_ostream FormattedOS(outs());
3351 Annotations.flush();
3352 StringRef AnnotationsStr = Annotations.str();
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003353 if (isThumb)
Kevin Enderbybf246f52014-09-24 23:08:22 +00003354 ThumbIP->printInst(&Inst, FormattedOS, AnnotationsStr);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003355 else
Kevin Enderbybf246f52014-09-24 23:08:22 +00003356 IP->printInst(&Inst, FormattedOS, AnnotationsStr);
3357 emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo);
Owen Andersond9243c42011-10-17 21:37:35 +00003358
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003359 // Print debug info.
3360 if (diContext) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003361 DILineInfo dli = diContext->getLineInfoForAddress(PC);
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003362 // Print valid line info if it changed.
Alexey Samsonovd0109992014-04-18 21:36:39 +00003363 if (dli != lastLine && dli.Line != 0)
3364 outs() << "\t## " << dli.FileName << ':' << dli.Line << ':'
3365 << dli.Column;
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003366 lastLine = dli;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003367 }
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003368 outs() << "\n";
3369 } else {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003370 unsigned int Arch = MachOOF->getArch();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003371 if (Arch == Triple::x86_64 || Arch == Triple::x86) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003372 outs() << format("\t.byte 0x%02x #bad opcode\n",
3373 *(Bytes.data() + Index) & 0xff);
3374 Size = 1; // skip exactly one illegible byte and move on.
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003375 } else if (Arch == Triple::aarch64) {
3376 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
3377 (*(Bytes.data() + Index + 1) & 0xff) << 8 |
3378 (*(Bytes.data() + Index + 2) & 0xff) << 16 |
3379 (*(Bytes.data() + Index + 3) & 0xff) << 24;
3380 outs() << format("\t.long\t0x%08x\n", opcode);
3381 Size = 4;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003382 } else {
3383 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
3384 if (Size == 0)
3385 Size = 1; // skip illegible bytes
3386 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003387 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003388 }
3389 }
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003390 if (!symbolTableWorked) {
Rafael Espindola80291272014-10-08 15:28:58 +00003391 // Reading the symbol table didn't work, disassemble the whole section.
3392 uint64_t SectAddress = Sections[SectIdx].getAddress();
3393 uint64_t SectSize = Sections[SectIdx].getSize();
Kevin Enderbybadd1002012-05-18 00:13:56 +00003394 uint64_t InstSize;
3395 for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
Bill Wendling4e68e062012-07-19 00:17:40 +00003396 MCInst Inst;
Kevin Enderbybadd1002012-05-18 00:13:56 +00003397
Kevin Enderbybf246f52014-09-24 23:08:22 +00003398 uint64_t PC = SectAddress + Index;
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003399 if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC,
3400 DebugOut, nulls())) {
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00003401 if (!NoLeadingAddr) {
3402 if (FullLeadingAddr) {
3403 if (MachOOF->is64Bit())
3404 outs() << format("%016" PRIx64, PC);
3405 else
3406 outs() << format("%08" PRIx64, PC);
3407 } else {
3408 outs() << format("%8" PRIx64 ":", PC);
3409 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00003410 }
3411 if (!NoShowRawInsn) {
3412 outs() << "\t";
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003413 DumpBytes(
3414 StringRef(reinterpret_cast<const char *>(Bytes.data()) + Index,
3415 InstSize));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003416 }
Bill Wendling4e68e062012-07-19 00:17:40 +00003417 IP->printInst(&Inst, outs(), "");
3418 outs() << "\n";
3419 } else {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003420 unsigned int Arch = MachOOF->getArch();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003421 if (Arch == Triple::x86_64 || Arch == Triple::x86) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003422 outs() << format("\t.byte 0x%02x #bad opcode\n",
3423 *(Bytes.data() + Index) & 0xff);
3424 InstSize = 1; // skip exactly one illegible byte and move on.
3425 } else {
3426 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
3427 if (InstSize == 0)
3428 InstSize = 1; // skip illegible bytes
3429 }
Bill Wendling4e68e062012-07-19 00:17:40 +00003430 }
Kevin Enderbybadd1002012-05-18 00:13:56 +00003431 }
3432 }
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00003433 // The TripleName's need to be reset if we are called again for a different
3434 // archtecture.
3435 TripleName = "";
3436 ThumbTripleName = "";
3437
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003438 if (SymbolizerInfo.method != nullptr)
3439 free(SymbolizerInfo.method);
Kevin Enderby04bf6932014-10-28 23:39:46 +00003440 if (SymbolizerInfo.demangled_name != nullptr)
3441 free(SymbolizerInfo.demangled_name);
Kevin Enderby078be602014-10-23 19:53:12 +00003442 if (SymbolizerInfo.bindtable != nullptr)
3443 delete SymbolizerInfo.bindtable;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003444 if (ThumbSymbolizerInfo.method != nullptr)
3445 free(ThumbSymbolizerInfo.method);
3446 if (ThumbSymbolizerInfo.demangled_name != nullptr)
3447 free(ThumbSymbolizerInfo.demangled_name);
3448 if (ThumbSymbolizerInfo.bindtable != nullptr)
3449 delete ThumbSymbolizerInfo.bindtable;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003450 }
3451}
Tim Northover4bd286a2014-08-01 13:07:19 +00003452
Tim Northover39c70bb2014-08-12 11:52:59 +00003453//===----------------------------------------------------------------------===//
3454// __compact_unwind section dumping
3455//===----------------------------------------------------------------------===//
3456
Tim Northover4bd286a2014-08-01 13:07:19 +00003457namespace {
Tim Northover39c70bb2014-08-12 11:52:59 +00003458
3459template <typename T> static uint64_t readNext(const char *&Buf) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003460 using llvm::support::little;
3461 using llvm::support::unaligned;
Tim Northover39c70bb2014-08-12 11:52:59 +00003462
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003463 uint64_t Val = support::endian::read<T, little, unaligned>(Buf);
3464 Buf += sizeof(T);
3465 return Val;
3466}
Tim Northover39c70bb2014-08-12 11:52:59 +00003467
Tim Northover4bd286a2014-08-01 13:07:19 +00003468struct CompactUnwindEntry {
3469 uint32_t OffsetInSection;
3470
3471 uint64_t FunctionAddr;
3472 uint32_t Length;
3473 uint32_t CompactEncoding;
3474 uint64_t PersonalityAddr;
3475 uint64_t LSDAAddr;
3476
3477 RelocationRef FunctionReloc;
3478 RelocationRef PersonalityReloc;
3479 RelocationRef LSDAReloc;
3480
3481 CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64)
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003482 : OffsetInSection(Offset) {
Tim Northover4bd286a2014-08-01 13:07:19 +00003483 if (Is64)
3484 read<uint64_t>(Contents.data() + Offset);
3485 else
3486 read<uint32_t>(Contents.data() + Offset);
3487 }
3488
3489private:
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003490 template <typename UIntPtr> void read(const char *Buf) {
Tim Northover4bd286a2014-08-01 13:07:19 +00003491 FunctionAddr = readNext<UIntPtr>(Buf);
3492 Length = readNext<uint32_t>(Buf);
3493 CompactEncoding = readNext<uint32_t>(Buf);
3494 PersonalityAddr = readNext<UIntPtr>(Buf);
3495 LSDAAddr = readNext<UIntPtr>(Buf);
3496 }
3497};
3498}
3499
3500/// Given a relocation from __compact_unwind, consisting of the RelocationRef
3501/// and data being relocated, determine the best base Name and Addend to use for
3502/// display purposes.
3503///
3504/// 1. An Extern relocation will directly reference a symbol (and the data is
3505/// then already an addend), so use that.
3506/// 2. Otherwise the data is an offset in the object file's layout; try to find
3507// a symbol before it in the same section, and use the offset from there.
3508/// 3. Finally, if all that fails, fall back to an offset from the start of the
3509/// referenced section.
3510static void findUnwindRelocNameAddend(const MachOObjectFile *Obj,
3511 std::map<uint64_t, SymbolRef> &Symbols,
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003512 const RelocationRef &Reloc, uint64_t Addr,
Tim Northover4bd286a2014-08-01 13:07:19 +00003513 StringRef &Name, uint64_t &Addend) {
3514 if (Reloc.getSymbol() != Obj->symbol_end()) {
3515 Reloc.getSymbol()->getName(Name);
3516 Addend = Addr;
3517 return;
3518 }
3519
3520 auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl());
3521 SectionRef RelocSection = Obj->getRelocationSection(RE);
3522
Rafael Espindola80291272014-10-08 15:28:58 +00003523 uint64_t SectionAddr = RelocSection.getAddress();
Tim Northover4bd286a2014-08-01 13:07:19 +00003524
3525 auto Sym = Symbols.upper_bound(Addr);
3526 if (Sym == Symbols.begin()) {
3527 // The first symbol in the object is after this reference, the best we can
3528 // do is section-relative notation.
3529 RelocSection.getName(Name);
3530 Addend = Addr - SectionAddr;
3531 return;
3532 }
3533
3534 // Go back one so that SymbolAddress <= Addr.
3535 --Sym;
3536
3537 section_iterator SymSection = Obj->section_end();
3538 Sym->second.getSection(SymSection);
3539 if (RelocSection == *SymSection) {
3540 // There's a valid symbol in the same section before this reference.
3541 Sym->second.getName(Name);
3542 Addend = Addr - Sym->first;
3543 return;
3544 }
3545
3546 // There is a symbol before this reference, but it's in a different
3547 // section. Probably not helpful to mention it, so use the section name.
3548 RelocSection.getName(Name);
3549 Addend = Addr - SectionAddr;
3550}
3551
3552static void printUnwindRelocDest(const MachOObjectFile *Obj,
3553 std::map<uint64_t, SymbolRef> &Symbols,
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003554 const RelocationRef &Reloc, uint64_t Addr) {
Tim Northover4bd286a2014-08-01 13:07:19 +00003555 StringRef Name;
3556 uint64_t Addend;
3557
Tim Northover0b0add52014-09-09 10:45:06 +00003558 if (!Reloc.getObjectFile())
3559 return;
3560
Tim Northover4bd286a2014-08-01 13:07:19 +00003561 findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend);
3562
3563 outs() << Name;
3564 if (Addend)
Tim Northover63a25622014-08-11 09:14:06 +00003565 outs() << " + " << format("0x%" PRIx64, Addend);
Tim Northover4bd286a2014-08-01 13:07:19 +00003566}
3567
3568static void
3569printMachOCompactUnwindSection(const MachOObjectFile *Obj,
3570 std::map<uint64_t, SymbolRef> &Symbols,
3571 const SectionRef &CompactUnwind) {
3572
3573 assert(Obj->isLittleEndian() &&
3574 "There should not be a big-endian .o with __compact_unwind");
3575
3576 bool Is64 = Obj->is64Bit();
3577 uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t);
3578 uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t);
3579
3580 StringRef Contents;
3581 CompactUnwind.getContents(Contents);
3582
3583 SmallVector<CompactUnwindEntry, 4> CompactUnwinds;
3584
3585 // First populate the initial raw offsets, encodings and so on from the entry.
3586 for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) {
3587 CompactUnwindEntry Entry(Contents.data(), Offset, Is64);
3588 CompactUnwinds.push_back(Entry);
3589 }
3590
3591 // Next we need to look at the relocations to find out what objects are
3592 // actually being referred to.
3593 for (const RelocationRef &Reloc : CompactUnwind.relocations()) {
3594 uint64_t RelocAddress;
3595 Reloc.getOffset(RelocAddress);
3596
3597 uint32_t EntryIdx = RelocAddress / EntrySize;
3598 uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize;
3599 CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx];
3600
3601 if (OffsetInEntry == 0)
3602 Entry.FunctionReloc = Reloc;
3603 else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t))
3604 Entry.PersonalityReloc = Reloc;
3605 else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t))
3606 Entry.LSDAReloc = Reloc;
3607 else
3608 llvm_unreachable("Unexpected relocation in __compact_unwind section");
3609 }
3610
3611 // Finally, we're ready to print the data we've gathered.
3612 outs() << "Contents of __compact_unwind section:\n";
3613 for (auto &Entry : CompactUnwinds) {
Tim Northover06af2602014-08-08 12:08:51 +00003614 outs() << " Entry at offset "
3615 << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n";
Tim Northover4bd286a2014-08-01 13:07:19 +00003616
3617 // 1. Start of the region this entry applies to.
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003618 outs() << " start: " << format("0x%" PRIx64,
3619 Entry.FunctionAddr) << ' ';
3620 printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr);
Tim Northover4bd286a2014-08-01 13:07:19 +00003621 outs() << '\n';
3622
3623 // 2. Length of the region this entry applies to.
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003624 outs() << " length: " << format("0x%" PRIx32, Entry.Length)
3625 << '\n';
Tim Northover4bd286a2014-08-01 13:07:19 +00003626 // 3. The 32-bit compact encoding.
3627 outs() << " compact encoding: "
Tim Northoverb911bf82014-08-08 12:00:09 +00003628 << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n';
Tim Northover4bd286a2014-08-01 13:07:19 +00003629
3630 // 4. The personality function, if present.
3631 if (Entry.PersonalityReloc.getObjectFile()) {
3632 outs() << " personality function: "
Tim Northoverb911bf82014-08-08 12:00:09 +00003633 << format("0x%" PRIx64, Entry.PersonalityAddr) << ' ';
Tim Northover4bd286a2014-08-01 13:07:19 +00003634 printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc,
3635 Entry.PersonalityAddr);
3636 outs() << '\n';
3637 }
3638
3639 // 5. This entry's language-specific data area.
3640 if (Entry.LSDAReloc.getObjectFile()) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003641 outs() << " LSDA: " << format("0x%" PRIx64,
3642 Entry.LSDAAddr) << ' ';
Tim Northover4bd286a2014-08-01 13:07:19 +00003643 printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr);
3644 outs() << '\n';
3645 }
3646 }
3647}
3648
Tim Northover39c70bb2014-08-12 11:52:59 +00003649//===----------------------------------------------------------------------===//
3650// __unwind_info section dumping
3651//===----------------------------------------------------------------------===//
3652
3653static void printRegularSecondLevelUnwindPage(const char *PageStart) {
3654 const char *Pos = PageStart;
3655 uint32_t Kind = readNext<uint32_t>(Pos);
3656 (void)Kind;
3657 assert(Kind == 2 && "kind for a regular 2nd level index should be 2");
3658
3659 uint16_t EntriesStart = readNext<uint16_t>(Pos);
3660 uint16_t NumEntries = readNext<uint16_t>(Pos);
3661
3662 Pos = PageStart + EntriesStart;
3663 for (unsigned i = 0; i < NumEntries; ++i) {
3664 uint32_t FunctionOffset = readNext<uint32_t>(Pos);
3665 uint32_t Encoding = readNext<uint32_t>(Pos);
3666
3667 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003668 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
3669 << ", "
3670 << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003671 }
3672}
3673
3674static void printCompressedSecondLevelUnwindPage(
3675 const char *PageStart, uint32_t FunctionBase,
3676 const SmallVectorImpl<uint32_t> &CommonEncodings) {
3677 const char *Pos = PageStart;
3678 uint32_t Kind = readNext<uint32_t>(Pos);
3679 (void)Kind;
3680 assert(Kind == 3 && "kind for a compressed 2nd level index should be 3");
3681
3682 uint16_t EntriesStart = readNext<uint16_t>(Pos);
3683 uint16_t NumEntries = readNext<uint16_t>(Pos);
3684
3685 uint16_t EncodingsStart = readNext<uint16_t>(Pos);
3686 readNext<uint16_t>(Pos);
Aaron Ballman80930af2014-08-14 13:53:19 +00003687 const auto *PageEncodings = reinterpret_cast<const support::ulittle32_t *>(
3688 PageStart + EncodingsStart);
Tim Northover39c70bb2014-08-12 11:52:59 +00003689
3690 Pos = PageStart + EntriesStart;
3691 for (unsigned i = 0; i < NumEntries; ++i) {
3692 uint32_t Entry = readNext<uint32_t>(Pos);
3693 uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff);
3694 uint32_t EncodingIdx = Entry >> 24;
3695
3696 uint32_t Encoding;
3697 if (EncodingIdx < CommonEncodings.size())
3698 Encoding = CommonEncodings[EncodingIdx];
3699 else
3700 Encoding = PageEncodings[EncodingIdx - CommonEncodings.size()];
3701
3702 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003703 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
3704 << ", "
3705 << "encoding[" << EncodingIdx
3706 << "]=" << format("0x%08" PRIx32, Encoding) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003707 }
3708}
3709
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003710static void printMachOUnwindInfoSection(const MachOObjectFile *Obj,
3711 std::map<uint64_t, SymbolRef> &Symbols,
3712 const SectionRef &UnwindInfo) {
Tim Northover39c70bb2014-08-12 11:52:59 +00003713
3714 assert(Obj->isLittleEndian() &&
3715 "There should not be a big-endian .o with __unwind_info");
3716
3717 outs() << "Contents of __unwind_info section:\n";
3718
3719 StringRef Contents;
3720 UnwindInfo.getContents(Contents);
3721 const char *Pos = Contents.data();
3722
3723 //===----------------------------------
3724 // Section header
3725 //===----------------------------------
3726
3727 uint32_t Version = readNext<uint32_t>(Pos);
3728 outs() << " Version: "
3729 << format("0x%" PRIx32, Version) << '\n';
3730 assert(Version == 1 && "only understand version 1");
3731
3732 uint32_t CommonEncodingsStart = readNext<uint32_t>(Pos);
3733 outs() << " Common encodings array section offset: "
3734 << format("0x%" PRIx32, CommonEncodingsStart) << '\n';
3735 uint32_t NumCommonEncodings = readNext<uint32_t>(Pos);
3736 outs() << " Number of common encodings in array: "
3737 << format("0x%" PRIx32, NumCommonEncodings) << '\n';
3738
3739 uint32_t PersonalitiesStart = readNext<uint32_t>(Pos);
3740 outs() << " Personality function array section offset: "
3741 << format("0x%" PRIx32, PersonalitiesStart) << '\n';
3742 uint32_t NumPersonalities = readNext<uint32_t>(Pos);
3743 outs() << " Number of personality functions in array: "
3744 << format("0x%" PRIx32, NumPersonalities) << '\n';
3745
3746 uint32_t IndicesStart = readNext<uint32_t>(Pos);
3747 outs() << " Index array section offset: "
3748 << format("0x%" PRIx32, IndicesStart) << '\n';
3749 uint32_t NumIndices = readNext<uint32_t>(Pos);
3750 outs() << " Number of indices in array: "
3751 << format("0x%" PRIx32, NumIndices) << '\n';
3752
3753 //===----------------------------------
3754 // A shared list of common encodings
3755 //===----------------------------------
3756
3757 // These occupy indices in the range [0, N] whenever an encoding is referenced
3758 // from a compressed 2nd level index table. In practice the linker only
3759 // creates ~128 of these, so that indices are available to embed encodings in
3760 // the 2nd level index.
3761
3762 SmallVector<uint32_t, 64> CommonEncodings;
3763 outs() << " Common encodings: (count = " << NumCommonEncodings << ")\n";
3764 Pos = Contents.data() + CommonEncodingsStart;
3765 for (unsigned i = 0; i < NumCommonEncodings; ++i) {
3766 uint32_t Encoding = readNext<uint32_t>(Pos);
3767 CommonEncodings.push_back(Encoding);
3768
3769 outs() << " encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding)
3770 << '\n';
3771 }
3772
Tim Northover39c70bb2014-08-12 11:52:59 +00003773 //===----------------------------------
3774 // Personality functions used in this executable
3775 //===----------------------------------
3776
3777 // There should be only a handful of these (one per source language,
3778 // roughly). Particularly since they only get 2 bits in the compact encoding.
3779
3780 outs() << " Personality functions: (count = " << NumPersonalities << ")\n";
3781 Pos = Contents.data() + PersonalitiesStart;
3782 for (unsigned i = 0; i < NumPersonalities; ++i) {
3783 uint32_t PersonalityFn = readNext<uint32_t>(Pos);
3784 outs() << " personality[" << i + 1
3785 << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n';
3786 }
3787
3788 //===----------------------------------
3789 // The level 1 index entries
3790 //===----------------------------------
3791
3792 // These specify an approximate place to start searching for the more detailed
3793 // information, sorted by PC.
3794
3795 struct IndexEntry {
3796 uint32_t FunctionOffset;
3797 uint32_t SecondLevelPageStart;
3798 uint32_t LSDAStart;
3799 };
3800
3801 SmallVector<IndexEntry, 4> IndexEntries;
3802
3803 outs() << " Top level indices: (count = " << NumIndices << ")\n";
3804 Pos = Contents.data() + IndicesStart;
3805 for (unsigned i = 0; i < NumIndices; ++i) {
3806 IndexEntry Entry;
3807
3808 Entry.FunctionOffset = readNext<uint32_t>(Pos);
3809 Entry.SecondLevelPageStart = readNext<uint32_t>(Pos);
3810 Entry.LSDAStart = readNext<uint32_t>(Pos);
3811 IndexEntries.push_back(Entry);
3812
3813 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003814 << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset)
3815 << ", "
Tim Northover39c70bb2014-08-12 11:52:59 +00003816 << "2nd level page offset="
3817 << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003818 << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003819 }
3820
Tim Northover39c70bb2014-08-12 11:52:59 +00003821 //===----------------------------------
3822 // Next come the LSDA tables
3823 //===----------------------------------
3824
3825 // The LSDA layout is rather implicit: it's a contiguous array of entries from
3826 // the first top-level index's LSDAOffset to the last (sentinel).
3827
3828 outs() << " LSDA descriptors:\n";
3829 Pos = Contents.data() + IndexEntries[0].LSDAStart;
3830 int NumLSDAs = (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) /
3831 (2 * sizeof(uint32_t));
3832 for (int i = 0; i < NumLSDAs; ++i) {
3833 uint32_t FunctionOffset = readNext<uint32_t>(Pos);
3834 uint32_t LSDAOffset = readNext<uint32_t>(Pos);
3835 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003836 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
3837 << ", "
3838 << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003839 }
3840
3841 //===----------------------------------
3842 // Finally, the 2nd level indices
3843 //===----------------------------------
3844
3845 // Generally these are 4K in size, and have 2 possible forms:
3846 // + Regular stores up to 511 entries with disparate encodings
3847 // + Compressed stores up to 1021 entries if few enough compact encoding
3848 // values are used.
3849 outs() << " Second level indices:\n";
3850 for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) {
3851 // The final sentinel top-level index has no associated 2nd level page
3852 if (IndexEntries[i].SecondLevelPageStart == 0)
3853 break;
3854
3855 outs() << " Second level index[" << i << "]: "
3856 << "offset in section="
3857 << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart)
3858 << ", "
3859 << "base function offset="
3860 << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n';
3861
3862 Pos = Contents.data() + IndexEntries[i].SecondLevelPageStart;
Aaron Ballman80930af2014-08-14 13:53:19 +00003863 uint32_t Kind = *reinterpret_cast<const support::ulittle32_t *>(Pos);
Tim Northover39c70bb2014-08-12 11:52:59 +00003864 if (Kind == 2)
3865 printRegularSecondLevelUnwindPage(Pos);
3866 else if (Kind == 3)
3867 printCompressedSecondLevelUnwindPage(Pos, IndexEntries[i].FunctionOffset,
3868 CommonEncodings);
3869 else
3870 llvm_unreachable("Do not know how to print this kind of 2nd level page");
Tim Northover39c70bb2014-08-12 11:52:59 +00003871 }
3872}
3873
Tim Northover4bd286a2014-08-01 13:07:19 +00003874void llvm::printMachOUnwindInfo(const MachOObjectFile *Obj) {
3875 std::map<uint64_t, SymbolRef> Symbols;
3876 for (const SymbolRef &SymRef : Obj->symbols()) {
3877 // Discard any undefined or absolute symbols. They're not going to take part
3878 // in the convenience lookup for unwind info and just take up resources.
3879 section_iterator Section = Obj->section_end();
3880 SymRef.getSection(Section);
3881 if (Section == Obj->section_end())
3882 continue;
3883
3884 uint64_t Addr;
3885 SymRef.getAddress(Addr);
3886 Symbols.insert(std::make_pair(Addr, SymRef));
3887 }
3888
3889 for (const SectionRef &Section : Obj->sections()) {
3890 StringRef SectName;
3891 Section.getName(SectName);
3892 if (SectName == "__compact_unwind")
3893 printMachOCompactUnwindSection(Obj, Symbols, Section);
3894 else if (SectName == "__unwind_info")
Tim Northover39c70bb2014-08-12 11:52:59 +00003895 printMachOUnwindInfoSection(Obj, Symbols, Section);
Tim Northover4bd286a2014-08-01 13:07:19 +00003896 else if (SectName == "__eh_frame")
3897 outs() << "llvm-objdump: warning: unhandled __eh_frame section\n";
Tim Northover4bd286a2014-08-01 13:07:19 +00003898 }
3899}
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003900
3901static void PrintMachHeader(uint32_t magic, uint32_t cputype,
3902 uint32_t cpusubtype, uint32_t filetype,
3903 uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags,
3904 bool verbose) {
3905 outs() << "Mach header\n";
3906 outs() << " magic cputype cpusubtype caps filetype ncmds "
3907 "sizeofcmds flags\n";
3908 if (verbose) {
3909 if (magic == MachO::MH_MAGIC)
3910 outs() << " MH_MAGIC";
3911 else if (magic == MachO::MH_MAGIC_64)
3912 outs() << "MH_MAGIC_64";
3913 else
3914 outs() << format(" 0x%08" PRIx32, magic);
3915 switch (cputype) {
3916 case MachO::CPU_TYPE_I386:
3917 outs() << " I386";
3918 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3919 case MachO::CPU_SUBTYPE_I386_ALL:
3920 outs() << " ALL";
3921 break;
3922 default:
3923 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3924 break;
3925 }
3926 break;
3927 case MachO::CPU_TYPE_X86_64:
3928 outs() << " X86_64";
Kevin Enderby131d1772015-01-09 19:22:37 +00003929 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3930 case MachO::CPU_SUBTYPE_X86_64_ALL:
3931 outs() << " ALL";
3932 break;
3933 case MachO::CPU_SUBTYPE_X86_64_H:
3934 outs() << " Haswell";
3935 break;
3936 default:
3937 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3938 break;
3939 }
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003940 break;
3941 case MachO::CPU_TYPE_ARM:
3942 outs() << " ARM";
3943 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3944 case MachO::CPU_SUBTYPE_ARM_ALL:
3945 outs() << " ALL";
3946 break;
3947 case MachO::CPU_SUBTYPE_ARM_V4T:
3948 outs() << " V4T";
3949 break;
3950 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
3951 outs() << " V5TEJ";
3952 break;
3953 case MachO::CPU_SUBTYPE_ARM_XSCALE:
3954 outs() << " XSCALE";
3955 break;
3956 case MachO::CPU_SUBTYPE_ARM_V6:
3957 outs() << " V6";
3958 break;
3959 case MachO::CPU_SUBTYPE_ARM_V6M:
3960 outs() << " V6M";
3961 break;
3962 case MachO::CPU_SUBTYPE_ARM_V7:
3963 outs() << " V7";
3964 break;
3965 case MachO::CPU_SUBTYPE_ARM_V7EM:
3966 outs() << " V7EM";
3967 break;
3968 case MachO::CPU_SUBTYPE_ARM_V7K:
3969 outs() << " V7K";
3970 break;
3971 case MachO::CPU_SUBTYPE_ARM_V7M:
3972 outs() << " V7M";
3973 break;
3974 case MachO::CPU_SUBTYPE_ARM_V7S:
3975 outs() << " V7S";
3976 break;
3977 default:
3978 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3979 break;
3980 }
3981 break;
3982 case MachO::CPU_TYPE_ARM64:
3983 outs() << " ARM64";
3984 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3985 case MachO::CPU_SUBTYPE_ARM64_ALL:
3986 outs() << " ALL";
3987 break;
3988 default:
3989 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3990 break;
3991 }
3992 break;
3993 case MachO::CPU_TYPE_POWERPC:
3994 outs() << " PPC";
3995 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3996 case MachO::CPU_SUBTYPE_POWERPC_ALL:
3997 outs() << " ALL";
3998 break;
3999 default:
4000 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
4001 break;
4002 }
4003 break;
4004 case MachO::CPU_TYPE_POWERPC64:
4005 outs() << " PPC64";
4006 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
4007 case MachO::CPU_SUBTYPE_POWERPC_ALL:
4008 outs() << " ALL";
4009 break;
4010 default:
4011 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
4012 break;
4013 }
4014 break;
4015 }
4016 if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004017 outs() << " LIB64";
Kevin Enderbyb76d3862014-08-22 20:35:18 +00004018 } else {
4019 outs() << format(" 0x%02" PRIx32,
4020 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
4021 }
4022 switch (filetype) {
4023 case MachO::MH_OBJECT:
4024 outs() << " OBJECT";
4025 break;
4026 case MachO::MH_EXECUTE:
4027 outs() << " EXECUTE";
4028 break;
4029 case MachO::MH_FVMLIB:
4030 outs() << " FVMLIB";
4031 break;
4032 case MachO::MH_CORE:
4033 outs() << " CORE";
4034 break;
4035 case MachO::MH_PRELOAD:
4036 outs() << " PRELOAD";
4037 break;
4038 case MachO::MH_DYLIB:
4039 outs() << " DYLIB";
4040 break;
4041 case MachO::MH_DYLIB_STUB:
4042 outs() << " DYLIB_STUB";
4043 break;
4044 case MachO::MH_DYLINKER:
4045 outs() << " DYLINKER";
4046 break;
4047 case MachO::MH_BUNDLE:
4048 outs() << " BUNDLE";
4049 break;
4050 case MachO::MH_DSYM:
4051 outs() << " DSYM";
4052 break;
4053 case MachO::MH_KEXT_BUNDLE:
4054 outs() << " KEXTBUNDLE";
4055 break;
4056 default:
4057 outs() << format(" %10u", filetype);
4058 break;
4059 }
4060 outs() << format(" %5u", ncmds);
4061 outs() << format(" %10u", sizeofcmds);
4062 uint32_t f = flags;
4063 if (f & MachO::MH_NOUNDEFS) {
4064 outs() << " NOUNDEFS";
4065 f &= ~MachO::MH_NOUNDEFS;
4066 }
4067 if (f & MachO::MH_INCRLINK) {
4068 outs() << " INCRLINK";
4069 f &= ~MachO::MH_INCRLINK;
4070 }
4071 if (f & MachO::MH_DYLDLINK) {
4072 outs() << " DYLDLINK";
4073 f &= ~MachO::MH_DYLDLINK;
4074 }
4075 if (f & MachO::MH_BINDATLOAD) {
4076 outs() << " BINDATLOAD";
4077 f &= ~MachO::MH_BINDATLOAD;
4078 }
4079 if (f & MachO::MH_PREBOUND) {
4080 outs() << " PREBOUND";
4081 f &= ~MachO::MH_PREBOUND;
4082 }
4083 if (f & MachO::MH_SPLIT_SEGS) {
4084 outs() << " SPLIT_SEGS";
4085 f &= ~MachO::MH_SPLIT_SEGS;
4086 }
4087 if (f & MachO::MH_LAZY_INIT) {
4088 outs() << " LAZY_INIT";
4089 f &= ~MachO::MH_LAZY_INIT;
4090 }
4091 if (f & MachO::MH_TWOLEVEL) {
4092 outs() << " TWOLEVEL";
4093 f &= ~MachO::MH_TWOLEVEL;
4094 }
4095 if (f & MachO::MH_FORCE_FLAT) {
4096 outs() << " FORCE_FLAT";
4097 f &= ~MachO::MH_FORCE_FLAT;
4098 }
4099 if (f & MachO::MH_NOMULTIDEFS) {
4100 outs() << " NOMULTIDEFS";
4101 f &= ~MachO::MH_NOMULTIDEFS;
4102 }
4103 if (f & MachO::MH_NOFIXPREBINDING) {
4104 outs() << " NOFIXPREBINDING";
4105 f &= ~MachO::MH_NOFIXPREBINDING;
4106 }
4107 if (f & MachO::MH_PREBINDABLE) {
4108 outs() << " PREBINDABLE";
4109 f &= ~MachO::MH_PREBINDABLE;
4110 }
4111 if (f & MachO::MH_ALLMODSBOUND) {
4112 outs() << " ALLMODSBOUND";
4113 f &= ~MachO::MH_ALLMODSBOUND;
4114 }
4115 if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) {
4116 outs() << " SUBSECTIONS_VIA_SYMBOLS";
4117 f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
4118 }
4119 if (f & MachO::MH_CANONICAL) {
4120 outs() << " CANONICAL";
4121 f &= ~MachO::MH_CANONICAL;
4122 }
4123 if (f & MachO::MH_WEAK_DEFINES) {
4124 outs() << " WEAK_DEFINES";
4125 f &= ~MachO::MH_WEAK_DEFINES;
4126 }
4127 if (f & MachO::MH_BINDS_TO_WEAK) {
4128 outs() << " BINDS_TO_WEAK";
4129 f &= ~MachO::MH_BINDS_TO_WEAK;
4130 }
4131 if (f & MachO::MH_ALLOW_STACK_EXECUTION) {
4132 outs() << " ALLOW_STACK_EXECUTION";
4133 f &= ~MachO::MH_ALLOW_STACK_EXECUTION;
4134 }
4135 if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) {
4136 outs() << " DEAD_STRIPPABLE_DYLIB";
4137 f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB;
4138 }
4139 if (f & MachO::MH_PIE) {
4140 outs() << " PIE";
4141 f &= ~MachO::MH_PIE;
4142 }
4143 if (f & MachO::MH_NO_REEXPORTED_DYLIBS) {
4144 outs() << " NO_REEXPORTED_DYLIBS";
4145 f &= ~MachO::MH_NO_REEXPORTED_DYLIBS;
4146 }
4147 if (f & MachO::MH_HAS_TLV_DESCRIPTORS) {
4148 outs() << " MH_HAS_TLV_DESCRIPTORS";
4149 f &= ~MachO::MH_HAS_TLV_DESCRIPTORS;
4150 }
4151 if (f & MachO::MH_NO_HEAP_EXECUTION) {
4152 outs() << " MH_NO_HEAP_EXECUTION";
4153 f &= ~MachO::MH_NO_HEAP_EXECUTION;
4154 }
4155 if (f & MachO::MH_APP_EXTENSION_SAFE) {
4156 outs() << " APP_EXTENSION_SAFE";
4157 f &= ~MachO::MH_APP_EXTENSION_SAFE;
4158 }
4159 if (f != 0 || flags == 0)
4160 outs() << format(" 0x%08" PRIx32, f);
4161 } else {
4162 outs() << format(" 0x%08" PRIx32, magic);
4163 outs() << format(" %7d", cputype);
4164 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
4165 outs() << format(" 0x%02" PRIx32,
4166 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
4167 outs() << format(" %10u", filetype);
4168 outs() << format(" %5u", ncmds);
4169 outs() << format(" %10u", sizeofcmds);
4170 outs() << format(" 0x%08" PRIx32, flags);
4171 }
4172 outs() << "\n";
4173}
4174
Kevin Enderby956366c2014-08-29 22:30:52 +00004175static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize,
4176 StringRef SegName, uint64_t vmaddr,
4177 uint64_t vmsize, uint64_t fileoff,
4178 uint64_t filesize, uint32_t maxprot,
4179 uint32_t initprot, uint32_t nsects,
4180 uint32_t flags, uint32_t object_size,
4181 bool verbose) {
4182 uint64_t expected_cmdsize;
4183 if (cmd == MachO::LC_SEGMENT) {
4184 outs() << " cmd LC_SEGMENT\n";
4185 expected_cmdsize = nsects;
4186 expected_cmdsize *= sizeof(struct MachO::section);
4187 expected_cmdsize += sizeof(struct MachO::segment_command);
4188 } else {
4189 outs() << " cmd LC_SEGMENT_64\n";
4190 expected_cmdsize = nsects;
4191 expected_cmdsize *= sizeof(struct MachO::section_64);
4192 expected_cmdsize += sizeof(struct MachO::segment_command_64);
4193 }
4194 outs() << " cmdsize " << cmdsize;
4195 if (cmdsize != expected_cmdsize)
4196 outs() << " Inconsistent size\n";
4197 else
4198 outs() << "\n";
4199 outs() << " segname " << SegName << "\n";
4200 if (cmd == MachO::LC_SEGMENT_64) {
4201 outs() << " vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n";
4202 outs() << " vmsize " << format("0x%016" PRIx64, vmsize) << "\n";
4203 } else {
Kevin Enderbyadb7c432014-12-16 18:58:11 +00004204 outs() << " vmaddr " << format("0x%08" PRIx64, vmaddr) << "\n";
4205 outs() << " vmsize " << format("0x%08" PRIx64, vmsize) << "\n";
Kevin Enderby956366c2014-08-29 22:30:52 +00004206 }
4207 outs() << " fileoff " << fileoff;
4208 if (fileoff > object_size)
4209 outs() << " (past end of file)\n";
4210 else
4211 outs() << "\n";
4212 outs() << " filesize " << filesize;
4213 if (fileoff + filesize > object_size)
4214 outs() << " (past end of file)\n";
4215 else
4216 outs() << "\n";
4217 if (verbose) {
4218 if ((maxprot &
4219 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
4220 MachO::VM_PROT_EXECUTE)) != 0)
4221 outs() << " maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n";
4222 else {
4223 if (maxprot & MachO::VM_PROT_READ)
4224 outs() << " maxprot r";
4225 else
4226 outs() << " maxprot -";
4227 if (maxprot & MachO::VM_PROT_WRITE)
4228 outs() << "w";
4229 else
4230 outs() << "-";
4231 if (maxprot & MachO::VM_PROT_EXECUTE)
4232 outs() << "x\n";
4233 else
4234 outs() << "-\n";
4235 }
4236 if ((initprot &
4237 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
4238 MachO::VM_PROT_EXECUTE)) != 0)
4239 outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n";
4240 else {
4241 if (initprot & MachO::VM_PROT_READ)
4242 outs() << " initprot r";
4243 else
4244 outs() << " initprot -";
4245 if (initprot & MachO::VM_PROT_WRITE)
4246 outs() << "w";
4247 else
4248 outs() << "-";
4249 if (initprot & MachO::VM_PROT_EXECUTE)
4250 outs() << "x\n";
4251 else
4252 outs() << "-\n";
4253 }
4254 } else {
4255 outs() << " maxprot " << format("0x%08" PRIx32, maxprot) << "\n";
4256 outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n";
4257 }
4258 outs() << " nsects " << nsects << "\n";
4259 if (verbose) {
4260 outs() << " flags";
4261 if (flags == 0)
4262 outs() << " (none)\n";
4263 else {
4264 if (flags & MachO::SG_HIGHVM) {
4265 outs() << " HIGHVM";
4266 flags &= ~MachO::SG_HIGHVM;
4267 }
4268 if (flags & MachO::SG_FVMLIB) {
4269 outs() << " FVMLIB";
4270 flags &= ~MachO::SG_FVMLIB;
4271 }
4272 if (flags & MachO::SG_NORELOC) {
4273 outs() << " NORELOC";
4274 flags &= ~MachO::SG_NORELOC;
4275 }
4276 if (flags & MachO::SG_PROTECTED_VERSION_1) {
4277 outs() << " PROTECTED_VERSION_1";
4278 flags &= ~MachO::SG_PROTECTED_VERSION_1;
4279 }
4280 if (flags)
4281 outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n";
4282 else
4283 outs() << "\n";
4284 }
4285 } else {
4286 outs() << " flags " << format("0x%" PRIx32, flags) << "\n";
4287 }
4288}
4289
4290static void PrintSection(const char *sectname, const char *segname,
4291 uint64_t addr, uint64_t size, uint32_t offset,
4292 uint32_t align, uint32_t reloff, uint32_t nreloc,
4293 uint32_t flags, uint32_t reserved1, uint32_t reserved2,
4294 uint32_t cmd, const char *sg_segname,
4295 uint32_t filetype, uint32_t object_size,
4296 bool verbose) {
4297 outs() << "Section\n";
4298 outs() << " sectname " << format("%.16s\n", sectname);
4299 outs() << " segname " << format("%.16s", segname);
4300 if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0)
4301 outs() << " (does not match segment)\n";
4302 else
4303 outs() << "\n";
4304 if (cmd == MachO::LC_SEGMENT_64) {
4305 outs() << " addr " << format("0x%016" PRIx64, addr) << "\n";
4306 outs() << " size " << format("0x%016" PRIx64, size);
4307 } else {
Kevin Enderby75594b62014-12-16 21:00:25 +00004308 outs() << " addr " << format("0x%08" PRIx64, addr) << "\n";
4309 outs() << " size " << format("0x%08" PRIx64, size);
Kevin Enderby956366c2014-08-29 22:30:52 +00004310 }
4311 if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size)
4312 outs() << " (past end of file)\n";
4313 else
4314 outs() << "\n";
4315 outs() << " offset " << offset;
4316 if (offset > object_size)
4317 outs() << " (past end of file)\n";
4318 else
4319 outs() << "\n";
4320 uint32_t align_shifted = 1 << align;
4321 outs() << " align 2^" << align << " (" << align_shifted << ")\n";
4322 outs() << " reloff " << reloff;
4323 if (reloff > object_size)
4324 outs() << " (past end of file)\n";
4325 else
4326 outs() << "\n";
4327 outs() << " nreloc " << nreloc;
4328 if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size)
4329 outs() << " (past end of file)\n";
4330 else
4331 outs() << "\n";
4332 uint32_t section_type = flags & MachO::SECTION_TYPE;
4333 if (verbose) {
4334 outs() << " type";
4335 if (section_type == MachO::S_REGULAR)
4336 outs() << " S_REGULAR\n";
4337 else if (section_type == MachO::S_ZEROFILL)
4338 outs() << " S_ZEROFILL\n";
4339 else if (section_type == MachO::S_CSTRING_LITERALS)
4340 outs() << " S_CSTRING_LITERALS\n";
4341 else if (section_type == MachO::S_4BYTE_LITERALS)
4342 outs() << " S_4BYTE_LITERALS\n";
4343 else if (section_type == MachO::S_8BYTE_LITERALS)
4344 outs() << " S_8BYTE_LITERALS\n";
4345 else if (section_type == MachO::S_16BYTE_LITERALS)
4346 outs() << " S_16BYTE_LITERALS\n";
4347 else if (section_type == MachO::S_LITERAL_POINTERS)
4348 outs() << " S_LITERAL_POINTERS\n";
4349 else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS)
4350 outs() << " S_NON_LAZY_SYMBOL_POINTERS\n";
4351 else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS)
4352 outs() << " S_LAZY_SYMBOL_POINTERS\n";
4353 else if (section_type == MachO::S_SYMBOL_STUBS)
4354 outs() << " S_SYMBOL_STUBS\n";
4355 else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS)
4356 outs() << " S_MOD_INIT_FUNC_POINTERS\n";
4357 else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS)
4358 outs() << " S_MOD_TERM_FUNC_POINTERS\n";
4359 else if (section_type == MachO::S_COALESCED)
4360 outs() << " S_COALESCED\n";
4361 else if (section_type == MachO::S_INTERPOSING)
4362 outs() << " S_INTERPOSING\n";
4363 else if (section_type == MachO::S_DTRACE_DOF)
4364 outs() << " S_DTRACE_DOF\n";
4365 else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS)
4366 outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n";
4367 else if (section_type == MachO::S_THREAD_LOCAL_REGULAR)
4368 outs() << " S_THREAD_LOCAL_REGULAR\n";
4369 else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL)
4370 outs() << " S_THREAD_LOCAL_ZEROFILL\n";
4371 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES)
4372 outs() << " S_THREAD_LOCAL_VARIABLES\n";
4373 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
4374 outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n";
4375 else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS)
4376 outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n";
4377 else
4378 outs() << format("0x%08" PRIx32, section_type) << "\n";
4379 outs() << "attributes";
4380 uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES;
4381 if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS)
4382 outs() << " PURE_INSTRUCTIONS";
4383 if (section_attributes & MachO::S_ATTR_NO_TOC)
4384 outs() << " NO_TOC";
4385 if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS)
4386 outs() << " STRIP_STATIC_SYMS";
4387 if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP)
4388 outs() << " NO_DEAD_STRIP";
4389 if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT)
4390 outs() << " LIVE_SUPPORT";
4391 if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE)
4392 outs() << " SELF_MODIFYING_CODE";
4393 if (section_attributes & MachO::S_ATTR_DEBUG)
4394 outs() << " DEBUG";
4395 if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS)
4396 outs() << " SOME_INSTRUCTIONS";
4397 if (section_attributes & MachO::S_ATTR_EXT_RELOC)
4398 outs() << " EXT_RELOC";
4399 if (section_attributes & MachO::S_ATTR_LOC_RELOC)
4400 outs() << " LOC_RELOC";
4401 if (section_attributes == 0)
4402 outs() << " (none)";
4403 outs() << "\n";
4404 } else
4405 outs() << " flags " << format("0x%08" PRIx32, flags) << "\n";
4406 outs() << " reserved1 " << reserved1;
4407 if (section_type == MachO::S_SYMBOL_STUBS ||
4408 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
4409 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
4410 section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
4411 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
4412 outs() << " (index into indirect symbol table)\n";
4413 else
4414 outs() << "\n";
4415 outs() << " reserved2 " << reserved2;
4416 if (section_type == MachO::S_SYMBOL_STUBS)
4417 outs() << " (size of stubs)\n";
4418 else
4419 outs() << "\n";
4420}
4421
David Majnemer73cc6ff2014-11-13 19:48:56 +00004422static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit,
Kevin Enderby956366c2014-08-29 22:30:52 +00004423 uint32_t object_size) {
4424 outs() << " cmd LC_SYMTAB\n";
4425 outs() << " cmdsize " << st.cmdsize;
4426 if (st.cmdsize != sizeof(struct MachO::symtab_command))
4427 outs() << " Incorrect size\n";
4428 else
4429 outs() << "\n";
4430 outs() << " symoff " << st.symoff;
4431 if (st.symoff > object_size)
4432 outs() << " (past end of file)\n";
4433 else
4434 outs() << "\n";
4435 outs() << " nsyms " << st.nsyms;
4436 uint64_t big_size;
David Majnemer73cc6ff2014-11-13 19:48:56 +00004437 if (Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00004438 big_size = st.nsyms;
4439 big_size *= sizeof(struct MachO::nlist_64);
4440 big_size += st.symoff;
4441 if (big_size > object_size)
4442 outs() << " (past end of file)\n";
4443 else
4444 outs() << "\n";
4445 } else {
4446 big_size = st.nsyms;
4447 big_size *= sizeof(struct MachO::nlist);
4448 big_size += st.symoff;
4449 if (big_size > object_size)
4450 outs() << " (past end of file)\n";
4451 else
4452 outs() << "\n";
4453 }
4454 outs() << " stroff " << st.stroff;
4455 if (st.stroff > object_size)
4456 outs() << " (past end of file)\n";
4457 else
4458 outs() << "\n";
4459 outs() << " strsize " << st.strsize;
4460 big_size = st.stroff;
4461 big_size += st.strsize;
4462 if (big_size > object_size)
4463 outs() << " (past end of file)\n";
4464 else
4465 outs() << "\n";
4466}
4467
4468static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,
4469 uint32_t nsyms, uint32_t object_size,
David Majnemer73cc6ff2014-11-13 19:48:56 +00004470 bool Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00004471 outs() << " cmd LC_DYSYMTAB\n";
4472 outs() << " cmdsize " << dyst.cmdsize;
4473 if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command))
4474 outs() << " Incorrect size\n";
4475 else
4476 outs() << "\n";
4477 outs() << " ilocalsym " << dyst.ilocalsym;
4478 if (dyst.ilocalsym > nsyms)
4479 outs() << " (greater than the number of symbols)\n";
4480 else
4481 outs() << "\n";
4482 outs() << " nlocalsym " << dyst.nlocalsym;
4483 uint64_t big_size;
4484 big_size = dyst.ilocalsym;
4485 big_size += dyst.nlocalsym;
4486 if (big_size > nsyms)
4487 outs() << " (past the end of the symbol table)\n";
4488 else
4489 outs() << "\n";
4490 outs() << " iextdefsym " << dyst.iextdefsym;
4491 if (dyst.iextdefsym > nsyms)
4492 outs() << " (greater than the number of symbols)\n";
4493 else
4494 outs() << "\n";
4495 outs() << " nextdefsym " << dyst.nextdefsym;
4496 big_size = dyst.iextdefsym;
4497 big_size += dyst.nextdefsym;
4498 if (big_size > nsyms)
4499 outs() << " (past the end of the symbol table)\n";
4500 else
4501 outs() << "\n";
4502 outs() << " iundefsym " << dyst.iundefsym;
4503 if (dyst.iundefsym > nsyms)
4504 outs() << " (greater than the number of symbols)\n";
4505 else
4506 outs() << "\n";
4507 outs() << " nundefsym " << dyst.nundefsym;
4508 big_size = dyst.iundefsym;
4509 big_size += dyst.nundefsym;
4510 if (big_size > nsyms)
4511 outs() << " (past the end of the symbol table)\n";
4512 else
4513 outs() << "\n";
4514 outs() << " tocoff " << dyst.tocoff;
4515 if (dyst.tocoff > object_size)
4516 outs() << " (past end of file)\n";
4517 else
4518 outs() << "\n";
4519 outs() << " ntoc " << dyst.ntoc;
4520 big_size = dyst.ntoc;
4521 big_size *= sizeof(struct MachO::dylib_table_of_contents);
4522 big_size += dyst.tocoff;
4523 if (big_size > object_size)
4524 outs() << " (past end of file)\n";
4525 else
4526 outs() << "\n";
4527 outs() << " modtaboff " << dyst.modtaboff;
4528 if (dyst.modtaboff > object_size)
4529 outs() << " (past end of file)\n";
4530 else
4531 outs() << "\n";
4532 outs() << " nmodtab " << dyst.nmodtab;
4533 uint64_t modtabend;
David Majnemer73cc6ff2014-11-13 19:48:56 +00004534 if (Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00004535 modtabend = dyst.nmodtab;
4536 modtabend *= sizeof(struct MachO::dylib_module_64);
4537 modtabend += dyst.modtaboff;
4538 } else {
4539 modtabend = dyst.nmodtab;
4540 modtabend *= sizeof(struct MachO::dylib_module);
4541 modtabend += dyst.modtaboff;
4542 }
4543 if (modtabend > object_size)
4544 outs() << " (past end of file)\n";
4545 else
4546 outs() << "\n";
4547 outs() << " extrefsymoff " << dyst.extrefsymoff;
4548 if (dyst.extrefsymoff > object_size)
4549 outs() << " (past end of file)\n";
4550 else
4551 outs() << "\n";
4552 outs() << " nextrefsyms " << dyst.nextrefsyms;
4553 big_size = dyst.nextrefsyms;
4554 big_size *= sizeof(struct MachO::dylib_reference);
4555 big_size += dyst.extrefsymoff;
4556 if (big_size > object_size)
4557 outs() << " (past end of file)\n";
4558 else
4559 outs() << "\n";
4560 outs() << " indirectsymoff " << dyst.indirectsymoff;
4561 if (dyst.indirectsymoff > object_size)
4562 outs() << " (past end of file)\n";
4563 else
4564 outs() << "\n";
4565 outs() << " nindirectsyms " << dyst.nindirectsyms;
4566 big_size = dyst.nindirectsyms;
4567 big_size *= sizeof(uint32_t);
4568 big_size += dyst.indirectsymoff;
4569 if (big_size > object_size)
4570 outs() << " (past end of file)\n";
4571 else
4572 outs() << "\n";
4573 outs() << " extreloff " << dyst.extreloff;
4574 if (dyst.extreloff > object_size)
4575 outs() << " (past end of file)\n";
4576 else
4577 outs() << "\n";
4578 outs() << " nextrel " << dyst.nextrel;
4579 big_size = dyst.nextrel;
4580 big_size *= sizeof(struct MachO::relocation_info);
4581 big_size += dyst.extreloff;
4582 if (big_size > object_size)
4583 outs() << " (past end of file)\n";
4584 else
4585 outs() << "\n";
4586 outs() << " locreloff " << dyst.locreloff;
4587 if (dyst.locreloff > object_size)
4588 outs() << " (past end of file)\n";
4589 else
4590 outs() << "\n";
4591 outs() << " nlocrel " << dyst.nlocrel;
4592 big_size = dyst.nlocrel;
4593 big_size *= sizeof(struct MachO::relocation_info);
4594 big_size += dyst.locreloff;
4595 if (big_size > object_size)
4596 outs() << " (past end of file)\n";
4597 else
4598 outs() << "\n";
4599}
4600
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004601static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,
4602 uint32_t object_size) {
4603 if (dc.cmd == MachO::LC_DYLD_INFO)
4604 outs() << " cmd LC_DYLD_INFO\n";
4605 else
4606 outs() << " cmd LC_DYLD_INFO_ONLY\n";
4607 outs() << " cmdsize " << dc.cmdsize;
4608 if (dc.cmdsize != sizeof(struct MachO::dyld_info_command))
4609 outs() << " Incorrect size\n";
4610 else
4611 outs() << "\n";
4612 outs() << " rebase_off " << dc.rebase_off;
4613 if (dc.rebase_off > object_size)
4614 outs() << " (past end of file)\n";
4615 else
4616 outs() << "\n";
4617 outs() << " rebase_size " << dc.rebase_size;
4618 uint64_t big_size;
4619 big_size = dc.rebase_off;
4620 big_size += dc.rebase_size;
4621 if (big_size > object_size)
4622 outs() << " (past end of file)\n";
4623 else
4624 outs() << "\n";
4625 outs() << " bind_off " << dc.bind_off;
4626 if (dc.bind_off > object_size)
4627 outs() << " (past end of file)\n";
4628 else
4629 outs() << "\n";
4630 outs() << " bind_size " << dc.bind_size;
4631 big_size = dc.bind_off;
4632 big_size += dc.bind_size;
4633 if (big_size > object_size)
4634 outs() << " (past end of file)\n";
4635 else
4636 outs() << "\n";
4637 outs() << " weak_bind_off " << dc.weak_bind_off;
4638 if (dc.weak_bind_off > object_size)
4639 outs() << " (past end of file)\n";
4640 else
4641 outs() << "\n";
4642 outs() << " weak_bind_size " << dc.weak_bind_size;
4643 big_size = dc.weak_bind_off;
4644 big_size += dc.weak_bind_size;
4645 if (big_size > object_size)
4646 outs() << " (past end of file)\n";
4647 else
4648 outs() << "\n";
4649 outs() << " lazy_bind_off " << dc.lazy_bind_off;
4650 if (dc.lazy_bind_off > object_size)
4651 outs() << " (past end of file)\n";
4652 else
4653 outs() << "\n";
4654 outs() << " lazy_bind_size " << dc.lazy_bind_size;
4655 big_size = dc.lazy_bind_off;
4656 big_size += dc.lazy_bind_size;
4657 if (big_size > object_size)
4658 outs() << " (past end of file)\n";
4659 else
4660 outs() << "\n";
4661 outs() << " export_off " << dc.export_off;
4662 if (dc.export_off > object_size)
4663 outs() << " (past end of file)\n";
4664 else
4665 outs() << "\n";
4666 outs() << " export_size " << dc.export_size;
4667 big_size = dc.export_off;
4668 big_size += dc.export_size;
4669 if (big_size > object_size)
4670 outs() << " (past end of file)\n";
4671 else
4672 outs() << "\n";
4673}
4674
4675static void PrintDyldLoadCommand(MachO::dylinker_command dyld,
4676 const char *Ptr) {
4677 if (dyld.cmd == MachO::LC_ID_DYLINKER)
4678 outs() << " cmd LC_ID_DYLINKER\n";
4679 else if (dyld.cmd == MachO::LC_LOAD_DYLINKER)
4680 outs() << " cmd LC_LOAD_DYLINKER\n";
4681 else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT)
4682 outs() << " cmd LC_DYLD_ENVIRONMENT\n";
4683 else
4684 outs() << " cmd ?(" << dyld.cmd << ")\n";
4685 outs() << " cmdsize " << dyld.cmdsize;
4686 if (dyld.cmdsize < sizeof(struct MachO::dylinker_command))
4687 outs() << " Incorrect size\n";
4688 else
4689 outs() << "\n";
4690 if (dyld.name >= dyld.cmdsize)
4691 outs() << " name ?(bad offset " << dyld.name << ")\n";
4692 else {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004693 const char *P = (const char *)(Ptr) + dyld.name;
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004694 outs() << " name " << P << " (offset " << dyld.name << ")\n";
4695 }
4696}
4697
4698static void PrintUuidLoadCommand(MachO::uuid_command uuid) {
4699 outs() << " cmd LC_UUID\n";
4700 outs() << " cmdsize " << uuid.cmdsize;
4701 if (uuid.cmdsize != sizeof(struct MachO::uuid_command))
4702 outs() << " Incorrect size\n";
4703 else
4704 outs() << "\n";
4705 outs() << " uuid ";
4706 outs() << format("%02" PRIX32, uuid.uuid[0]);
4707 outs() << format("%02" PRIX32, uuid.uuid[1]);
4708 outs() << format("%02" PRIX32, uuid.uuid[2]);
4709 outs() << format("%02" PRIX32, uuid.uuid[3]);
4710 outs() << "-";
4711 outs() << format("%02" PRIX32, uuid.uuid[4]);
4712 outs() << format("%02" PRIX32, uuid.uuid[5]);
4713 outs() << "-";
4714 outs() << format("%02" PRIX32, uuid.uuid[6]);
4715 outs() << format("%02" PRIX32, uuid.uuid[7]);
4716 outs() << "-";
4717 outs() << format("%02" PRIX32, uuid.uuid[8]);
4718 outs() << format("%02" PRIX32, uuid.uuid[9]);
4719 outs() << "-";
4720 outs() << format("%02" PRIX32, uuid.uuid[10]);
4721 outs() << format("%02" PRIX32, uuid.uuid[11]);
4722 outs() << format("%02" PRIX32, uuid.uuid[12]);
4723 outs() << format("%02" PRIX32, uuid.uuid[13]);
4724 outs() << format("%02" PRIX32, uuid.uuid[14]);
4725 outs() << format("%02" PRIX32, uuid.uuid[15]);
4726 outs() << "\n";
4727}
4728
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004729static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) {
Jean-Daniel Dupas00cc1f52014-12-04 07:37:02 +00004730 outs() << " cmd LC_RPATH\n";
4731 outs() << " cmdsize " << rpath.cmdsize;
4732 if (rpath.cmdsize < sizeof(struct MachO::rpath_command))
4733 outs() << " Incorrect size\n";
4734 else
4735 outs() << "\n";
4736 if (rpath.path >= rpath.cmdsize)
4737 outs() << " path ?(bad offset " << rpath.path << ")\n";
4738 else {
4739 const char *P = (const char *)(Ptr) + rpath.path;
4740 outs() << " path " << P << " (offset " << rpath.path << ")\n";
4741 }
4742}
4743
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004744static void PrintVersionMinLoadCommand(MachO::version_min_command vd) {
4745 if (vd.cmd == MachO::LC_VERSION_MIN_MACOSX)
4746 outs() << " cmd LC_VERSION_MIN_MACOSX\n";
4747 else if (vd.cmd == MachO::LC_VERSION_MIN_IPHONEOS)
4748 outs() << " cmd LC_VERSION_MIN_IPHONEOS\n";
4749 else
4750 outs() << " cmd " << vd.cmd << " (?)\n";
4751 outs() << " cmdsize " << vd.cmdsize;
4752 if (vd.cmdsize != sizeof(struct MachO::version_min_command))
4753 outs() << " Incorrect size\n";
4754 else
4755 outs() << "\n";
4756 outs() << " version " << ((vd.version >> 16) & 0xffff) << "."
4757 << ((vd.version >> 8) & 0xff);
4758 if ((vd.version & 0xff) != 0)
4759 outs() << "." << (vd.version & 0xff);
4760 outs() << "\n";
4761 if (vd.sdk == 0)
Kevin Enderby57538292014-12-17 01:01:30 +00004762 outs() << " sdk n/a";
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004763 else {
4764 outs() << " sdk " << ((vd.sdk >> 16) & 0xffff) << "."
4765 << ((vd.sdk >> 8) & 0xff);
4766 }
4767 if ((vd.sdk & 0xff) != 0)
4768 outs() << "." << (vd.sdk & 0xff);
4769 outs() << "\n";
4770}
4771
4772static void PrintSourceVersionCommand(MachO::source_version_command sd) {
4773 outs() << " cmd LC_SOURCE_VERSION\n";
4774 outs() << " cmdsize " << sd.cmdsize;
4775 if (sd.cmdsize != sizeof(struct MachO::source_version_command))
4776 outs() << " Incorrect size\n";
4777 else
4778 outs() << "\n";
4779 uint64_t a = (sd.version >> 40) & 0xffffff;
4780 uint64_t b = (sd.version >> 30) & 0x3ff;
4781 uint64_t c = (sd.version >> 20) & 0x3ff;
4782 uint64_t d = (sd.version >> 10) & 0x3ff;
4783 uint64_t e = sd.version & 0x3ff;
4784 outs() << " version " << a << "." << b;
4785 if (e != 0)
4786 outs() << "." << c << "." << d << "." << e;
4787 else if (d != 0)
4788 outs() << "." << c << "." << d;
4789 else if (c != 0)
4790 outs() << "." << c;
4791 outs() << "\n";
4792}
4793
4794static void PrintEntryPointCommand(MachO::entry_point_command ep) {
4795 outs() << " cmd LC_MAIN\n";
4796 outs() << " cmdsize " << ep.cmdsize;
4797 if (ep.cmdsize != sizeof(struct MachO::entry_point_command))
4798 outs() << " Incorrect size\n";
4799 else
4800 outs() << "\n";
4801 outs() << " entryoff " << ep.entryoff << "\n";
4802 outs() << " stacksize " << ep.stacksize << "\n";
4803}
4804
Kevin Enderby0804f4672014-12-16 23:25:52 +00004805static void PrintEncryptionInfoCommand(MachO::encryption_info_command ec,
4806 uint32_t object_size) {
4807 outs() << " cmd LC_ENCRYPTION_INFO\n";
4808 outs() << " cmdsize " << ec.cmdsize;
4809 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command))
4810 outs() << " Incorrect size\n";
4811 else
4812 outs() << "\n";
4813 outs() << " cryptoff " << ec.cryptoff;
4814 if (ec.cryptoff > object_size)
4815 outs() << " (past end of file)\n";
4816 else
4817 outs() << "\n";
4818 outs() << " cryptsize " << ec.cryptsize;
4819 if (ec.cryptsize > object_size)
4820 outs() << " (past end of file)\n";
4821 else
4822 outs() << "\n";
4823 outs() << " cryptid " << ec.cryptid << "\n";
4824}
4825
Kevin Enderby57538292014-12-17 01:01:30 +00004826static void PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004827 uint32_t object_size) {
Kevin Enderby57538292014-12-17 01:01:30 +00004828 outs() << " cmd LC_ENCRYPTION_INFO_64\n";
4829 outs() << " cmdsize " << ec.cmdsize;
4830 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command_64))
4831 outs() << " Incorrect size\n";
4832 else
4833 outs() << "\n";
4834 outs() << " cryptoff " << ec.cryptoff;
4835 if (ec.cryptoff > object_size)
4836 outs() << " (past end of file)\n";
4837 else
4838 outs() << "\n";
4839 outs() << " cryptsize " << ec.cryptsize;
4840 if (ec.cryptsize > object_size)
4841 outs() << " (past end of file)\n";
4842 else
4843 outs() << "\n";
4844 outs() << " cryptid " << ec.cryptid << "\n";
4845 outs() << " pad " << ec.pad << "\n";
4846}
4847
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00004848static void PrintLinkerOptionCommand(MachO::linker_option_command lo,
4849 const char *Ptr) {
4850 outs() << " cmd LC_LINKER_OPTION\n";
4851 outs() << " cmdsize " << lo.cmdsize;
4852 if (lo.cmdsize < sizeof(struct MachO::linker_option_command))
4853 outs() << " Incorrect size\n";
4854 else
4855 outs() << "\n";
4856 outs() << " count " << lo.count << "\n";
4857 const char *string = Ptr + sizeof(struct MachO::linker_option_command);
4858 uint32_t left = lo.cmdsize - sizeof(struct MachO::linker_option_command);
4859 uint32_t i = 0;
4860 while (left > 0) {
4861 while (*string == '\0' && left > 0) {
4862 string++;
4863 left--;
4864 }
4865 if (left > 0) {
4866 i++;
4867 outs() << " string #" << i << " " << format("%.*s\n", left, string);
David Majnemerd4449ed2014-12-20 08:24:43 +00004868 uint32_t NullPos = StringRef(string, left).find('\0');
4869 uint32_t len = std::min(NullPos, left) + 1;
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00004870 string += len;
4871 left -= len;
4872 }
4873 }
4874 if (lo.count != i)
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004875 outs() << " count " << lo.count << " does not match number of strings "
4876 << i << "\n";
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00004877}
4878
Kevin Enderbyb4b79312014-12-18 19:24:35 +00004879static void PrintSubFrameworkCommand(MachO::sub_framework_command sub,
4880 const char *Ptr) {
4881 outs() << " cmd LC_SUB_FRAMEWORK\n";
4882 outs() << " cmdsize " << sub.cmdsize;
4883 if (sub.cmdsize < sizeof(struct MachO::sub_framework_command))
4884 outs() << " Incorrect size\n";
4885 else
4886 outs() << "\n";
4887 if (sub.umbrella < sub.cmdsize) {
4888 const char *P = Ptr + sub.umbrella;
4889 outs() << " umbrella " << P << " (offset " << sub.umbrella << ")\n";
4890 } else {
4891 outs() << " umbrella ?(bad offset " << sub.umbrella << ")\n";
4892 }
4893}
4894
Kevin Enderbya2bd8d92014-12-18 23:13:26 +00004895static void PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,
4896 const char *Ptr) {
4897 outs() << " cmd LC_SUB_UMBRELLA\n";
4898 outs() << " cmdsize " << sub.cmdsize;
4899 if (sub.cmdsize < sizeof(struct MachO::sub_umbrella_command))
4900 outs() << " Incorrect size\n";
4901 else
4902 outs() << "\n";
4903 if (sub.sub_umbrella < sub.cmdsize) {
4904 const char *P = Ptr + sub.sub_umbrella;
4905 outs() << " sub_umbrella " << P << " (offset " << sub.sub_umbrella << ")\n";
4906 } else {
4907 outs() << " sub_umbrella ?(bad offset " << sub.sub_umbrella << ")\n";
4908 }
4909}
4910
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00004911static void PrintSubLibraryCommand(MachO::sub_library_command sub,
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004912 const char *Ptr) {
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00004913 outs() << " cmd LC_SUB_LIBRARY\n";
4914 outs() << " cmdsize " << sub.cmdsize;
4915 if (sub.cmdsize < sizeof(struct MachO::sub_library_command))
4916 outs() << " Incorrect size\n";
4917 else
4918 outs() << "\n";
4919 if (sub.sub_library < sub.cmdsize) {
4920 const char *P = Ptr + sub.sub_library;
4921 outs() << " sub_library " << P << " (offset " << sub.sub_library << ")\n";
4922 } else {
4923 outs() << " sub_library ?(bad offset " << sub.sub_library << ")\n";
4924 }
4925}
4926
Kevin Enderby186eac32014-12-19 21:06:24 +00004927static void PrintSubClientCommand(MachO::sub_client_command sub,
4928 const char *Ptr) {
4929 outs() << " cmd LC_SUB_CLIENT\n";
4930 outs() << " cmdsize " << sub.cmdsize;
4931 if (sub.cmdsize < sizeof(struct MachO::sub_client_command))
4932 outs() << " Incorrect size\n";
4933 else
4934 outs() << "\n";
4935 if (sub.client < sub.cmdsize) {
4936 const char *P = Ptr + sub.client;
4937 outs() << " client " << P << " (offset " << sub.client << ")\n";
4938 } else {
4939 outs() << " client ?(bad offset " << sub.client << ")\n";
4940 }
4941}
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00004942
Kevin Enderby52e4ce42014-12-19 22:25:22 +00004943static void PrintRoutinesCommand(MachO::routines_command r) {
4944 outs() << " cmd LC_ROUTINES\n";
4945 outs() << " cmdsize " << r.cmdsize;
4946 if (r.cmdsize != sizeof(struct MachO::routines_command))
4947 outs() << " Incorrect size\n";
4948 else
4949 outs() << "\n";
4950 outs() << " init_address " << format("0x%08" PRIx32, r.init_address) << "\n";
4951 outs() << " init_module " << r.init_module << "\n";
4952 outs() << " reserved1 " << r.reserved1 << "\n";
4953 outs() << " reserved2 " << r.reserved2 << "\n";
4954 outs() << " reserved3 " << r.reserved3 << "\n";
4955 outs() << " reserved4 " << r.reserved4 << "\n";
4956 outs() << " reserved5 " << r.reserved5 << "\n";
4957 outs() << " reserved6 " << r.reserved6 << "\n";
4958}
4959
4960static void PrintRoutinesCommand64(MachO::routines_command_64 r) {
4961 outs() << " cmd LC_ROUTINES_64\n";
4962 outs() << " cmdsize " << r.cmdsize;
4963 if (r.cmdsize != sizeof(struct MachO::routines_command_64))
4964 outs() << " Incorrect size\n";
4965 else
4966 outs() << "\n";
4967 outs() << " init_address " << format("0x%016" PRIx64, r.init_address) << "\n";
4968 outs() << " init_module " << r.init_module << "\n";
4969 outs() << " reserved1 " << r.reserved1 << "\n";
4970 outs() << " reserved2 " << r.reserved2 << "\n";
4971 outs() << " reserved3 " << r.reserved3 << "\n";
4972 outs() << " reserved4 " << r.reserved4 << "\n";
4973 outs() << " reserved5 " << r.reserved5 << "\n";
4974 outs() << " reserved6 " << r.reserved6 << "\n";
4975}
4976
Kevin Enderby48ef5342014-12-23 22:56:39 +00004977static void Print_x86_thread_state64_t(MachO::x86_thread_state64_t &cpu64) {
4978 outs() << " rax " << format("0x%016" PRIx64, cpu64.rax);
4979 outs() << " rbx " << format("0x%016" PRIx64, cpu64.rbx);
4980 outs() << " rcx " << format("0x%016" PRIx64, cpu64.rcx) << "\n";
4981 outs() << " rdx " << format("0x%016" PRIx64, cpu64.rdx);
4982 outs() << " rdi " << format("0x%016" PRIx64, cpu64.rdi);
4983 outs() << " rsi " << format("0x%016" PRIx64, cpu64.rsi) << "\n";
4984 outs() << " rbp " << format("0x%016" PRIx64, cpu64.rbp);
4985 outs() << " rsp " << format("0x%016" PRIx64, cpu64.rsp);
4986 outs() << " r8 " << format("0x%016" PRIx64, cpu64.r8) << "\n";
4987 outs() << " r9 " << format("0x%016" PRIx64, cpu64.r9);
4988 outs() << " r10 " << format("0x%016" PRIx64, cpu64.r10);
4989 outs() << " r11 " << format("0x%016" PRIx64, cpu64.r11) << "\n";
4990 outs() << " r12 " << format("0x%016" PRIx64, cpu64.r12);
4991 outs() << " r13 " << format("0x%016" PRIx64, cpu64.r13);
4992 outs() << " r14 " << format("0x%016" PRIx64, cpu64.r14) << "\n";
4993 outs() << " r15 " << format("0x%016" PRIx64, cpu64.r15);
4994 outs() << " rip " << format("0x%016" PRIx64, cpu64.rip) << "\n";
4995 outs() << "rflags " << format("0x%016" PRIx64, cpu64.rflags);
4996 outs() << " cs " << format("0x%016" PRIx64, cpu64.cs);
4997 outs() << " fs " << format("0x%016" PRIx64, cpu64.fs) << "\n";
4998 outs() << " gs " << format("0x%016" PRIx64, cpu64.gs) << "\n";
4999}
5000
Kevin Enderby227df342014-12-23 23:43:59 +00005001static void Print_mmst_reg(MachO::mmst_reg_t &r) {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005002 uint32_t f;
5003 outs() << "\t mmst_reg ";
5004 for (f = 0; f < 10; f++)
5005 outs() << format("%02" PRIx32, (r.mmst_reg[f] & 0xff)) << " ";
5006 outs() << "\n";
5007 outs() << "\t mmst_rsrv ";
5008 for (f = 0; f < 6; f++)
5009 outs() << format("%02" PRIx32, (r.mmst_rsrv[f] & 0xff)) << " ";
5010 outs() << "\n";
5011}
5012
Kevin Enderbyaefb0032014-12-24 00:16:51 +00005013static void Print_xmm_reg(MachO::xmm_reg_t &r) {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005014 uint32_t f;
5015 outs() << "\t xmm_reg ";
5016 for (f = 0; f < 16; f++)
5017 outs() << format("%02" PRIx32, (r.xmm_reg[f] & 0xff)) << " ";
5018 outs() << "\n";
5019}
5020
5021static void Print_x86_float_state_t(MachO::x86_float_state64_t &fpu) {
5022 outs() << "\t fpu_reserved[0] " << fpu.fpu_reserved[0];
5023 outs() << " fpu_reserved[1] " << fpu.fpu_reserved[1] << "\n";
5024 outs() << "\t control: invalid " << fpu.fpu_fcw.invalid;
5025 outs() << " denorm " << fpu.fpu_fcw.denorm;
5026 outs() << " zdiv " << fpu.fpu_fcw.zdiv;
5027 outs() << " ovrfl " << fpu.fpu_fcw.ovrfl;
5028 outs() << " undfl " << fpu.fpu_fcw.undfl;
5029 outs() << " precis " << fpu.fpu_fcw.precis << "\n";
5030 outs() << "\t\t pc ";
5031 if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_24B)
5032 outs() << "FP_PREC_24B ";
5033 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_53B)
5034 outs() << "FP_PREC_53B ";
5035 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_64B)
5036 outs() << "FP_PREC_64B ";
5037 else
5038 outs() << fpu.fpu_fcw.pc << " ";
5039 outs() << "rc ";
5040 if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_NEAR)
5041 outs() << "FP_RND_NEAR ";
5042 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_DOWN)
5043 outs() << "FP_RND_DOWN ";
5044 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_UP)
5045 outs() << "FP_RND_UP ";
5046 else if (fpu.fpu_fcw.rc == MachO::x86_FP_CHOP)
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005047 outs() << "FP_CHOP ";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005048 outs() << "\n";
5049 outs() << "\t status: invalid " << fpu.fpu_fsw.invalid;
5050 outs() << " denorm " << fpu.fpu_fsw.denorm;
5051 outs() << " zdiv " << fpu.fpu_fsw.zdiv;
5052 outs() << " ovrfl " << fpu.fpu_fsw.ovrfl;
5053 outs() << " undfl " << fpu.fpu_fsw.undfl;
5054 outs() << " precis " << fpu.fpu_fsw.precis;
5055 outs() << " stkflt " << fpu.fpu_fsw.stkflt << "\n";
5056 outs() << "\t errsumm " << fpu.fpu_fsw.errsumm;
5057 outs() << " c0 " << fpu.fpu_fsw.c0;
5058 outs() << " c1 " << fpu.fpu_fsw.c1;
5059 outs() << " c2 " << fpu.fpu_fsw.c2;
5060 outs() << " tos " << fpu.fpu_fsw.tos;
5061 outs() << " c3 " << fpu.fpu_fsw.c3;
5062 outs() << " busy " << fpu.fpu_fsw.busy << "\n";
5063 outs() << "\t fpu_ftw " << format("0x%02" PRIx32, fpu.fpu_ftw);
5064 outs() << " fpu_rsrv1 " << format("0x%02" PRIx32, fpu.fpu_rsrv1);
5065 outs() << " fpu_fop " << format("0x%04" PRIx32, fpu.fpu_fop);
5066 outs() << " fpu_ip " << format("0x%08" PRIx32, fpu.fpu_ip) << "\n";
5067 outs() << "\t fpu_cs " << format("0x%04" PRIx32, fpu.fpu_cs);
5068 outs() << " fpu_rsrv2 " << format("0x%04" PRIx32, fpu.fpu_rsrv2);
5069 outs() << " fpu_dp " << format("0x%08" PRIx32, fpu.fpu_dp);
5070 outs() << " fpu_ds " << format("0x%04" PRIx32, fpu.fpu_ds) << "\n";
5071 outs() << "\t fpu_rsrv3 " << format("0x%04" PRIx32, fpu.fpu_rsrv3);
5072 outs() << " fpu_mxcsr " << format("0x%08" PRIx32, fpu.fpu_mxcsr);
5073 outs() << " fpu_mxcsrmask " << format("0x%08" PRIx32, fpu.fpu_mxcsrmask);
5074 outs() << "\n";
5075 outs() << "\t fpu_stmm0:\n";
5076 Print_mmst_reg(fpu.fpu_stmm0);
5077 outs() << "\t fpu_stmm1:\n";
5078 Print_mmst_reg(fpu.fpu_stmm1);
5079 outs() << "\t fpu_stmm2:\n";
5080 Print_mmst_reg(fpu.fpu_stmm2);
5081 outs() << "\t fpu_stmm3:\n";
5082 Print_mmst_reg(fpu.fpu_stmm3);
5083 outs() << "\t fpu_stmm4:\n";
5084 Print_mmst_reg(fpu.fpu_stmm4);
5085 outs() << "\t fpu_stmm5:\n";
5086 Print_mmst_reg(fpu.fpu_stmm5);
5087 outs() << "\t fpu_stmm6:\n";
5088 Print_mmst_reg(fpu.fpu_stmm6);
5089 outs() << "\t fpu_stmm7:\n";
5090 Print_mmst_reg(fpu.fpu_stmm7);
5091 outs() << "\t fpu_xmm0:\n";
5092 Print_xmm_reg(fpu.fpu_xmm0);
5093 outs() << "\t fpu_xmm1:\n";
5094 Print_xmm_reg(fpu.fpu_xmm1);
5095 outs() << "\t fpu_xmm2:\n";
5096 Print_xmm_reg(fpu.fpu_xmm2);
5097 outs() << "\t fpu_xmm3:\n";
5098 Print_xmm_reg(fpu.fpu_xmm3);
5099 outs() << "\t fpu_xmm4:\n";
5100 Print_xmm_reg(fpu.fpu_xmm4);
5101 outs() << "\t fpu_xmm5:\n";
5102 Print_xmm_reg(fpu.fpu_xmm5);
5103 outs() << "\t fpu_xmm6:\n";
5104 Print_xmm_reg(fpu.fpu_xmm6);
5105 outs() << "\t fpu_xmm7:\n";
5106 Print_xmm_reg(fpu.fpu_xmm7);
5107 outs() << "\t fpu_xmm8:\n";
5108 Print_xmm_reg(fpu.fpu_xmm8);
5109 outs() << "\t fpu_xmm9:\n";
5110 Print_xmm_reg(fpu.fpu_xmm9);
5111 outs() << "\t fpu_xmm10:\n";
5112 Print_xmm_reg(fpu.fpu_xmm10);
5113 outs() << "\t fpu_xmm11:\n";
5114 Print_xmm_reg(fpu.fpu_xmm11);
5115 outs() << "\t fpu_xmm12:\n";
5116 Print_xmm_reg(fpu.fpu_xmm12);
5117 outs() << "\t fpu_xmm13:\n";
5118 Print_xmm_reg(fpu.fpu_xmm13);
5119 outs() << "\t fpu_xmm14:\n";
5120 Print_xmm_reg(fpu.fpu_xmm14);
5121 outs() << "\t fpu_xmm15:\n";
5122 Print_xmm_reg(fpu.fpu_xmm15);
5123 outs() << "\t fpu_rsrv4:\n";
5124 for (uint32_t f = 0; f < 6; f++) {
5125 outs() << "\t ";
5126 for (uint32_t g = 0; g < 16; g++)
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005127 outs() << format("%02" PRIx32, fpu.fpu_rsrv4[f * g]) << " ";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005128 outs() << "\n";
5129 }
5130 outs() << "\t fpu_reserved1 " << format("0x%08" PRIx32, fpu.fpu_reserved1);
5131 outs() << "\n";
5132}
5133
5134static void Print_x86_exception_state_t(MachO::x86_exception_state64_t &exc64) {
5135 outs() << "\t trapno " << format("0x%08" PRIx32, exc64.trapno);
5136 outs() << " err " << format("0x%08" PRIx32, exc64.err);
5137 outs() << " faultvaddr " << format("0x%016" PRIx64, exc64.faultvaddr) << "\n";
5138}
5139
5140static void PrintThreadCommand(MachO::thread_command t, const char *Ptr,
5141 bool isLittleEndian, uint32_t cputype) {
5142 if (t.cmd == MachO::LC_THREAD)
5143 outs() << " cmd LC_THREAD\n";
5144 else if (t.cmd == MachO::LC_UNIXTHREAD)
5145 outs() << " cmd LC_UNIXTHREAD\n";
5146 else
5147 outs() << " cmd " << t.cmd << " (unknown)\n";
5148 outs() << " cmdsize " << t.cmdsize;
5149 if (t.cmdsize < sizeof(struct MachO::thread_command) + 2 * sizeof(uint32_t))
5150 outs() << " Incorrect size\n";
5151 else
5152 outs() << "\n";
5153
5154 const char *begin = Ptr + sizeof(struct MachO::thread_command);
5155 const char *end = Ptr + t.cmdsize;
5156 uint32_t flavor, count, left;
5157 if (cputype == MachO::CPU_TYPE_X86_64) {
5158 while (begin < end) {
5159 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5160 memcpy((char *)&flavor, begin, sizeof(uint32_t));
5161 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005162 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005163 flavor = 0;
5164 begin = end;
5165 }
5166 if (isLittleEndian != sys::IsLittleEndianHost)
5167 sys::swapByteOrder(flavor);
5168 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5169 memcpy((char *)&count, begin, sizeof(uint32_t));
5170 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005171 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005172 count = 0;
5173 begin = end;
5174 }
5175 if (isLittleEndian != sys::IsLittleEndianHost)
5176 sys::swapByteOrder(count);
5177 if (flavor == MachO::x86_THREAD_STATE64) {
5178 outs() << " flavor x86_THREAD_STATE64\n";
5179 if (count == MachO::x86_THREAD_STATE64_COUNT)
5180 outs() << " count x86_THREAD_STATE64_COUNT\n";
5181 else
5182 outs() << " count " << count
5183 << " (not x86_THREAD_STATE64_COUNT)\n";
5184 MachO::x86_thread_state64_t cpu64;
5185 left = end - begin;
5186 if (left >= sizeof(MachO::x86_thread_state64_t)) {
5187 memcpy(&cpu64, begin, sizeof(MachO::x86_thread_state64_t));
5188 begin += sizeof(MachO::x86_thread_state64_t);
5189 } else {
5190 memset(&cpu64, '\0', sizeof(MachO::x86_thread_state64_t));
5191 memcpy(&cpu64, begin, left);
5192 begin += left;
5193 }
5194 if (isLittleEndian != sys::IsLittleEndianHost)
5195 swapStruct(cpu64);
5196 Print_x86_thread_state64_t(cpu64);
5197 } else if (flavor == MachO::x86_THREAD_STATE) {
5198 outs() << " flavor x86_THREAD_STATE\n";
5199 if (count == MachO::x86_THREAD_STATE_COUNT)
5200 outs() << " count x86_THREAD_STATE_COUNT\n";
5201 else
5202 outs() << " count " << count
5203 << " (not x86_THREAD_STATE_COUNT)\n";
5204 struct MachO::x86_thread_state_t ts;
5205 left = end - begin;
5206 if (left >= sizeof(MachO::x86_thread_state_t)) {
5207 memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
5208 begin += sizeof(MachO::x86_thread_state_t);
5209 } else {
5210 memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
5211 memcpy(&ts, begin, left);
5212 begin += left;
5213 }
5214 if (isLittleEndian != sys::IsLittleEndianHost)
5215 swapStruct(ts);
5216 if (ts.tsh.flavor == MachO::x86_THREAD_STATE64) {
5217 outs() << "\t tsh.flavor x86_THREAD_STATE64 ";
5218 if (ts.tsh.count == MachO::x86_THREAD_STATE64_COUNT)
5219 outs() << "tsh.count x86_THREAD_STATE64_COUNT\n";
5220 else
5221 outs() << "tsh.count " << ts.tsh.count
5222 << " (not x86_THREAD_STATE64_COUNT\n";
5223 Print_x86_thread_state64_t(ts.uts.ts64);
5224 } else {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005225 outs() << "\t tsh.flavor " << ts.tsh.flavor << " tsh.count "
5226 << ts.tsh.count << "\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005227 }
5228 } else if (flavor == MachO::x86_FLOAT_STATE) {
5229 outs() << " flavor x86_FLOAT_STATE\n";
5230 if (count == MachO::x86_FLOAT_STATE_COUNT)
5231 outs() << " count x86_FLOAT_STATE_COUNT\n";
5232 else
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005233 outs() << " count " << count << " (not x86_FLOAT_STATE_COUNT)\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005234 struct MachO::x86_float_state_t fs;
5235 left = end - begin;
5236 if (left >= sizeof(MachO::x86_float_state_t)) {
5237 memcpy(&fs, begin, sizeof(MachO::x86_float_state_t));
5238 begin += sizeof(MachO::x86_float_state_t);
5239 } else {
5240 memset(&fs, '\0', sizeof(MachO::x86_float_state_t));
5241 memcpy(&fs, begin, left);
5242 begin += left;
5243 }
5244 if (isLittleEndian != sys::IsLittleEndianHost)
5245 swapStruct(fs);
5246 if (fs.fsh.flavor == MachO::x86_FLOAT_STATE64) {
5247 outs() << "\t fsh.flavor x86_FLOAT_STATE64 ";
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005248 if (fs.fsh.count == MachO::x86_FLOAT_STATE64_COUNT)
Kevin Enderby48ef5342014-12-23 22:56:39 +00005249 outs() << "fsh.count x86_FLOAT_STATE64_COUNT\n";
5250 else
5251 outs() << "fsh.count " << fs.fsh.count
5252 << " (not x86_FLOAT_STATE64_COUNT\n";
5253 Print_x86_float_state_t(fs.ufs.fs64);
5254 } else {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005255 outs() << "\t fsh.flavor " << fs.fsh.flavor << " fsh.count "
5256 << fs.fsh.count << "\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005257 }
5258 } else if (flavor == MachO::x86_EXCEPTION_STATE) {
5259 outs() << " flavor x86_EXCEPTION_STATE\n";
5260 if (count == MachO::x86_EXCEPTION_STATE_COUNT)
5261 outs() << " count x86_EXCEPTION_STATE_COUNT\n";
5262 else
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005263 outs() << " count " << count
Kevin Enderby48ef5342014-12-23 22:56:39 +00005264 << " (not x86_EXCEPTION_STATE_COUNT)\n";
5265 struct MachO::x86_exception_state_t es;
5266 left = end - begin;
5267 if (left >= sizeof(MachO::x86_exception_state_t)) {
5268 memcpy(&es, begin, sizeof(MachO::x86_exception_state_t));
5269 begin += sizeof(MachO::x86_exception_state_t);
5270 } else {
5271 memset(&es, '\0', sizeof(MachO::x86_exception_state_t));
5272 memcpy(&es, begin, left);
5273 begin += left;
5274 }
5275 if (isLittleEndian != sys::IsLittleEndianHost)
5276 swapStruct(es);
5277 if (es.esh.flavor == MachO::x86_EXCEPTION_STATE64) {
5278 outs() << "\t esh.flavor x86_EXCEPTION_STATE64\n";
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005279 if (es.esh.count == MachO::x86_EXCEPTION_STATE64_COUNT)
Kevin Enderby48ef5342014-12-23 22:56:39 +00005280 outs() << "\t esh.count x86_EXCEPTION_STATE64_COUNT\n";
5281 else
5282 outs() << "\t esh.count " << es.esh.count
5283 << " (not x86_EXCEPTION_STATE64_COUNT\n";
5284 Print_x86_exception_state_t(es.ues.es64);
5285 } else {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005286 outs() << "\t esh.flavor " << es.esh.flavor << " esh.count "
5287 << es.esh.count << "\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005288 }
5289 } else {
5290 outs() << " flavor " << flavor << " (unknown)\n";
5291 outs() << " count " << count << "\n";
5292 outs() << " state (unknown)\n";
5293 begin += count * sizeof(uint32_t);
5294 }
5295 }
5296 } else {
5297 while (begin < end) {
5298 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5299 memcpy((char *)&flavor, begin, sizeof(uint32_t));
5300 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005301 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005302 flavor = 0;
5303 begin = end;
5304 }
5305 if (isLittleEndian != sys::IsLittleEndianHost)
5306 sys::swapByteOrder(flavor);
5307 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5308 memcpy((char *)&count, begin, sizeof(uint32_t));
5309 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005310 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005311 count = 0;
5312 begin = end;
5313 }
5314 if (isLittleEndian != sys::IsLittleEndianHost)
5315 sys::swapByteOrder(count);
5316 outs() << " flavor " << flavor << "\n";
5317 outs() << " count " << count << "\n";
5318 outs() << " state (Unknown cputype/cpusubtype)\n";
5319 begin += count * sizeof(uint32_t);
5320 }
5321 }
5322}
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005323
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005324static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) {
5325 if (dl.cmd == MachO::LC_ID_DYLIB)
5326 outs() << " cmd LC_ID_DYLIB\n";
5327 else if (dl.cmd == MachO::LC_LOAD_DYLIB)
5328 outs() << " cmd LC_LOAD_DYLIB\n";
5329 else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB)
5330 outs() << " cmd LC_LOAD_WEAK_DYLIB\n";
5331 else if (dl.cmd == MachO::LC_REEXPORT_DYLIB)
5332 outs() << " cmd LC_REEXPORT_DYLIB\n";
5333 else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB)
5334 outs() << " cmd LC_LAZY_LOAD_DYLIB\n";
5335 else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
5336 outs() << " cmd LC_LOAD_UPWARD_DYLIB\n";
5337 else
5338 outs() << " cmd " << dl.cmd << " (unknown)\n";
5339 outs() << " cmdsize " << dl.cmdsize;
5340 if (dl.cmdsize < sizeof(struct MachO::dylib_command))
5341 outs() << " Incorrect size\n";
5342 else
5343 outs() << "\n";
5344 if (dl.dylib.name < dl.cmdsize) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005345 const char *P = (const char *)(Ptr) + dl.dylib.name;
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005346 outs() << " name " << P << " (offset " << dl.dylib.name << ")\n";
5347 } else {
5348 outs() << " name ?(bad offset " << dl.dylib.name << ")\n";
5349 }
5350 outs() << " time stamp " << dl.dylib.timestamp << " ";
5351 time_t t = dl.dylib.timestamp;
5352 outs() << ctime(&t);
5353 outs() << " current version ";
5354 if (dl.dylib.current_version == 0xffffffff)
5355 outs() << "n/a\n";
5356 else
5357 outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "."
5358 << ((dl.dylib.current_version >> 8) & 0xff) << "."
5359 << (dl.dylib.current_version & 0xff) << "\n";
5360 outs() << "compatibility version ";
5361 if (dl.dylib.compatibility_version == 0xffffffff)
5362 outs() << "n/a\n";
5363 else
5364 outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
5365 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
5366 << (dl.dylib.compatibility_version & 0xff) << "\n";
5367}
5368
5369static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld,
5370 uint32_t object_size) {
5371 if (ld.cmd == MachO::LC_CODE_SIGNATURE)
5372 outs() << " cmd LC_FUNCTION_STARTS\n";
5373 else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO)
5374 outs() << " cmd LC_SEGMENT_SPLIT_INFO\n";
5375 else if (ld.cmd == MachO::LC_FUNCTION_STARTS)
5376 outs() << " cmd LC_FUNCTION_STARTS\n";
5377 else if (ld.cmd == MachO::LC_DATA_IN_CODE)
5378 outs() << " cmd LC_DATA_IN_CODE\n";
5379 else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS)
5380 outs() << " cmd LC_DYLIB_CODE_SIGN_DRS\n";
5381 else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT)
5382 outs() << " cmd LC_LINKER_OPTIMIZATION_HINT\n";
5383 else
5384 outs() << " cmd " << ld.cmd << " (?)\n";
5385 outs() << " cmdsize " << ld.cmdsize;
5386 if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command))
5387 outs() << " Incorrect size\n";
5388 else
5389 outs() << "\n";
5390 outs() << " dataoff " << ld.dataoff;
5391 if (ld.dataoff > object_size)
5392 outs() << " (past end of file)\n";
5393 else
5394 outs() << "\n";
5395 outs() << " datasize " << ld.datasize;
5396 uint64_t big_size = ld.dataoff;
5397 big_size += ld.datasize;
5398 if (big_size > object_size)
5399 outs() << " (past end of file)\n";
5400 else
5401 outs() << "\n";
5402}
5403
Kevin Enderby956366c2014-08-29 22:30:52 +00005404static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t ncmds,
5405 uint32_t filetype, uint32_t cputype,
5406 bool verbose) {
Filipe Cabecinhase71bd0c2015-01-06 17:08:26 +00005407 if (ncmds == 0)
5408 return;
Kevin Enderby956366c2014-08-29 22:30:52 +00005409 StringRef Buf = Obj->getData();
5410 MachOObjectFile::LoadCommandInfo Command = Obj->getFirstLoadCommandInfo();
5411 for (unsigned i = 0;; ++i) {
5412 outs() << "Load command " << i << "\n";
5413 if (Command.C.cmd == MachO::LC_SEGMENT) {
5414 MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command);
5415 const char *sg_segname = SLC.segname;
5416 PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr,
5417 SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot,
5418 SLC.initprot, SLC.nsects, SLC.flags, Buf.size(),
5419 verbose);
5420 for (unsigned j = 0; j < SLC.nsects; j++) {
Kevin Enderbyc9713382014-12-16 01:14:45 +00005421 MachO::section S = Obj->getSection(Command, j);
Kevin Enderby956366c2014-08-29 22:30:52 +00005422 PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align,
5423 S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2,
5424 SLC.cmd, sg_segname, filetype, Buf.size(), verbose);
5425 }
5426 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
5427 MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command);
5428 const char *sg_segname = SLC_64.segname;
5429 PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname,
5430 SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff,
5431 SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot,
5432 SLC_64.nsects, SLC_64.flags, Buf.size(), verbose);
5433 for (unsigned j = 0; j < SLC_64.nsects; j++) {
5434 MachO::section_64 S_64 = Obj->getSection64(Command, j);
5435 PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size,
5436 S_64.offset, S_64.align, S_64.reloff, S_64.nreloc,
5437 S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd,
5438 sg_segname, filetype, Buf.size(), verbose);
5439 }
5440 } else if (Command.C.cmd == MachO::LC_SYMTAB) {
5441 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
David Majnemer73cc6ff2014-11-13 19:48:56 +00005442 PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size());
Kevin Enderby956366c2014-08-29 22:30:52 +00005443 } else if (Command.C.cmd == MachO::LC_DYSYMTAB) {
5444 MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand();
5445 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
David Majnemer73cc6ff2014-11-13 19:48:56 +00005446 PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(),
5447 Obj->is64Bit());
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005448 } else if (Command.C.cmd == MachO::LC_DYLD_INFO ||
5449 Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) {
5450 MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command);
5451 PrintDyldInfoLoadCommand(DyldInfo, Buf.size());
5452 } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER ||
5453 Command.C.cmd == MachO::LC_ID_DYLINKER ||
5454 Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) {
5455 MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command);
5456 PrintDyldLoadCommand(Dyld, Command.Ptr);
5457 } else if (Command.C.cmd == MachO::LC_UUID) {
5458 MachO::uuid_command Uuid = Obj->getUuidCommand(Command);
5459 PrintUuidLoadCommand(Uuid);
Jean-Daniel Dupas00cc1f52014-12-04 07:37:02 +00005460 } else if (Command.C.cmd == MachO::LC_RPATH) {
5461 MachO::rpath_command Rpath = Obj->getRpathCommand(Command);
5462 PrintRpathLoadCommand(Rpath, Command.Ptr);
Kevin Enderby1ff0ecc2014-12-16 21:48:27 +00005463 } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX ||
5464 Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005465 MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command);
5466 PrintVersionMinLoadCommand(Vd);
5467 } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) {
5468 MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command);
5469 PrintSourceVersionCommand(Sd);
5470 } else if (Command.C.cmd == MachO::LC_MAIN) {
5471 MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command);
5472 PrintEntryPointCommand(Ep);
Kevin Enderby0804f4672014-12-16 23:25:52 +00005473 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO) {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005474 MachO::encryption_info_command Ei =
5475 Obj->getEncryptionInfoCommand(Command);
Kevin Enderby0804f4672014-12-16 23:25:52 +00005476 PrintEncryptionInfoCommand(Ei, Buf.size());
Kevin Enderby57538292014-12-17 01:01:30 +00005477 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO_64) {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005478 MachO::encryption_info_command_64 Ei =
5479 Obj->getEncryptionInfoCommand64(Command);
Kevin Enderby57538292014-12-17 01:01:30 +00005480 PrintEncryptionInfoCommand64(Ei, Buf.size());
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00005481 } else if (Command.C.cmd == MachO::LC_LINKER_OPTION) {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005482 MachO::linker_option_command Lo =
5483 Obj->getLinkerOptionLoadCommand(Command);
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00005484 PrintLinkerOptionCommand(Lo, Command.Ptr);
Kevin Enderbyb4b79312014-12-18 19:24:35 +00005485 } else if (Command.C.cmd == MachO::LC_SUB_FRAMEWORK) {
5486 MachO::sub_framework_command Sf = Obj->getSubFrameworkCommand(Command);
5487 PrintSubFrameworkCommand(Sf, Command.Ptr);
Kevin Enderbya2bd8d92014-12-18 23:13:26 +00005488 } else if (Command.C.cmd == MachO::LC_SUB_UMBRELLA) {
5489 MachO::sub_umbrella_command Sf = Obj->getSubUmbrellaCommand(Command);
5490 PrintSubUmbrellaCommand(Sf, Command.Ptr);
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00005491 } else if (Command.C.cmd == MachO::LC_SUB_LIBRARY) {
5492 MachO::sub_library_command Sl = Obj->getSubLibraryCommand(Command);
5493 PrintSubLibraryCommand(Sl, Command.Ptr);
Kevin Enderby186eac32014-12-19 21:06:24 +00005494 } else if (Command.C.cmd == MachO::LC_SUB_CLIENT) {
5495 MachO::sub_client_command Sc = Obj->getSubClientCommand(Command);
5496 PrintSubClientCommand(Sc, Command.Ptr);
Kevin Enderby52e4ce42014-12-19 22:25:22 +00005497 } else if (Command.C.cmd == MachO::LC_ROUTINES) {
5498 MachO::routines_command Rc = Obj->getRoutinesCommand(Command);
5499 PrintRoutinesCommand(Rc);
5500 } else if (Command.C.cmd == MachO::LC_ROUTINES_64) {
5501 MachO::routines_command_64 Rc = Obj->getRoutinesCommand64(Command);
5502 PrintRoutinesCommand64(Rc);
Kevin Enderby48ef5342014-12-23 22:56:39 +00005503 } else if (Command.C.cmd == MachO::LC_THREAD ||
5504 Command.C.cmd == MachO::LC_UNIXTHREAD) {
5505 MachO::thread_command Tc = Obj->getThreadCommand(Command);
5506 PrintThreadCommand(Tc, Command.Ptr, Obj->isLittleEndian(), cputype);
Nick Kledzik15558912014-10-16 18:58:20 +00005507 } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB ||
5508 Command.C.cmd == MachO::LC_ID_DYLIB ||
5509 Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
5510 Command.C.cmd == MachO::LC_REEXPORT_DYLIB ||
5511 Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
5512 Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005513 MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command);
5514 PrintDylibCommand(Dl, Command.Ptr);
5515 } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE ||
5516 Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO ||
5517 Command.C.cmd == MachO::LC_FUNCTION_STARTS ||
5518 Command.C.cmd == MachO::LC_DATA_IN_CODE ||
5519 Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS ||
5520 Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) {
5521 MachO::linkedit_data_command Ld =
5522 Obj->getLinkeditDataLoadCommand(Command);
5523 PrintLinkEditDataCommand(Ld, Buf.size());
Kevin Enderby956366c2014-08-29 22:30:52 +00005524 } else {
5525 outs() << " cmd ?(" << format("0x%08" PRIx32, Command.C.cmd)
5526 << ")\n";
5527 outs() << " cmdsize " << Command.C.cmdsize << "\n";
5528 // TODO: get and print the raw bytes of the load command.
5529 }
5530 // TODO: print all the other kinds of load commands.
5531 if (i == ncmds - 1)
5532 break;
5533 else
5534 Command = Obj->getNextLoadCommandInfo(Command);
5535 }
5536}
5537
5538static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds,
5539 uint32_t &filetype, uint32_t &cputype,
5540 bool verbose) {
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005541 if (Obj->is64Bit()) {
5542 MachO::mach_header_64 H_64;
5543 H_64 = Obj->getHeader64();
5544 PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype,
5545 H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose);
Kevin Enderby956366c2014-08-29 22:30:52 +00005546 ncmds = H_64.ncmds;
5547 filetype = H_64.filetype;
5548 cputype = H_64.cputype;
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005549 } else {
5550 MachO::mach_header H;
5551 H = Obj->getHeader();
5552 PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds,
5553 H.sizeofcmds, H.flags, verbose);
Kevin Enderby956366c2014-08-29 22:30:52 +00005554 ncmds = H.ncmds;
5555 filetype = H.filetype;
5556 cputype = H.cputype;
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005557 }
5558}
5559
5560void llvm::printMachOFileHeader(const object::ObjectFile *Obj) {
5561 const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
Kevin Enderby956366c2014-08-29 22:30:52 +00005562 uint32_t ncmds = 0;
5563 uint32_t filetype = 0;
5564 uint32_t cputype = 0;
Kevin Enderbyf0640752015-03-13 17:56:32 +00005565 getAndPrintMachHeader(file, ncmds, filetype, cputype, !NonVerbose);
5566 PrintLoadCommands(file, ncmds, filetype, cputype, !NonVerbose);
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005567}
Nick Kledzikd04bc352014-08-30 00:20:14 +00005568
5569//===----------------------------------------------------------------------===//
5570// export trie dumping
5571//===----------------------------------------------------------------------===//
5572
5573void llvm::printMachOExportsTrie(const object::MachOObjectFile *Obj) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005574 for (const llvm::object::ExportEntry &Entry : Obj->exports()) {
5575 uint64_t Flags = Entry.flags();
Nick Kledzikd04bc352014-08-30 00:20:14 +00005576 bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT);
5577 bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION);
5578 bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
5579 MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL);
5580 bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
5581 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE);
5582 bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER);
5583 if (ReExport)
5584 outs() << "[re-export] ";
5585 else
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005586 outs() << format("0x%08llX ",
5587 Entry.address()); // FIXME:add in base address
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005588 outs() << Entry.name();
Nick Kledzikd04bc352014-08-30 00:20:14 +00005589 if (WeakDef || ThreadLocal || Resolver || Abs) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005590 bool NeedsComma = false;
Nick Kledzik1d1ac4b2014-09-03 01:12:52 +00005591 outs() << " [";
Nick Kledzikd04bc352014-08-30 00:20:14 +00005592 if (WeakDef) {
5593 outs() << "weak_def";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005594 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005595 }
5596 if (ThreadLocal) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005597 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00005598 outs() << ", ";
5599 outs() << "per-thread";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005600 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005601 }
5602 if (Abs) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005603 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00005604 outs() << ", ";
5605 outs() << "absolute";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005606 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005607 }
5608 if (Resolver) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005609 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00005610 outs() << ", ";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005611 outs() << format("resolver=0x%08llX", Entry.other());
5612 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005613 }
5614 outs() << "]";
5615 }
5616 if (ReExport) {
5617 StringRef DylibName = "unknown";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005618 int Ordinal = Entry.other() - 1;
5619 Obj->getLibraryShortNameByIndex(Ordinal, DylibName);
5620 if (Entry.otherName().empty())
Nick Kledzikd04bc352014-08-30 00:20:14 +00005621 outs() << " (from " << DylibName << ")";
5622 else
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005623 outs() << " (" << Entry.otherName() << " from " << DylibName << ")";
Nick Kledzikd04bc352014-08-30 00:20:14 +00005624 }
5625 outs() << "\n";
5626 }
5627}
Nick Kledzikac431442014-09-12 21:34:15 +00005628
Nick Kledzikac431442014-09-12 21:34:15 +00005629//===----------------------------------------------------------------------===//
5630// rebase table dumping
5631//===----------------------------------------------------------------------===//
5632
5633namespace {
5634class SegInfo {
5635public:
5636 SegInfo(const object::MachOObjectFile *Obj);
5637
5638 StringRef segmentName(uint32_t SegIndex);
5639 StringRef sectionName(uint32_t SegIndex, uint64_t SegOffset);
5640 uint64_t address(uint32_t SegIndex, uint64_t SegOffset);
5641
5642private:
5643 struct SectionInfo {
5644 uint64_t Address;
5645 uint64_t Size;
5646 StringRef SectionName;
5647 StringRef SegmentName;
5648 uint64_t OffsetInSegment;
5649 uint64_t SegmentStartAddress;
5650 uint32_t SegmentIndex;
5651 };
5652 const SectionInfo &findSection(uint32_t SegIndex, uint64_t SegOffset);
5653 SmallVector<SectionInfo, 32> Sections;
5654};
5655}
5656
5657SegInfo::SegInfo(const object::MachOObjectFile *Obj) {
5658 // Build table of sections so segIndex/offset pairs can be translated.
Nick Kledzik56ebef42014-09-16 01:41:51 +00005659 uint32_t CurSegIndex = Obj->hasPageZeroSegment() ? 1 : 0;
Nick Kledzikac431442014-09-12 21:34:15 +00005660 StringRef CurSegName;
5661 uint64_t CurSegAddress;
5662 for (const SectionRef &Section : Obj->sections()) {
5663 SectionInfo Info;
5664 if (error(Section.getName(Info.SectionName)))
5665 return;
Rafael Espindola80291272014-10-08 15:28:58 +00005666 Info.Address = Section.getAddress();
5667 Info.Size = Section.getSize();
Nick Kledzikac431442014-09-12 21:34:15 +00005668 Info.SegmentName =
5669 Obj->getSectionFinalSegmentName(Section.getRawDataRefImpl());
5670 if (!Info.SegmentName.equals(CurSegName)) {
5671 ++CurSegIndex;
5672 CurSegName = Info.SegmentName;
5673 CurSegAddress = Info.Address;
5674 }
5675 Info.SegmentIndex = CurSegIndex - 1;
5676 Info.OffsetInSegment = Info.Address - CurSegAddress;
5677 Info.SegmentStartAddress = CurSegAddress;
5678 Sections.push_back(Info);
5679 }
5680}
5681
5682StringRef SegInfo::segmentName(uint32_t SegIndex) {
5683 for (const SectionInfo &SI : Sections) {
5684 if (SI.SegmentIndex == SegIndex)
5685 return SI.SegmentName;
5686 }
5687 llvm_unreachable("invalid segIndex");
5688}
5689
5690const SegInfo::SectionInfo &SegInfo::findSection(uint32_t SegIndex,
5691 uint64_t OffsetInSeg) {
5692 for (const SectionInfo &SI : Sections) {
5693 if (SI.SegmentIndex != SegIndex)
5694 continue;
5695 if (SI.OffsetInSegment > OffsetInSeg)
5696 continue;
5697 if (OffsetInSeg >= (SI.OffsetInSegment + SI.Size))
5698 continue;
5699 return SI;
5700 }
5701 llvm_unreachable("segIndex and offset not in any section");
5702}
5703
5704StringRef SegInfo::sectionName(uint32_t SegIndex, uint64_t OffsetInSeg) {
5705 return findSection(SegIndex, OffsetInSeg).SectionName;
5706}
5707
5708uint64_t SegInfo::address(uint32_t SegIndex, uint64_t OffsetInSeg) {
5709 const SectionInfo &SI = findSection(SegIndex, OffsetInSeg);
5710 return SI.SegmentStartAddress + OffsetInSeg;
5711}
5712
5713void llvm::printMachORebaseTable(const object::MachOObjectFile *Obj) {
5714 // Build table of sections so names can used in final output.
5715 SegInfo sectionTable(Obj);
5716
5717 outs() << "segment section address type\n";
5718 for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable()) {
5719 uint32_t SegIndex = Entry.segmentIndex();
5720 uint64_t OffsetInSeg = Entry.segmentOffset();
5721 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5722 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5723 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5724
5725 // Table lines look like: __DATA __nl_symbol_ptr 0x0000F00C pointer
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005726 outs() << format("%-8s %-18s 0x%08" PRIX64 " %s\n",
5727 SegmentName.str().c_str(), SectionName.str().c_str(),
5728 Address, Entry.typeName().str().c_str());
Nick Kledzikac431442014-09-12 21:34:15 +00005729 }
5730}
Nick Kledzik56ebef42014-09-16 01:41:51 +00005731
5732static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) {
5733 StringRef DylibName;
5734 switch (Ordinal) {
5735 case MachO::BIND_SPECIAL_DYLIB_SELF:
5736 return "this-image";
5737 case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE:
5738 return "main-executable";
5739 case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP:
5740 return "flat-namespace";
5741 default:
Nick Kledzikabd29872014-09-16 22:03:13 +00005742 if (Ordinal > 0) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005743 std::error_code EC =
5744 Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName);
Nick Kledzikabd29872014-09-16 22:03:13 +00005745 if (EC)
Nick Kledzik51d2c2b2014-10-14 23:29:38 +00005746 return "<<bad library ordinal>>";
Nick Kledzikabd29872014-09-16 22:03:13 +00005747 return DylibName;
5748 }
Nick Kledzik56ebef42014-09-16 01:41:51 +00005749 }
Nick Kledzikabd29872014-09-16 22:03:13 +00005750 return "<<unknown special ordinal>>";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005751}
5752
5753//===----------------------------------------------------------------------===//
5754// bind table dumping
5755//===----------------------------------------------------------------------===//
5756
5757void llvm::printMachOBindTable(const object::MachOObjectFile *Obj) {
5758 // Build table of sections so names can used in final output.
5759 SegInfo sectionTable(Obj);
5760
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005761 outs() << "segment section address type "
5762 "addend dylib symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005763 for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable()) {
5764 uint32_t SegIndex = Entry.segmentIndex();
5765 uint64_t OffsetInSeg = Entry.segmentOffset();
5766 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5767 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5768 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5769
5770 // Table lines look like:
5771 // __DATA __got 0x00012010 pointer 0 libSystem ___stack_chk_guard
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005772 StringRef Attr;
Nick Kledzik56ebef42014-09-16 01:41:51 +00005773 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT)
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005774 Attr = " (weak_import)";
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005775 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005776 << left_justify(SectionName, 18) << " "
5777 << format_hex(Address, 10, true) << " "
5778 << left_justify(Entry.typeName(), 8) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005779 << format_decimal(Entry.addend(), 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005780 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005781 << Entry.symbolName() << Attr << "\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005782 }
5783}
5784
5785//===----------------------------------------------------------------------===//
5786// lazy bind table dumping
5787//===----------------------------------------------------------------------===//
5788
5789void llvm::printMachOLazyBindTable(const object::MachOObjectFile *Obj) {
5790 // Build table of sections so names can used in final output.
5791 SegInfo sectionTable(Obj);
5792
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005793 outs() << "segment section address "
5794 "dylib symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005795 for (const llvm::object::MachOBindEntry &Entry : Obj->lazyBindTable()) {
5796 uint32_t SegIndex = Entry.segmentIndex();
5797 uint64_t OffsetInSeg = Entry.segmentOffset();
5798 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5799 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5800 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5801
5802 // Table lines look like:
5803 // __DATA __got 0x00012010 libSystem ___stack_chk_guard
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005804 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005805 << left_justify(SectionName, 18) << " "
5806 << format_hex(Address, 10, true) << " "
5807 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
Nick Kledzik56ebef42014-09-16 01:41:51 +00005808 << Entry.symbolName() << "\n";
5809 }
5810}
5811
Nick Kledzik56ebef42014-09-16 01:41:51 +00005812//===----------------------------------------------------------------------===//
5813// weak bind table dumping
5814//===----------------------------------------------------------------------===//
5815
5816void llvm::printMachOWeakBindTable(const object::MachOObjectFile *Obj) {
5817 // Build table of sections so names can used in final output.
5818 SegInfo sectionTable(Obj);
5819
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005820 outs() << "segment section address "
5821 "type addend symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005822 for (const llvm::object::MachOBindEntry &Entry : Obj->weakBindTable()) {
5823 // Strong symbols don't have a location to update.
5824 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) {
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005825 outs() << " strong "
Nick Kledzik56ebef42014-09-16 01:41:51 +00005826 << Entry.symbolName() << "\n";
5827 continue;
5828 }
5829 uint32_t SegIndex = Entry.segmentIndex();
5830 uint64_t OffsetInSeg = Entry.segmentOffset();
5831 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5832 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5833 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5834
5835 // Table lines look like:
5836 // __DATA __data 0x00001000 pointer 0 _foo
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005837 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005838 << left_justify(SectionName, 18) << " "
5839 << format_hex(Address, 10, true) << " "
5840 << left_justify(Entry.typeName(), 8) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005841 << format_decimal(Entry.addend(), 8) << " " << Entry.symbolName()
5842 << "\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005843 }
5844}
5845
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005846// get_dyld_bind_info_symbolname() is used for disassembly and passed an
5847// address, ReferenceValue, in the Mach-O file and looks in the dyld bind
5848// information for that address. If the address is found its binding symbol
5849// name is returned. If not nullptr is returned.
5850static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
5851 struct DisassembleInfo *info) {
Kevin Enderby078be602014-10-23 19:53:12 +00005852 if (info->bindtable == nullptr) {
5853 info->bindtable = new (BindTable);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005854 SegInfo sectionTable(info->O);
5855 for (const llvm::object::MachOBindEntry &Entry : info->O->bindTable()) {
5856 uint32_t SegIndex = Entry.segmentIndex();
5857 uint64_t OffsetInSeg = Entry.segmentOffset();
5858 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5859 const char *SymbolName = nullptr;
5860 StringRef name = Entry.symbolName();
5861 if (!name.empty())
5862 SymbolName = name.data();
Kevin Enderby078be602014-10-23 19:53:12 +00005863 info->bindtable->push_back(std::make_pair(Address, SymbolName));
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005864 }
5865 }
Kevin Enderby078be602014-10-23 19:53:12 +00005866 for (bind_table_iterator BI = info->bindtable->begin(),
5867 BE = info->bindtable->end();
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005868 BI != BE; ++BI) {
5869 uint64_t Address = BI->first;
5870 if (ReferenceValue == Address) {
5871 const char *SymbolName = BI->second;
5872 return SymbolName;
5873 }
5874 }
5875 return nullptr;
5876}