blob: 2b0b268e1bce388f12a3bd173cd24c54dd4cfeae [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
69
Kevin Enderbybf246f52014-09-24 23:08:22 +000070static cl::opt<bool>
71 PrintImmHex("print-imm-hex",
72 cl::desc("Use hex format for immediate values"));
73
Kevin Enderby13023a12015-01-15 23:19:11 +000074cl::opt<bool> llvm::UniversalHeaders("universal-headers",
75 cl::desc("Print Mach-O universal headers "
76 "(requires -macho)"));
77
Kevin Enderby131d1772015-01-09 19:22:37 +000078cl::opt<bool>
Kevin Enderby13023a12015-01-15 23:19:11 +000079 llvm::ArchiveHeaders("archive-headers",
80 cl::desc("Print archive headers for Mach-O archives "
81 "(requires -macho)"));
Kevin Enderby131d1772015-01-09 19:22:37 +000082
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +000083cl::opt<bool>
84 llvm::IndirectSymbols("indirect-symbols",
85 cl::desc("Print indirect symbol table for Mach-O "
86 "objects (requires -macho)"));
87
Kevin Enderby69fe98d2015-01-23 18:52:17 +000088cl::opt<bool>
89 llvm::DataInCode("data-in-code",
90 cl::desc("Print the data in code table for Mach-O objects "
91 "(requires -macho)"));
92
Kevin Enderby9a509442015-01-27 21:28:24 +000093cl::opt<bool>
94 llvm::LinkOptHints("link-opt-hints",
95 cl::desc("Print the linker optimization hints for "
96 "Mach-O objects (requires -macho)"));
97
Kevin Enderbyf6d25852015-01-31 00:37:11 +000098cl::list<std::string>
99 llvm::DumpSections("section",
100 cl::desc("Prints the specified segment,section for "
101 "Mach-O objects (requires -macho)"));
102
Kevin Enderbycd66be52015-03-11 22:06:32 +0000103cl::opt<bool>
104 llvm::InfoPlist("info-plist",
105 cl::desc("Print the info plist section as strings for "
106 "Mach-O objects (requires -macho)"));
107
Kevin Enderbyf0640752015-03-13 17:56:32 +0000108cl::opt<bool>
Kevin Enderbybc847fa2015-03-16 20:08:09 +0000109 llvm::DylibsUsed("dylibs-used",
110 cl::desc("Print the shared libraries used for linked "
111 "Mach-O files (requires -macho)"));
112
113cl::opt<bool>
114 llvm::DylibId("dylib-id",
115 cl::desc("Print the shared library's id for the dylib Mach-O "
116 "file (requires -macho)"));
117
118cl::opt<bool>
Kevin Enderbyf0640752015-03-13 17:56:32 +0000119 llvm::NonVerbose("non-verbose",
120 cl::desc("Print the info for Mach-O objects in "
121 "non-verbose or numeric form (requires -macho)"));
122
Kevin Enderby6a221752015-03-17 17:10:57 +0000123cl::opt<std::string> llvm::DisSymName(
124 "dis-symname",
125 cl::desc("disassemble just this symbol's instructions (requires -macho"));
126
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +0000127static cl::list<std::string>
128 ArchFlags("arch", cl::desc("architecture(s) from a Mach-O file to dump"),
129 cl::ZeroOrMore);
130bool ArchAll = false;
131
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000132static std::string ThumbTripleName;
133
134static const Target *GetTarget(const MachOObjectFile *MachOObj,
135 const char **McpuDefault,
136 const Target **ThumbTarget) {
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000137 // Figure out the target triple.
Cameron Zwarich88cc16a2012-02-03 06:35:22 +0000138 if (TripleName.empty()) {
139 llvm::Triple TT("unknown-unknown-unknown");
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000140 llvm::Triple ThumbTriple = Triple();
141 TT = MachOObj->getArch(McpuDefault, &ThumbTriple);
Cameron Zwarich88cc16a2012-02-03 06:35:22 +0000142 TripleName = TT.str();
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000143 ThumbTripleName = ThumbTriple.str();
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000144 }
145
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000146 // Get the target specific parser.
147 std::string Error;
148 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000149 if (TheTarget && ThumbTripleName.empty())
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000150 return TheTarget;
151
Kevin Enderbyec5ca032014-08-18 20:21:02 +0000152 *ThumbTarget = TargetRegistry::lookupTarget(ThumbTripleName, Error);
153 if (*ThumbTarget)
154 return TheTarget;
155
156 errs() << "llvm-objdump: error: unable to get target for '";
157 if (!TheTarget)
158 errs() << TripleName;
159 else
160 errs() << ThumbTripleName;
161 errs() << "', see --version and --triple.\n";
Craig Toppere6cb63e2014-04-25 04:24:47 +0000162 return nullptr;
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000163}
164
Owen Andersond9243c42011-10-17 21:37:35 +0000165struct SymbolSorter {
166 bool operator()(const SymbolRef &A, const SymbolRef &B) {
167 SymbolRef::Type AType, BType;
168 A.getType(AType);
169 B.getType(BType);
170
171 uint64_t AAddr, BAddr;
172 if (AType != SymbolRef::ST_Function)
173 AAddr = 0;
174 else
175 A.getAddress(AAddr);
176 if (BType != SymbolRef::ST_Function)
177 BAddr = 0;
178 else
179 B.getAddress(BAddr);
180 return AAddr < BAddr;
181 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000182};
183
Kevin Enderby273ae012013-06-06 17:20:50 +0000184// Types for the storted data in code table that is built before disassembly
185// and the predicate function to sort them.
186typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
187typedef std::vector<DiceTableEntry> DiceTable;
188typedef DiceTable::iterator dice_table_iterator;
189
Kevin Enderby930fdc72014-11-06 19:00:13 +0000190// This is used to search for a data in code table entry for the PC being
191// disassembled. The j parameter has the PC in j.first. A single data in code
192// table entry can cover many bytes for each of its Kind's. So if the offset,
193// aka the i.first value, of the data in code table entry plus its Length
194// covers the PC being searched for this will return true. If not it will
195// return false.
David Majnemerea9b8ee2014-11-04 08:41:48 +0000196static bool compareDiceTableEntries(const DiceTableEntry &i,
197 const DiceTableEntry &j) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000198 uint16_t Length;
199 i.second.getLength(Length);
200
201 return j.first >= i.first && j.first < i.first + Length;
Kevin Enderby273ae012013-06-06 17:20:50 +0000202}
203
Kevin Enderby930fdc72014-11-06 19:00:13 +0000204static uint64_t DumpDataInCode(const char *bytes, uint64_t Length,
205 unsigned short Kind) {
206 uint32_t Value, Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000207
208 switch (Kind) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000209 default:
Charles Davis8bdfafd2013-09-01 04:28:48 +0000210 case MachO::DICE_KIND_DATA:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000211 if (Length >= 4) {
212 if (!NoShowRawInsn)
213 DumpBytes(StringRef(bytes, 4));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000214 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
Kevin Enderby273ae012013-06-06 17:20:50 +0000215 outs() << "\t.long " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000216 Size = 4;
217 } else if (Length >= 2) {
218 if (!NoShowRawInsn)
219 DumpBytes(StringRef(bytes, 2));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000220 Value = bytes[1] << 8 | bytes[0];
Kevin Enderby273ae012013-06-06 17:20:50 +0000221 outs() << "\t.short " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000222 Size = 2;
223 } else {
224 if (!NoShowRawInsn)
225 DumpBytes(StringRef(bytes, 2));
Kevin Enderby273ae012013-06-06 17:20:50 +0000226 Value = bytes[0];
227 outs() << "\t.byte " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000228 Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000229 }
Kevin Enderby930fdc72014-11-06 19:00:13 +0000230 if (Kind == MachO::DICE_KIND_DATA)
231 outs() << "\t@ KIND_DATA\n";
232 else
233 outs() << "\t@ data in code kind = " << Kind << "\n";
Kevin Enderby273ae012013-06-06 17:20:50 +0000234 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000235 case MachO::DICE_KIND_JUMP_TABLE8:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000236 if (!NoShowRawInsn)
237 DumpBytes(StringRef(bytes, 1));
Kevin Enderby273ae012013-06-06 17:20:50 +0000238 Value = bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000239 outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n";
240 Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000241 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000242 case MachO::DICE_KIND_JUMP_TABLE16:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000243 if (!NoShowRawInsn)
244 DumpBytes(StringRef(bytes, 2));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000245 Value = bytes[1] << 8 | bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000246 outs() << "\t.short " << format("%5u", Value & 0xffff)
247 << "\t@ KIND_JUMP_TABLE16\n";
248 Size = 2;
Kevin Enderby273ae012013-06-06 17:20:50 +0000249 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000250 case MachO::DICE_KIND_JUMP_TABLE32:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000251 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
252 if (!NoShowRawInsn)
253 DumpBytes(StringRef(bytes, 4));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000254 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000255 outs() << "\t.long " << Value;
256 if (Kind == MachO::DICE_KIND_JUMP_TABLE32)
257 outs() << "\t@ KIND_JUMP_TABLE32\n";
258 else
259 outs() << "\t@ KIND_ABS_JUMP_TABLE32\n";
260 Size = 4;
Kevin Enderby273ae012013-06-06 17:20:50 +0000261 break;
262 }
Kevin Enderby930fdc72014-11-06 19:00:13 +0000263 return Size;
Kevin Enderby273ae012013-06-06 17:20:50 +0000264}
265
Alexey Samsonov464d2e42014-03-17 07:28:19 +0000266static void getSectionsAndSymbols(const MachO::mach_header Header,
267 MachOObjectFile *MachOObj,
268 std::vector<SectionRef> &Sections,
269 std::vector<SymbolRef> &Symbols,
270 SmallVectorImpl<uint64_t> &FoundFns,
271 uint64_t &BaseSegmentAddress) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +0000272 for (const SymbolRef &Symbol : MachOObj->symbols()) {
273 StringRef SymName;
274 Symbol.getName(SymName);
275 if (!SymName.startswith("ltmp"))
276 Symbols.push_back(Symbol);
277 }
Owen Andersond9243c42011-10-17 21:37:35 +0000278
Alexey Samsonov48803e52014-03-13 14:37:36 +0000279 for (const SectionRef &Section : MachOObj->sections()) {
Owen Andersond9243c42011-10-17 21:37:35 +0000280 StringRef SectName;
Alexey Samsonov48803e52014-03-13 14:37:36 +0000281 Section.getName(SectName);
282 Sections.push_back(Section);
Owen Andersond9243c42011-10-17 21:37:35 +0000283 }
284
Rafael Espindola56f976f2013-04-18 18:08:55 +0000285 MachOObjectFile::LoadCommandInfo Command =
Alexey Samsonov48803e52014-03-13 14:37:36 +0000286 MachOObj->getFirstLoadCommandInfo();
Kevin Enderby273ae012013-06-06 17:20:50 +0000287 bool BaseSegmentAddressSet = false;
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000288 for (unsigned i = 0;; ++i) {
Charles Davis8bdfafd2013-09-01 04:28:48 +0000289 if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) {
Benjamin Kramer699128e2011-09-21 01:13:19 +0000290 // We found a function starts segment, parse the addresses for later
291 // consumption.
Charles Davis8bdfafd2013-09-01 04:28:48 +0000292 MachO::linkedit_data_command LLC =
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000293 MachOObj->getLinkeditDataLoadCommand(Command);
Benjamin Kramer699128e2011-09-21 01:13:19 +0000294
Charles Davis8bdfafd2013-09-01 04:28:48 +0000295 MachOObj->ReadULEB128s(LLC.dataoff, FoundFns);
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000296 } else if (Command.C.cmd == MachO::LC_SEGMENT) {
297 MachO::segment_command SLC = MachOObj->getSegmentLoadCommand(Command);
Charles Davis8bdfafd2013-09-01 04:28:48 +0000298 StringRef SegName = SLC.segname;
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000299 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
Kevin Enderby273ae012013-06-06 17:20:50 +0000300 BaseSegmentAddressSet = true;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000301 BaseSegmentAddress = SLC.vmaddr;
Kevin Enderby273ae012013-06-06 17:20:50 +0000302 }
303 }
Rafael Espindolafeef8c22013-04-19 11:36:47 +0000304
Charles Davis8bdfafd2013-09-01 04:28:48 +0000305 if (i == Header.ncmds - 1)
Rafael Espindolafeef8c22013-04-19 11:36:47 +0000306 break;
307 else
308 Command = MachOObj->getNextLoadCommandInfo(Command);
Benjamin Kramer8a529dc2011-09-21 22:16:43 +0000309 }
Benjamin Kramer699128e2011-09-21 01:13:19 +0000310}
311
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +0000312static void PrintIndirectSymbolTable(MachOObjectFile *O, bool verbose,
313 uint32_t n, uint32_t count,
314 uint32_t stride, uint64_t addr) {
315 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
316 uint32_t nindirectsyms = Dysymtab.nindirectsyms;
317 if (n > nindirectsyms)
318 outs() << " (entries start past the end of the indirect symbol "
319 "table) (reserved1 field greater than the table size)";
320 else if (n + count > nindirectsyms)
321 outs() << " (entries extends past the end of the indirect symbol "
322 "table)";
323 outs() << "\n";
324 uint32_t cputype = O->getHeader().cputype;
325 if (cputype & MachO::CPU_ARCH_ABI64)
326 outs() << "address index";
327 else
328 outs() << "address index";
329 if (verbose)
330 outs() << " name\n";
331 else
332 outs() << "\n";
333 for (uint32_t j = 0; j < count && n + j < nindirectsyms; j++) {
334 if (cputype & MachO::CPU_ARCH_ABI64)
335 outs() << format("0x%016" PRIx64, addr + j * stride) << " ";
336 else
337 outs() << format("0x%08" PRIx32, addr + j * stride) << " ";
338 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand();
339 uint32_t indirect_symbol = O->getIndirectSymbolTableEntry(Dysymtab, n + j);
340 if (indirect_symbol == MachO::INDIRECT_SYMBOL_LOCAL) {
341 outs() << "LOCAL\n";
342 continue;
343 }
344 if (indirect_symbol ==
345 (MachO::INDIRECT_SYMBOL_LOCAL | MachO::INDIRECT_SYMBOL_ABS)) {
346 outs() << "LOCAL ABSOLUTE\n";
347 continue;
348 }
349 if (indirect_symbol == MachO::INDIRECT_SYMBOL_ABS) {
350 outs() << "ABSOLUTE\n";
351 continue;
352 }
353 outs() << format("%5u ", indirect_symbol);
Kevin Enderbyf0640752015-03-13 17:56:32 +0000354 if (verbose) {
355 MachO::symtab_command Symtab = O->getSymtabLoadCommand();
356 if (indirect_symbol < Symtab.nsyms) {
357 symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol);
358 SymbolRef Symbol = *Sym;
359 StringRef SymName;
360 Symbol.getName(SymName);
361 outs() << SymName;
362 } else {
363 outs() << "?";
364 }
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +0000365 }
366 outs() << "\n";
367 }
368}
369
370static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) {
371 uint32_t LoadCommandCount = O->getHeader().ncmds;
372 MachOObjectFile::LoadCommandInfo Load = O->getFirstLoadCommandInfo();
373 for (unsigned I = 0;; ++I) {
374 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
375 MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load);
376 for (unsigned J = 0; J < Seg.nsects; ++J) {
377 MachO::section_64 Sec = O->getSection64(Load, J);
378 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
379 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
380 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
381 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
382 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
383 section_type == MachO::S_SYMBOL_STUBS) {
384 uint32_t stride;
385 if (section_type == MachO::S_SYMBOL_STUBS)
386 stride = Sec.reserved2;
387 else
388 stride = 8;
389 if (stride == 0) {
390 outs() << "Can't print indirect symbols for (" << Sec.segname << ","
391 << Sec.sectname << ") "
392 << "(size of stubs in reserved2 field is zero)\n";
393 continue;
394 }
395 uint32_t count = Sec.size / stride;
396 outs() << "Indirect symbols for (" << Sec.segname << ","
397 << Sec.sectname << ") " << count << " entries";
398 uint32_t n = Sec.reserved1;
399 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
400 }
401 }
402 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
403 MachO::segment_command Seg = O->getSegmentLoadCommand(Load);
404 for (unsigned J = 0; J < Seg.nsects; ++J) {
405 MachO::section Sec = O->getSection(Load, J);
406 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
407 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
408 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
409 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
410 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
411 section_type == MachO::S_SYMBOL_STUBS) {
412 uint32_t stride;
413 if (section_type == MachO::S_SYMBOL_STUBS)
414 stride = Sec.reserved2;
415 else
416 stride = 4;
417 if (stride == 0) {
418 outs() << "Can't print indirect symbols for (" << Sec.segname << ","
419 << Sec.sectname << ") "
420 << "(size of stubs in reserved2 field is zero)\n";
421 continue;
422 }
423 uint32_t count = Sec.size / stride;
424 outs() << "Indirect symbols for (" << Sec.segname << ","
425 << Sec.sectname << ") " << count << " entries";
426 uint32_t n = Sec.reserved1;
427 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr);
428 }
429 }
430 }
431 if (I == LoadCommandCount - 1)
432 break;
433 else
434 Load = O->getNextLoadCommandInfo(Load);
435 }
436}
437
Kevin Enderby69fe98d2015-01-23 18:52:17 +0000438static void PrintDataInCodeTable(MachOObjectFile *O, bool verbose) {
439 MachO::linkedit_data_command DIC = O->getDataInCodeLoadCommand();
440 uint32_t nentries = DIC.datasize / sizeof(struct MachO::data_in_code_entry);
441 outs() << "Data in code table (" << nentries << " entries)\n";
442 outs() << "offset length kind\n";
443 for (dice_iterator DI = O->begin_dices(), DE = O->end_dices(); DI != DE;
444 ++DI) {
445 uint32_t Offset;
446 DI->getOffset(Offset);
447 outs() << format("0x%08" PRIx32, Offset) << " ";
448 uint16_t Length;
449 DI->getLength(Length);
450 outs() << format("%6u", Length) << " ";
451 uint16_t Kind;
452 DI->getKind(Kind);
453 if (verbose) {
454 switch (Kind) {
455 case MachO::DICE_KIND_DATA:
456 outs() << "DATA";
457 break;
458 case MachO::DICE_KIND_JUMP_TABLE8:
459 outs() << "JUMP_TABLE8";
460 break;
461 case MachO::DICE_KIND_JUMP_TABLE16:
462 outs() << "JUMP_TABLE16";
463 break;
464 case MachO::DICE_KIND_JUMP_TABLE32:
465 outs() << "JUMP_TABLE32";
466 break;
467 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
468 outs() << "ABS_JUMP_TABLE32";
469 break;
470 default:
471 outs() << format("0x%04" PRIx32, Kind);
472 break;
473 }
474 } else
475 outs() << format("0x%04" PRIx32, Kind);
476 outs() << "\n";
477 }
478}
479
Kevin Enderby9a509442015-01-27 21:28:24 +0000480static void PrintLinkOptHints(MachOObjectFile *O) {
481 MachO::linkedit_data_command LohLC = O->getLinkOptHintsLoadCommand();
482 const char *loh = O->getData().substr(LohLC.dataoff, 1).data();
483 uint32_t nloh = LohLC.datasize;
484 outs() << "Linker optimiztion hints (" << nloh << " total bytes)\n";
485 for (uint32_t i = 0; i < nloh;) {
486 unsigned n;
487 uint64_t identifier = decodeULEB128((const uint8_t *)(loh + i), &n);
488 i += n;
489 outs() << " identifier " << identifier << " ";
490 if (i >= nloh)
491 return;
492 switch (identifier) {
493 case 1:
494 outs() << "AdrpAdrp\n";
495 break;
496 case 2:
497 outs() << "AdrpLdr\n";
498 break;
499 case 3:
500 outs() << "AdrpAddLdr\n";
501 break;
502 case 4:
503 outs() << "AdrpLdrGotLdr\n";
504 break;
505 case 5:
506 outs() << "AdrpAddStr\n";
507 break;
508 case 6:
509 outs() << "AdrpLdrGotStr\n";
510 break;
511 case 7:
512 outs() << "AdrpAdd\n";
513 break;
514 case 8:
515 outs() << "AdrpLdrGot\n";
516 break;
517 default:
518 outs() << "Unknown identifier value\n";
519 break;
520 }
521 uint64_t narguments = decodeULEB128((const uint8_t *)(loh + i), &n);
522 i += n;
523 outs() << " narguments " << narguments << "\n";
524 if (i >= nloh)
525 return;
526
527 for (uint32_t j = 0; j < narguments; j++) {
528 uint64_t value = decodeULEB128((const uint8_t *)(loh + i), &n);
529 i += n;
530 outs() << "\tvalue " << format("0x%" PRIx64, value) << "\n";
531 if (i >= nloh)
532 return;
533 }
534 }
535}
536
Kevin Enderbybc847fa2015-03-16 20:08:09 +0000537static void PrintDylibs(MachOObjectFile *O, bool JustId) {
538 uint32_t LoadCommandCount = O->getHeader().ncmds;
539 MachOObjectFile::LoadCommandInfo Load = O->getFirstLoadCommandInfo();
540 for (unsigned I = 0;; ++I) {
541 if ((JustId && Load.C.cmd == MachO::LC_ID_DYLIB) ||
542 (!JustId && (Load.C.cmd == MachO::LC_ID_DYLIB ||
543 Load.C.cmd == MachO::LC_LOAD_DYLIB ||
544 Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
545 Load.C.cmd == MachO::LC_REEXPORT_DYLIB ||
546 Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
547 Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB))) {
548 MachO::dylib_command dl = O->getDylibIDLoadCommand(Load);
549 if (dl.dylib.name < dl.cmdsize) {
550 const char *p = (const char *)(Load.Ptr) + dl.dylib.name;
551 if (JustId)
552 outs() << p << "\n";
553 else {
554 outs() << "\t" << p;
555 outs() << " (compatibility version "
556 << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
557 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
558 << (dl.dylib.compatibility_version & 0xff) << ",";
559 outs() << " current version "
560 << ((dl.dylib.current_version >> 16) & 0xffff) << "."
561 << ((dl.dylib.current_version >> 8) & 0xff) << "."
562 << (dl.dylib.current_version & 0xff) << ")\n";
563 }
564 } else {
565 outs() << "\tBad offset (" << dl.dylib.name << ") for name of ";
566 if (Load.C.cmd == MachO::LC_ID_DYLIB)
567 outs() << "LC_ID_DYLIB ";
568 else if (Load.C.cmd == MachO::LC_LOAD_DYLIB)
569 outs() << "LC_LOAD_DYLIB ";
570 else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB)
571 outs() << "LC_LOAD_WEAK_DYLIB ";
572 else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB)
573 outs() << "LC_LAZY_LOAD_DYLIB ";
574 else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB)
575 outs() << "LC_REEXPORT_DYLIB ";
576 else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
577 outs() << "LC_LOAD_UPWARD_DYLIB ";
578 else
579 outs() << "LC_??? ";
580 outs() << "command " << I << "\n";
581 }
582 }
583 if (I == LoadCommandCount - 1)
584 break;
585 else
586 Load = O->getNextLoadCommandInfo(Load);
587 }
588}
589
Kevin Enderbyf6d25852015-01-31 00:37:11 +0000590typedef DenseMap<uint64_t, StringRef> SymbolAddressMap;
591
592static void CreateSymbolAddressMap(MachOObjectFile *O,
593 SymbolAddressMap *AddrMap) {
594 // Create a map of symbol addresses to symbol names.
595 for (const SymbolRef &Symbol : O->symbols()) {
596 SymbolRef::Type ST;
597 Symbol.getType(ST);
598 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
599 ST == SymbolRef::ST_Other) {
600 uint64_t Address;
601 Symbol.getAddress(Address);
602 StringRef SymName;
603 Symbol.getName(SymName);
604 (*AddrMap)[Address] = SymName;
605 }
606 }
607}
608
609// GuessSymbolName is passed the address of what might be a symbol and a
610// pointer to the SymbolAddressMap. It returns the name of a symbol
611// with that address or nullptr if no symbol is found with that address.
612static const char *GuessSymbolName(uint64_t value, SymbolAddressMap *AddrMap) {
613 const char *SymbolName = nullptr;
614 // A DenseMap can't lookup up some values.
615 if (value != 0xffffffffffffffffULL && value != 0xfffffffffffffffeULL) {
616 StringRef name = AddrMap->lookup(value);
617 if (!name.empty())
618 SymbolName = name.data();
619 }
620 return SymbolName;
621}
622
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000623static void DumpCstringChar(const char c) {
624 char p[2];
625 p[0] = c;
626 p[1] = '\0';
627 outs().write_escaped(p);
628}
629
Kevin Enderby10ba0412015-02-04 21:38:42 +0000630static void DumpCstringSection(MachOObjectFile *O, const char *sect,
631 uint32_t sect_size, uint64_t sect_addr,
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000632 bool print_addresses) {
Kevin Enderby10ba0412015-02-04 21:38:42 +0000633 for (uint32_t i = 0; i < sect_size; i++) {
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000634 if (print_addresses) {
635 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000636 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000637 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000638 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000639 }
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000640 for (; i < sect_size && sect[i] != '\0'; i++)
641 DumpCstringChar(sect[i]);
Kevin Enderby10ba0412015-02-04 21:38:42 +0000642 if (i < sect_size && sect[i] == '\0')
643 outs() << "\n";
644 }
645}
646
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000647static void DumpLiteral4(uint32_t l, float f) {
648 outs() << format("0x%08" PRIx32, l);
649 if ((l & 0x7f800000) != 0x7f800000)
650 outs() << format(" (%.16e)\n", f);
651 else {
652 if (l == 0x7f800000)
653 outs() << " (+Infinity)\n";
654 else if (l == 0xff800000)
655 outs() << " (-Infinity)\n";
656 else if ((l & 0x00400000) == 0x00400000)
657 outs() << " (non-signaling Not-a-Number)\n";
658 else
659 outs() << " (signaling Not-a-Number)\n";
660 }
661}
662
663static void DumpLiteral4Section(MachOObjectFile *O, const char *sect,
664 uint32_t sect_size, uint64_t sect_addr,
665 bool print_addresses) {
666 for (uint32_t i = 0; i < sect_size; i += sizeof(float)) {
667 if (print_addresses) {
668 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000669 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000670 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000671 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000672 }
673 float f;
674 memcpy(&f, sect + i, sizeof(float));
675 if (O->isLittleEndian() != sys::IsLittleEndianHost)
676 sys::swapByteOrder(f);
677 uint32_t l;
678 memcpy(&l, sect + i, sizeof(uint32_t));
679 if (O->isLittleEndian() != sys::IsLittleEndianHost)
680 sys::swapByteOrder(l);
681 DumpLiteral4(l, f);
682 }
683}
684
685static void DumpLiteral8(MachOObjectFile *O, uint32_t l0, uint32_t l1,
686 double d) {
687 outs() << format("0x%08" PRIx32, l0) << " " << format("0x%08" PRIx32, l1);
688 uint32_t Hi, Lo;
689 if (O->isLittleEndian()) {
690 Hi = l1;
691 Lo = l0;
692 } else {
693 Hi = l0;
694 Lo = l1;
695 }
696 // Hi is the high word, so this is equivalent to if(isfinite(d))
697 if ((Hi & 0x7ff00000) != 0x7ff00000)
698 outs() << format(" (%.16e)\n", d);
699 else {
700 if (Hi == 0x7ff00000 && Lo == 0)
701 outs() << " (+Infinity)\n";
702 else if (Hi == 0xfff00000 && Lo == 0)
703 outs() << " (-Infinity)\n";
704 else if ((Hi & 0x00080000) == 0x00080000)
705 outs() << " (non-signaling Not-a-Number)\n";
706 else
707 outs() << " (signaling Not-a-Number)\n";
708 }
709}
710
711static void DumpLiteral8Section(MachOObjectFile *O, const char *sect,
712 uint32_t sect_size, uint64_t sect_addr,
713 bool print_addresses) {
714 for (uint32_t i = 0; i < sect_size; i += sizeof(double)) {
715 if (print_addresses) {
716 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000717 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000718 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000719 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000720 }
721 double d;
722 memcpy(&d, sect + i, sizeof(double));
723 if (O->isLittleEndian() != sys::IsLittleEndianHost)
724 sys::swapByteOrder(d);
725 uint32_t l0, l1;
726 memcpy(&l0, sect + i, sizeof(uint32_t));
727 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
728 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
729 sys::swapByteOrder(l0);
730 sys::swapByteOrder(l1);
731 }
732 DumpLiteral8(O, l0, l1, d);
733 }
734}
735
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000736static void DumpLiteral16(uint32_t l0, uint32_t l1, uint32_t l2, uint32_t l3) {
737 outs() << format("0x%08" PRIx32, l0) << " ";
738 outs() << format("0x%08" PRIx32, l1) << " ";
739 outs() << format("0x%08" PRIx32, l2) << " ";
740 outs() << format("0x%08" PRIx32, l3) << "\n";
741}
742
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000743static void DumpLiteral16Section(MachOObjectFile *O, const char *sect,
744 uint32_t sect_size, uint64_t sect_addr,
745 bool print_addresses) {
746 for (uint32_t i = 0; i < sect_size; i += 16) {
747 if (print_addresses) {
748 if (O->is64Bit())
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000749 outs() << format("%016" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000750 else
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000751 outs() << format("%08" PRIx64, sect_addr + i) << " ";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000752 }
753 uint32_t l0, l1, l2, l3;
754 memcpy(&l0, sect + i, sizeof(uint32_t));
755 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t));
756 memcpy(&l2, sect + i + 2 * sizeof(uint32_t), sizeof(uint32_t));
757 memcpy(&l3, sect + i + 3 * sizeof(uint32_t), sizeof(uint32_t));
758 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
759 sys::swapByteOrder(l0);
760 sys::swapByteOrder(l1);
761 sys::swapByteOrder(l2);
762 sys::swapByteOrder(l3);
763 }
Kevin Enderby578fe5a2015-02-17 21:35:48 +0000764 DumpLiteral16(l0, l1, l2, l3);
765 }
766}
767
768static void DumpLiteralPointerSection(MachOObjectFile *O,
769 const SectionRef &Section,
770 const char *sect, uint32_t sect_size,
771 uint64_t sect_addr,
772 bool print_addresses) {
773 // Collect the literal sections in this Mach-O file.
774 std::vector<SectionRef> LiteralSections;
775 for (const SectionRef &Section : O->sections()) {
776 DataRefImpl Ref = Section.getRawDataRefImpl();
777 uint32_t section_type;
778 if (O->is64Bit()) {
779 const MachO::section_64 Sec = O->getSection64(Ref);
780 section_type = Sec.flags & MachO::SECTION_TYPE;
781 } else {
782 const MachO::section Sec = O->getSection(Ref);
783 section_type = Sec.flags & MachO::SECTION_TYPE;
784 }
785 if (section_type == MachO::S_CSTRING_LITERALS ||
786 section_type == MachO::S_4BYTE_LITERALS ||
787 section_type == MachO::S_8BYTE_LITERALS ||
788 section_type == MachO::S_16BYTE_LITERALS)
789 LiteralSections.push_back(Section);
790 }
791
792 // Set the size of the literal pointer.
793 uint32_t lp_size = O->is64Bit() ? 8 : 4;
794
795 // Collect the external relocation symbols for the the literal pointers.
796 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
797 for (const RelocationRef &Reloc : Section.relocations()) {
798 DataRefImpl Rel;
799 MachO::any_relocation_info RE;
800 bool isExtern = false;
801 Rel = Reloc.getRawDataRefImpl();
802 RE = O->getRelocation(Rel);
803 isExtern = O->getPlainRelocationExternal(RE);
804 if (isExtern) {
805 uint64_t RelocOffset;
806 Reloc.getOffset(RelocOffset);
807 symbol_iterator RelocSym = Reloc.getSymbol();
808 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
809 }
810 }
811 array_pod_sort(Relocs.begin(), Relocs.end());
812
813 // Dump each literal pointer.
814 for (uint32_t i = 0; i < sect_size; i += lp_size) {
815 if (print_addresses) {
816 if (O->is64Bit())
817 outs() << format("%016" PRIx64, sect_addr + i) << " ";
818 else
819 outs() << format("%08" PRIx64, sect_addr + i) << " ";
820 }
821 uint64_t lp;
822 if (O->is64Bit()) {
823 memcpy(&lp, sect + i, sizeof(uint64_t));
824 if (O->isLittleEndian() != sys::IsLittleEndianHost)
825 sys::swapByteOrder(lp);
826 } else {
827 uint32_t li;
828 memcpy(&li, sect + i, sizeof(uint32_t));
829 if (O->isLittleEndian() != sys::IsLittleEndianHost)
830 sys::swapByteOrder(li);
831 lp = li;
832 }
833
834 // First look for an external relocation entry for this literal pointer.
835 bool reloc_found = false;
836 for (unsigned j = 0, e = Relocs.size(); j != e; ++j) {
837 if (Relocs[i].first == i) {
838 symbol_iterator RelocSym = Relocs[j].second;
839 StringRef SymName;
840 RelocSym->getName(SymName);
841 outs() << "external relocation entry for symbol:" << SymName << "\n";
842 reloc_found = true;
843 }
844 }
845 if (reloc_found == true)
846 continue;
847
848 // For local references see what the section the literal pointer points to.
849 bool found = false;
850 for (unsigned SectIdx = 0; SectIdx != LiteralSections.size(); SectIdx++) {
851 uint64_t SectAddress = LiteralSections[SectIdx].getAddress();
852 uint64_t SectSize = LiteralSections[SectIdx].getSize();
853 if (lp >= SectAddress && lp < SectAddress + SectSize) {
854 found = true;
855
856 StringRef SectName;
857 LiteralSections[SectIdx].getName(SectName);
858 DataRefImpl Ref = LiteralSections[SectIdx].getRawDataRefImpl();
859 StringRef SegmentName = O->getSectionFinalSegmentName(Ref);
860 outs() << SegmentName << ":" << SectName << ":";
861
862 uint32_t section_type;
863 if (O->is64Bit()) {
864 const MachO::section_64 Sec = O->getSection64(Ref);
865 section_type = Sec.flags & MachO::SECTION_TYPE;
866 } else {
867 const MachO::section Sec = O->getSection(Ref);
868 section_type = Sec.flags & MachO::SECTION_TYPE;
869 }
870
871 StringRef BytesStr;
872 LiteralSections[SectIdx].getContents(BytesStr);
873 const char *Contents = reinterpret_cast<const char *>(BytesStr.data());
874
875 switch (section_type) {
876 case MachO::S_CSTRING_LITERALS:
877 for (uint64_t i = lp - SectAddress;
878 i < SectSize && Contents[i] != '\0'; i++) {
879 DumpCstringChar(Contents[i]);
880 }
881 outs() << "\n";
882 break;
883 case MachO::S_4BYTE_LITERALS:
884 float f;
885 memcpy(&f, Contents + (lp - SectAddress), sizeof(float));
886 uint32_t l;
887 memcpy(&l, Contents + (lp - SectAddress), sizeof(uint32_t));
888 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
889 sys::swapByteOrder(f);
890 sys::swapByteOrder(l);
891 }
892 DumpLiteral4(l, f);
893 break;
894 case MachO::S_8BYTE_LITERALS: {
895 double d;
896 memcpy(&d, Contents + (lp - SectAddress), sizeof(double));
897 uint32_t l0, l1;
898 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
899 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
900 sizeof(uint32_t));
901 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
902 sys::swapByteOrder(f);
903 sys::swapByteOrder(l0);
904 sys::swapByteOrder(l1);
905 }
906 DumpLiteral8(O, l0, l1, d);
907 break;
908 }
909 case MachO::S_16BYTE_LITERALS: {
910 uint32_t l0, l1, l2, l3;
911 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t));
912 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t),
913 sizeof(uint32_t));
914 memcpy(&l2, Contents + (lp - SectAddress) + 2 * sizeof(uint32_t),
915 sizeof(uint32_t));
916 memcpy(&l3, Contents + (lp - SectAddress) + 3 * sizeof(uint32_t),
917 sizeof(uint32_t));
918 if (O->isLittleEndian() != sys::IsLittleEndianHost) {
919 sys::swapByteOrder(l0);
920 sys::swapByteOrder(l1);
921 sys::swapByteOrder(l2);
922 sys::swapByteOrder(l3);
923 }
924 DumpLiteral16(l0, l1, l2, l3);
925 break;
926 }
927 }
928 }
929 }
930 if (found == false)
931 outs() << format("0x%" PRIx64, lp) << " (not in a literal section)\n";
Kevin Enderby74b43cb2015-02-06 23:25:38 +0000932 }
933}
934
Kevin Enderbyf6d25852015-01-31 00:37:11 +0000935static void DumpInitTermPointerSection(MachOObjectFile *O, const char *sect,
936 uint32_t sect_size, uint64_t sect_addr,
937 SymbolAddressMap *AddrMap,
938 bool verbose) {
939 uint32_t stride;
940 if (O->is64Bit())
941 stride = sizeof(uint64_t);
942 else
943 stride = sizeof(uint32_t);
944 for (uint32_t i = 0; i < sect_size; i += stride) {
945 const char *SymbolName = nullptr;
946 if (O->is64Bit()) {
947 outs() << format("0x%016" PRIx64, sect_addr + i * stride) << " ";
948 uint64_t pointer_value;
949 memcpy(&pointer_value, sect + i, stride);
950 if (O->isLittleEndian() != sys::IsLittleEndianHost)
951 sys::swapByteOrder(pointer_value);
952 outs() << format("0x%016" PRIx64, pointer_value);
953 if (verbose)
954 SymbolName = GuessSymbolName(pointer_value, AddrMap);
955 } else {
956 outs() << format("0x%08" PRIx64, sect_addr + i * stride) << " ";
957 uint32_t pointer_value;
958 memcpy(&pointer_value, sect + i, stride);
959 if (O->isLittleEndian() != sys::IsLittleEndianHost)
960 sys::swapByteOrder(pointer_value);
961 outs() << format("0x%08" PRIx32, pointer_value);
962 if (verbose)
963 SymbolName = GuessSymbolName(pointer_value, AddrMap);
964 }
965 if (SymbolName)
966 outs() << " " << SymbolName;
967 outs() << "\n";
968 }
969}
970
971static void DumpRawSectionContents(MachOObjectFile *O, const char *sect,
972 uint32_t size, uint64_t addr) {
973 uint32_t cputype = O->getHeader().cputype;
974 if (cputype == MachO::CPU_TYPE_I386 || cputype == MachO::CPU_TYPE_X86_64) {
975 uint32_t j;
976 for (uint32_t i = 0; i < size; i += j, addr += j) {
977 if (O->is64Bit())
978 outs() << format("%016" PRIx64, addr) << "\t";
979 else
Kevin Enderbyf0640752015-03-13 17:56:32 +0000980 outs() << format("%08" PRIx64, addr) << "\t";
Kevin Enderbyf6d25852015-01-31 00:37:11 +0000981 for (j = 0; j < 16 && i + j < size; j++) {
982 uint8_t byte_word = *(sect + i + j);
983 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
984 }
985 outs() << "\n";
986 }
987 } else {
988 uint32_t j;
989 for (uint32_t i = 0; i < size; i += j, addr += j) {
990 if (O->is64Bit())
991 outs() << format("%016" PRIx64, addr) << "\t";
992 else
993 outs() << format("%08" PRIx64, sect) << "\t";
994 for (j = 0; j < 4 * sizeof(int32_t) && i + j < size;
995 j += sizeof(int32_t)) {
996 if (i + j + sizeof(int32_t) < size) {
997 uint32_t long_word;
998 memcpy(&long_word, sect + i + j, sizeof(int32_t));
999 if (O->isLittleEndian() != sys::IsLittleEndianHost)
1000 sys::swapByteOrder(long_word);
1001 outs() << format("%08" PRIx32, long_word) << " ";
1002 } else {
1003 for (uint32_t k = 0; i + j + k < size; k++) {
1004 uint8_t byte_word = *(sect + i + j);
1005 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " ";
1006 }
1007 }
1008 }
1009 outs() << "\n";
1010 }
1011 }
1012}
1013
Kevin Enderby95df54c2015-02-04 01:01:38 +00001014static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
1015 StringRef DisSegName, StringRef DisSectName);
1016
1017static void DumpSectionContents(StringRef Filename, MachOObjectFile *O,
1018 bool verbose) {
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001019 SymbolAddressMap AddrMap;
1020 if (verbose)
1021 CreateSymbolAddressMap(O, &AddrMap);
1022
1023 for (unsigned i = 0; i < DumpSections.size(); ++i) {
1024 StringRef DumpSection = DumpSections[i];
1025 std::pair<StringRef, StringRef> DumpSegSectName;
1026 DumpSegSectName = DumpSection.split(',');
1027 StringRef DumpSegName, DumpSectName;
1028 if (DumpSegSectName.second.size()) {
1029 DumpSegName = DumpSegSectName.first;
1030 DumpSectName = DumpSegSectName.second;
1031 } else {
1032 DumpSegName = "";
1033 DumpSectName = DumpSegSectName.first;
1034 }
1035 for (const SectionRef &Section : O->sections()) {
1036 StringRef SectName;
1037 Section.getName(SectName);
1038 DataRefImpl Ref = Section.getRawDataRefImpl();
1039 StringRef SegName = O->getSectionFinalSegmentName(Ref);
1040 if ((DumpSegName.empty() || SegName == DumpSegName) &&
1041 (SectName == DumpSectName)) {
1042 outs() << "Contents of (" << SegName << "," << SectName
1043 << ") section\n";
Kevin Enderby95df54c2015-02-04 01:01:38 +00001044 uint32_t section_flags;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001045 if (O->is64Bit()) {
1046 const MachO::section_64 Sec = O->getSection64(Ref);
Kevin Enderby95df54c2015-02-04 01:01:38 +00001047 section_flags = Sec.flags;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001048
1049 } else {
1050 const MachO::section Sec = O->getSection(Ref);
Kevin Enderby95df54c2015-02-04 01:01:38 +00001051 section_flags = Sec.flags;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001052 }
Kevin Enderby95df54c2015-02-04 01:01:38 +00001053 uint32_t section_type = section_flags & MachO::SECTION_TYPE;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001054
1055 StringRef BytesStr;
1056 Section.getContents(BytesStr);
1057 const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1058 uint32_t sect_size = BytesStr.size();
1059 uint64_t sect_addr = Section.getAddress();
1060
1061 if (verbose) {
Kevin Enderby95df54c2015-02-04 01:01:38 +00001062 if ((section_flags & MachO::S_ATTR_PURE_INSTRUCTIONS) ||
1063 (section_flags & MachO::S_ATTR_SOME_INSTRUCTIONS)) {
1064 DisassembleMachO(Filename, O, SegName, SectName);
1065 continue;
1066 }
Kevin Enderbycd66be52015-03-11 22:06:32 +00001067 if (SegName == "__TEXT" && SectName == "__info_plist") {
1068 outs() << sect;
1069 continue;
1070 }
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001071 switch (section_type) {
1072 case MachO::S_REGULAR:
1073 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1074 break;
1075 case MachO::S_ZEROFILL:
1076 outs() << "zerofill section and has no contents in the file\n";
1077 break;
Kevin Enderby10ba0412015-02-04 21:38:42 +00001078 case MachO::S_CSTRING_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001079 DumpCstringSection(O, sect, sect_size, sect_addr, !NoLeadingAddr);
Kevin Enderby10ba0412015-02-04 21:38:42 +00001080 break;
Kevin Enderby74b43cb2015-02-06 23:25:38 +00001081 case MachO::S_4BYTE_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001082 DumpLiteral4Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
Kevin Enderby74b43cb2015-02-06 23:25:38 +00001083 break;
1084 case MachO::S_8BYTE_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001085 DumpLiteral8Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
Kevin Enderby74b43cb2015-02-06 23:25:38 +00001086 break;
1087 case MachO::S_16BYTE_LITERALS:
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001088 DumpLiteral16Section(O, sect, sect_size, sect_addr, !NoLeadingAddr);
1089 break;
Kevin Enderby578fe5a2015-02-17 21:35:48 +00001090 case MachO::S_LITERAL_POINTERS:
1091 DumpLiteralPointerSection(O, Section, sect, sect_size, sect_addr,
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00001092 !NoLeadingAddr);
Kevin Enderby578fe5a2015-02-17 21:35:48 +00001093 break;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001094 case MachO::S_MOD_INIT_FUNC_POINTERS:
1095 case MachO::S_MOD_TERM_FUNC_POINTERS:
1096 DumpInitTermPointerSection(O, sect, sect_size, sect_addr, &AddrMap,
1097 verbose);
1098 break;
1099 default:
1100 outs() << "Unknown section type ("
1101 << format("0x%08" PRIx32, section_type) << ")\n";
1102 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1103 break;
1104 }
1105 } else {
1106 if (section_type == MachO::S_ZEROFILL)
1107 outs() << "zerofill section and has no contents in the file\n";
1108 else
1109 DumpRawSectionContents(O, sect, sect_size, sect_addr);
1110 }
1111 }
1112 }
1113 }
1114}
1115
Kevin Enderbycd66be52015-03-11 22:06:32 +00001116static void DumpInfoPlistSectionContents(StringRef Filename,
1117 MachOObjectFile *O) {
1118 for (const SectionRef &Section : O->sections()) {
1119 StringRef SectName;
1120 Section.getName(SectName);
1121 DataRefImpl Ref = Section.getRawDataRefImpl();
1122 StringRef SegName = O->getSectionFinalSegmentName(Ref);
1123 if (SegName == "__TEXT" && SectName == "__info_plist") {
1124 outs() << "Contents of (" << SegName << "," << SectName << ") section\n";
1125 StringRef BytesStr;
1126 Section.getContents(BytesStr);
1127 const char *sect = reinterpret_cast<const char *>(BytesStr.data());
1128 outs() << sect;
1129 return;
1130 }
1131 }
1132}
1133
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001134// checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file
1135// and if it is and there is a list of architecture flags is specified then
1136// check to make sure this Mach-O file is one of those architectures or all
1137// architectures were specified. If not then an error is generated and this
1138// routine returns false. Else it returns true.
1139static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) {
1140 if (isa<MachOObjectFile>(O) && !ArchAll && ArchFlags.size() != 0) {
1141 MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(O);
1142 bool ArchFound = false;
1143 MachO::mach_header H;
1144 MachO::mach_header_64 H_64;
1145 Triple T;
1146 if (MachO->is64Bit()) {
1147 H_64 = MachO->MachOObjectFile::getHeader64();
1148 T = MachOObjectFile::getArch(H_64.cputype, H_64.cpusubtype);
1149 } else {
1150 H = MachO->MachOObjectFile::getHeader();
1151 T = MachOObjectFile::getArch(H.cputype, H.cpusubtype);
1152 }
1153 unsigned i;
1154 for (i = 0; i < ArchFlags.size(); ++i) {
1155 if (ArchFlags[i] == T.getArchName())
1156 ArchFound = true;
1157 break;
1158 }
1159 if (!ArchFound) {
1160 errs() << "llvm-objdump: file: " + Filename + " does not contain "
1161 << "architecture: " + ArchFlags[i] + "\n";
1162 return false;
1163 }
1164 }
1165 return true;
1166}
1167
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001168// ProcessMachO() is passed a single opened Mach-O file, which may be an
1169// archive member and or in a slice of a universal file. It prints the
1170// the file name and header info and then processes it according to the
1171// command line options.
1172static void ProcessMachO(StringRef Filename, MachOObjectFile *MachOOF,
1173 StringRef ArchiveMemberName = StringRef(),
1174 StringRef ArchitectureName = StringRef()) {
Kevin Enderby131d1772015-01-09 19:22:37 +00001175 // If we are doing some processing here on the Mach-O file print the header
1176 // info. And don't print it otherwise like in the case of printing the
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +00001177 // UniversalHeaders or ArchiveHeaders.
Kevin Enderby131d1772015-01-09 19:22:37 +00001178 if (Disassemble || PrivateHeaders || ExportsTrie || Rebase || Bind ||
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001179 LazyBind || WeakBind || IndirectSymbols || DataInCode || LinkOptHints ||
Kevin Enderbybc847fa2015-03-16 20:08:09 +00001180 DylibsUsed || DylibId || DumpSections.size() != 0) {
Kevin Enderby131d1772015-01-09 19:22:37 +00001181 outs() << Filename;
1182 if (!ArchiveMemberName.empty())
1183 outs() << '(' << ArchiveMemberName << ')';
1184 if (!ArchitectureName.empty())
1185 outs() << " (architecture " << ArchitectureName << ")";
1186 outs() << ":\n";
1187 }
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001188
1189 if (Disassemble)
Kevin Enderby95df54c2015-02-04 01:01:38 +00001190 DisassembleMachO(Filename, MachOOF, "__TEXT", "__text");
Kevin Enderbya7bdc7e2015-01-22 18:55:27 +00001191 if (IndirectSymbols)
Kevin Enderbyf0640752015-03-13 17:56:32 +00001192 PrintIndirectSymbols(MachOOF, !NonVerbose);
Kevin Enderby69fe98d2015-01-23 18:52:17 +00001193 if (DataInCode)
Kevin Enderbyf0640752015-03-13 17:56:32 +00001194 PrintDataInCodeTable(MachOOF, !NonVerbose);
Kevin Enderby9a509442015-01-27 21:28:24 +00001195 if (LinkOptHints)
1196 PrintLinkOptHints(MachOOF);
Kevin Enderby98da6132015-01-20 21:47:46 +00001197 if (Relocations)
1198 PrintRelocations(MachOOF);
1199 if (SectionHeaders)
1200 PrintSectionHeaders(MachOOF);
1201 if (SectionContents)
1202 PrintSectionContents(MachOOF);
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001203 if (DumpSections.size() != 0)
Kevin Enderbyf0640752015-03-13 17:56:32 +00001204 DumpSectionContents(Filename, MachOOF, !NonVerbose);
Kevin Enderbycd66be52015-03-11 22:06:32 +00001205 if (InfoPlist)
1206 DumpInfoPlistSectionContents(Filename, MachOOF);
Kevin Enderbybc847fa2015-03-16 20:08:09 +00001207 if (DylibsUsed)
1208 PrintDylibs(MachOOF, false);
1209 if (DylibId)
1210 PrintDylibs(MachOOF, true);
Kevin Enderby98da6132015-01-20 21:47:46 +00001211 if (SymbolTable)
1212 PrintSymbolTable(MachOOF);
1213 if (UnwindInfo)
1214 printMachOUnwindInfo(MachOOF);
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001215 if (PrivateHeaders)
1216 printMachOFileHeader(MachOOF);
1217 if (ExportsTrie)
1218 printExportsTrie(MachOOF);
1219 if (Rebase)
1220 printRebaseTable(MachOOF);
1221 if (Bind)
1222 printBindTable(MachOOF);
1223 if (LazyBind)
1224 printLazyBindTable(MachOOF);
1225 if (WeakBind)
1226 printWeakBindTable(MachOOF);
1227}
1228
Kevin Enderby131d1772015-01-09 19:22:37 +00001229// printUnknownCPUType() helps print_fat_headers for unknown CPU's.
1230static void printUnknownCPUType(uint32_t cputype, uint32_t cpusubtype) {
1231 outs() << " cputype (" << cputype << ")\n";
1232 outs() << " cpusubtype (" << cpusubtype << ")\n";
1233}
1234
1235// printCPUType() helps print_fat_headers by printing the cputype and
1236// pusubtype (symbolically for the one's it knows about).
1237static void printCPUType(uint32_t cputype, uint32_t cpusubtype) {
1238 switch (cputype) {
1239 case MachO::CPU_TYPE_I386:
1240 switch (cpusubtype) {
1241 case MachO::CPU_SUBTYPE_I386_ALL:
1242 outs() << " cputype CPU_TYPE_I386\n";
1243 outs() << " cpusubtype CPU_SUBTYPE_I386_ALL\n";
1244 break;
1245 default:
1246 printUnknownCPUType(cputype, cpusubtype);
1247 break;
1248 }
1249 break;
1250 case MachO::CPU_TYPE_X86_64:
1251 switch (cpusubtype) {
1252 case MachO::CPU_SUBTYPE_X86_64_ALL:
1253 outs() << " cputype CPU_TYPE_X86_64\n";
1254 outs() << " cpusubtype CPU_SUBTYPE_X86_64_ALL\n";
1255 break;
1256 case MachO::CPU_SUBTYPE_X86_64_H:
1257 outs() << " cputype CPU_TYPE_X86_64\n";
1258 outs() << " cpusubtype CPU_SUBTYPE_X86_64_H\n";
1259 break;
1260 default:
1261 printUnknownCPUType(cputype, cpusubtype);
1262 break;
1263 }
1264 break;
1265 case MachO::CPU_TYPE_ARM:
1266 switch (cpusubtype) {
1267 case MachO::CPU_SUBTYPE_ARM_ALL:
1268 outs() << " cputype CPU_TYPE_ARM\n";
1269 outs() << " cpusubtype CPU_SUBTYPE_ARM_ALL\n";
1270 break;
1271 case MachO::CPU_SUBTYPE_ARM_V4T:
1272 outs() << " cputype CPU_TYPE_ARM\n";
1273 outs() << " cpusubtype CPU_SUBTYPE_ARM_V4T\n";
1274 break;
1275 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
1276 outs() << " cputype CPU_TYPE_ARM\n";
1277 outs() << " cpusubtype CPU_SUBTYPE_ARM_V5TEJ\n";
1278 break;
1279 case MachO::CPU_SUBTYPE_ARM_XSCALE:
1280 outs() << " cputype CPU_TYPE_ARM\n";
1281 outs() << " cpusubtype CPU_SUBTYPE_ARM_XSCALE\n";
1282 break;
1283 case MachO::CPU_SUBTYPE_ARM_V6:
1284 outs() << " cputype CPU_TYPE_ARM\n";
1285 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6\n";
1286 break;
1287 case MachO::CPU_SUBTYPE_ARM_V6M:
1288 outs() << " cputype CPU_TYPE_ARM\n";
1289 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6M\n";
1290 break;
1291 case MachO::CPU_SUBTYPE_ARM_V7:
1292 outs() << " cputype CPU_TYPE_ARM\n";
1293 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7\n";
1294 break;
1295 case MachO::CPU_SUBTYPE_ARM_V7EM:
1296 outs() << " cputype CPU_TYPE_ARM\n";
1297 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7EM\n";
1298 break;
1299 case MachO::CPU_SUBTYPE_ARM_V7K:
1300 outs() << " cputype CPU_TYPE_ARM\n";
1301 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7K\n";
1302 break;
1303 case MachO::CPU_SUBTYPE_ARM_V7M:
1304 outs() << " cputype CPU_TYPE_ARM\n";
1305 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7M\n";
1306 break;
1307 case MachO::CPU_SUBTYPE_ARM_V7S:
1308 outs() << " cputype CPU_TYPE_ARM\n";
1309 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7S\n";
1310 break;
1311 default:
1312 printUnknownCPUType(cputype, cpusubtype);
1313 break;
1314 }
1315 break;
1316 case MachO::CPU_TYPE_ARM64:
1317 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
1318 case MachO::CPU_SUBTYPE_ARM64_ALL:
1319 outs() << " cputype CPU_TYPE_ARM64\n";
1320 outs() << " cpusubtype CPU_SUBTYPE_ARM64_ALL\n";
1321 break;
1322 default:
1323 printUnknownCPUType(cputype, cpusubtype);
1324 break;
1325 }
1326 break;
1327 default:
1328 printUnknownCPUType(cputype, cpusubtype);
1329 break;
1330 }
1331}
1332
1333static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB,
1334 bool verbose) {
1335 outs() << "Fat headers\n";
1336 if (verbose)
1337 outs() << "fat_magic FAT_MAGIC\n";
1338 else
1339 outs() << "fat_magic " << format("0x%" PRIx32, MachO::FAT_MAGIC) << "\n";
1340
1341 uint32_t nfat_arch = UB->getNumberOfObjects();
1342 StringRef Buf = UB->getData();
1343 uint64_t size = Buf.size();
1344 uint64_t big_size = sizeof(struct MachO::fat_header) +
1345 nfat_arch * sizeof(struct MachO::fat_arch);
1346 outs() << "nfat_arch " << UB->getNumberOfObjects();
1347 if (nfat_arch == 0)
1348 outs() << " (malformed, contains zero architecture types)\n";
1349 else if (big_size > size)
1350 outs() << " (malformed, architectures past end of file)\n";
1351 else
1352 outs() << "\n";
1353
1354 for (uint32_t i = 0; i < nfat_arch; ++i) {
1355 MachOUniversalBinary::ObjectForArch OFA(UB, i);
1356 uint32_t cputype = OFA.getCPUType();
1357 uint32_t cpusubtype = OFA.getCPUSubType();
1358 outs() << "architecture ";
1359 for (uint32_t j = 0; i != 0 && j <= i - 1; j++) {
1360 MachOUniversalBinary::ObjectForArch other_OFA(UB, j);
1361 uint32_t other_cputype = other_OFA.getCPUType();
1362 uint32_t other_cpusubtype = other_OFA.getCPUSubType();
Kevin Enderby0512bd72015-01-09 21:55:03 +00001363 if (cputype != 0 && cpusubtype != 0 && cputype == other_cputype &&
Kevin Enderby131d1772015-01-09 19:22:37 +00001364 (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) ==
Kevin Enderby0512bd72015-01-09 21:55:03 +00001365 (other_cpusubtype & ~MachO::CPU_SUBTYPE_MASK)) {
Kevin Enderby131d1772015-01-09 19:22:37 +00001366 outs() << "(illegal duplicate architecture) ";
1367 break;
Kevin Enderby0512bd72015-01-09 21:55:03 +00001368 }
Kevin Enderby131d1772015-01-09 19:22:37 +00001369 }
1370 if (verbose) {
1371 outs() << OFA.getArchTypeName() << "\n";
1372 printCPUType(cputype, cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
1373 } else {
1374 outs() << i << "\n";
1375 outs() << " cputype " << cputype << "\n";
1376 outs() << " cpusubtype " << (cpusubtype & ~MachO::CPU_SUBTYPE_MASK)
1377 << "\n";
1378 }
1379 if (verbose &&
1380 (cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64)
1381 outs() << " capabilities CPU_SUBTYPE_LIB64\n";
1382 else
1383 outs() << " capabilities "
1384 << format("0x%" PRIx32,
1385 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24) << "\n";
1386 outs() << " offset " << OFA.getOffset();
1387 if (OFA.getOffset() > size)
1388 outs() << " (past end of file)";
1389 if (OFA.getOffset() % (1 << OFA.getAlign()) != 0)
1390 outs() << " (not aligned on it's alignment (2^" << OFA.getAlign() << ")";
1391 outs() << "\n";
1392 outs() << " size " << OFA.getSize();
1393 big_size = OFA.getOffset() + OFA.getSize();
1394 if (big_size > size)
1395 outs() << " (past end of file)";
1396 outs() << "\n";
1397 outs() << " align 2^" << OFA.getAlign() << " (" << (1 << OFA.getAlign())
1398 << ")\n";
1399 }
1400}
1401
Kevin Enderby13023a12015-01-15 23:19:11 +00001402static void printArchiveChild(Archive::Child &C, bool verbose,
1403 bool print_offset) {
1404 if (print_offset)
1405 outs() << C.getChildOffset() << "\t";
1406 sys::fs::perms Mode = C.getAccessMode();
1407 if (verbose) {
1408 // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG.
1409 // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG.
1410 outs() << "-";
1411 if (Mode & sys::fs::owner_read)
1412 outs() << "r";
1413 else
1414 outs() << "-";
1415 if (Mode & sys::fs::owner_write)
1416 outs() << "w";
1417 else
1418 outs() << "-";
1419 if (Mode & sys::fs::owner_exe)
1420 outs() << "x";
1421 else
1422 outs() << "-";
1423 if (Mode & sys::fs::group_read)
1424 outs() << "r";
1425 else
1426 outs() << "-";
1427 if (Mode & sys::fs::group_write)
1428 outs() << "w";
1429 else
1430 outs() << "-";
1431 if (Mode & sys::fs::group_exe)
1432 outs() << "x";
1433 else
1434 outs() << "-";
1435 if (Mode & sys::fs::others_read)
1436 outs() << "r";
1437 else
1438 outs() << "-";
1439 if (Mode & sys::fs::others_write)
1440 outs() << "w";
1441 else
1442 outs() << "-";
1443 if (Mode & sys::fs::others_exe)
1444 outs() << "x";
1445 else
1446 outs() << "-";
1447 } else {
1448 outs() << format("0%o ", Mode);
1449 }
1450
1451 unsigned UID = C.getUID();
1452 outs() << format("%3d/", UID);
1453 unsigned GID = C.getGID();
1454 outs() << format("%-3d ", GID);
Kevin Enderbyc12718932015-01-16 22:10:36 +00001455 uint64_t Size = C.getRawSize();
Kevin Enderby479ee612015-01-23 21:02:44 +00001456 outs() << format("%5" PRId64, Size) << " ";
Kevin Enderby13023a12015-01-15 23:19:11 +00001457
1458 StringRef RawLastModified = C.getRawLastModified();
1459 if (verbose) {
1460 unsigned Seconds;
1461 if (RawLastModified.getAsInteger(10, Seconds))
1462 outs() << "(date: \"%s\" contains non-decimal chars) " << RawLastModified;
1463 else {
1464 // Since cime(3) returns a 26 character string of the form:
1465 // "Sun Sep 16 01:03:52 1973\n\0"
1466 // just print 24 characters.
1467 time_t t = Seconds;
1468 outs() << format("%.24s ", ctime(&t));
1469 }
1470 } else {
1471 outs() << RawLastModified << " ";
1472 }
1473
1474 if (verbose) {
1475 ErrorOr<StringRef> NameOrErr = C.getName();
1476 if (NameOrErr.getError()) {
1477 StringRef RawName = C.getRawName();
1478 outs() << RawName << "\n";
1479 } else {
1480 StringRef Name = NameOrErr.get();
1481 outs() << Name << "\n";
1482 }
1483 } else {
1484 StringRef RawName = C.getRawName();
1485 outs() << RawName << "\n";
1486 }
1487}
1488
1489static void printArchiveHeaders(Archive *A, bool verbose, bool print_offset) {
1490 if (A->hasSymbolTable()) {
1491 Archive::child_iterator S = A->getSymbolTableChild();
1492 Archive::Child C = *S;
1493 printArchiveChild(C, verbose, print_offset);
1494 }
1495 for (Archive::child_iterator I = A->child_begin(), E = A->child_end(); I != E;
1496 ++I) {
1497 Archive::Child C = *I;
1498 printArchiveChild(C, verbose, print_offset);
1499 }
1500}
1501
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001502// ParseInputMachO() parses the named Mach-O file in Filename and handles the
1503// -arch flags selecting just those slices as specified by them and also parses
1504// archive files. Then for each individual Mach-O file ProcessMachO() is
1505// called to process the file based on the command line options.
1506void llvm::ParseInputMachO(StringRef Filename) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001507 // Check for -arch all and verifiy the -arch flags are valid.
1508 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
1509 if (ArchFlags[i] == "all") {
1510 ArchAll = true;
1511 } else {
1512 if (!MachOObjectFile::isValidArch(ArchFlags[i])) {
1513 errs() << "llvm-objdump: Unknown architecture named '" + ArchFlags[i] +
1514 "'for the -arch option\n";
1515 return;
1516 }
1517 }
1518 }
1519
1520 // Attempt to open the binary.
1521 ErrorOr<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename);
1522 if (std::error_code EC = BinaryOrErr.getError()) {
1523 errs() << "llvm-objdump: '" << Filename << "': " << EC.message() << ".\n";
Rafael Espindolade882cd2014-12-03 23:29:34 +00001524 return;
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001525 }
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001526 Binary &Bin = *BinaryOrErr.get().getBinary();
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001527
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001528 if (Archive *A = dyn_cast<Archive>(&Bin)) {
1529 outs() << "Archive : " << Filename << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001530 if (ArchiveHeaders)
1531 printArchiveHeaders(A, true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001532 for (Archive::child_iterator I = A->child_begin(), E = A->child_end();
1533 I != E; ++I) {
1534 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = I->getAsBinary();
1535 if (ChildOrErr.getError())
1536 continue;
1537 if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
1538 if (!checkMachOAndArchFlags(O, Filename))
1539 return;
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001540 ProcessMachO(Filename, O, O->getFileName());
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001541 }
1542 }
1543 return;
1544 }
Kevin Enderby131d1772015-01-09 19:22:37 +00001545 if (UniversalHeaders) {
1546 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin))
Kevin Enderbyf0640752015-03-13 17:56:32 +00001547 printMachOUniversalHeaders(UB, !NonVerbose);
Kevin Enderby131d1772015-01-09 19:22:37 +00001548 }
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001549 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) {
1550 // If we have a list of architecture flags specified dump only those.
1551 if (!ArchAll && ArchFlags.size() != 0) {
1552 // Look for a slice in the universal binary that matches each ArchFlag.
1553 bool ArchFound;
1554 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
1555 ArchFound = false;
1556 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
1557 E = UB->end_objects();
1558 I != E; ++I) {
1559 if (ArchFlags[i] == I->getArchTypeName()) {
1560 ArchFound = true;
1561 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr =
1562 I->getAsObjectFile();
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001563 std::string ArchitectureName = "";
1564 if (ArchFlags.size() > 1)
1565 ArchitectureName = I->getArchTypeName();
1566 if (ObjOrErr) {
1567 ObjectFile &O = *ObjOrErr.get();
1568 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001569 ProcessMachO(Filename, MachOOF, "", ArchitectureName);
Rafael Espindola0bfe8282014-12-09 21:05:36 +00001570 } else if (ErrorOr<std::unique_ptr<Archive>> AOrErr =
1571 I->getAsArchive()) {
1572 std::unique_ptr<Archive> &A = *AOrErr;
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001573 outs() << "Archive : " << Filename;
1574 if (!ArchitectureName.empty())
1575 outs() << " (architecture " << ArchitectureName << ")";
1576 outs() << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001577 if (ArchiveHeaders)
1578 printArchiveHeaders(A.get(), true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001579 for (Archive::child_iterator AI = A->child_begin(),
1580 AE = A->child_end();
1581 AI != AE; ++AI) {
1582 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
1583 if (ChildOrErr.getError())
1584 continue;
1585 if (MachOObjectFile *O =
1586 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001587 ProcessMachO(Filename, O, O->getFileName(), ArchitectureName);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001588 }
1589 }
1590 }
1591 }
1592 if (!ArchFound) {
1593 errs() << "llvm-objdump: file: " + Filename + " does not contain "
1594 << "architecture: " + ArchFlags[i] + "\n";
1595 return;
1596 }
1597 }
1598 return;
1599 }
1600 // No architecture flags were specified so if this contains a slice that
1601 // matches the host architecture dump only that.
1602 if (!ArchAll) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001603 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
1604 E = UB->end_objects();
1605 I != E; ++I) {
Kevin Enderby0512bd72015-01-09 21:55:03 +00001606 if (MachOObjectFile::getHostArch().getArchName() ==
1607 I->getArchTypeName()) {
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001608 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001609 std::string ArchiveName;
1610 ArchiveName.clear();
1611 if (ObjOrErr) {
1612 ObjectFile &O = *ObjOrErr.get();
1613 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001614 ProcessMachO(Filename, MachOOF);
Rafael Espindola0bfe8282014-12-09 21:05:36 +00001615 } else if (ErrorOr<std::unique_ptr<Archive>> AOrErr =
1616 I->getAsArchive()) {
1617 std::unique_ptr<Archive> &A = *AOrErr;
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001618 outs() << "Archive : " << Filename << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001619 if (ArchiveHeaders)
1620 printArchiveHeaders(A.get(), true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001621 for (Archive::child_iterator AI = A->child_begin(),
1622 AE = A->child_end();
1623 AI != AE; ++AI) {
1624 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
1625 if (ChildOrErr.getError())
1626 continue;
1627 if (MachOObjectFile *O =
1628 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001629 ProcessMachO(Filename, O, O->getFileName());
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001630 }
1631 }
1632 return;
1633 }
1634 }
1635 }
1636 // Either all architectures have been specified or none have been specified
1637 // and this does not contain the host architecture so dump all the slices.
1638 bool moreThanOneArch = UB->getNumberOfObjects() > 1;
1639 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
1640 E = UB->end_objects();
1641 I != E; ++I) {
1642 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001643 std::string ArchitectureName = "";
1644 if (moreThanOneArch)
1645 ArchitectureName = I->getArchTypeName();
1646 if (ObjOrErr) {
1647 ObjectFile &Obj = *ObjOrErr.get();
1648 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001649 ProcessMachO(Filename, MachOOF, "", ArchitectureName);
Rafael Espindola0bfe8282014-12-09 21:05:36 +00001650 } else if (ErrorOr<std::unique_ptr<Archive>> AOrErr = I->getAsArchive()) {
1651 std::unique_ptr<Archive> &A = *AOrErr;
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001652 outs() << "Archive : " << Filename;
1653 if (!ArchitectureName.empty())
1654 outs() << " (architecture " << ArchitectureName << ")";
1655 outs() << "\n";
Kevin Enderby13023a12015-01-15 23:19:11 +00001656 if (ArchiveHeaders)
1657 printArchiveHeaders(A.get(), true, false);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001658 for (Archive::child_iterator AI = A->child_begin(), AE = A->child_end();
1659 AI != AE; ++AI) {
1660 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
1661 if (ChildOrErr.getError())
1662 continue;
1663 if (MachOObjectFile *O =
1664 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
1665 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O))
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001666 ProcessMachO(Filename, MachOOF, MachOOF->getFileName(),
1667 ArchitectureName);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001668 }
1669 }
1670 }
1671 }
1672 return;
1673 }
1674 if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) {
1675 if (!checkMachOAndArchFlags(O, Filename))
1676 return;
1677 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O)) {
Kevin Enderbye2297dd2015-01-07 21:02:18 +00001678 ProcessMachO(Filename, MachOOF);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00001679 } else
1680 errs() << "llvm-objdump: '" << Filename << "': "
1681 << "Object is not a Mach-O file type.\n";
1682 } else
1683 errs() << "llvm-objdump: '" << Filename << "': "
1684 << "Unrecognized file type.\n";
Rafael Espindola9b709252013-04-13 01:45:40 +00001685}
1686
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001687typedef std::pair<uint64_t, const char *> BindInfoEntry;
1688typedef std::vector<BindInfoEntry> BindTable;
1689typedef BindTable::iterator bind_table_iterator;
Kevin Enderbybf246f52014-09-24 23:08:22 +00001690
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001691// The block of info used by the Symbolizer call backs.
1692struct DisassembleInfo {
1693 bool verbose;
1694 MachOObjectFile *O;
1695 SectionRef S;
Kevin Enderbybf246f52014-09-24 23:08:22 +00001696 SymbolAddressMap *AddrMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001697 std::vector<SectionRef> *Sections;
1698 const char *class_name;
1699 const char *selector_name;
1700 char *method;
Kevin Enderby04bf6932014-10-28 23:39:46 +00001701 char *demangled_name;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001702 uint64_t adrp_addr;
1703 uint32_t adrp_inst;
Kevin Enderby078be602014-10-23 19:53:12 +00001704 BindTable *bindtable;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001705};
1706
1707// SymbolizerGetOpInfo() is the operand information call back function.
1708// This is called to get the symbolic information for operand(s) of an
1709// instruction when it is being done. This routine does this from
1710// the relocation information, symbol table, etc. That block of information
1711// is a pointer to the struct DisassembleInfo that was passed when the
1712// disassembler context was created and passed to back to here when
1713// called back by the disassembler for instruction operands that could have
1714// relocation information. The address of the instruction containing operand is
1715// at the Pc parameter. The immediate value the operand has is passed in
1716// op_info->Value and is at Offset past the start of the instruction and has a
1717// byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the
1718// LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol
1719// names and addends of the symbolic expression to add for the operand. The
1720// value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic
1721// information is returned then this function returns 1 else it returns 0.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00001722static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset,
1723 uint64_t Size, int TagType, void *TagBuf) {
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001724 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
1725 struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001726 uint64_t value = op_info->Value;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001727
1728 // Make sure all fields returned are zero if we don't set them.
1729 memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1));
1730 op_info->Value = value;
1731
1732 // If the TagType is not the value 1 which it code knows about or if no
1733 // verbose symbolic information is wanted then just return 0, indicating no
1734 // information is being returned.
1735 if (TagType != 1 || info->verbose == false)
1736 return 0;
1737
1738 unsigned int Arch = info->O->getArch();
1739 if (Arch == Triple::x86) {
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001740 if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
1741 return 0;
1742 // First search the section's relocation entries (if any) for an entry
1743 // for this section offset.
1744 uint32_t sect_addr = info->S.getAddress();
1745 uint32_t sect_offset = (Pc + Offset) - sect_addr;
1746 bool reloc_found = false;
1747 DataRefImpl Rel;
1748 MachO::any_relocation_info RE;
1749 bool isExtern = false;
1750 SymbolRef Symbol;
1751 bool r_scattered = false;
1752 uint32_t r_value, pair_r_value, r_type;
1753 for (const RelocationRef &Reloc : info->S.relocations()) {
1754 uint64_t RelocOffset;
1755 Reloc.getOffset(RelocOffset);
1756 if (RelocOffset == sect_offset) {
1757 Rel = Reloc.getRawDataRefImpl();
1758 RE = info->O->getRelocation(Rel);
Kevin Enderby3eb73e12014-11-11 19:16:45 +00001759 r_type = info->O->getAnyRelocationType(RE);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001760 r_scattered = info->O->isRelocationScattered(RE);
1761 if (r_scattered) {
1762 r_value = info->O->getScatteredRelocationValue(RE);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001763 if (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
1764 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) {
1765 DataRefImpl RelNext = Rel;
1766 info->O->moveRelocationNext(RelNext);
1767 MachO::any_relocation_info RENext;
1768 RENext = info->O->getRelocation(RelNext);
1769 if (info->O->isRelocationScattered(RENext))
Kevin Enderby930fdc72014-11-06 19:00:13 +00001770 pair_r_value = info->O->getScatteredRelocationValue(RENext);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001771 else
1772 return 0;
1773 }
1774 } else {
1775 isExtern = info->O->getPlainRelocationExternal(RE);
1776 if (isExtern) {
1777 symbol_iterator RelocSym = Reloc.getSymbol();
1778 Symbol = *RelocSym;
1779 }
1780 }
1781 reloc_found = true;
1782 break;
1783 }
1784 }
1785 if (reloc_found && isExtern) {
1786 StringRef SymName;
1787 Symbol.getName(SymName);
1788 const char *name = SymName.data();
1789 op_info->AddSymbol.Present = 1;
1790 op_info->AddSymbol.Name = name;
1791 // For i386 extern relocation entries the value in the instruction is
1792 // the offset from the symbol, and value is already set in op_info->Value.
1793 return 1;
1794 }
1795 if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
1796 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) {
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001797 const char *add = GuessSymbolName(r_value, info->AddrMap);
1798 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001799 uint32_t offset = value - (r_value - pair_r_value);
1800 op_info->AddSymbol.Present = 1;
1801 if (add != nullptr)
1802 op_info->AddSymbol.Name = add;
1803 else
1804 op_info->AddSymbol.Value = r_value;
1805 op_info->SubtractSymbol.Present = 1;
1806 if (sub != nullptr)
1807 op_info->SubtractSymbol.Name = sub;
1808 else
1809 op_info->SubtractSymbol.Value = pair_r_value;
1810 op_info->Value = offset;
1811 return 1;
1812 }
1813 // TODO:
1814 // Second search the external relocation entries of a fully linked image
1815 // (if any) for an entry that matches this segment offset.
1816 // uint32_t seg_offset = (Pc + Offset);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001817 return 0;
1818 } else if (Arch == Triple::x86_64) {
1819 if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
1820 return 0;
1821 // First search the section's relocation entries (if any) for an entry
1822 // for this section offset.
Rafael Espindola80291272014-10-08 15:28:58 +00001823 uint64_t sect_addr = info->S.getAddress();
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001824 uint64_t sect_offset = (Pc + Offset) - sect_addr;
1825 bool reloc_found = false;
1826 DataRefImpl Rel;
1827 MachO::any_relocation_info RE;
1828 bool isExtern = false;
1829 SymbolRef Symbol;
1830 for (const RelocationRef &Reloc : info->S.relocations()) {
1831 uint64_t RelocOffset;
1832 Reloc.getOffset(RelocOffset);
1833 if (RelocOffset == sect_offset) {
1834 Rel = Reloc.getRawDataRefImpl();
1835 RE = info->O->getRelocation(Rel);
1836 // NOTE: Scattered relocations don't exist on x86_64.
1837 isExtern = info->O->getPlainRelocationExternal(RE);
1838 if (isExtern) {
1839 symbol_iterator RelocSym = Reloc.getSymbol();
1840 Symbol = *RelocSym;
1841 }
1842 reloc_found = true;
1843 break;
1844 }
1845 }
1846 if (reloc_found && isExtern) {
1847 // The Value passed in will be adjusted by the Pc if the instruction
1848 // adds the Pc. But for x86_64 external relocation entries the Value
1849 // is the offset from the external symbol.
1850 if (info->O->getAnyRelocationPCRel(RE))
1851 op_info->Value -= Pc + Offset + Size;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001852 StringRef SymName;
1853 Symbol.getName(SymName);
1854 const char *name = SymName.data();
1855 unsigned Type = info->O->getAnyRelocationType(RE);
1856 if (Type == MachO::X86_64_RELOC_SUBTRACTOR) {
1857 DataRefImpl RelNext = Rel;
1858 info->O->moveRelocationNext(RelNext);
1859 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
1860 unsigned TypeNext = info->O->getAnyRelocationType(RENext);
1861 bool isExternNext = info->O->getPlainRelocationExternal(RENext);
1862 unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext);
1863 if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) {
1864 op_info->SubtractSymbol.Present = 1;
1865 op_info->SubtractSymbol.Name = name;
1866 symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum);
1867 Symbol = *RelocSymNext;
1868 StringRef SymNameNext;
1869 Symbol.getName(SymNameNext);
1870 name = SymNameNext.data();
1871 }
1872 }
1873 // TODO: add the VariantKinds to op_info->VariantKind for relocation types
1874 // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT.
1875 op_info->AddSymbol.Present = 1;
1876 op_info->AddSymbol.Name = name;
1877 return 1;
1878 }
1879 // TODO:
1880 // Second search the external relocation entries of a fully linked image
1881 // (if any) for an entry that matches this segment offset.
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001882 // uint64_t seg_offset = (Pc + Offset);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001883 return 0;
1884 } else if (Arch == Triple::arm) {
Kevin Enderby930fdc72014-11-06 19:00:13 +00001885 if (Offset != 0 || (Size != 4 && Size != 2))
1886 return 0;
1887 // First search the section's relocation entries (if any) for an entry
1888 // for this section offset.
1889 uint32_t sect_addr = info->S.getAddress();
1890 uint32_t sect_offset = (Pc + Offset) - sect_addr;
1891 bool reloc_found = false;
1892 DataRefImpl Rel;
1893 MachO::any_relocation_info RE;
1894 bool isExtern = false;
1895 SymbolRef Symbol;
1896 bool r_scattered = false;
1897 uint32_t r_value, pair_r_value, r_type, r_length, other_half;
1898 for (const RelocationRef &Reloc : info->S.relocations()) {
1899 uint64_t RelocOffset;
1900 Reloc.getOffset(RelocOffset);
1901 if (RelocOffset == sect_offset) {
1902 Rel = Reloc.getRawDataRefImpl();
1903 RE = info->O->getRelocation(Rel);
1904 r_length = info->O->getAnyRelocationLength(RE);
1905 r_scattered = info->O->isRelocationScattered(RE);
1906 if (r_scattered) {
1907 r_value = info->O->getScatteredRelocationValue(RE);
1908 r_type = info->O->getScatteredRelocationType(RE);
1909 } else {
1910 r_type = info->O->getAnyRelocationType(RE);
1911 isExtern = info->O->getPlainRelocationExternal(RE);
1912 if (isExtern) {
1913 symbol_iterator RelocSym = Reloc.getSymbol();
1914 Symbol = *RelocSym;
1915 }
1916 }
1917 if (r_type == MachO::ARM_RELOC_HALF ||
1918 r_type == MachO::ARM_RELOC_SECTDIFF ||
1919 r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
1920 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1921 DataRefImpl RelNext = Rel;
1922 info->O->moveRelocationNext(RelNext);
1923 MachO::any_relocation_info RENext;
1924 RENext = info->O->getRelocation(RelNext);
1925 other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff;
1926 if (info->O->isRelocationScattered(RENext))
1927 pair_r_value = info->O->getScatteredRelocationValue(RENext);
1928 }
1929 reloc_found = true;
1930 break;
1931 }
1932 }
1933 if (reloc_found && isExtern) {
1934 StringRef SymName;
1935 Symbol.getName(SymName);
1936 const char *name = SymName.data();
1937 op_info->AddSymbol.Present = 1;
1938 op_info->AddSymbol.Name = name;
Sylvestre Ledru648cced2015-02-05 17:00:23 +00001939 switch (r_type) {
1940 case MachO::ARM_RELOC_HALF:
1941 if ((r_length & 0x1) == 1) {
1942 op_info->Value = value << 16 | other_half;
1943 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
1944 } else {
1945 op_info->Value = other_half << 16 | value;
1946 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
Sylvestre Ledrufe0c7ad2015-02-05 16:35:44 +00001947 }
Sylvestre Ledru648cced2015-02-05 17:00:23 +00001948 break;
1949 default:
1950 break;
Kevin Enderby930fdc72014-11-06 19:00:13 +00001951 }
1952 return 1;
1953 }
1954 // If we have a branch that is not an external relocation entry then
1955 // return 0 so the code in tryAddingSymbolicOperand() can use the
1956 // SymbolLookUp call back with the branch target address to look up the
1957 // symbol and possiblity add an annotation for a symbol stub.
1958 if (reloc_found && isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 ||
1959 r_type == MachO::ARM_THUMB_RELOC_BR22))
1960 return 0;
1961
1962 uint32_t offset = 0;
1963 if (reloc_found) {
1964 if (r_type == MachO::ARM_RELOC_HALF ||
1965 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1966 if ((r_length & 0x1) == 1)
1967 value = value << 16 | other_half;
1968 else
1969 value = other_half << 16 | value;
1970 }
1971 if (r_scattered && (r_type != MachO::ARM_RELOC_HALF &&
1972 r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) {
1973 offset = value - r_value;
1974 value = r_value;
1975 }
1976 }
1977
1978 if (reloc_found && r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1979 if ((r_length & 0x1) == 1)
1980 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
1981 else
1982 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
Kevin Enderbyf6d25852015-01-31 00:37:11 +00001983 const char *add = GuessSymbolName(r_value, info->AddrMap);
1984 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap);
Kevin Enderby930fdc72014-11-06 19:00:13 +00001985 int32_t offset = value - (r_value - pair_r_value);
1986 op_info->AddSymbol.Present = 1;
1987 if (add != nullptr)
1988 op_info->AddSymbol.Name = add;
1989 else
1990 op_info->AddSymbol.Value = r_value;
1991 op_info->SubtractSymbol.Present = 1;
1992 if (sub != nullptr)
1993 op_info->SubtractSymbol.Name = sub;
1994 else
1995 op_info->SubtractSymbol.Value = pair_r_value;
1996 op_info->Value = offset;
1997 return 1;
1998 }
1999
2000 if (reloc_found == false)
2001 return 0;
2002
2003 op_info->AddSymbol.Present = 1;
2004 op_info->Value = offset;
2005 if (reloc_found) {
2006 if (r_type == MachO::ARM_RELOC_HALF) {
2007 if ((r_length & 0x1) == 1)
2008 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
2009 else
2010 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
2011 }
2012 }
Kevin Enderbyf6d25852015-01-31 00:37:11 +00002013 const char *add = GuessSymbolName(value, info->AddrMap);
Kevin Enderby930fdc72014-11-06 19:00:13 +00002014 if (add != nullptr) {
2015 op_info->AddSymbol.Name = add;
2016 return 1;
2017 }
2018 op_info->AddSymbol.Value = value;
2019 return 1;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002020 } else if (Arch == Triple::aarch64) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002021 if (Offset != 0 || Size != 4)
2022 return 0;
2023 // First search the section's relocation entries (if any) for an entry
2024 // for this section offset.
2025 uint64_t sect_addr = info->S.getAddress();
2026 uint64_t sect_offset = (Pc + Offset) - sect_addr;
2027 bool reloc_found = false;
2028 DataRefImpl Rel;
2029 MachO::any_relocation_info RE;
2030 bool isExtern = false;
2031 SymbolRef Symbol;
2032 uint32_t r_type = 0;
2033 for (const RelocationRef &Reloc : info->S.relocations()) {
2034 uint64_t RelocOffset;
2035 Reloc.getOffset(RelocOffset);
2036 if (RelocOffset == sect_offset) {
2037 Rel = Reloc.getRawDataRefImpl();
2038 RE = info->O->getRelocation(Rel);
2039 r_type = info->O->getAnyRelocationType(RE);
2040 if (r_type == MachO::ARM64_RELOC_ADDEND) {
2041 DataRefImpl RelNext = Rel;
2042 info->O->moveRelocationNext(RelNext);
2043 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
2044 if (value == 0) {
2045 value = info->O->getPlainRelocationSymbolNum(RENext);
2046 op_info->Value = value;
2047 }
2048 }
2049 // NOTE: Scattered relocations don't exist on arm64.
2050 isExtern = info->O->getPlainRelocationExternal(RE);
2051 if (isExtern) {
2052 symbol_iterator RelocSym = Reloc.getSymbol();
2053 Symbol = *RelocSym;
2054 }
2055 reloc_found = true;
2056 break;
2057 }
2058 }
2059 if (reloc_found && isExtern) {
2060 StringRef SymName;
2061 Symbol.getName(SymName);
2062 const char *name = SymName.data();
2063 op_info->AddSymbol.Present = 1;
2064 op_info->AddSymbol.Name = name;
2065
2066 switch (r_type) {
2067 case MachO::ARM64_RELOC_PAGE21:
2068 /* @page */
2069 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE;
2070 break;
2071 case MachO::ARM64_RELOC_PAGEOFF12:
2072 /* @pageoff */
2073 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF;
2074 break;
2075 case MachO::ARM64_RELOC_GOT_LOAD_PAGE21:
2076 /* @gotpage */
2077 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE;
2078 break;
2079 case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12:
2080 /* @gotpageoff */
2081 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF;
2082 break;
2083 case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21:
2084 /* @tvlppage is not implemented in llvm-mc */
2085 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP;
2086 break;
2087 case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12:
2088 /* @tvlppageoff is not implemented in llvm-mc */
2089 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF;
2090 break;
2091 default:
2092 case MachO::ARM64_RELOC_BRANCH26:
2093 op_info->VariantKind = LLVMDisassembler_VariantKind_None;
2094 break;
2095 }
2096 return 1;
2097 }
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002098 return 0;
2099 } else {
2100 return 0;
2101 }
2102}
2103
Kevin Enderbybf246f52014-09-24 23:08:22 +00002104// GuessCstringPointer is passed the address of what might be a pointer to a
2105// literal string in a cstring section. If that address is in a cstring section
2106// it returns a pointer to that string. Else it returns nullptr.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002107static const char *GuessCstringPointer(uint64_t ReferenceValue,
2108 struct DisassembleInfo *info) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002109 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
2110 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
2111 for (unsigned I = 0;; ++I) {
2112 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
2113 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
2114 for (unsigned J = 0; J < Seg.nsects; ++J) {
2115 MachO::section_64 Sec = info->O->getSection64(Load, J);
2116 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2117 if (section_type == MachO::S_CSTRING_LITERALS &&
2118 ReferenceValue >= Sec.addr &&
2119 ReferenceValue < Sec.addr + Sec.size) {
2120 uint64_t sect_offset = ReferenceValue - Sec.addr;
2121 uint64_t object_offset = Sec.offset + sect_offset;
2122 StringRef MachOContents = info->O->getData();
2123 uint64_t object_size = MachOContents.size();
2124 const char *object_addr = (const char *)MachOContents.data();
2125 if (object_offset < object_size) {
2126 const char *name = object_addr + object_offset;
2127 return name;
2128 } else {
2129 return nullptr;
2130 }
2131 }
2132 }
2133 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
2134 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
2135 for (unsigned J = 0; J < Seg.nsects; ++J) {
2136 MachO::section Sec = info->O->getSection(Load, J);
2137 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2138 if (section_type == MachO::S_CSTRING_LITERALS &&
2139 ReferenceValue >= Sec.addr &&
2140 ReferenceValue < Sec.addr + Sec.size) {
2141 uint64_t sect_offset = ReferenceValue - Sec.addr;
2142 uint64_t object_offset = Sec.offset + sect_offset;
2143 StringRef MachOContents = info->O->getData();
2144 uint64_t object_size = MachOContents.size();
2145 const char *object_addr = (const char *)MachOContents.data();
2146 if (object_offset < object_size) {
2147 const char *name = object_addr + object_offset;
2148 return name;
2149 } else {
2150 return nullptr;
2151 }
2152 }
2153 }
2154 }
2155 if (I == LoadCommandCount - 1)
2156 break;
2157 else
2158 Load = info->O->getNextLoadCommandInfo(Load);
2159 }
2160 return nullptr;
2161}
2162
Kevin Enderby85974882014-09-26 22:20:44 +00002163// GuessIndirectSymbol returns the name of the indirect symbol for the
2164// ReferenceValue passed in or nullptr. This is used when ReferenceValue maybe
2165// an address of a symbol stub or a lazy or non-lazy pointer to associate the
2166// symbol name being referenced by the stub or pointer.
2167static const char *GuessIndirectSymbol(uint64_t ReferenceValue,
2168 struct DisassembleInfo *info) {
2169 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
2170 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
2171 MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand();
2172 MachO::symtab_command Symtab = info->O->getSymtabLoadCommand();
2173 for (unsigned I = 0;; ++I) {
2174 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
2175 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
2176 for (unsigned J = 0; J < Seg.nsects; ++J) {
2177 MachO::section_64 Sec = info->O->getSection64(Load, J);
2178 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2179 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
2180 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
2181 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
2182 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
2183 section_type == MachO::S_SYMBOL_STUBS) &&
2184 ReferenceValue >= Sec.addr &&
2185 ReferenceValue < Sec.addr + Sec.size) {
2186 uint32_t stride;
2187 if (section_type == MachO::S_SYMBOL_STUBS)
2188 stride = Sec.reserved2;
2189 else
2190 stride = 8;
2191 if (stride == 0)
2192 return nullptr;
2193 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
2194 if (index < Dysymtab.nindirectsyms) {
2195 uint32_t indirect_symbol =
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002196 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
Kevin Enderby85974882014-09-26 22:20:44 +00002197 if (indirect_symbol < Symtab.nsyms) {
2198 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
2199 SymbolRef Symbol = *Sym;
2200 StringRef SymName;
2201 Symbol.getName(SymName);
2202 const char *name = SymName.data();
2203 return name;
2204 }
2205 }
2206 }
2207 }
2208 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
2209 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
2210 for (unsigned J = 0; J < Seg.nsects; ++J) {
2211 MachO::section Sec = info->O->getSection(Load, J);
2212 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
2213 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
2214 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
2215 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
2216 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
2217 section_type == MachO::S_SYMBOL_STUBS) &&
2218 ReferenceValue >= Sec.addr &&
2219 ReferenceValue < Sec.addr + Sec.size) {
2220 uint32_t stride;
2221 if (section_type == MachO::S_SYMBOL_STUBS)
2222 stride = Sec.reserved2;
2223 else
2224 stride = 4;
2225 if (stride == 0)
2226 return nullptr;
2227 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
2228 if (index < Dysymtab.nindirectsyms) {
2229 uint32_t indirect_symbol =
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002230 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
Kevin Enderby85974882014-09-26 22:20:44 +00002231 if (indirect_symbol < Symtab.nsyms) {
2232 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
2233 SymbolRef Symbol = *Sym;
2234 StringRef SymName;
2235 Symbol.getName(SymName);
2236 const char *name = SymName.data();
2237 return name;
2238 }
2239 }
2240 }
2241 }
2242 }
2243 if (I == LoadCommandCount - 1)
2244 break;
2245 else
2246 Load = info->O->getNextLoadCommandInfo(Load);
2247 }
2248 return nullptr;
2249}
2250
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002251// method_reference() is called passing it the ReferenceName that might be
2252// a reference it to an Objective-C method call. If so then it allocates and
2253// assembles a method call string with the values last seen and saved in
2254// the DisassembleInfo's class_name and selector_name fields. This is saved
2255// into the method field of the info and any previous string is free'ed.
2256// Then the class_name field in the info is set to nullptr. The method call
2257// string is set into ReferenceName and ReferenceType is set to
2258// LLVMDisassembler_ReferenceType_Out_Objc_Message. If this not a method call
2259// then both ReferenceType and ReferenceName are left unchanged.
2260static void method_reference(struct DisassembleInfo *info,
2261 uint64_t *ReferenceType,
2262 const char **ReferenceName) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002263 unsigned int Arch = info->O->getArch();
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002264 if (*ReferenceName != nullptr) {
2265 if (strcmp(*ReferenceName, "_objc_msgSend") == 0) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002266 if (info->selector_name != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002267 if (info->method != nullptr)
2268 free(info->method);
2269 if (info->class_name != nullptr) {
2270 info->method = (char *)malloc(5 + strlen(info->class_name) +
2271 strlen(info->selector_name));
2272 if (info->method != nullptr) {
2273 strcpy(info->method, "+[");
2274 strcat(info->method, info->class_name);
2275 strcat(info->method, " ");
2276 strcat(info->method, info->selector_name);
2277 strcat(info->method, "]");
2278 *ReferenceName = info->method;
2279 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
2280 }
2281 } else {
2282 info->method = (char *)malloc(9 + strlen(info->selector_name));
2283 if (info->method != nullptr) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002284 if (Arch == Triple::x86_64)
2285 strcpy(info->method, "-[%rdi ");
2286 else if (Arch == Triple::aarch64)
2287 strcpy(info->method, "-[x0 ");
2288 else
2289 strcpy(info->method, "-[r? ");
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002290 strcat(info->method, info->selector_name);
2291 strcat(info->method, "]");
2292 *ReferenceName = info->method;
2293 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
2294 }
2295 }
2296 info->class_name = nullptr;
2297 }
2298 } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002299 if (info->selector_name != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002300 if (info->method != nullptr)
2301 free(info->method);
2302 info->method = (char *)malloc(17 + strlen(info->selector_name));
2303 if (info->method != nullptr) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002304 if (Arch == Triple::x86_64)
2305 strcpy(info->method, "-[[%rdi super] ");
2306 else if (Arch == Triple::aarch64)
2307 strcpy(info->method, "-[[x0 super] ");
2308 else
2309 strcpy(info->method, "-[[r? super] ");
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002310 strcat(info->method, info->selector_name);
2311 strcat(info->method, "]");
2312 *ReferenceName = info->method;
2313 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
2314 }
2315 info->class_name = nullptr;
2316 }
2317 }
2318 }
2319}
2320
2321// GuessPointerPointer() is passed the address of what might be a pointer to
2322// a reference to an Objective-C class, selector, message ref or cfstring.
2323// If so the value of the pointer is returned and one of the booleans are set
2324// to true. If not zero is returned and all the booleans are set to false.
2325static uint64_t GuessPointerPointer(uint64_t ReferenceValue,
2326 struct DisassembleInfo *info,
2327 bool &classref, bool &selref, bool &msgref,
2328 bool &cfstring) {
2329 classref = false;
2330 selref = false;
2331 msgref = false;
2332 cfstring = false;
2333 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
2334 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
2335 for (unsigned I = 0;; ++I) {
2336 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
2337 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
2338 for (unsigned J = 0; J < Seg.nsects; ++J) {
2339 MachO::section_64 Sec = info->O->getSection64(Load, J);
2340 if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 ||
2341 strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
2342 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 ||
2343 strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 ||
2344 strncmp(Sec.sectname, "__cfstring", 16) == 0) &&
2345 ReferenceValue >= Sec.addr &&
2346 ReferenceValue < Sec.addr + Sec.size) {
2347 uint64_t sect_offset = ReferenceValue - Sec.addr;
2348 uint64_t object_offset = Sec.offset + sect_offset;
2349 StringRef MachOContents = info->O->getData();
2350 uint64_t object_size = MachOContents.size();
2351 const char *object_addr = (const char *)MachOContents.data();
2352 if (object_offset < object_size) {
2353 uint64_t pointer_value;
2354 memcpy(&pointer_value, object_addr + object_offset,
2355 sizeof(uint64_t));
2356 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2357 sys::swapByteOrder(pointer_value);
2358 if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0)
2359 selref = true;
2360 else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
2361 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0)
2362 classref = true;
2363 else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 &&
2364 ReferenceValue + 8 < Sec.addr + Sec.size) {
2365 msgref = true;
2366 memcpy(&pointer_value, object_addr + object_offset + 8,
2367 sizeof(uint64_t));
2368 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2369 sys::swapByteOrder(pointer_value);
2370 } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0)
2371 cfstring = true;
2372 return pointer_value;
2373 } else {
2374 return 0;
2375 }
2376 }
2377 }
2378 }
2379 // TODO: Look for LC_SEGMENT for 32-bit Mach-O files.
2380 if (I == LoadCommandCount - 1)
2381 break;
2382 else
2383 Load = info->O->getNextLoadCommandInfo(Load);
2384 }
2385 return 0;
2386}
2387
2388// get_pointer_64 returns a pointer to the bytes in the object file at the
2389// Address from a section in the Mach-O file. And indirectly returns the
2390// offset into the section, number of bytes left in the section past the offset
2391// and which section is was being referenced. If the Address is not in a
2392// section nullptr is returned.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002393static const char *get_pointer_64(uint64_t Address, uint32_t &offset,
2394 uint32_t &left, SectionRef &S,
2395 DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002396 offset = 0;
2397 left = 0;
2398 S = SectionRef();
2399 for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) {
2400 uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress();
2401 uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize();
2402 if (Address >= SectAddress && Address < SectAddress + SectSize) {
2403 S = (*(info->Sections))[SectIdx];
2404 offset = Address - SectAddress;
2405 left = SectSize - offset;
2406 StringRef SectContents;
2407 ((*(info->Sections))[SectIdx]).getContents(SectContents);
2408 return SectContents.data() + offset;
2409 }
2410 }
2411 return nullptr;
2412}
2413
2414// get_symbol_64() returns the name of a symbol (or nullptr) and the address of
2415// the symbol indirectly through n_value. Based on the relocation information
2416// for the specified section offset in the specified section reference.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002417static const char *get_symbol_64(uint32_t sect_offset, SectionRef S,
2418 DisassembleInfo *info, uint64_t &n_value) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002419 n_value = 0;
2420 if (info->verbose == false)
2421 return nullptr;
2422
2423 // See if there is an external relocation entry at the sect_offset.
2424 bool reloc_found = false;
2425 DataRefImpl Rel;
2426 MachO::any_relocation_info RE;
2427 bool isExtern = false;
2428 SymbolRef Symbol;
2429 for (const RelocationRef &Reloc : S.relocations()) {
2430 uint64_t RelocOffset;
2431 Reloc.getOffset(RelocOffset);
2432 if (RelocOffset == sect_offset) {
2433 Rel = Reloc.getRawDataRefImpl();
2434 RE = info->O->getRelocation(Rel);
2435 if (info->O->isRelocationScattered(RE))
2436 continue;
2437 isExtern = info->O->getPlainRelocationExternal(RE);
2438 if (isExtern) {
2439 symbol_iterator RelocSym = Reloc.getSymbol();
2440 Symbol = *RelocSym;
2441 }
2442 reloc_found = true;
2443 break;
2444 }
2445 }
2446 // If there is an external relocation entry for a symbol in this section
2447 // at this section_offset then use that symbol's value for the n_value
2448 // and return its name.
2449 const char *SymbolName = nullptr;
2450 if (reloc_found && isExtern) {
2451 Symbol.getAddress(n_value);
2452 StringRef name;
2453 Symbol.getName(name);
2454 if (!name.empty()) {
2455 SymbolName = name.data();
2456 return SymbolName;
2457 }
2458 }
2459
2460 // TODO: For fully linked images, look through the external relocation
2461 // entries off the dynamic symtab command. For these the r_offset is from the
2462 // start of the first writeable segment in the Mach-O file. So the offset
2463 // to this section from that segment is passed to this routine by the caller,
2464 // as the database_offset. Which is the difference of the section's starting
2465 // address and the first writable segment.
2466 //
2467 // NOTE: need add passing the database_offset to this routine.
2468
2469 // TODO: We did not find an external relocation entry so look up the
2470 // ReferenceValue as an address of a symbol and if found return that symbol's
2471 // name.
2472 //
2473 // NOTE: need add passing the ReferenceValue to this routine. Then that code
2474 // would simply be this:
Kevin Enderbyf6d25852015-01-31 00:37:11 +00002475 // SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002476
2477 return SymbolName;
2478}
2479
2480// These are structs in the Objective-C meta data and read to produce the
2481// comments for disassembly. While these are part of the ABI they are no
2482// public defintions. So the are here not in include/llvm/Support/MachO.h .
2483
2484// The cfstring object in a 64-bit Mach-O file.
2485struct cfstring64_t {
2486 uint64_t isa; // class64_t * (64-bit pointer)
2487 uint64_t flags; // flag bits
2488 uint64_t characters; // char * (64-bit pointer)
2489 uint64_t length; // number of non-NULL characters in above
2490};
2491
2492// The class object in a 64-bit Mach-O file.
2493struct class64_t {
2494 uint64_t isa; // class64_t * (64-bit pointer)
2495 uint64_t superclass; // class64_t * (64-bit pointer)
2496 uint64_t cache; // Cache (64-bit pointer)
2497 uint64_t vtable; // IMP * (64-bit pointer)
2498 uint64_t data; // class_ro64_t * (64-bit pointer)
2499};
2500
2501struct class_ro64_t {
2502 uint32_t flags;
2503 uint32_t instanceStart;
2504 uint32_t instanceSize;
2505 uint32_t reserved;
2506 uint64_t ivarLayout; // const uint8_t * (64-bit pointer)
2507 uint64_t name; // const char * (64-bit pointer)
2508 uint64_t baseMethods; // const method_list_t * (64-bit pointer)
2509 uint64_t baseProtocols; // const protocol_list_t * (64-bit pointer)
2510 uint64_t ivars; // const ivar_list_t * (64-bit pointer)
2511 uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer)
2512 uint64_t baseProperties; // const struct objc_property_list (64-bit pointer)
2513};
2514
2515inline void swapStruct(struct cfstring64_t &cfs) {
2516 sys::swapByteOrder(cfs.isa);
2517 sys::swapByteOrder(cfs.flags);
2518 sys::swapByteOrder(cfs.characters);
2519 sys::swapByteOrder(cfs.length);
2520}
2521
2522inline void swapStruct(struct class64_t &c) {
2523 sys::swapByteOrder(c.isa);
2524 sys::swapByteOrder(c.superclass);
2525 sys::swapByteOrder(c.cache);
2526 sys::swapByteOrder(c.vtable);
2527 sys::swapByteOrder(c.data);
2528}
2529
2530inline void swapStruct(struct class_ro64_t &cro) {
2531 sys::swapByteOrder(cro.flags);
2532 sys::swapByteOrder(cro.instanceStart);
2533 sys::swapByteOrder(cro.instanceSize);
2534 sys::swapByteOrder(cro.reserved);
2535 sys::swapByteOrder(cro.ivarLayout);
2536 sys::swapByteOrder(cro.name);
2537 sys::swapByteOrder(cro.baseMethods);
2538 sys::swapByteOrder(cro.baseProtocols);
2539 sys::swapByteOrder(cro.ivars);
2540 sys::swapByteOrder(cro.weakIvarLayout);
2541 sys::swapByteOrder(cro.baseProperties);
2542}
2543
2544static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
2545 struct DisassembleInfo *info);
2546
2547// get_objc2_64bit_class_name() is used for disassembly and is passed a pointer
2548// to an Objective-C class and returns the class name. It is also passed the
2549// address of the pointer, so when the pointer is zero as it can be in an .o
2550// file, that is used to look for an external relocation entry with a symbol
2551// name.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002552static const char *get_objc2_64bit_class_name(uint64_t pointer_value,
2553 uint64_t ReferenceValue,
2554 struct DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002555 const char *r;
2556 uint32_t offset, left;
2557 SectionRef S;
2558
2559 // The pointer_value can be 0 in an object file and have a relocation
2560 // entry for the class symbol at the ReferenceValue (the address of the
2561 // pointer).
2562 if (pointer_value == 0) {
2563 r = get_pointer_64(ReferenceValue, offset, left, S, info);
2564 if (r == nullptr || left < sizeof(uint64_t))
2565 return nullptr;
2566 uint64_t n_value;
2567 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
2568 if (symbol_name == nullptr)
2569 return nullptr;
Hans Wennborgdb53e302014-10-23 21:59:17 +00002570 const char *class_name = strrchr(symbol_name, '$');
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002571 if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0')
2572 return class_name + 2;
2573 else
2574 return nullptr;
2575 }
2576
2577 // The case were the pointer_value is non-zero and points to a class defined
2578 // in this Mach-O file.
2579 r = get_pointer_64(pointer_value, offset, left, S, info);
2580 if (r == nullptr || left < sizeof(struct class64_t))
2581 return nullptr;
2582 struct class64_t c;
2583 memcpy(&c, r, sizeof(struct class64_t));
2584 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2585 swapStruct(c);
2586 if (c.data == 0)
2587 return nullptr;
2588 r = get_pointer_64(c.data, offset, left, S, info);
2589 if (r == nullptr || left < sizeof(struct class_ro64_t))
2590 return nullptr;
2591 struct class_ro64_t cro;
2592 memcpy(&cro, r, sizeof(struct class_ro64_t));
2593 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2594 swapStruct(cro);
2595 if (cro.name == 0)
2596 return nullptr;
2597 const char *name = get_pointer_64(cro.name, offset, left, S, info);
2598 return name;
2599}
2600
2601// get_objc2_64bit_cfstring_name is used for disassembly and is passed a
2602// pointer to a cfstring and returns its name or nullptr.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002603static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,
2604 struct DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002605 const char *r, *name;
2606 uint32_t offset, left;
2607 SectionRef S;
2608 struct cfstring64_t cfs;
2609 uint64_t cfs_characters;
2610
2611 r = get_pointer_64(ReferenceValue, offset, left, S, info);
2612 if (r == nullptr || left < sizeof(struct cfstring64_t))
2613 return nullptr;
2614 memcpy(&cfs, r, sizeof(struct cfstring64_t));
2615 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
2616 swapStruct(cfs);
2617 if (cfs.characters == 0) {
2618 uint64_t n_value;
2619 const char *symbol_name = get_symbol_64(
2620 offset + offsetof(struct cfstring64_t, characters), S, info, n_value);
2621 if (symbol_name == nullptr)
2622 return nullptr;
2623 cfs_characters = n_value;
2624 } else
2625 cfs_characters = cfs.characters;
2626 name = get_pointer_64(cfs_characters, offset, left, S, info);
2627
2628 return name;
2629}
2630
2631// get_objc2_64bit_selref() is used for disassembly and is passed a the address
2632// of a pointer to an Objective-C selector reference when the pointer value is
2633// zero as in a .o file and is likely to have a external relocation entry with
2634// who's symbol's n_value is the real pointer to the selector name. If that is
2635// the case the real pointer to the selector name is returned else 0 is
2636// returned
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002637static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue,
2638 struct DisassembleInfo *info) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002639 uint32_t offset, left;
2640 SectionRef S;
2641
2642 const char *r = get_pointer_64(ReferenceValue, offset, left, S, info);
2643 if (r == nullptr || left < sizeof(uint64_t))
2644 return 0;
2645 uint64_t n_value;
2646 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
2647 if (symbol_name == nullptr)
2648 return 0;
2649 return n_value;
2650}
2651
Kevin Enderbybf246f52014-09-24 23:08:22 +00002652// GuessLiteralPointer returns a string which for the item in the Mach-O file
2653// for the address passed in as ReferenceValue for printing as a comment with
2654// the instruction and also returns the corresponding type of that item
2655// indirectly through ReferenceType.
2656//
2657// If ReferenceValue is an address of literal cstring then a pointer to the
2658// cstring is returned and ReferenceType is set to
2659// LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr .
2660//
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002661// If ReferenceValue is an address of an Objective-C CFString, Selector ref or
2662// Class ref that name is returned and the ReferenceType is set accordingly.
2663//
2664// Lastly, literals which are Symbol address in a literal pool are looked for
2665// and if found the symbol name is returned and ReferenceType is set to
2666// LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr .
2667//
2668// If there is no item in the Mach-O file for the address passed in as
2669// ReferenceValue nullptr is returned and ReferenceType is unchanged.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002670static const char *GuessLiteralPointer(uint64_t ReferenceValue,
2671 uint64_t ReferencePC,
2672 uint64_t *ReferenceType,
2673 struct DisassembleInfo *info) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002674 // First see if there is an external relocation entry at the ReferencePC.
Rafael Espindola80291272014-10-08 15:28:58 +00002675 uint64_t sect_addr = info->S.getAddress();
Kevin Enderbybf246f52014-09-24 23:08:22 +00002676 uint64_t sect_offset = ReferencePC - sect_addr;
2677 bool reloc_found = false;
2678 DataRefImpl Rel;
2679 MachO::any_relocation_info RE;
2680 bool isExtern = false;
2681 SymbolRef Symbol;
2682 for (const RelocationRef &Reloc : info->S.relocations()) {
2683 uint64_t RelocOffset;
2684 Reloc.getOffset(RelocOffset);
2685 if (RelocOffset == sect_offset) {
2686 Rel = Reloc.getRawDataRefImpl();
2687 RE = info->O->getRelocation(Rel);
2688 if (info->O->isRelocationScattered(RE))
2689 continue;
2690 isExtern = info->O->getPlainRelocationExternal(RE);
2691 if (isExtern) {
2692 symbol_iterator RelocSym = Reloc.getSymbol();
2693 Symbol = *RelocSym;
2694 }
2695 reloc_found = true;
2696 break;
2697 }
2698 }
2699 // If there is an external relocation entry for a symbol in a section
2700 // then used that symbol's value for the value of the reference.
2701 if (reloc_found && isExtern) {
2702 if (info->O->getAnyRelocationPCRel(RE)) {
2703 unsigned Type = info->O->getAnyRelocationType(RE);
2704 if (Type == MachO::X86_64_RELOC_SIGNED) {
2705 Symbol.getAddress(ReferenceValue);
2706 }
2707 }
2708 }
2709
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002710 // Look for literals such as Objective-C CFStrings refs, Selector refs,
2711 // Message refs and Class refs.
2712 bool classref, selref, msgref, cfstring;
2713 uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref,
2714 selref, msgref, cfstring);
2715 if (classref == true && pointer_value == 0) {
2716 // Note the ReferenceValue is a pointer into the __objc_classrefs section.
2717 // And the pointer_value in that section is typically zero as it will be
2718 // set by dyld as part of the "bind information".
2719 const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info);
2720 if (name != nullptr) {
2721 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
Hans Wennborgdb53e302014-10-23 21:59:17 +00002722 const char *class_name = strrchr(name, '$');
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002723 if (class_name != nullptr && class_name[1] == '_' &&
2724 class_name[2] != '\0') {
2725 info->class_name = class_name + 2;
2726 return name;
2727 }
2728 }
2729 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00002730
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002731 if (classref == true) {
2732 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
2733 const char *name =
2734 get_objc2_64bit_class_name(pointer_value, ReferenceValue, info);
2735 if (name != nullptr)
2736 info->class_name = name;
2737 else
2738 name = "bad class ref";
Kevin Enderbybf246f52014-09-24 23:08:22 +00002739 return name;
2740 }
2741
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002742 if (cfstring == true) {
2743 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref;
2744 const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info);
2745 return name;
2746 }
2747
2748 if (selref == true && pointer_value == 0)
2749 pointer_value = get_objc2_64bit_selref(ReferenceValue, info);
2750
2751 if (pointer_value != 0)
2752 ReferenceValue = pointer_value;
2753
2754 const char *name = GuessCstringPointer(ReferenceValue, info);
2755 if (name) {
2756 if (pointer_value != 0 && selref == true) {
2757 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref;
2758 info->selector_name = name;
2759 } else if (pointer_value != 0 && msgref == true) {
2760 info->class_name = nullptr;
2761 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref;
2762 info->selector_name = name;
2763 } else
2764 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr;
2765 return name;
2766 }
2767
2768 // Lastly look for an indirect symbol with this ReferenceValue which is in
2769 // a literal pool. If found return that symbol name.
2770 name = GuessIndirectSymbol(ReferenceValue, info);
2771 if (name) {
2772 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr;
2773 return name;
2774 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00002775
2776 return nullptr;
2777}
2778
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002779// SymbolizerSymbolLookUp is the symbol lookup function passed when creating
Kevin Enderbybf246f52014-09-24 23:08:22 +00002780// the Symbolizer. It looks up the ReferenceValue using the info passed via the
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002781// pointer to the struct DisassembleInfo that was passed when MCSymbolizer
2782// is created and returns the symbol name that matches the ReferenceValue or
2783// nullptr if none. The ReferenceType is passed in for the IN type of
2784// reference the instruction is making from the values in defined in the header
2785// "llvm-c/Disassembler.h". On return the ReferenceType can set to a specific
2786// Out type and the ReferenceName will also be set which is added as a comment
2787// to the disassembled instruction.
2788//
Kevin Enderby04bf6932014-10-28 23:39:46 +00002789#if HAVE_CXXABI_H
2790// If the symbol name is a C++ mangled name then the demangled name is
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002791// returned through ReferenceName and ReferenceType is set to
2792// LLVMDisassembler_ReferenceType_DeMangled_Name .
Kevin Enderby04bf6932014-10-28 23:39:46 +00002793#endif
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002794//
2795// When this is called to get a symbol name for a branch target then the
2796// ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then
2797// SymbolValue will be looked for in the indirect symbol table to determine if
2798// it is an address for a symbol stub. If so then the symbol name for that
2799// stub is returned indirectly through ReferenceName and then ReferenceType is
2800// set to LLVMDisassembler_ReferenceType_Out_SymbolStub.
2801//
Kevin Enderbybf246f52014-09-24 23:08:22 +00002802// When this is called with an value loaded via a PC relative load then
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002803// ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the
2804// SymbolValue is checked to be an address of literal pointer, symbol pointer,
2805// or an Objective-C meta data reference. If so the output ReferenceType is
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002806// set to correspond to that as well as setting the ReferenceName.
Benjamin Kramerf044d3f2015-03-09 16:23:46 +00002807static const char *SymbolizerSymbolLookUp(void *DisInfo,
2808 uint64_t ReferenceValue,
2809 uint64_t *ReferenceType,
2810 uint64_t ReferencePC,
2811 const char **ReferenceName) {
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002812 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
Kevin Enderbybf246f52014-09-24 23:08:22 +00002813 // If no verbose symbolic information is wanted then just return nullptr.
2814 if (info->verbose == false) {
2815 *ReferenceName = nullptr;
2816 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002817 return nullptr;
2818 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00002819
Kevin Enderbyf6d25852015-01-31 00:37:11 +00002820 const char *SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap);
Kevin Enderbybf246f52014-09-24 23:08:22 +00002821
Kevin Enderby85974882014-09-26 22:20:44 +00002822 if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) {
2823 *ReferenceName = GuessIndirectSymbol(ReferenceValue, info);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002824 if (*ReferenceName != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002825 method_reference(info, ReferenceType, ReferenceName);
2826 if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message)
2827 *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub;
2828 } else
Kevin Enderby04bf6932014-10-28 23:39:46 +00002829#if HAVE_CXXABI_H
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002830 if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
Kevin Enderby04bf6932014-10-28 23:39:46 +00002831 if (info->demangled_name != nullptr)
2832 free(info->demangled_name);
2833 int status;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002834 info->demangled_name =
2835 abi::__cxa_demangle(SymbolName + 1, nullptr, nullptr, &status);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002836 if (info->demangled_name != nullptr) {
2837 *ReferenceName = info->demangled_name;
2838 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
2839 } else
2840 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2841 } else
2842#endif
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002843 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2844 } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) {
2845 *ReferenceName =
2846 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
Kevin Enderby85974882014-09-26 22:20:44 +00002847 if (*ReferenceName)
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002848 method_reference(info, ReferenceType, ReferenceName);
Kevin Enderby85974882014-09-26 22:20:44 +00002849 else
2850 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002851 // If this is arm64 and the reference is an adrp instruction save the
2852 // instruction, passed in ReferenceValue and the address of the instruction
2853 // for use later if we see and add immediate instruction.
2854 } else if (info->O->getArch() == Triple::aarch64 &&
2855 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
2856 info->adrp_inst = ReferenceValue;
2857 info->adrp_addr = ReferencePC;
2858 SymbolName = nullptr;
2859 *ReferenceName = nullptr;
2860 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2861 // If this is arm64 and reference is an add immediate instruction and we
2862 // have
2863 // seen an adrp instruction just before it and the adrp's Xd register
2864 // matches
2865 // this add's Xn register reconstruct the value being referenced and look to
2866 // see if it is a literal pointer. Note the add immediate instruction is
2867 // passed in ReferenceValue.
2868 } else if (info->O->getArch() == Triple::aarch64 &&
2869 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri &&
2870 ReferencePC - 4 == info->adrp_addr &&
2871 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
2872 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
2873 uint32_t addxri_inst;
2874 uint64_t adrp_imm, addxri_imm;
2875
2876 adrp_imm =
2877 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
2878 if (info->adrp_inst & 0x0200000)
2879 adrp_imm |= 0xfffffffffc000000LL;
2880
2881 addxri_inst = ReferenceValue;
2882 addxri_imm = (addxri_inst >> 10) & 0xfff;
2883 if (((addxri_inst >> 22) & 0x3) == 1)
2884 addxri_imm <<= 12;
2885
2886 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
2887 (adrp_imm << 12) + addxri_imm;
2888
2889 *ReferenceName =
2890 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
2891 if (*ReferenceName == nullptr)
2892 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2893 // If this is arm64 and the reference is a load register instruction and we
2894 // have seen an adrp instruction just before it and the adrp's Xd register
2895 // matches this add's Xn register reconstruct the value being referenced and
2896 // look to see if it is a literal pointer. Note the load register
2897 // instruction is passed in ReferenceValue.
2898 } else if (info->O->getArch() == Triple::aarch64 &&
2899 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui &&
2900 ReferencePC - 4 == info->adrp_addr &&
2901 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
2902 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
2903 uint32_t ldrxui_inst;
2904 uint64_t adrp_imm, ldrxui_imm;
2905
2906 adrp_imm =
2907 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
2908 if (info->adrp_inst & 0x0200000)
2909 adrp_imm |= 0xfffffffffc000000LL;
2910
2911 ldrxui_inst = ReferenceValue;
2912 ldrxui_imm = (ldrxui_inst >> 10) & 0xfff;
2913
2914 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
2915 (adrp_imm << 12) + (ldrxui_imm << 3);
2916
2917 *ReferenceName =
2918 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
2919 if (*ReferenceName == nullptr)
2920 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2921 }
2922 // If this arm64 and is an load register (PC-relative) instruction the
2923 // ReferenceValue is the PC plus the immediate value.
2924 else if (info->O->getArch() == Triple::aarch64 &&
2925 (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl ||
2926 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) {
2927 *ReferenceName =
2928 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
2929 if (*ReferenceName == nullptr)
2930 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderby85974882014-09-26 22:20:44 +00002931 }
Kevin Enderby04bf6932014-10-28 23:39:46 +00002932#if HAVE_CXXABI_H
2933 else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
2934 if (info->demangled_name != nullptr)
2935 free(info->demangled_name);
2936 int status;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002937 info->demangled_name =
2938 abi::__cxa_demangle(SymbolName + 1, nullptr, nullptr, &status);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002939 if (info->demangled_name != nullptr) {
2940 *ReferenceName = info->demangled_name;
2941 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
2942 }
2943 }
2944#endif
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002945 else {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002946 *ReferenceName = nullptr;
2947 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
2948 }
2949
2950 return SymbolName;
2951}
2952
Kevin Enderbybf246f52014-09-24 23:08:22 +00002953/// \brief Emits the comments that are stored in the CommentStream.
2954/// Each comment in the CommentStream must end with a newline.
2955static void emitComments(raw_svector_ostream &CommentStream,
2956 SmallString<128> &CommentsToEmit,
2957 formatted_raw_ostream &FormattedOS,
2958 const MCAsmInfo &MAI) {
2959 // Flush the stream before taking its content.
2960 CommentStream.flush();
2961 StringRef Comments = CommentsToEmit.str();
2962 // Get the default information for printing a comment.
2963 const char *CommentBegin = MAI.getCommentString();
2964 unsigned CommentColumn = MAI.getCommentColumn();
2965 bool IsFirst = true;
2966 while (!Comments.empty()) {
2967 if (!IsFirst)
2968 FormattedOS << '\n';
2969 // Emit a line of comments.
2970 FormattedOS.PadToColumn(CommentColumn);
2971 size_t Position = Comments.find('\n');
2972 FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position);
2973 // Move after the newline character.
2974 Comments = Comments.substr(Position + 1);
2975 IsFirst = false;
2976 }
2977 FormattedOS.flush();
2978
2979 // Tell the comment stream that the vector changed underneath it.
2980 CommentsToEmit.clear();
2981 CommentStream.resync();
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002982}
2983
Kevin Enderby95df54c2015-02-04 01:01:38 +00002984static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF,
2985 StringRef DisSegName, StringRef DisSectName) {
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002986 const char *McpuDefault = nullptr;
2987 const Target *ThumbTarget = nullptr;
2988 const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget);
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002989 if (!TheTarget) {
2990 // GetTarget prints out stuff.
2991 return;
2992 }
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002993 if (MCPU.empty() && McpuDefault)
2994 MCPU = McpuDefault;
2995
Ahmed Charles56440fd2014-03-06 05:51:42 +00002996 std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002997 std::unique_ptr<const MCInstrInfo> ThumbInstrInfo;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002998 if (ThumbTarget)
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002999 ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo());
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003000
Kevin Enderbyc9595622014-08-06 23:24:41 +00003001 // Package up features to be passed to target/subtarget
3002 std::string FeaturesStr;
3003 if (MAttrs.size()) {
3004 SubtargetFeatures Features;
3005 for (unsigned i = 0; i != MAttrs.size(); ++i)
3006 Features.AddFeature(MAttrs[i]);
3007 FeaturesStr = Features.getString();
3008 }
3009
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003010 // Set up disassembler.
Ahmed Charles56440fd2014-03-06 05:51:42 +00003011 std::unique_ptr<const MCRegisterInfo> MRI(
3012 TheTarget->createMCRegInfo(TripleName));
3013 std::unique_ptr<const MCAsmInfo> AsmInfo(
Rafael Espindola227144c2013-05-13 01:16:13 +00003014 TheTarget->createMCAsmInfo(*MRI, TripleName));
Ahmed Charles56440fd2014-03-06 05:51:42 +00003015 std::unique_ptr<const MCSubtargetInfo> STI(
Kevin Enderbyc9595622014-08-06 23:24:41 +00003016 TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
Craig Toppere6cb63e2014-04-25 04:24:47 +00003017 MCContext Ctx(AsmInfo.get(), MRI.get(), nullptr);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003018 std::unique_ptr<MCDisassembler> DisAsm(
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003019 TheTarget->createMCDisassembler(*STI, Ctx));
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003020 std::unique_ptr<MCSymbolizer> Symbolizer;
3021 struct DisassembleInfo SymbolizerInfo;
3022 std::unique_ptr<MCRelocationInfo> RelInfo(
3023 TheTarget->createMCRelocationInfo(TripleName, Ctx));
3024 if (RelInfo) {
3025 Symbolizer.reset(TheTarget->createMCSymbolizer(
3026 TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
David Blaikie186db432015-01-18 20:45:48 +00003027 &SymbolizerInfo, &Ctx, std::move(RelInfo)));
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003028 DisAsm->setSymbolizer(std::move(Symbolizer));
3029 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003030 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
Ahmed Charles56440fd2014-03-06 05:51:42 +00003031 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
3032 AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI, *STI));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003033 // Set the display preference for hex vs. decimal immediates.
3034 IP->setPrintImmHex(PrintImmHex);
3035 // Comment stream and backing vector.
3036 SmallString<128> CommentsToEmit;
3037 raw_svector_ostream CommentStream(CommentsToEmit);
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00003038 // FIXME: Setting the CommentStream in the InstPrinter is problematic in that
3039 // if it is done then arm64 comments for string literals don't get printed
3040 // and some constant get printed instead and not setting it causes intel
3041 // (32-bit and 64-bit) comments printed with different spacing before the
3042 // comment causing different diffs with the 'C' disassembler library API.
3043 // IP->setCommentStream(CommentStream);
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003044
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003045 if (!AsmInfo || !STI || !DisAsm || !IP) {
Michael J. Spencerc1363cf2011-10-07 19:25:47 +00003046 errs() << "error: couldn't initialize disassembler for target "
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003047 << TripleName << '\n';
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003048 return;
3049 }
3050
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003051 // Set up thumb disassembler.
3052 std::unique_ptr<const MCRegisterInfo> ThumbMRI;
3053 std::unique_ptr<const MCAsmInfo> ThumbAsmInfo;
3054 std::unique_ptr<const MCSubtargetInfo> ThumbSTI;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003055 std::unique_ptr<MCDisassembler> ThumbDisAsm;
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003056 std::unique_ptr<MCInstPrinter> ThumbIP;
3057 std::unique_ptr<MCContext> ThumbCtx;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003058 std::unique_ptr<MCSymbolizer> ThumbSymbolizer;
3059 struct DisassembleInfo ThumbSymbolizerInfo;
3060 std::unique_ptr<MCRelocationInfo> ThumbRelInfo;
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003061 if (ThumbTarget) {
3062 ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName));
3063 ThumbAsmInfo.reset(
3064 ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName));
3065 ThumbSTI.reset(
3066 ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MCPU, FeaturesStr));
3067 ThumbCtx.reset(new MCContext(ThumbAsmInfo.get(), ThumbMRI.get(), nullptr));
3068 ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx));
Kevin Enderby930fdc72014-11-06 19:00:13 +00003069 MCContext *PtrThumbCtx = ThumbCtx.get();
3070 ThumbRelInfo.reset(
3071 ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx));
3072 if (ThumbRelInfo) {
3073 ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer(
3074 ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
David Blaikie186db432015-01-18 20:45:48 +00003075 &ThumbSymbolizerInfo, PtrThumbCtx, std::move(ThumbRelInfo)));
Kevin Enderby930fdc72014-11-06 19:00:13 +00003076 ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer));
3077 }
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003078 int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect();
3079 ThumbIP.reset(ThumbTarget->createMCInstPrinter(
3080 ThumbAsmPrinterVariant, *ThumbAsmInfo, *ThumbInstrInfo, *ThumbMRI,
3081 *ThumbSTI));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003082 // Set the display preference for hex vs. decimal immediates.
3083 ThumbIP->setPrintImmHex(PrintImmHex);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003084 }
3085
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003086 if (ThumbTarget && (!ThumbAsmInfo || !ThumbSTI || !ThumbDisAsm || !ThumbIP)) {
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003087 errs() << "error: couldn't initialize disassembler for target "
3088 << ThumbTripleName << '\n';
3089 return;
3090 }
3091
Charles Davis8bdfafd2013-09-01 04:28:48 +00003092 MachO::mach_header Header = MachOOF->getHeader();
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003093
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003094 // FIXME: Using the -cfg command line option, this code used to be able to
3095 // annotate relocations with the referenced symbol's name, and if this was
3096 // inside a __[cf]string section, the data it points to. This is now replaced
3097 // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
Owen Andersond9243c42011-10-17 21:37:35 +00003098 std::vector<SectionRef> Sections;
3099 std::vector<SymbolRef> Symbols;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003100 SmallVector<uint64_t, 8> FoundFns;
Kevin Enderby273ae012013-06-06 17:20:50 +00003101 uint64_t BaseSegmentAddress;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003102
Kevin Enderby273ae012013-06-06 17:20:50 +00003103 getSectionsAndSymbols(Header, MachOOF, Sections, Symbols, FoundFns,
3104 BaseSegmentAddress);
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003105
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003106 // Sort the symbols by address, just in case they didn't come in that way.
Owen Andersond9243c42011-10-17 21:37:35 +00003107 std::sort(Symbols.begin(), Symbols.end(), SymbolSorter());
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003108
Kevin Enderby273ae012013-06-06 17:20:50 +00003109 // Build a data in code table that is sorted on by the address of each entry.
3110 uint64_t BaseAddress = 0;
Charles Davis8bdfafd2013-09-01 04:28:48 +00003111 if (Header.filetype == MachO::MH_OBJECT)
Rafael Espindola80291272014-10-08 15:28:58 +00003112 BaseAddress = Sections[0].getAddress();
Kevin Enderby273ae012013-06-06 17:20:50 +00003113 else
3114 BaseAddress = BaseSegmentAddress;
3115 DiceTable Dices;
Kevin Enderby273ae012013-06-06 17:20:50 +00003116 for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
Rafael Espindola5e812af2014-01-30 02:49:50 +00003117 DI != DE; ++DI) {
Kevin Enderby273ae012013-06-06 17:20:50 +00003118 uint32_t Offset;
3119 DI->getOffset(Offset);
3120 Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
3121 }
3122 array_pod_sort(Dices.begin(), Dices.end());
3123
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003124#ifndef NDEBUG
3125 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
3126#else
3127 raw_ostream &DebugOut = nulls();
3128#endif
3129
Ahmed Charles56440fd2014-03-06 05:51:42 +00003130 std::unique_ptr<DIContext> diContext;
Rafael Espindola9b709252013-04-13 01:45:40 +00003131 ObjectFile *DbgObj = MachOOF;
Benjamin Kramer699128e2011-09-21 01:13:19 +00003132 // Try to find debug info and set up the DIContext for it.
3133 if (UseDbg) {
Benjamin Kramer699128e2011-09-21 01:13:19 +00003134 // A separate DSym file path was specified, parse it as a macho file,
3135 // get the sections and supply it to the section name parsing machinery.
3136 if (!DSYMFile.empty()) {
Rafael Espindola48af1c22014-08-19 18:44:46 +00003137 ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr =
Rafael Espindolaadf21f22014-07-06 17:43:13 +00003138 MemoryBuffer::getFileOrSTDIN(DSYMFile);
Rafael Espindola48af1c22014-08-19 18:44:46 +00003139 if (std::error_code EC = BufOrErr.getError()) {
Rafael Espindolaadf21f22014-07-06 17:43:13 +00003140 errs() << "llvm-objdump: " << Filename << ": " << EC.message() << '\n';
Benjamin Kramer699128e2011-09-21 01:13:19 +00003141 return;
3142 }
Rafael Espindola48af1c22014-08-19 18:44:46 +00003143 DbgObj =
3144 ObjectFile::createMachOObjectFile(BufOrErr.get()->getMemBufferRef())
3145 .get()
3146 .release();
Benjamin Kramer699128e2011-09-21 01:13:19 +00003147 }
3148
Eric Christopher7370b552012-11-12 21:40:38 +00003149 // Setup the DIContext
Rafael Espindolaa04bb5b2014-07-31 20:19:36 +00003150 diContext.reset(DIContext::getDWARFContext(*DbgObj));
Benjamin Kramer699128e2011-09-21 01:13:19 +00003151 }
3152
Kevin Enderby95df54c2015-02-04 01:01:38 +00003153 if (DumpSections.size() == 0)
3154 outs() << "(" << DisSegName << "," << DisSectName << ") section\n";
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00003155
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003156 for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
Owen Andersond9243c42011-10-17 21:37:35 +00003157 StringRef SectName;
Kevin Enderby95df54c2015-02-04 01:01:38 +00003158 if (Sections[SectIdx].getName(SectName) || SectName != DisSectName)
3159 continue;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003160
Rafael Espindolaa9f810b2012-12-21 03:47:03 +00003161 DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003162
Rafael Espindolab0f76a42013-04-05 15:15:22 +00003163 StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
Kevin Enderby95df54c2015-02-04 01:01:38 +00003164 if (SegmentName != DisSegName)
Rafael Espindolaa9f810b2012-12-21 03:47:03 +00003165 continue;
3166
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003167 StringRef BytesStr;
3168 Sections[SectIdx].getContents(BytesStr);
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003169 ArrayRef<uint8_t> Bytes(reinterpret_cast<const uint8_t *>(BytesStr.data()),
3170 BytesStr.size());
Rafael Espindola80291272014-10-08 15:28:58 +00003171 uint64_t SectAddress = Sections[SectIdx].getAddress();
Rafael Espindolabd604f22014-11-07 00:52:15 +00003172
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003173 bool symbolTableWorked = false;
3174
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003175 // Parse relocations.
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00003176 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
3177 for (const RelocationRef &Reloc : Sections[SectIdx].relocations()) {
Rafael Espindola80291272014-10-08 15:28:58 +00003178 uint64_t RelocOffset;
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00003179 Reloc.getOffset(RelocOffset);
Rafael Espindola80291272014-10-08 15:28:58 +00003180 uint64_t SectionAddress = Sections[SectIdx].getAddress();
Owen Andersond9243c42011-10-17 21:37:35 +00003181 RelocOffset -= SectionAddress;
3182
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00003183 symbol_iterator RelocSym = Reloc.getSymbol();
Owen Andersond9243c42011-10-17 21:37:35 +00003184
Rafael Espindola806f0062013-06-05 01:33:53 +00003185 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003186 }
3187 array_pod_sort(Relocs.begin(), Relocs.end());
3188
Kevin Enderbybf246f52014-09-24 23:08:22 +00003189 // Create a map of symbol addresses to symbol names for use by
3190 // the SymbolizerSymbolLookUp() routine.
3191 SymbolAddressMap AddrMap;
Kevin Enderby6a221752015-03-17 17:10:57 +00003192 bool DisSymNameFound = false;
Kevin Enderbybf246f52014-09-24 23:08:22 +00003193 for (const SymbolRef &Symbol : MachOOF->symbols()) {
3194 SymbolRef::Type ST;
3195 Symbol.getType(ST);
3196 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
3197 ST == SymbolRef::ST_Other) {
3198 uint64_t Address;
3199 Symbol.getAddress(Address);
3200 StringRef SymName;
3201 Symbol.getName(SymName);
3202 AddrMap[Address] = SymName;
Kevin Enderby6a221752015-03-17 17:10:57 +00003203 if (!DisSymName.empty() && DisSymName == SymName)
3204 DisSymNameFound = true;
Kevin Enderbybf246f52014-09-24 23:08:22 +00003205 }
3206 }
Kevin Enderby6a221752015-03-17 17:10:57 +00003207 if (!DisSymName.empty() && DisSymNameFound == false) {
3208 outs() << "Can't find -dis-symname: " << DisSymName << "\n";
3209 return;
3210 }
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003211 // Set up the block of info used by the Symbolizer call backs.
3212 SymbolizerInfo.verbose = true;
3213 SymbolizerInfo.O = MachOOF;
3214 SymbolizerInfo.S = Sections[SectIdx];
Kevin Enderbybf246f52014-09-24 23:08:22 +00003215 SymbolizerInfo.AddrMap = &AddrMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003216 SymbolizerInfo.Sections = &Sections;
3217 SymbolizerInfo.class_name = nullptr;
3218 SymbolizerInfo.selector_name = nullptr;
3219 SymbolizerInfo.method = nullptr;
Kevin Enderby04bf6932014-10-28 23:39:46 +00003220 SymbolizerInfo.demangled_name = nullptr;
Kevin Enderby078be602014-10-23 19:53:12 +00003221 SymbolizerInfo.bindtable = nullptr;
Kevin Enderby10738222014-11-19 20:20:16 +00003222 SymbolizerInfo.adrp_addr = 0;
3223 SymbolizerInfo.adrp_inst = 0;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003224 // Same for the ThumbSymbolizer
3225 ThumbSymbolizerInfo.verbose = true;
3226 ThumbSymbolizerInfo.O = MachOOF;
3227 ThumbSymbolizerInfo.S = Sections[SectIdx];
3228 ThumbSymbolizerInfo.AddrMap = &AddrMap;
3229 ThumbSymbolizerInfo.Sections = &Sections;
3230 ThumbSymbolizerInfo.class_name = nullptr;
3231 ThumbSymbolizerInfo.selector_name = nullptr;
3232 ThumbSymbolizerInfo.method = nullptr;
3233 ThumbSymbolizerInfo.demangled_name = nullptr;
3234 ThumbSymbolizerInfo.bindtable = nullptr;
Kevin Enderby10738222014-11-19 20:20:16 +00003235 ThumbSymbolizerInfo.adrp_addr = 0;
3236 ThumbSymbolizerInfo.adrp_inst = 0;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00003237
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003238 // Disassemble symbol by symbol.
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003239 for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
Owen Andersond9243c42011-10-17 21:37:35 +00003240 StringRef SymName;
3241 Symbols[SymIdx].getName(SymName);
3242
3243 SymbolRef::Type ST;
3244 Symbols[SymIdx].getType(ST);
3245 if (ST != SymbolRef::ST_Function)
3246 continue;
3247
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003248 // Make sure the symbol is defined in this section.
Rafael Espindola80291272014-10-08 15:28:58 +00003249 bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]);
Owen Andersond9243c42011-10-17 21:37:35 +00003250 if (!containsSym)
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003251 continue;
3252
Kevin Enderby6a221752015-03-17 17:10:57 +00003253 // If we are only disassembling one symbol see if this is that symbol.
3254 if (!DisSymName.empty() && DisSymName != SymName)
3255 continue;
3256
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003257 // Start at the address of the symbol relative to the section's address.
Owen Andersond9243c42011-10-17 21:37:35 +00003258 uint64_t Start = 0;
Rafael Espindola80291272014-10-08 15:28:58 +00003259 uint64_t SectionAddress = Sections[SectIdx].getAddress();
Danil Malyshevcbe72fc2011-11-29 17:40:10 +00003260 Symbols[SymIdx].getAddress(Start);
Cameron Zwarich54478a52012-02-03 05:42:17 +00003261 Start -= SectionAddress;
Owen Andersond9243c42011-10-17 21:37:35 +00003262
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00003263 // Stop disassembling either at the beginning of the next symbol or at
3264 // the end of the section.
Kevin Enderbyedd58722012-05-15 18:57:14 +00003265 bool containsNextSym = false;
Owen Andersond9243c42011-10-17 21:37:35 +00003266 uint64_t NextSym = 0;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003267 uint64_t NextSymIdx = SymIdx + 1;
Owen Andersond9243c42011-10-17 21:37:35 +00003268 while (Symbols.size() > NextSymIdx) {
3269 SymbolRef::Type NextSymType;
3270 Symbols[NextSymIdx].getType(NextSymType);
3271 if (NextSymType == SymbolRef::ST_Function) {
Rafael Espindola80291272014-10-08 15:28:58 +00003272 containsNextSym =
3273 Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]);
Danil Malyshevcbe72fc2011-11-29 17:40:10 +00003274 Symbols[NextSymIdx].getAddress(NextSym);
Cameron Zwarich54478a52012-02-03 05:42:17 +00003275 NextSym -= SectionAddress;
Owen Andersond9243c42011-10-17 21:37:35 +00003276 break;
3277 }
3278 ++NextSymIdx;
3279 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003280
Rafael Espindola80291272014-10-08 15:28:58 +00003281 uint64_t SectSize = Sections[SectIdx].getSize();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003282 uint64_t End = containsNextSym ? NextSym : SectSize;
Owen Andersond9243c42011-10-17 21:37:35 +00003283 uint64_t Size;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003284
3285 symbolTableWorked = true;
Rafael Espindolabd604f22014-11-07 00:52:15 +00003286
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003287 DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl();
3288 bool isThumb =
3289 (MachOOF->getSymbolFlags(Symb) & SymbolRef::SF_Thumb) && ThumbTarget;
3290
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003291 outs() << SymName << ":\n";
3292 DILineInfo lastLine;
3293 for (uint64_t Index = Start; Index < End; Index += Size) {
3294 MCInst Inst;
Owen Andersond9243c42011-10-17 21:37:35 +00003295
Kevin Enderbybf246f52014-09-24 23:08:22 +00003296 uint64_t PC = SectAddress + Index;
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00003297 if (!NoLeadingAddr) {
3298 if (FullLeadingAddr) {
3299 if (MachOOF->is64Bit())
3300 outs() << format("%016" PRIx64, PC);
3301 else
3302 outs() << format("%08" PRIx64, PC);
3303 } else {
3304 outs() << format("%8" PRIx64 ":", PC);
3305 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00003306 }
3307 if (!NoShowRawInsn)
3308 outs() << "\t";
Kevin Enderby273ae012013-06-06 17:20:50 +00003309
3310 // Check the data in code table here to see if this is data not an
3311 // instruction to be disassembled.
3312 DiceTable Dice;
Kevin Enderbybf246f52014-09-24 23:08:22 +00003313 Dice.push_back(std::make_pair(PC, DiceRef()));
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003314 dice_table_iterator DTI =
3315 std::search(Dices.begin(), Dices.end(), Dice.begin(), Dice.end(),
3316 compareDiceTableEntries);
3317 if (DTI != Dices.end()) {
Kevin Enderby273ae012013-06-06 17:20:50 +00003318 uint16_t Length;
3319 DTI->second.getLength(Length);
Kevin Enderby273ae012013-06-06 17:20:50 +00003320 uint16_t Kind;
3321 DTI->second.getKind(Kind);
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003322 Size = DumpDataInCode(reinterpret_cast<const char *>(Bytes.data()) +
3323 Index,
3324 Length, Kind);
Kevin Enderby930fdc72014-11-06 19:00:13 +00003325 if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) &&
3326 (PC == (DTI->first + Length - 1)) && (Length & 1))
3327 Size++;
Kevin Enderby273ae012013-06-06 17:20:50 +00003328 continue;
3329 }
3330
Kevin Enderbybf246f52014-09-24 23:08:22 +00003331 SmallVector<char, 64> AnnotationsBytes;
3332 raw_svector_ostream Annotations(AnnotationsBytes);
3333
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003334 bool gotInst;
3335 if (isThumb)
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003336 gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index),
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003337 PC, DebugOut, Annotations);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003338 else
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003339 gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC,
Kevin Enderbybf246f52014-09-24 23:08:22 +00003340 DebugOut, Annotations);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003341 if (gotInst) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00003342 if (!NoShowRawInsn) {
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003343 DumpBytes(StringRef(
3344 reinterpret_cast<const char *>(Bytes.data()) + Index, Size));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003345 }
3346 formatted_raw_ostream FormattedOS(outs());
3347 Annotations.flush();
3348 StringRef AnnotationsStr = Annotations.str();
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003349 if (isThumb)
Kevin Enderbybf246f52014-09-24 23:08:22 +00003350 ThumbIP->printInst(&Inst, FormattedOS, AnnotationsStr);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00003351 else
Kevin Enderbybf246f52014-09-24 23:08:22 +00003352 IP->printInst(&Inst, FormattedOS, AnnotationsStr);
3353 emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo);
Owen Andersond9243c42011-10-17 21:37:35 +00003354
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003355 // Print debug info.
3356 if (diContext) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003357 DILineInfo dli = diContext->getLineInfoForAddress(PC);
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003358 // Print valid line info if it changed.
Alexey Samsonovd0109992014-04-18 21:36:39 +00003359 if (dli != lastLine && dli.Line != 0)
3360 outs() << "\t## " << dli.FileName << ':' << dli.Line << ':'
3361 << dli.Column;
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003362 lastLine = dli;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003363 }
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003364 outs() << "\n";
3365 } else {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003366 unsigned int Arch = MachOOF->getArch();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003367 if (Arch == Triple::x86_64 || Arch == Triple::x86) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003368 outs() << format("\t.byte 0x%02x #bad opcode\n",
3369 *(Bytes.data() + Index) & 0xff);
3370 Size = 1; // skip exactly one illegible byte and move on.
Kevin Enderbyae3c1262014-11-14 21:52:18 +00003371 } else if (Arch == Triple::aarch64) {
3372 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
3373 (*(Bytes.data() + Index + 1) & 0xff) << 8 |
3374 (*(Bytes.data() + Index + 2) & 0xff) << 16 |
3375 (*(Bytes.data() + Index + 3) & 0xff) << 24;
3376 outs() << format("\t.long\t0x%08x\n", opcode);
3377 Size = 4;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003378 } else {
3379 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
3380 if (Size == 0)
3381 Size = 1; // skip illegible bytes
3382 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003383 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003384 }
3385 }
Ahmed Bougachaaa790682013-05-24 01:07:04 +00003386 if (!symbolTableWorked) {
Rafael Espindola80291272014-10-08 15:28:58 +00003387 // Reading the symbol table didn't work, disassemble the whole section.
3388 uint64_t SectAddress = Sections[SectIdx].getAddress();
3389 uint64_t SectSize = Sections[SectIdx].getSize();
Kevin Enderbybadd1002012-05-18 00:13:56 +00003390 uint64_t InstSize;
3391 for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
Bill Wendling4e68e062012-07-19 00:17:40 +00003392 MCInst Inst;
Kevin Enderbybadd1002012-05-18 00:13:56 +00003393
Kevin Enderbybf246f52014-09-24 23:08:22 +00003394 uint64_t PC = SectAddress + Index;
Rafael Espindola7fc5b872014-11-12 02:04:27 +00003395 if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC,
3396 DebugOut, nulls())) {
Kevin Enderbyab5e6c92015-03-17 21:07:39 +00003397 if (!NoLeadingAddr) {
3398 if (FullLeadingAddr) {
3399 if (MachOOF->is64Bit())
3400 outs() << format("%016" PRIx64, PC);
3401 else
3402 outs() << format("%08" PRIx64, PC);
3403 } else {
3404 outs() << format("%8" PRIx64 ":", PC);
3405 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00003406 }
3407 if (!NoShowRawInsn) {
3408 outs() << "\t";
Aaron Ballman106fd7b2014-11-12 14:01:17 +00003409 DumpBytes(
3410 StringRef(reinterpret_cast<const char *>(Bytes.data()) + Index,
3411 InstSize));
Kevin Enderbybf246f52014-09-24 23:08:22 +00003412 }
Bill Wendling4e68e062012-07-19 00:17:40 +00003413 IP->printInst(&Inst, outs(), "");
3414 outs() << "\n";
3415 } else {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003416 unsigned int Arch = MachOOF->getArch();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003417 if (Arch == Triple::x86_64 || Arch == Triple::x86) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003418 outs() << format("\t.byte 0x%02x #bad opcode\n",
3419 *(Bytes.data() + Index) & 0xff);
3420 InstSize = 1; // skip exactly one illegible byte and move on.
3421 } else {
3422 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
3423 if (InstSize == 0)
3424 InstSize = 1; // skip illegible bytes
3425 }
Bill Wendling4e68e062012-07-19 00:17:40 +00003426 }
Kevin Enderbybadd1002012-05-18 00:13:56 +00003427 }
3428 }
Kevin Enderbyef3ad2f2014-12-04 23:56:27 +00003429 // The TripleName's need to be reset if we are called again for a different
3430 // archtecture.
3431 TripleName = "";
3432 ThumbTripleName = "";
3433
Kevin Enderby6f326ce2014-10-23 19:37:31 +00003434 if (SymbolizerInfo.method != nullptr)
3435 free(SymbolizerInfo.method);
Kevin Enderby04bf6932014-10-28 23:39:46 +00003436 if (SymbolizerInfo.demangled_name != nullptr)
3437 free(SymbolizerInfo.demangled_name);
Kevin Enderby078be602014-10-23 19:53:12 +00003438 if (SymbolizerInfo.bindtable != nullptr)
3439 delete SymbolizerInfo.bindtable;
Kevin Enderby930fdc72014-11-06 19:00:13 +00003440 if (ThumbSymbolizerInfo.method != nullptr)
3441 free(ThumbSymbolizerInfo.method);
3442 if (ThumbSymbolizerInfo.demangled_name != nullptr)
3443 free(ThumbSymbolizerInfo.demangled_name);
3444 if (ThumbSymbolizerInfo.bindtable != nullptr)
3445 delete ThumbSymbolizerInfo.bindtable;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00003446 }
3447}
Tim Northover4bd286a2014-08-01 13:07:19 +00003448
Tim Northover39c70bb2014-08-12 11:52:59 +00003449//===----------------------------------------------------------------------===//
3450// __compact_unwind section dumping
3451//===----------------------------------------------------------------------===//
3452
Tim Northover4bd286a2014-08-01 13:07:19 +00003453namespace {
Tim Northover39c70bb2014-08-12 11:52:59 +00003454
3455template <typename T> static uint64_t readNext(const char *&Buf) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003456 using llvm::support::little;
3457 using llvm::support::unaligned;
Tim Northover39c70bb2014-08-12 11:52:59 +00003458
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003459 uint64_t Val = support::endian::read<T, little, unaligned>(Buf);
3460 Buf += sizeof(T);
3461 return Val;
3462}
Tim Northover39c70bb2014-08-12 11:52:59 +00003463
Tim Northover4bd286a2014-08-01 13:07:19 +00003464struct CompactUnwindEntry {
3465 uint32_t OffsetInSection;
3466
3467 uint64_t FunctionAddr;
3468 uint32_t Length;
3469 uint32_t CompactEncoding;
3470 uint64_t PersonalityAddr;
3471 uint64_t LSDAAddr;
3472
3473 RelocationRef FunctionReloc;
3474 RelocationRef PersonalityReloc;
3475 RelocationRef LSDAReloc;
3476
3477 CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64)
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003478 : OffsetInSection(Offset) {
Tim Northover4bd286a2014-08-01 13:07:19 +00003479 if (Is64)
3480 read<uint64_t>(Contents.data() + Offset);
3481 else
3482 read<uint32_t>(Contents.data() + Offset);
3483 }
3484
3485private:
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003486 template <typename UIntPtr> void read(const char *Buf) {
Tim Northover4bd286a2014-08-01 13:07:19 +00003487 FunctionAddr = readNext<UIntPtr>(Buf);
3488 Length = readNext<uint32_t>(Buf);
3489 CompactEncoding = readNext<uint32_t>(Buf);
3490 PersonalityAddr = readNext<UIntPtr>(Buf);
3491 LSDAAddr = readNext<UIntPtr>(Buf);
3492 }
3493};
3494}
3495
3496/// Given a relocation from __compact_unwind, consisting of the RelocationRef
3497/// and data being relocated, determine the best base Name and Addend to use for
3498/// display purposes.
3499///
3500/// 1. An Extern relocation will directly reference a symbol (and the data is
3501/// then already an addend), so use that.
3502/// 2. Otherwise the data is an offset in the object file's layout; try to find
3503// a symbol before it in the same section, and use the offset from there.
3504/// 3. Finally, if all that fails, fall back to an offset from the start of the
3505/// referenced section.
3506static void findUnwindRelocNameAddend(const MachOObjectFile *Obj,
3507 std::map<uint64_t, SymbolRef> &Symbols,
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003508 const RelocationRef &Reloc, uint64_t Addr,
Tim Northover4bd286a2014-08-01 13:07:19 +00003509 StringRef &Name, uint64_t &Addend) {
3510 if (Reloc.getSymbol() != Obj->symbol_end()) {
3511 Reloc.getSymbol()->getName(Name);
3512 Addend = Addr;
3513 return;
3514 }
3515
3516 auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl());
3517 SectionRef RelocSection = Obj->getRelocationSection(RE);
3518
Rafael Espindola80291272014-10-08 15:28:58 +00003519 uint64_t SectionAddr = RelocSection.getAddress();
Tim Northover4bd286a2014-08-01 13:07:19 +00003520
3521 auto Sym = Symbols.upper_bound(Addr);
3522 if (Sym == Symbols.begin()) {
3523 // The first symbol in the object is after this reference, the best we can
3524 // do is section-relative notation.
3525 RelocSection.getName(Name);
3526 Addend = Addr - SectionAddr;
3527 return;
3528 }
3529
3530 // Go back one so that SymbolAddress <= Addr.
3531 --Sym;
3532
3533 section_iterator SymSection = Obj->section_end();
3534 Sym->second.getSection(SymSection);
3535 if (RelocSection == *SymSection) {
3536 // There's a valid symbol in the same section before this reference.
3537 Sym->second.getName(Name);
3538 Addend = Addr - Sym->first;
3539 return;
3540 }
3541
3542 // There is a symbol before this reference, but it's in a different
3543 // section. Probably not helpful to mention it, so use the section name.
3544 RelocSection.getName(Name);
3545 Addend = Addr - SectionAddr;
3546}
3547
3548static void printUnwindRelocDest(const MachOObjectFile *Obj,
3549 std::map<uint64_t, SymbolRef> &Symbols,
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003550 const RelocationRef &Reloc, uint64_t Addr) {
Tim Northover4bd286a2014-08-01 13:07:19 +00003551 StringRef Name;
3552 uint64_t Addend;
3553
Tim Northover0b0add52014-09-09 10:45:06 +00003554 if (!Reloc.getObjectFile())
3555 return;
3556
Tim Northover4bd286a2014-08-01 13:07:19 +00003557 findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend);
3558
3559 outs() << Name;
3560 if (Addend)
Tim Northover63a25622014-08-11 09:14:06 +00003561 outs() << " + " << format("0x%" PRIx64, Addend);
Tim Northover4bd286a2014-08-01 13:07:19 +00003562}
3563
3564static void
3565printMachOCompactUnwindSection(const MachOObjectFile *Obj,
3566 std::map<uint64_t, SymbolRef> &Symbols,
3567 const SectionRef &CompactUnwind) {
3568
3569 assert(Obj->isLittleEndian() &&
3570 "There should not be a big-endian .o with __compact_unwind");
3571
3572 bool Is64 = Obj->is64Bit();
3573 uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t);
3574 uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t);
3575
3576 StringRef Contents;
3577 CompactUnwind.getContents(Contents);
3578
3579 SmallVector<CompactUnwindEntry, 4> CompactUnwinds;
3580
3581 // First populate the initial raw offsets, encodings and so on from the entry.
3582 for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) {
3583 CompactUnwindEntry Entry(Contents.data(), Offset, Is64);
3584 CompactUnwinds.push_back(Entry);
3585 }
3586
3587 // Next we need to look at the relocations to find out what objects are
3588 // actually being referred to.
3589 for (const RelocationRef &Reloc : CompactUnwind.relocations()) {
3590 uint64_t RelocAddress;
3591 Reloc.getOffset(RelocAddress);
3592
3593 uint32_t EntryIdx = RelocAddress / EntrySize;
3594 uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize;
3595 CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx];
3596
3597 if (OffsetInEntry == 0)
3598 Entry.FunctionReloc = Reloc;
3599 else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t))
3600 Entry.PersonalityReloc = Reloc;
3601 else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t))
3602 Entry.LSDAReloc = Reloc;
3603 else
3604 llvm_unreachable("Unexpected relocation in __compact_unwind section");
3605 }
3606
3607 // Finally, we're ready to print the data we've gathered.
3608 outs() << "Contents of __compact_unwind section:\n";
3609 for (auto &Entry : CompactUnwinds) {
Tim Northover06af2602014-08-08 12:08:51 +00003610 outs() << " Entry at offset "
3611 << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n";
Tim Northover4bd286a2014-08-01 13:07:19 +00003612
3613 // 1. Start of the region this entry applies to.
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003614 outs() << " start: " << format("0x%" PRIx64,
3615 Entry.FunctionAddr) << ' ';
3616 printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr);
Tim Northover4bd286a2014-08-01 13:07:19 +00003617 outs() << '\n';
3618
3619 // 2. Length of the region this entry applies to.
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003620 outs() << " length: " << format("0x%" PRIx32, Entry.Length)
3621 << '\n';
Tim Northover4bd286a2014-08-01 13:07:19 +00003622 // 3. The 32-bit compact encoding.
3623 outs() << " compact encoding: "
Tim Northoverb911bf82014-08-08 12:00:09 +00003624 << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n';
Tim Northover4bd286a2014-08-01 13:07:19 +00003625
3626 // 4. The personality function, if present.
3627 if (Entry.PersonalityReloc.getObjectFile()) {
3628 outs() << " personality function: "
Tim Northoverb911bf82014-08-08 12:00:09 +00003629 << format("0x%" PRIx64, Entry.PersonalityAddr) << ' ';
Tim Northover4bd286a2014-08-01 13:07:19 +00003630 printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc,
3631 Entry.PersonalityAddr);
3632 outs() << '\n';
3633 }
3634
3635 // 5. This entry's language-specific data area.
3636 if (Entry.LSDAReloc.getObjectFile()) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003637 outs() << " LSDA: " << format("0x%" PRIx64,
3638 Entry.LSDAAddr) << ' ';
Tim Northover4bd286a2014-08-01 13:07:19 +00003639 printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr);
3640 outs() << '\n';
3641 }
3642 }
3643}
3644
Tim Northover39c70bb2014-08-12 11:52:59 +00003645//===----------------------------------------------------------------------===//
3646// __unwind_info section dumping
3647//===----------------------------------------------------------------------===//
3648
3649static void printRegularSecondLevelUnwindPage(const char *PageStart) {
3650 const char *Pos = PageStart;
3651 uint32_t Kind = readNext<uint32_t>(Pos);
3652 (void)Kind;
3653 assert(Kind == 2 && "kind for a regular 2nd level index should be 2");
3654
3655 uint16_t EntriesStart = readNext<uint16_t>(Pos);
3656 uint16_t NumEntries = readNext<uint16_t>(Pos);
3657
3658 Pos = PageStart + EntriesStart;
3659 for (unsigned i = 0; i < NumEntries; ++i) {
3660 uint32_t FunctionOffset = readNext<uint32_t>(Pos);
3661 uint32_t Encoding = readNext<uint32_t>(Pos);
3662
3663 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003664 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
3665 << ", "
3666 << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003667 }
3668}
3669
3670static void printCompressedSecondLevelUnwindPage(
3671 const char *PageStart, uint32_t FunctionBase,
3672 const SmallVectorImpl<uint32_t> &CommonEncodings) {
3673 const char *Pos = PageStart;
3674 uint32_t Kind = readNext<uint32_t>(Pos);
3675 (void)Kind;
3676 assert(Kind == 3 && "kind for a compressed 2nd level index should be 3");
3677
3678 uint16_t EntriesStart = readNext<uint16_t>(Pos);
3679 uint16_t NumEntries = readNext<uint16_t>(Pos);
3680
3681 uint16_t EncodingsStart = readNext<uint16_t>(Pos);
3682 readNext<uint16_t>(Pos);
Aaron Ballman80930af2014-08-14 13:53:19 +00003683 const auto *PageEncodings = reinterpret_cast<const support::ulittle32_t *>(
3684 PageStart + EncodingsStart);
Tim Northover39c70bb2014-08-12 11:52:59 +00003685
3686 Pos = PageStart + EntriesStart;
3687 for (unsigned i = 0; i < NumEntries; ++i) {
3688 uint32_t Entry = readNext<uint32_t>(Pos);
3689 uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff);
3690 uint32_t EncodingIdx = Entry >> 24;
3691
3692 uint32_t Encoding;
3693 if (EncodingIdx < CommonEncodings.size())
3694 Encoding = CommonEncodings[EncodingIdx];
3695 else
3696 Encoding = PageEncodings[EncodingIdx - CommonEncodings.size()];
3697
3698 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003699 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
3700 << ", "
3701 << "encoding[" << EncodingIdx
3702 << "]=" << format("0x%08" PRIx32, Encoding) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003703 }
3704}
3705
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003706static void printMachOUnwindInfoSection(const MachOObjectFile *Obj,
3707 std::map<uint64_t, SymbolRef> &Symbols,
3708 const SectionRef &UnwindInfo) {
Tim Northover39c70bb2014-08-12 11:52:59 +00003709
3710 assert(Obj->isLittleEndian() &&
3711 "There should not be a big-endian .o with __unwind_info");
3712
3713 outs() << "Contents of __unwind_info section:\n";
3714
3715 StringRef Contents;
3716 UnwindInfo.getContents(Contents);
3717 const char *Pos = Contents.data();
3718
3719 //===----------------------------------
3720 // Section header
3721 //===----------------------------------
3722
3723 uint32_t Version = readNext<uint32_t>(Pos);
3724 outs() << " Version: "
3725 << format("0x%" PRIx32, Version) << '\n';
3726 assert(Version == 1 && "only understand version 1");
3727
3728 uint32_t CommonEncodingsStart = readNext<uint32_t>(Pos);
3729 outs() << " Common encodings array section offset: "
3730 << format("0x%" PRIx32, CommonEncodingsStart) << '\n';
3731 uint32_t NumCommonEncodings = readNext<uint32_t>(Pos);
3732 outs() << " Number of common encodings in array: "
3733 << format("0x%" PRIx32, NumCommonEncodings) << '\n';
3734
3735 uint32_t PersonalitiesStart = readNext<uint32_t>(Pos);
3736 outs() << " Personality function array section offset: "
3737 << format("0x%" PRIx32, PersonalitiesStart) << '\n';
3738 uint32_t NumPersonalities = readNext<uint32_t>(Pos);
3739 outs() << " Number of personality functions in array: "
3740 << format("0x%" PRIx32, NumPersonalities) << '\n';
3741
3742 uint32_t IndicesStart = readNext<uint32_t>(Pos);
3743 outs() << " Index array section offset: "
3744 << format("0x%" PRIx32, IndicesStart) << '\n';
3745 uint32_t NumIndices = readNext<uint32_t>(Pos);
3746 outs() << " Number of indices in array: "
3747 << format("0x%" PRIx32, NumIndices) << '\n';
3748
3749 //===----------------------------------
3750 // A shared list of common encodings
3751 //===----------------------------------
3752
3753 // These occupy indices in the range [0, N] whenever an encoding is referenced
3754 // from a compressed 2nd level index table. In practice the linker only
3755 // creates ~128 of these, so that indices are available to embed encodings in
3756 // the 2nd level index.
3757
3758 SmallVector<uint32_t, 64> CommonEncodings;
3759 outs() << " Common encodings: (count = " << NumCommonEncodings << ")\n";
3760 Pos = Contents.data() + CommonEncodingsStart;
3761 for (unsigned i = 0; i < NumCommonEncodings; ++i) {
3762 uint32_t Encoding = readNext<uint32_t>(Pos);
3763 CommonEncodings.push_back(Encoding);
3764
3765 outs() << " encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding)
3766 << '\n';
3767 }
3768
Tim Northover39c70bb2014-08-12 11:52:59 +00003769 //===----------------------------------
3770 // Personality functions used in this executable
3771 //===----------------------------------
3772
3773 // There should be only a handful of these (one per source language,
3774 // roughly). Particularly since they only get 2 bits in the compact encoding.
3775
3776 outs() << " Personality functions: (count = " << NumPersonalities << ")\n";
3777 Pos = Contents.data() + PersonalitiesStart;
3778 for (unsigned i = 0; i < NumPersonalities; ++i) {
3779 uint32_t PersonalityFn = readNext<uint32_t>(Pos);
3780 outs() << " personality[" << i + 1
3781 << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n';
3782 }
3783
3784 //===----------------------------------
3785 // The level 1 index entries
3786 //===----------------------------------
3787
3788 // These specify an approximate place to start searching for the more detailed
3789 // information, sorted by PC.
3790
3791 struct IndexEntry {
3792 uint32_t FunctionOffset;
3793 uint32_t SecondLevelPageStart;
3794 uint32_t LSDAStart;
3795 };
3796
3797 SmallVector<IndexEntry, 4> IndexEntries;
3798
3799 outs() << " Top level indices: (count = " << NumIndices << ")\n";
3800 Pos = Contents.data() + IndicesStart;
3801 for (unsigned i = 0; i < NumIndices; ++i) {
3802 IndexEntry Entry;
3803
3804 Entry.FunctionOffset = readNext<uint32_t>(Pos);
3805 Entry.SecondLevelPageStart = readNext<uint32_t>(Pos);
3806 Entry.LSDAStart = readNext<uint32_t>(Pos);
3807 IndexEntries.push_back(Entry);
3808
3809 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003810 << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset)
3811 << ", "
Tim Northover39c70bb2014-08-12 11:52:59 +00003812 << "2nd level page offset="
3813 << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003814 << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003815 }
3816
Tim Northover39c70bb2014-08-12 11:52:59 +00003817 //===----------------------------------
3818 // Next come the LSDA tables
3819 //===----------------------------------
3820
3821 // The LSDA layout is rather implicit: it's a contiguous array of entries from
3822 // the first top-level index's LSDAOffset to the last (sentinel).
3823
3824 outs() << " LSDA descriptors:\n";
3825 Pos = Contents.data() + IndexEntries[0].LSDAStart;
3826 int NumLSDAs = (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) /
3827 (2 * sizeof(uint32_t));
3828 for (int i = 0; i < NumLSDAs; ++i) {
3829 uint32_t FunctionOffset = readNext<uint32_t>(Pos);
3830 uint32_t LSDAOffset = readNext<uint32_t>(Pos);
3831 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003832 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
3833 << ", "
3834 << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00003835 }
3836
3837 //===----------------------------------
3838 // Finally, the 2nd level indices
3839 //===----------------------------------
3840
3841 // Generally these are 4K in size, and have 2 possible forms:
3842 // + Regular stores up to 511 entries with disparate encodings
3843 // + Compressed stores up to 1021 entries if few enough compact encoding
3844 // values are used.
3845 outs() << " Second level indices:\n";
3846 for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) {
3847 // The final sentinel top-level index has no associated 2nd level page
3848 if (IndexEntries[i].SecondLevelPageStart == 0)
3849 break;
3850
3851 outs() << " Second level index[" << i << "]: "
3852 << "offset in section="
3853 << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart)
3854 << ", "
3855 << "base function offset="
3856 << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n';
3857
3858 Pos = Contents.data() + IndexEntries[i].SecondLevelPageStart;
Aaron Ballman80930af2014-08-14 13:53:19 +00003859 uint32_t Kind = *reinterpret_cast<const support::ulittle32_t *>(Pos);
Tim Northover39c70bb2014-08-12 11:52:59 +00003860 if (Kind == 2)
3861 printRegularSecondLevelUnwindPage(Pos);
3862 else if (Kind == 3)
3863 printCompressedSecondLevelUnwindPage(Pos, IndexEntries[i].FunctionOffset,
3864 CommonEncodings);
3865 else
3866 llvm_unreachable("Do not know how to print this kind of 2nd level page");
Tim Northover39c70bb2014-08-12 11:52:59 +00003867 }
3868}
3869
Tim Northover4bd286a2014-08-01 13:07:19 +00003870void llvm::printMachOUnwindInfo(const MachOObjectFile *Obj) {
3871 std::map<uint64_t, SymbolRef> Symbols;
3872 for (const SymbolRef &SymRef : Obj->symbols()) {
3873 // Discard any undefined or absolute symbols. They're not going to take part
3874 // in the convenience lookup for unwind info and just take up resources.
3875 section_iterator Section = Obj->section_end();
3876 SymRef.getSection(Section);
3877 if (Section == Obj->section_end())
3878 continue;
3879
3880 uint64_t Addr;
3881 SymRef.getAddress(Addr);
3882 Symbols.insert(std::make_pair(Addr, SymRef));
3883 }
3884
3885 for (const SectionRef &Section : Obj->sections()) {
3886 StringRef SectName;
3887 Section.getName(SectName);
3888 if (SectName == "__compact_unwind")
3889 printMachOCompactUnwindSection(Obj, Symbols, Section);
3890 else if (SectName == "__unwind_info")
Tim Northover39c70bb2014-08-12 11:52:59 +00003891 printMachOUnwindInfoSection(Obj, Symbols, Section);
Tim Northover4bd286a2014-08-01 13:07:19 +00003892 else if (SectName == "__eh_frame")
3893 outs() << "llvm-objdump: warning: unhandled __eh_frame section\n";
Tim Northover4bd286a2014-08-01 13:07:19 +00003894 }
3895}
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003896
3897static void PrintMachHeader(uint32_t magic, uint32_t cputype,
3898 uint32_t cpusubtype, uint32_t filetype,
3899 uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags,
3900 bool verbose) {
3901 outs() << "Mach header\n";
3902 outs() << " magic cputype cpusubtype caps filetype ncmds "
3903 "sizeofcmds flags\n";
3904 if (verbose) {
3905 if (magic == MachO::MH_MAGIC)
3906 outs() << " MH_MAGIC";
3907 else if (magic == MachO::MH_MAGIC_64)
3908 outs() << "MH_MAGIC_64";
3909 else
3910 outs() << format(" 0x%08" PRIx32, magic);
3911 switch (cputype) {
3912 case MachO::CPU_TYPE_I386:
3913 outs() << " I386";
3914 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3915 case MachO::CPU_SUBTYPE_I386_ALL:
3916 outs() << " ALL";
3917 break;
3918 default:
3919 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3920 break;
3921 }
3922 break;
3923 case MachO::CPU_TYPE_X86_64:
3924 outs() << " X86_64";
Kevin Enderby131d1772015-01-09 19:22:37 +00003925 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3926 case MachO::CPU_SUBTYPE_X86_64_ALL:
3927 outs() << " ALL";
3928 break;
3929 case MachO::CPU_SUBTYPE_X86_64_H:
3930 outs() << " Haswell";
3931 break;
3932 default:
3933 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3934 break;
3935 }
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003936 break;
3937 case MachO::CPU_TYPE_ARM:
3938 outs() << " ARM";
3939 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3940 case MachO::CPU_SUBTYPE_ARM_ALL:
3941 outs() << " ALL";
3942 break;
3943 case MachO::CPU_SUBTYPE_ARM_V4T:
3944 outs() << " V4T";
3945 break;
3946 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
3947 outs() << " V5TEJ";
3948 break;
3949 case MachO::CPU_SUBTYPE_ARM_XSCALE:
3950 outs() << " XSCALE";
3951 break;
3952 case MachO::CPU_SUBTYPE_ARM_V6:
3953 outs() << " V6";
3954 break;
3955 case MachO::CPU_SUBTYPE_ARM_V6M:
3956 outs() << " V6M";
3957 break;
3958 case MachO::CPU_SUBTYPE_ARM_V7:
3959 outs() << " V7";
3960 break;
3961 case MachO::CPU_SUBTYPE_ARM_V7EM:
3962 outs() << " V7EM";
3963 break;
3964 case MachO::CPU_SUBTYPE_ARM_V7K:
3965 outs() << " V7K";
3966 break;
3967 case MachO::CPU_SUBTYPE_ARM_V7M:
3968 outs() << " V7M";
3969 break;
3970 case MachO::CPU_SUBTYPE_ARM_V7S:
3971 outs() << " V7S";
3972 break;
3973 default:
3974 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3975 break;
3976 }
3977 break;
3978 case MachO::CPU_TYPE_ARM64:
3979 outs() << " ARM64";
3980 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3981 case MachO::CPU_SUBTYPE_ARM64_ALL:
3982 outs() << " ALL";
3983 break;
3984 default:
3985 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3986 break;
3987 }
3988 break;
3989 case MachO::CPU_TYPE_POWERPC:
3990 outs() << " PPC";
3991 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
3992 case MachO::CPU_SUBTYPE_POWERPC_ALL:
3993 outs() << " ALL";
3994 break;
3995 default:
3996 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
3997 break;
3998 }
3999 break;
4000 case MachO::CPU_TYPE_POWERPC64:
4001 outs() << " PPC64";
4002 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
4003 case MachO::CPU_SUBTYPE_POWERPC_ALL:
4004 outs() << " ALL";
4005 break;
4006 default:
4007 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
4008 break;
4009 }
4010 break;
4011 }
4012 if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004013 outs() << " LIB64";
Kevin Enderbyb76d3862014-08-22 20:35:18 +00004014 } else {
4015 outs() << format(" 0x%02" PRIx32,
4016 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
4017 }
4018 switch (filetype) {
4019 case MachO::MH_OBJECT:
4020 outs() << " OBJECT";
4021 break;
4022 case MachO::MH_EXECUTE:
4023 outs() << " EXECUTE";
4024 break;
4025 case MachO::MH_FVMLIB:
4026 outs() << " FVMLIB";
4027 break;
4028 case MachO::MH_CORE:
4029 outs() << " CORE";
4030 break;
4031 case MachO::MH_PRELOAD:
4032 outs() << " PRELOAD";
4033 break;
4034 case MachO::MH_DYLIB:
4035 outs() << " DYLIB";
4036 break;
4037 case MachO::MH_DYLIB_STUB:
4038 outs() << " DYLIB_STUB";
4039 break;
4040 case MachO::MH_DYLINKER:
4041 outs() << " DYLINKER";
4042 break;
4043 case MachO::MH_BUNDLE:
4044 outs() << " BUNDLE";
4045 break;
4046 case MachO::MH_DSYM:
4047 outs() << " DSYM";
4048 break;
4049 case MachO::MH_KEXT_BUNDLE:
4050 outs() << " KEXTBUNDLE";
4051 break;
4052 default:
4053 outs() << format(" %10u", filetype);
4054 break;
4055 }
4056 outs() << format(" %5u", ncmds);
4057 outs() << format(" %10u", sizeofcmds);
4058 uint32_t f = flags;
4059 if (f & MachO::MH_NOUNDEFS) {
4060 outs() << " NOUNDEFS";
4061 f &= ~MachO::MH_NOUNDEFS;
4062 }
4063 if (f & MachO::MH_INCRLINK) {
4064 outs() << " INCRLINK";
4065 f &= ~MachO::MH_INCRLINK;
4066 }
4067 if (f & MachO::MH_DYLDLINK) {
4068 outs() << " DYLDLINK";
4069 f &= ~MachO::MH_DYLDLINK;
4070 }
4071 if (f & MachO::MH_BINDATLOAD) {
4072 outs() << " BINDATLOAD";
4073 f &= ~MachO::MH_BINDATLOAD;
4074 }
4075 if (f & MachO::MH_PREBOUND) {
4076 outs() << " PREBOUND";
4077 f &= ~MachO::MH_PREBOUND;
4078 }
4079 if (f & MachO::MH_SPLIT_SEGS) {
4080 outs() << " SPLIT_SEGS";
4081 f &= ~MachO::MH_SPLIT_SEGS;
4082 }
4083 if (f & MachO::MH_LAZY_INIT) {
4084 outs() << " LAZY_INIT";
4085 f &= ~MachO::MH_LAZY_INIT;
4086 }
4087 if (f & MachO::MH_TWOLEVEL) {
4088 outs() << " TWOLEVEL";
4089 f &= ~MachO::MH_TWOLEVEL;
4090 }
4091 if (f & MachO::MH_FORCE_FLAT) {
4092 outs() << " FORCE_FLAT";
4093 f &= ~MachO::MH_FORCE_FLAT;
4094 }
4095 if (f & MachO::MH_NOMULTIDEFS) {
4096 outs() << " NOMULTIDEFS";
4097 f &= ~MachO::MH_NOMULTIDEFS;
4098 }
4099 if (f & MachO::MH_NOFIXPREBINDING) {
4100 outs() << " NOFIXPREBINDING";
4101 f &= ~MachO::MH_NOFIXPREBINDING;
4102 }
4103 if (f & MachO::MH_PREBINDABLE) {
4104 outs() << " PREBINDABLE";
4105 f &= ~MachO::MH_PREBINDABLE;
4106 }
4107 if (f & MachO::MH_ALLMODSBOUND) {
4108 outs() << " ALLMODSBOUND";
4109 f &= ~MachO::MH_ALLMODSBOUND;
4110 }
4111 if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) {
4112 outs() << " SUBSECTIONS_VIA_SYMBOLS";
4113 f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
4114 }
4115 if (f & MachO::MH_CANONICAL) {
4116 outs() << " CANONICAL";
4117 f &= ~MachO::MH_CANONICAL;
4118 }
4119 if (f & MachO::MH_WEAK_DEFINES) {
4120 outs() << " WEAK_DEFINES";
4121 f &= ~MachO::MH_WEAK_DEFINES;
4122 }
4123 if (f & MachO::MH_BINDS_TO_WEAK) {
4124 outs() << " BINDS_TO_WEAK";
4125 f &= ~MachO::MH_BINDS_TO_WEAK;
4126 }
4127 if (f & MachO::MH_ALLOW_STACK_EXECUTION) {
4128 outs() << " ALLOW_STACK_EXECUTION";
4129 f &= ~MachO::MH_ALLOW_STACK_EXECUTION;
4130 }
4131 if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) {
4132 outs() << " DEAD_STRIPPABLE_DYLIB";
4133 f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB;
4134 }
4135 if (f & MachO::MH_PIE) {
4136 outs() << " PIE";
4137 f &= ~MachO::MH_PIE;
4138 }
4139 if (f & MachO::MH_NO_REEXPORTED_DYLIBS) {
4140 outs() << " NO_REEXPORTED_DYLIBS";
4141 f &= ~MachO::MH_NO_REEXPORTED_DYLIBS;
4142 }
4143 if (f & MachO::MH_HAS_TLV_DESCRIPTORS) {
4144 outs() << " MH_HAS_TLV_DESCRIPTORS";
4145 f &= ~MachO::MH_HAS_TLV_DESCRIPTORS;
4146 }
4147 if (f & MachO::MH_NO_HEAP_EXECUTION) {
4148 outs() << " MH_NO_HEAP_EXECUTION";
4149 f &= ~MachO::MH_NO_HEAP_EXECUTION;
4150 }
4151 if (f & MachO::MH_APP_EXTENSION_SAFE) {
4152 outs() << " APP_EXTENSION_SAFE";
4153 f &= ~MachO::MH_APP_EXTENSION_SAFE;
4154 }
4155 if (f != 0 || flags == 0)
4156 outs() << format(" 0x%08" PRIx32, f);
4157 } else {
4158 outs() << format(" 0x%08" PRIx32, magic);
4159 outs() << format(" %7d", cputype);
4160 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
4161 outs() << format(" 0x%02" PRIx32,
4162 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
4163 outs() << format(" %10u", filetype);
4164 outs() << format(" %5u", ncmds);
4165 outs() << format(" %10u", sizeofcmds);
4166 outs() << format(" 0x%08" PRIx32, flags);
4167 }
4168 outs() << "\n";
4169}
4170
Kevin Enderby956366c2014-08-29 22:30:52 +00004171static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize,
4172 StringRef SegName, uint64_t vmaddr,
4173 uint64_t vmsize, uint64_t fileoff,
4174 uint64_t filesize, uint32_t maxprot,
4175 uint32_t initprot, uint32_t nsects,
4176 uint32_t flags, uint32_t object_size,
4177 bool verbose) {
4178 uint64_t expected_cmdsize;
4179 if (cmd == MachO::LC_SEGMENT) {
4180 outs() << " cmd LC_SEGMENT\n";
4181 expected_cmdsize = nsects;
4182 expected_cmdsize *= sizeof(struct MachO::section);
4183 expected_cmdsize += sizeof(struct MachO::segment_command);
4184 } else {
4185 outs() << " cmd LC_SEGMENT_64\n";
4186 expected_cmdsize = nsects;
4187 expected_cmdsize *= sizeof(struct MachO::section_64);
4188 expected_cmdsize += sizeof(struct MachO::segment_command_64);
4189 }
4190 outs() << " cmdsize " << cmdsize;
4191 if (cmdsize != expected_cmdsize)
4192 outs() << " Inconsistent size\n";
4193 else
4194 outs() << "\n";
4195 outs() << " segname " << SegName << "\n";
4196 if (cmd == MachO::LC_SEGMENT_64) {
4197 outs() << " vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n";
4198 outs() << " vmsize " << format("0x%016" PRIx64, vmsize) << "\n";
4199 } else {
Kevin Enderbyadb7c432014-12-16 18:58:11 +00004200 outs() << " vmaddr " << format("0x%08" PRIx64, vmaddr) << "\n";
4201 outs() << " vmsize " << format("0x%08" PRIx64, vmsize) << "\n";
Kevin Enderby956366c2014-08-29 22:30:52 +00004202 }
4203 outs() << " fileoff " << fileoff;
4204 if (fileoff > object_size)
4205 outs() << " (past end of file)\n";
4206 else
4207 outs() << "\n";
4208 outs() << " filesize " << filesize;
4209 if (fileoff + filesize > object_size)
4210 outs() << " (past end of file)\n";
4211 else
4212 outs() << "\n";
4213 if (verbose) {
4214 if ((maxprot &
4215 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
4216 MachO::VM_PROT_EXECUTE)) != 0)
4217 outs() << " maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n";
4218 else {
4219 if (maxprot & MachO::VM_PROT_READ)
4220 outs() << " maxprot r";
4221 else
4222 outs() << " maxprot -";
4223 if (maxprot & MachO::VM_PROT_WRITE)
4224 outs() << "w";
4225 else
4226 outs() << "-";
4227 if (maxprot & MachO::VM_PROT_EXECUTE)
4228 outs() << "x\n";
4229 else
4230 outs() << "-\n";
4231 }
4232 if ((initprot &
4233 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
4234 MachO::VM_PROT_EXECUTE)) != 0)
4235 outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n";
4236 else {
4237 if (initprot & MachO::VM_PROT_READ)
4238 outs() << " initprot r";
4239 else
4240 outs() << " initprot -";
4241 if (initprot & MachO::VM_PROT_WRITE)
4242 outs() << "w";
4243 else
4244 outs() << "-";
4245 if (initprot & MachO::VM_PROT_EXECUTE)
4246 outs() << "x\n";
4247 else
4248 outs() << "-\n";
4249 }
4250 } else {
4251 outs() << " maxprot " << format("0x%08" PRIx32, maxprot) << "\n";
4252 outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n";
4253 }
4254 outs() << " nsects " << nsects << "\n";
4255 if (verbose) {
4256 outs() << " flags";
4257 if (flags == 0)
4258 outs() << " (none)\n";
4259 else {
4260 if (flags & MachO::SG_HIGHVM) {
4261 outs() << " HIGHVM";
4262 flags &= ~MachO::SG_HIGHVM;
4263 }
4264 if (flags & MachO::SG_FVMLIB) {
4265 outs() << " FVMLIB";
4266 flags &= ~MachO::SG_FVMLIB;
4267 }
4268 if (flags & MachO::SG_NORELOC) {
4269 outs() << " NORELOC";
4270 flags &= ~MachO::SG_NORELOC;
4271 }
4272 if (flags & MachO::SG_PROTECTED_VERSION_1) {
4273 outs() << " PROTECTED_VERSION_1";
4274 flags &= ~MachO::SG_PROTECTED_VERSION_1;
4275 }
4276 if (flags)
4277 outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n";
4278 else
4279 outs() << "\n";
4280 }
4281 } else {
4282 outs() << " flags " << format("0x%" PRIx32, flags) << "\n";
4283 }
4284}
4285
4286static void PrintSection(const char *sectname, const char *segname,
4287 uint64_t addr, uint64_t size, uint32_t offset,
4288 uint32_t align, uint32_t reloff, uint32_t nreloc,
4289 uint32_t flags, uint32_t reserved1, uint32_t reserved2,
4290 uint32_t cmd, const char *sg_segname,
4291 uint32_t filetype, uint32_t object_size,
4292 bool verbose) {
4293 outs() << "Section\n";
4294 outs() << " sectname " << format("%.16s\n", sectname);
4295 outs() << " segname " << format("%.16s", segname);
4296 if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0)
4297 outs() << " (does not match segment)\n";
4298 else
4299 outs() << "\n";
4300 if (cmd == MachO::LC_SEGMENT_64) {
4301 outs() << " addr " << format("0x%016" PRIx64, addr) << "\n";
4302 outs() << " size " << format("0x%016" PRIx64, size);
4303 } else {
Kevin Enderby75594b62014-12-16 21:00:25 +00004304 outs() << " addr " << format("0x%08" PRIx64, addr) << "\n";
4305 outs() << " size " << format("0x%08" PRIx64, size);
Kevin Enderby956366c2014-08-29 22:30:52 +00004306 }
4307 if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size)
4308 outs() << " (past end of file)\n";
4309 else
4310 outs() << "\n";
4311 outs() << " offset " << offset;
4312 if (offset > object_size)
4313 outs() << " (past end of file)\n";
4314 else
4315 outs() << "\n";
4316 uint32_t align_shifted = 1 << align;
4317 outs() << " align 2^" << align << " (" << align_shifted << ")\n";
4318 outs() << " reloff " << reloff;
4319 if (reloff > object_size)
4320 outs() << " (past end of file)\n";
4321 else
4322 outs() << "\n";
4323 outs() << " nreloc " << nreloc;
4324 if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size)
4325 outs() << " (past end of file)\n";
4326 else
4327 outs() << "\n";
4328 uint32_t section_type = flags & MachO::SECTION_TYPE;
4329 if (verbose) {
4330 outs() << " type";
4331 if (section_type == MachO::S_REGULAR)
4332 outs() << " S_REGULAR\n";
4333 else if (section_type == MachO::S_ZEROFILL)
4334 outs() << " S_ZEROFILL\n";
4335 else if (section_type == MachO::S_CSTRING_LITERALS)
4336 outs() << " S_CSTRING_LITERALS\n";
4337 else if (section_type == MachO::S_4BYTE_LITERALS)
4338 outs() << " S_4BYTE_LITERALS\n";
4339 else if (section_type == MachO::S_8BYTE_LITERALS)
4340 outs() << " S_8BYTE_LITERALS\n";
4341 else if (section_type == MachO::S_16BYTE_LITERALS)
4342 outs() << " S_16BYTE_LITERALS\n";
4343 else if (section_type == MachO::S_LITERAL_POINTERS)
4344 outs() << " S_LITERAL_POINTERS\n";
4345 else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS)
4346 outs() << " S_NON_LAZY_SYMBOL_POINTERS\n";
4347 else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS)
4348 outs() << " S_LAZY_SYMBOL_POINTERS\n";
4349 else if (section_type == MachO::S_SYMBOL_STUBS)
4350 outs() << " S_SYMBOL_STUBS\n";
4351 else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS)
4352 outs() << " S_MOD_INIT_FUNC_POINTERS\n";
4353 else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS)
4354 outs() << " S_MOD_TERM_FUNC_POINTERS\n";
4355 else if (section_type == MachO::S_COALESCED)
4356 outs() << " S_COALESCED\n";
4357 else if (section_type == MachO::S_INTERPOSING)
4358 outs() << " S_INTERPOSING\n";
4359 else if (section_type == MachO::S_DTRACE_DOF)
4360 outs() << " S_DTRACE_DOF\n";
4361 else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS)
4362 outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n";
4363 else if (section_type == MachO::S_THREAD_LOCAL_REGULAR)
4364 outs() << " S_THREAD_LOCAL_REGULAR\n";
4365 else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL)
4366 outs() << " S_THREAD_LOCAL_ZEROFILL\n";
4367 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES)
4368 outs() << " S_THREAD_LOCAL_VARIABLES\n";
4369 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
4370 outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n";
4371 else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS)
4372 outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n";
4373 else
4374 outs() << format("0x%08" PRIx32, section_type) << "\n";
4375 outs() << "attributes";
4376 uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES;
4377 if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS)
4378 outs() << " PURE_INSTRUCTIONS";
4379 if (section_attributes & MachO::S_ATTR_NO_TOC)
4380 outs() << " NO_TOC";
4381 if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS)
4382 outs() << " STRIP_STATIC_SYMS";
4383 if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP)
4384 outs() << " NO_DEAD_STRIP";
4385 if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT)
4386 outs() << " LIVE_SUPPORT";
4387 if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE)
4388 outs() << " SELF_MODIFYING_CODE";
4389 if (section_attributes & MachO::S_ATTR_DEBUG)
4390 outs() << " DEBUG";
4391 if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS)
4392 outs() << " SOME_INSTRUCTIONS";
4393 if (section_attributes & MachO::S_ATTR_EXT_RELOC)
4394 outs() << " EXT_RELOC";
4395 if (section_attributes & MachO::S_ATTR_LOC_RELOC)
4396 outs() << " LOC_RELOC";
4397 if (section_attributes == 0)
4398 outs() << " (none)";
4399 outs() << "\n";
4400 } else
4401 outs() << " flags " << format("0x%08" PRIx32, flags) << "\n";
4402 outs() << " reserved1 " << reserved1;
4403 if (section_type == MachO::S_SYMBOL_STUBS ||
4404 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
4405 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
4406 section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
4407 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
4408 outs() << " (index into indirect symbol table)\n";
4409 else
4410 outs() << "\n";
4411 outs() << " reserved2 " << reserved2;
4412 if (section_type == MachO::S_SYMBOL_STUBS)
4413 outs() << " (size of stubs)\n";
4414 else
4415 outs() << "\n";
4416}
4417
David Majnemer73cc6ff2014-11-13 19:48:56 +00004418static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit,
Kevin Enderby956366c2014-08-29 22:30:52 +00004419 uint32_t object_size) {
4420 outs() << " cmd LC_SYMTAB\n";
4421 outs() << " cmdsize " << st.cmdsize;
4422 if (st.cmdsize != sizeof(struct MachO::symtab_command))
4423 outs() << " Incorrect size\n";
4424 else
4425 outs() << "\n";
4426 outs() << " symoff " << st.symoff;
4427 if (st.symoff > object_size)
4428 outs() << " (past end of file)\n";
4429 else
4430 outs() << "\n";
4431 outs() << " nsyms " << st.nsyms;
4432 uint64_t big_size;
David Majnemer73cc6ff2014-11-13 19:48:56 +00004433 if (Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00004434 big_size = st.nsyms;
4435 big_size *= sizeof(struct MachO::nlist_64);
4436 big_size += st.symoff;
4437 if (big_size > object_size)
4438 outs() << " (past end of file)\n";
4439 else
4440 outs() << "\n";
4441 } else {
4442 big_size = st.nsyms;
4443 big_size *= sizeof(struct MachO::nlist);
4444 big_size += st.symoff;
4445 if (big_size > object_size)
4446 outs() << " (past end of file)\n";
4447 else
4448 outs() << "\n";
4449 }
4450 outs() << " stroff " << st.stroff;
4451 if (st.stroff > object_size)
4452 outs() << " (past end of file)\n";
4453 else
4454 outs() << "\n";
4455 outs() << " strsize " << st.strsize;
4456 big_size = st.stroff;
4457 big_size += st.strsize;
4458 if (big_size > object_size)
4459 outs() << " (past end of file)\n";
4460 else
4461 outs() << "\n";
4462}
4463
4464static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,
4465 uint32_t nsyms, uint32_t object_size,
David Majnemer73cc6ff2014-11-13 19:48:56 +00004466 bool Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00004467 outs() << " cmd LC_DYSYMTAB\n";
4468 outs() << " cmdsize " << dyst.cmdsize;
4469 if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command))
4470 outs() << " Incorrect size\n";
4471 else
4472 outs() << "\n";
4473 outs() << " ilocalsym " << dyst.ilocalsym;
4474 if (dyst.ilocalsym > nsyms)
4475 outs() << " (greater than the number of symbols)\n";
4476 else
4477 outs() << "\n";
4478 outs() << " nlocalsym " << dyst.nlocalsym;
4479 uint64_t big_size;
4480 big_size = dyst.ilocalsym;
4481 big_size += dyst.nlocalsym;
4482 if (big_size > nsyms)
4483 outs() << " (past the end of the symbol table)\n";
4484 else
4485 outs() << "\n";
4486 outs() << " iextdefsym " << dyst.iextdefsym;
4487 if (dyst.iextdefsym > nsyms)
4488 outs() << " (greater than the number of symbols)\n";
4489 else
4490 outs() << "\n";
4491 outs() << " nextdefsym " << dyst.nextdefsym;
4492 big_size = dyst.iextdefsym;
4493 big_size += dyst.nextdefsym;
4494 if (big_size > nsyms)
4495 outs() << " (past the end of the symbol table)\n";
4496 else
4497 outs() << "\n";
4498 outs() << " iundefsym " << dyst.iundefsym;
4499 if (dyst.iundefsym > nsyms)
4500 outs() << " (greater than the number of symbols)\n";
4501 else
4502 outs() << "\n";
4503 outs() << " nundefsym " << dyst.nundefsym;
4504 big_size = dyst.iundefsym;
4505 big_size += dyst.nundefsym;
4506 if (big_size > nsyms)
4507 outs() << " (past the end of the symbol table)\n";
4508 else
4509 outs() << "\n";
4510 outs() << " tocoff " << dyst.tocoff;
4511 if (dyst.tocoff > object_size)
4512 outs() << " (past end of file)\n";
4513 else
4514 outs() << "\n";
4515 outs() << " ntoc " << dyst.ntoc;
4516 big_size = dyst.ntoc;
4517 big_size *= sizeof(struct MachO::dylib_table_of_contents);
4518 big_size += dyst.tocoff;
4519 if (big_size > object_size)
4520 outs() << " (past end of file)\n";
4521 else
4522 outs() << "\n";
4523 outs() << " modtaboff " << dyst.modtaboff;
4524 if (dyst.modtaboff > object_size)
4525 outs() << " (past end of file)\n";
4526 else
4527 outs() << "\n";
4528 outs() << " nmodtab " << dyst.nmodtab;
4529 uint64_t modtabend;
David Majnemer73cc6ff2014-11-13 19:48:56 +00004530 if (Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00004531 modtabend = dyst.nmodtab;
4532 modtabend *= sizeof(struct MachO::dylib_module_64);
4533 modtabend += dyst.modtaboff;
4534 } else {
4535 modtabend = dyst.nmodtab;
4536 modtabend *= sizeof(struct MachO::dylib_module);
4537 modtabend += dyst.modtaboff;
4538 }
4539 if (modtabend > object_size)
4540 outs() << " (past end of file)\n";
4541 else
4542 outs() << "\n";
4543 outs() << " extrefsymoff " << dyst.extrefsymoff;
4544 if (dyst.extrefsymoff > object_size)
4545 outs() << " (past end of file)\n";
4546 else
4547 outs() << "\n";
4548 outs() << " nextrefsyms " << dyst.nextrefsyms;
4549 big_size = dyst.nextrefsyms;
4550 big_size *= sizeof(struct MachO::dylib_reference);
4551 big_size += dyst.extrefsymoff;
4552 if (big_size > object_size)
4553 outs() << " (past end of file)\n";
4554 else
4555 outs() << "\n";
4556 outs() << " indirectsymoff " << dyst.indirectsymoff;
4557 if (dyst.indirectsymoff > object_size)
4558 outs() << " (past end of file)\n";
4559 else
4560 outs() << "\n";
4561 outs() << " nindirectsyms " << dyst.nindirectsyms;
4562 big_size = dyst.nindirectsyms;
4563 big_size *= sizeof(uint32_t);
4564 big_size += dyst.indirectsymoff;
4565 if (big_size > object_size)
4566 outs() << " (past end of file)\n";
4567 else
4568 outs() << "\n";
4569 outs() << " extreloff " << dyst.extreloff;
4570 if (dyst.extreloff > object_size)
4571 outs() << " (past end of file)\n";
4572 else
4573 outs() << "\n";
4574 outs() << " nextrel " << dyst.nextrel;
4575 big_size = dyst.nextrel;
4576 big_size *= sizeof(struct MachO::relocation_info);
4577 big_size += dyst.extreloff;
4578 if (big_size > object_size)
4579 outs() << " (past end of file)\n";
4580 else
4581 outs() << "\n";
4582 outs() << " locreloff " << dyst.locreloff;
4583 if (dyst.locreloff > object_size)
4584 outs() << " (past end of file)\n";
4585 else
4586 outs() << "\n";
4587 outs() << " nlocrel " << dyst.nlocrel;
4588 big_size = dyst.nlocrel;
4589 big_size *= sizeof(struct MachO::relocation_info);
4590 big_size += dyst.locreloff;
4591 if (big_size > object_size)
4592 outs() << " (past end of file)\n";
4593 else
4594 outs() << "\n";
4595}
4596
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004597static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,
4598 uint32_t object_size) {
4599 if (dc.cmd == MachO::LC_DYLD_INFO)
4600 outs() << " cmd LC_DYLD_INFO\n";
4601 else
4602 outs() << " cmd LC_DYLD_INFO_ONLY\n";
4603 outs() << " cmdsize " << dc.cmdsize;
4604 if (dc.cmdsize != sizeof(struct MachO::dyld_info_command))
4605 outs() << " Incorrect size\n";
4606 else
4607 outs() << "\n";
4608 outs() << " rebase_off " << dc.rebase_off;
4609 if (dc.rebase_off > object_size)
4610 outs() << " (past end of file)\n";
4611 else
4612 outs() << "\n";
4613 outs() << " rebase_size " << dc.rebase_size;
4614 uint64_t big_size;
4615 big_size = dc.rebase_off;
4616 big_size += dc.rebase_size;
4617 if (big_size > object_size)
4618 outs() << " (past end of file)\n";
4619 else
4620 outs() << "\n";
4621 outs() << " bind_off " << dc.bind_off;
4622 if (dc.bind_off > object_size)
4623 outs() << " (past end of file)\n";
4624 else
4625 outs() << "\n";
4626 outs() << " bind_size " << dc.bind_size;
4627 big_size = dc.bind_off;
4628 big_size += dc.bind_size;
4629 if (big_size > object_size)
4630 outs() << " (past end of file)\n";
4631 else
4632 outs() << "\n";
4633 outs() << " weak_bind_off " << dc.weak_bind_off;
4634 if (dc.weak_bind_off > object_size)
4635 outs() << " (past end of file)\n";
4636 else
4637 outs() << "\n";
4638 outs() << " weak_bind_size " << dc.weak_bind_size;
4639 big_size = dc.weak_bind_off;
4640 big_size += dc.weak_bind_size;
4641 if (big_size > object_size)
4642 outs() << " (past end of file)\n";
4643 else
4644 outs() << "\n";
4645 outs() << " lazy_bind_off " << dc.lazy_bind_off;
4646 if (dc.lazy_bind_off > object_size)
4647 outs() << " (past end of file)\n";
4648 else
4649 outs() << "\n";
4650 outs() << " lazy_bind_size " << dc.lazy_bind_size;
4651 big_size = dc.lazy_bind_off;
4652 big_size += dc.lazy_bind_size;
4653 if (big_size > object_size)
4654 outs() << " (past end of file)\n";
4655 else
4656 outs() << "\n";
4657 outs() << " export_off " << dc.export_off;
4658 if (dc.export_off > object_size)
4659 outs() << " (past end of file)\n";
4660 else
4661 outs() << "\n";
4662 outs() << " export_size " << dc.export_size;
4663 big_size = dc.export_off;
4664 big_size += dc.export_size;
4665 if (big_size > object_size)
4666 outs() << " (past end of file)\n";
4667 else
4668 outs() << "\n";
4669}
4670
4671static void PrintDyldLoadCommand(MachO::dylinker_command dyld,
4672 const char *Ptr) {
4673 if (dyld.cmd == MachO::LC_ID_DYLINKER)
4674 outs() << " cmd LC_ID_DYLINKER\n";
4675 else if (dyld.cmd == MachO::LC_LOAD_DYLINKER)
4676 outs() << " cmd LC_LOAD_DYLINKER\n";
4677 else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT)
4678 outs() << " cmd LC_DYLD_ENVIRONMENT\n";
4679 else
4680 outs() << " cmd ?(" << dyld.cmd << ")\n";
4681 outs() << " cmdsize " << dyld.cmdsize;
4682 if (dyld.cmdsize < sizeof(struct MachO::dylinker_command))
4683 outs() << " Incorrect size\n";
4684 else
4685 outs() << "\n";
4686 if (dyld.name >= dyld.cmdsize)
4687 outs() << " name ?(bad offset " << dyld.name << ")\n";
4688 else {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004689 const char *P = (const char *)(Ptr) + dyld.name;
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004690 outs() << " name " << P << " (offset " << dyld.name << ")\n";
4691 }
4692}
4693
4694static void PrintUuidLoadCommand(MachO::uuid_command uuid) {
4695 outs() << " cmd LC_UUID\n";
4696 outs() << " cmdsize " << uuid.cmdsize;
4697 if (uuid.cmdsize != sizeof(struct MachO::uuid_command))
4698 outs() << " Incorrect size\n";
4699 else
4700 outs() << "\n";
4701 outs() << " uuid ";
4702 outs() << format("%02" PRIX32, uuid.uuid[0]);
4703 outs() << format("%02" PRIX32, uuid.uuid[1]);
4704 outs() << format("%02" PRIX32, uuid.uuid[2]);
4705 outs() << format("%02" PRIX32, uuid.uuid[3]);
4706 outs() << "-";
4707 outs() << format("%02" PRIX32, uuid.uuid[4]);
4708 outs() << format("%02" PRIX32, uuid.uuid[5]);
4709 outs() << "-";
4710 outs() << format("%02" PRIX32, uuid.uuid[6]);
4711 outs() << format("%02" PRIX32, uuid.uuid[7]);
4712 outs() << "-";
4713 outs() << format("%02" PRIX32, uuid.uuid[8]);
4714 outs() << format("%02" PRIX32, uuid.uuid[9]);
4715 outs() << "-";
4716 outs() << format("%02" PRIX32, uuid.uuid[10]);
4717 outs() << format("%02" PRIX32, uuid.uuid[11]);
4718 outs() << format("%02" PRIX32, uuid.uuid[12]);
4719 outs() << format("%02" PRIX32, uuid.uuid[13]);
4720 outs() << format("%02" PRIX32, uuid.uuid[14]);
4721 outs() << format("%02" PRIX32, uuid.uuid[15]);
4722 outs() << "\n";
4723}
4724
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004725static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) {
Jean-Daniel Dupas00cc1f52014-12-04 07:37:02 +00004726 outs() << " cmd LC_RPATH\n";
4727 outs() << " cmdsize " << rpath.cmdsize;
4728 if (rpath.cmdsize < sizeof(struct MachO::rpath_command))
4729 outs() << " Incorrect size\n";
4730 else
4731 outs() << "\n";
4732 if (rpath.path >= rpath.cmdsize)
4733 outs() << " path ?(bad offset " << rpath.path << ")\n";
4734 else {
4735 const char *P = (const char *)(Ptr) + rpath.path;
4736 outs() << " path " << P << " (offset " << rpath.path << ")\n";
4737 }
4738}
4739
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004740static void PrintVersionMinLoadCommand(MachO::version_min_command vd) {
4741 if (vd.cmd == MachO::LC_VERSION_MIN_MACOSX)
4742 outs() << " cmd LC_VERSION_MIN_MACOSX\n";
4743 else if (vd.cmd == MachO::LC_VERSION_MIN_IPHONEOS)
4744 outs() << " cmd LC_VERSION_MIN_IPHONEOS\n";
4745 else
4746 outs() << " cmd " << vd.cmd << " (?)\n";
4747 outs() << " cmdsize " << vd.cmdsize;
4748 if (vd.cmdsize != sizeof(struct MachO::version_min_command))
4749 outs() << " Incorrect size\n";
4750 else
4751 outs() << "\n";
4752 outs() << " version " << ((vd.version >> 16) & 0xffff) << "."
4753 << ((vd.version >> 8) & 0xff);
4754 if ((vd.version & 0xff) != 0)
4755 outs() << "." << (vd.version & 0xff);
4756 outs() << "\n";
4757 if (vd.sdk == 0)
Kevin Enderby57538292014-12-17 01:01:30 +00004758 outs() << " sdk n/a";
Kevin Enderby8ae63c12014-09-04 16:54:47 +00004759 else {
4760 outs() << " sdk " << ((vd.sdk >> 16) & 0xffff) << "."
4761 << ((vd.sdk >> 8) & 0xff);
4762 }
4763 if ((vd.sdk & 0xff) != 0)
4764 outs() << "." << (vd.sdk & 0xff);
4765 outs() << "\n";
4766}
4767
4768static void PrintSourceVersionCommand(MachO::source_version_command sd) {
4769 outs() << " cmd LC_SOURCE_VERSION\n";
4770 outs() << " cmdsize " << sd.cmdsize;
4771 if (sd.cmdsize != sizeof(struct MachO::source_version_command))
4772 outs() << " Incorrect size\n";
4773 else
4774 outs() << "\n";
4775 uint64_t a = (sd.version >> 40) & 0xffffff;
4776 uint64_t b = (sd.version >> 30) & 0x3ff;
4777 uint64_t c = (sd.version >> 20) & 0x3ff;
4778 uint64_t d = (sd.version >> 10) & 0x3ff;
4779 uint64_t e = sd.version & 0x3ff;
4780 outs() << " version " << a << "." << b;
4781 if (e != 0)
4782 outs() << "." << c << "." << d << "." << e;
4783 else if (d != 0)
4784 outs() << "." << c << "." << d;
4785 else if (c != 0)
4786 outs() << "." << c;
4787 outs() << "\n";
4788}
4789
4790static void PrintEntryPointCommand(MachO::entry_point_command ep) {
4791 outs() << " cmd LC_MAIN\n";
4792 outs() << " cmdsize " << ep.cmdsize;
4793 if (ep.cmdsize != sizeof(struct MachO::entry_point_command))
4794 outs() << " Incorrect size\n";
4795 else
4796 outs() << "\n";
4797 outs() << " entryoff " << ep.entryoff << "\n";
4798 outs() << " stacksize " << ep.stacksize << "\n";
4799}
4800
Kevin Enderby0804f4672014-12-16 23:25:52 +00004801static void PrintEncryptionInfoCommand(MachO::encryption_info_command ec,
4802 uint32_t object_size) {
4803 outs() << " cmd LC_ENCRYPTION_INFO\n";
4804 outs() << " cmdsize " << ec.cmdsize;
4805 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command))
4806 outs() << " Incorrect size\n";
4807 else
4808 outs() << "\n";
4809 outs() << " cryptoff " << ec.cryptoff;
4810 if (ec.cryptoff > object_size)
4811 outs() << " (past end of file)\n";
4812 else
4813 outs() << "\n";
4814 outs() << " cryptsize " << ec.cryptsize;
4815 if (ec.cryptsize > object_size)
4816 outs() << " (past end of file)\n";
4817 else
4818 outs() << "\n";
4819 outs() << " cryptid " << ec.cryptid << "\n";
4820}
4821
Kevin Enderby57538292014-12-17 01:01:30 +00004822static void PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec,
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004823 uint32_t object_size) {
Kevin Enderby57538292014-12-17 01:01:30 +00004824 outs() << " cmd LC_ENCRYPTION_INFO_64\n";
4825 outs() << " cmdsize " << ec.cmdsize;
4826 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command_64))
4827 outs() << " Incorrect size\n";
4828 else
4829 outs() << "\n";
4830 outs() << " cryptoff " << ec.cryptoff;
4831 if (ec.cryptoff > object_size)
4832 outs() << " (past end of file)\n";
4833 else
4834 outs() << "\n";
4835 outs() << " cryptsize " << ec.cryptsize;
4836 if (ec.cryptsize > object_size)
4837 outs() << " (past end of file)\n";
4838 else
4839 outs() << "\n";
4840 outs() << " cryptid " << ec.cryptid << "\n";
4841 outs() << " pad " << ec.pad << "\n";
4842}
4843
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00004844static void PrintLinkerOptionCommand(MachO::linker_option_command lo,
4845 const char *Ptr) {
4846 outs() << " cmd LC_LINKER_OPTION\n";
4847 outs() << " cmdsize " << lo.cmdsize;
4848 if (lo.cmdsize < sizeof(struct MachO::linker_option_command))
4849 outs() << " Incorrect size\n";
4850 else
4851 outs() << "\n";
4852 outs() << " count " << lo.count << "\n";
4853 const char *string = Ptr + sizeof(struct MachO::linker_option_command);
4854 uint32_t left = lo.cmdsize - sizeof(struct MachO::linker_option_command);
4855 uint32_t i = 0;
4856 while (left > 0) {
4857 while (*string == '\0' && left > 0) {
4858 string++;
4859 left--;
4860 }
4861 if (left > 0) {
4862 i++;
4863 outs() << " string #" << i << " " << format("%.*s\n", left, string);
David Majnemerd4449ed2014-12-20 08:24:43 +00004864 uint32_t NullPos = StringRef(string, left).find('\0');
4865 uint32_t len = std::min(NullPos, left) + 1;
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00004866 string += len;
4867 left -= len;
4868 }
4869 }
4870 if (lo.count != i)
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004871 outs() << " count " << lo.count << " does not match number of strings "
4872 << i << "\n";
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00004873}
4874
Kevin Enderbyb4b79312014-12-18 19:24:35 +00004875static void PrintSubFrameworkCommand(MachO::sub_framework_command sub,
4876 const char *Ptr) {
4877 outs() << " cmd LC_SUB_FRAMEWORK\n";
4878 outs() << " cmdsize " << sub.cmdsize;
4879 if (sub.cmdsize < sizeof(struct MachO::sub_framework_command))
4880 outs() << " Incorrect size\n";
4881 else
4882 outs() << "\n";
4883 if (sub.umbrella < sub.cmdsize) {
4884 const char *P = Ptr + sub.umbrella;
4885 outs() << " umbrella " << P << " (offset " << sub.umbrella << ")\n";
4886 } else {
4887 outs() << " umbrella ?(bad offset " << sub.umbrella << ")\n";
4888 }
4889}
4890
Kevin Enderbya2bd8d92014-12-18 23:13:26 +00004891static void PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub,
4892 const char *Ptr) {
4893 outs() << " cmd LC_SUB_UMBRELLA\n";
4894 outs() << " cmdsize " << sub.cmdsize;
4895 if (sub.cmdsize < sizeof(struct MachO::sub_umbrella_command))
4896 outs() << " Incorrect size\n";
4897 else
4898 outs() << "\n";
4899 if (sub.sub_umbrella < sub.cmdsize) {
4900 const char *P = Ptr + sub.sub_umbrella;
4901 outs() << " sub_umbrella " << P << " (offset " << sub.sub_umbrella << ")\n";
4902 } else {
4903 outs() << " sub_umbrella ?(bad offset " << sub.sub_umbrella << ")\n";
4904 }
4905}
4906
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00004907static void PrintSubLibraryCommand(MachO::sub_library_command sub,
Kevin Enderby66d51fc2015-01-08 00:25:24 +00004908 const char *Ptr) {
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00004909 outs() << " cmd LC_SUB_LIBRARY\n";
4910 outs() << " cmdsize " << sub.cmdsize;
4911 if (sub.cmdsize < sizeof(struct MachO::sub_library_command))
4912 outs() << " Incorrect size\n";
4913 else
4914 outs() << "\n";
4915 if (sub.sub_library < sub.cmdsize) {
4916 const char *P = Ptr + sub.sub_library;
4917 outs() << " sub_library " << P << " (offset " << sub.sub_library << ")\n";
4918 } else {
4919 outs() << " sub_library ?(bad offset " << sub.sub_library << ")\n";
4920 }
4921}
4922
Kevin Enderby186eac32014-12-19 21:06:24 +00004923static void PrintSubClientCommand(MachO::sub_client_command sub,
4924 const char *Ptr) {
4925 outs() << " cmd LC_SUB_CLIENT\n";
4926 outs() << " cmdsize " << sub.cmdsize;
4927 if (sub.cmdsize < sizeof(struct MachO::sub_client_command))
4928 outs() << " Incorrect size\n";
4929 else
4930 outs() << "\n";
4931 if (sub.client < sub.cmdsize) {
4932 const char *P = Ptr + sub.client;
4933 outs() << " client " << P << " (offset " << sub.client << ")\n";
4934 } else {
4935 outs() << " client ?(bad offset " << sub.client << ")\n";
4936 }
4937}
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00004938
Kevin Enderby52e4ce42014-12-19 22:25:22 +00004939static void PrintRoutinesCommand(MachO::routines_command r) {
4940 outs() << " cmd LC_ROUTINES\n";
4941 outs() << " cmdsize " << r.cmdsize;
4942 if (r.cmdsize != sizeof(struct MachO::routines_command))
4943 outs() << " Incorrect size\n";
4944 else
4945 outs() << "\n";
4946 outs() << " init_address " << format("0x%08" PRIx32, r.init_address) << "\n";
4947 outs() << " init_module " << r.init_module << "\n";
4948 outs() << " reserved1 " << r.reserved1 << "\n";
4949 outs() << " reserved2 " << r.reserved2 << "\n";
4950 outs() << " reserved3 " << r.reserved3 << "\n";
4951 outs() << " reserved4 " << r.reserved4 << "\n";
4952 outs() << " reserved5 " << r.reserved5 << "\n";
4953 outs() << " reserved6 " << r.reserved6 << "\n";
4954}
4955
4956static void PrintRoutinesCommand64(MachO::routines_command_64 r) {
4957 outs() << " cmd LC_ROUTINES_64\n";
4958 outs() << " cmdsize " << r.cmdsize;
4959 if (r.cmdsize != sizeof(struct MachO::routines_command_64))
4960 outs() << " Incorrect size\n";
4961 else
4962 outs() << "\n";
4963 outs() << " init_address " << format("0x%016" PRIx64, r.init_address) << "\n";
4964 outs() << " init_module " << r.init_module << "\n";
4965 outs() << " reserved1 " << r.reserved1 << "\n";
4966 outs() << " reserved2 " << r.reserved2 << "\n";
4967 outs() << " reserved3 " << r.reserved3 << "\n";
4968 outs() << " reserved4 " << r.reserved4 << "\n";
4969 outs() << " reserved5 " << r.reserved5 << "\n";
4970 outs() << " reserved6 " << r.reserved6 << "\n";
4971}
4972
Kevin Enderby48ef5342014-12-23 22:56:39 +00004973static void Print_x86_thread_state64_t(MachO::x86_thread_state64_t &cpu64) {
4974 outs() << " rax " << format("0x%016" PRIx64, cpu64.rax);
4975 outs() << " rbx " << format("0x%016" PRIx64, cpu64.rbx);
4976 outs() << " rcx " << format("0x%016" PRIx64, cpu64.rcx) << "\n";
4977 outs() << " rdx " << format("0x%016" PRIx64, cpu64.rdx);
4978 outs() << " rdi " << format("0x%016" PRIx64, cpu64.rdi);
4979 outs() << " rsi " << format("0x%016" PRIx64, cpu64.rsi) << "\n";
4980 outs() << " rbp " << format("0x%016" PRIx64, cpu64.rbp);
4981 outs() << " rsp " << format("0x%016" PRIx64, cpu64.rsp);
4982 outs() << " r8 " << format("0x%016" PRIx64, cpu64.r8) << "\n";
4983 outs() << " r9 " << format("0x%016" PRIx64, cpu64.r9);
4984 outs() << " r10 " << format("0x%016" PRIx64, cpu64.r10);
4985 outs() << " r11 " << format("0x%016" PRIx64, cpu64.r11) << "\n";
4986 outs() << " r12 " << format("0x%016" PRIx64, cpu64.r12);
4987 outs() << " r13 " << format("0x%016" PRIx64, cpu64.r13);
4988 outs() << " r14 " << format("0x%016" PRIx64, cpu64.r14) << "\n";
4989 outs() << " r15 " << format("0x%016" PRIx64, cpu64.r15);
4990 outs() << " rip " << format("0x%016" PRIx64, cpu64.rip) << "\n";
4991 outs() << "rflags " << format("0x%016" PRIx64, cpu64.rflags);
4992 outs() << " cs " << format("0x%016" PRIx64, cpu64.cs);
4993 outs() << " fs " << format("0x%016" PRIx64, cpu64.fs) << "\n";
4994 outs() << " gs " << format("0x%016" PRIx64, cpu64.gs) << "\n";
4995}
4996
Kevin Enderby227df342014-12-23 23:43:59 +00004997static void Print_mmst_reg(MachO::mmst_reg_t &r) {
Kevin Enderby48ef5342014-12-23 22:56:39 +00004998 uint32_t f;
4999 outs() << "\t mmst_reg ";
5000 for (f = 0; f < 10; f++)
5001 outs() << format("%02" PRIx32, (r.mmst_reg[f] & 0xff)) << " ";
5002 outs() << "\n";
5003 outs() << "\t mmst_rsrv ";
5004 for (f = 0; f < 6; f++)
5005 outs() << format("%02" PRIx32, (r.mmst_rsrv[f] & 0xff)) << " ";
5006 outs() << "\n";
5007}
5008
Kevin Enderbyaefb0032014-12-24 00:16:51 +00005009static void Print_xmm_reg(MachO::xmm_reg_t &r) {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005010 uint32_t f;
5011 outs() << "\t xmm_reg ";
5012 for (f = 0; f < 16; f++)
5013 outs() << format("%02" PRIx32, (r.xmm_reg[f] & 0xff)) << " ";
5014 outs() << "\n";
5015}
5016
5017static void Print_x86_float_state_t(MachO::x86_float_state64_t &fpu) {
5018 outs() << "\t fpu_reserved[0] " << fpu.fpu_reserved[0];
5019 outs() << " fpu_reserved[1] " << fpu.fpu_reserved[1] << "\n";
5020 outs() << "\t control: invalid " << fpu.fpu_fcw.invalid;
5021 outs() << " denorm " << fpu.fpu_fcw.denorm;
5022 outs() << " zdiv " << fpu.fpu_fcw.zdiv;
5023 outs() << " ovrfl " << fpu.fpu_fcw.ovrfl;
5024 outs() << " undfl " << fpu.fpu_fcw.undfl;
5025 outs() << " precis " << fpu.fpu_fcw.precis << "\n";
5026 outs() << "\t\t pc ";
5027 if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_24B)
5028 outs() << "FP_PREC_24B ";
5029 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_53B)
5030 outs() << "FP_PREC_53B ";
5031 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_64B)
5032 outs() << "FP_PREC_64B ";
5033 else
5034 outs() << fpu.fpu_fcw.pc << " ";
5035 outs() << "rc ";
5036 if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_NEAR)
5037 outs() << "FP_RND_NEAR ";
5038 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_DOWN)
5039 outs() << "FP_RND_DOWN ";
5040 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_UP)
5041 outs() << "FP_RND_UP ";
5042 else if (fpu.fpu_fcw.rc == MachO::x86_FP_CHOP)
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005043 outs() << "FP_CHOP ";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005044 outs() << "\n";
5045 outs() << "\t status: invalid " << fpu.fpu_fsw.invalid;
5046 outs() << " denorm " << fpu.fpu_fsw.denorm;
5047 outs() << " zdiv " << fpu.fpu_fsw.zdiv;
5048 outs() << " ovrfl " << fpu.fpu_fsw.ovrfl;
5049 outs() << " undfl " << fpu.fpu_fsw.undfl;
5050 outs() << " precis " << fpu.fpu_fsw.precis;
5051 outs() << " stkflt " << fpu.fpu_fsw.stkflt << "\n";
5052 outs() << "\t errsumm " << fpu.fpu_fsw.errsumm;
5053 outs() << " c0 " << fpu.fpu_fsw.c0;
5054 outs() << " c1 " << fpu.fpu_fsw.c1;
5055 outs() << " c2 " << fpu.fpu_fsw.c2;
5056 outs() << " tos " << fpu.fpu_fsw.tos;
5057 outs() << " c3 " << fpu.fpu_fsw.c3;
5058 outs() << " busy " << fpu.fpu_fsw.busy << "\n";
5059 outs() << "\t fpu_ftw " << format("0x%02" PRIx32, fpu.fpu_ftw);
5060 outs() << " fpu_rsrv1 " << format("0x%02" PRIx32, fpu.fpu_rsrv1);
5061 outs() << " fpu_fop " << format("0x%04" PRIx32, fpu.fpu_fop);
5062 outs() << " fpu_ip " << format("0x%08" PRIx32, fpu.fpu_ip) << "\n";
5063 outs() << "\t fpu_cs " << format("0x%04" PRIx32, fpu.fpu_cs);
5064 outs() << " fpu_rsrv2 " << format("0x%04" PRIx32, fpu.fpu_rsrv2);
5065 outs() << " fpu_dp " << format("0x%08" PRIx32, fpu.fpu_dp);
5066 outs() << " fpu_ds " << format("0x%04" PRIx32, fpu.fpu_ds) << "\n";
5067 outs() << "\t fpu_rsrv3 " << format("0x%04" PRIx32, fpu.fpu_rsrv3);
5068 outs() << " fpu_mxcsr " << format("0x%08" PRIx32, fpu.fpu_mxcsr);
5069 outs() << " fpu_mxcsrmask " << format("0x%08" PRIx32, fpu.fpu_mxcsrmask);
5070 outs() << "\n";
5071 outs() << "\t fpu_stmm0:\n";
5072 Print_mmst_reg(fpu.fpu_stmm0);
5073 outs() << "\t fpu_stmm1:\n";
5074 Print_mmst_reg(fpu.fpu_stmm1);
5075 outs() << "\t fpu_stmm2:\n";
5076 Print_mmst_reg(fpu.fpu_stmm2);
5077 outs() << "\t fpu_stmm3:\n";
5078 Print_mmst_reg(fpu.fpu_stmm3);
5079 outs() << "\t fpu_stmm4:\n";
5080 Print_mmst_reg(fpu.fpu_stmm4);
5081 outs() << "\t fpu_stmm5:\n";
5082 Print_mmst_reg(fpu.fpu_stmm5);
5083 outs() << "\t fpu_stmm6:\n";
5084 Print_mmst_reg(fpu.fpu_stmm6);
5085 outs() << "\t fpu_stmm7:\n";
5086 Print_mmst_reg(fpu.fpu_stmm7);
5087 outs() << "\t fpu_xmm0:\n";
5088 Print_xmm_reg(fpu.fpu_xmm0);
5089 outs() << "\t fpu_xmm1:\n";
5090 Print_xmm_reg(fpu.fpu_xmm1);
5091 outs() << "\t fpu_xmm2:\n";
5092 Print_xmm_reg(fpu.fpu_xmm2);
5093 outs() << "\t fpu_xmm3:\n";
5094 Print_xmm_reg(fpu.fpu_xmm3);
5095 outs() << "\t fpu_xmm4:\n";
5096 Print_xmm_reg(fpu.fpu_xmm4);
5097 outs() << "\t fpu_xmm5:\n";
5098 Print_xmm_reg(fpu.fpu_xmm5);
5099 outs() << "\t fpu_xmm6:\n";
5100 Print_xmm_reg(fpu.fpu_xmm6);
5101 outs() << "\t fpu_xmm7:\n";
5102 Print_xmm_reg(fpu.fpu_xmm7);
5103 outs() << "\t fpu_xmm8:\n";
5104 Print_xmm_reg(fpu.fpu_xmm8);
5105 outs() << "\t fpu_xmm9:\n";
5106 Print_xmm_reg(fpu.fpu_xmm9);
5107 outs() << "\t fpu_xmm10:\n";
5108 Print_xmm_reg(fpu.fpu_xmm10);
5109 outs() << "\t fpu_xmm11:\n";
5110 Print_xmm_reg(fpu.fpu_xmm11);
5111 outs() << "\t fpu_xmm12:\n";
5112 Print_xmm_reg(fpu.fpu_xmm12);
5113 outs() << "\t fpu_xmm13:\n";
5114 Print_xmm_reg(fpu.fpu_xmm13);
5115 outs() << "\t fpu_xmm14:\n";
5116 Print_xmm_reg(fpu.fpu_xmm14);
5117 outs() << "\t fpu_xmm15:\n";
5118 Print_xmm_reg(fpu.fpu_xmm15);
5119 outs() << "\t fpu_rsrv4:\n";
5120 for (uint32_t f = 0; f < 6; f++) {
5121 outs() << "\t ";
5122 for (uint32_t g = 0; g < 16; g++)
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005123 outs() << format("%02" PRIx32, fpu.fpu_rsrv4[f * g]) << " ";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005124 outs() << "\n";
5125 }
5126 outs() << "\t fpu_reserved1 " << format("0x%08" PRIx32, fpu.fpu_reserved1);
5127 outs() << "\n";
5128}
5129
5130static void Print_x86_exception_state_t(MachO::x86_exception_state64_t &exc64) {
5131 outs() << "\t trapno " << format("0x%08" PRIx32, exc64.trapno);
5132 outs() << " err " << format("0x%08" PRIx32, exc64.err);
5133 outs() << " faultvaddr " << format("0x%016" PRIx64, exc64.faultvaddr) << "\n";
5134}
5135
5136static void PrintThreadCommand(MachO::thread_command t, const char *Ptr,
5137 bool isLittleEndian, uint32_t cputype) {
5138 if (t.cmd == MachO::LC_THREAD)
5139 outs() << " cmd LC_THREAD\n";
5140 else if (t.cmd == MachO::LC_UNIXTHREAD)
5141 outs() << " cmd LC_UNIXTHREAD\n";
5142 else
5143 outs() << " cmd " << t.cmd << " (unknown)\n";
5144 outs() << " cmdsize " << t.cmdsize;
5145 if (t.cmdsize < sizeof(struct MachO::thread_command) + 2 * sizeof(uint32_t))
5146 outs() << " Incorrect size\n";
5147 else
5148 outs() << "\n";
5149
5150 const char *begin = Ptr + sizeof(struct MachO::thread_command);
5151 const char *end = Ptr + t.cmdsize;
5152 uint32_t flavor, count, left;
5153 if (cputype == MachO::CPU_TYPE_X86_64) {
5154 while (begin < end) {
5155 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5156 memcpy((char *)&flavor, begin, sizeof(uint32_t));
5157 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005158 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005159 flavor = 0;
5160 begin = end;
5161 }
5162 if (isLittleEndian != sys::IsLittleEndianHost)
5163 sys::swapByteOrder(flavor);
5164 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5165 memcpy((char *)&count, begin, sizeof(uint32_t));
5166 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005167 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005168 count = 0;
5169 begin = end;
5170 }
5171 if (isLittleEndian != sys::IsLittleEndianHost)
5172 sys::swapByteOrder(count);
5173 if (flavor == MachO::x86_THREAD_STATE64) {
5174 outs() << " flavor x86_THREAD_STATE64\n";
5175 if (count == MachO::x86_THREAD_STATE64_COUNT)
5176 outs() << " count x86_THREAD_STATE64_COUNT\n";
5177 else
5178 outs() << " count " << count
5179 << " (not x86_THREAD_STATE64_COUNT)\n";
5180 MachO::x86_thread_state64_t cpu64;
5181 left = end - begin;
5182 if (left >= sizeof(MachO::x86_thread_state64_t)) {
5183 memcpy(&cpu64, begin, sizeof(MachO::x86_thread_state64_t));
5184 begin += sizeof(MachO::x86_thread_state64_t);
5185 } else {
5186 memset(&cpu64, '\0', sizeof(MachO::x86_thread_state64_t));
5187 memcpy(&cpu64, begin, left);
5188 begin += left;
5189 }
5190 if (isLittleEndian != sys::IsLittleEndianHost)
5191 swapStruct(cpu64);
5192 Print_x86_thread_state64_t(cpu64);
5193 } else if (flavor == MachO::x86_THREAD_STATE) {
5194 outs() << " flavor x86_THREAD_STATE\n";
5195 if (count == MachO::x86_THREAD_STATE_COUNT)
5196 outs() << " count x86_THREAD_STATE_COUNT\n";
5197 else
5198 outs() << " count " << count
5199 << " (not x86_THREAD_STATE_COUNT)\n";
5200 struct MachO::x86_thread_state_t ts;
5201 left = end - begin;
5202 if (left >= sizeof(MachO::x86_thread_state_t)) {
5203 memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t));
5204 begin += sizeof(MachO::x86_thread_state_t);
5205 } else {
5206 memset(&ts, '\0', sizeof(MachO::x86_thread_state_t));
5207 memcpy(&ts, begin, left);
5208 begin += left;
5209 }
5210 if (isLittleEndian != sys::IsLittleEndianHost)
5211 swapStruct(ts);
5212 if (ts.tsh.flavor == MachO::x86_THREAD_STATE64) {
5213 outs() << "\t tsh.flavor x86_THREAD_STATE64 ";
5214 if (ts.tsh.count == MachO::x86_THREAD_STATE64_COUNT)
5215 outs() << "tsh.count x86_THREAD_STATE64_COUNT\n";
5216 else
5217 outs() << "tsh.count " << ts.tsh.count
5218 << " (not x86_THREAD_STATE64_COUNT\n";
5219 Print_x86_thread_state64_t(ts.uts.ts64);
5220 } else {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005221 outs() << "\t tsh.flavor " << ts.tsh.flavor << " tsh.count "
5222 << ts.tsh.count << "\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005223 }
5224 } else if (flavor == MachO::x86_FLOAT_STATE) {
5225 outs() << " flavor x86_FLOAT_STATE\n";
5226 if (count == MachO::x86_FLOAT_STATE_COUNT)
5227 outs() << " count x86_FLOAT_STATE_COUNT\n";
5228 else
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005229 outs() << " count " << count << " (not x86_FLOAT_STATE_COUNT)\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005230 struct MachO::x86_float_state_t fs;
5231 left = end - begin;
5232 if (left >= sizeof(MachO::x86_float_state_t)) {
5233 memcpy(&fs, begin, sizeof(MachO::x86_float_state_t));
5234 begin += sizeof(MachO::x86_float_state_t);
5235 } else {
5236 memset(&fs, '\0', sizeof(MachO::x86_float_state_t));
5237 memcpy(&fs, begin, left);
5238 begin += left;
5239 }
5240 if (isLittleEndian != sys::IsLittleEndianHost)
5241 swapStruct(fs);
5242 if (fs.fsh.flavor == MachO::x86_FLOAT_STATE64) {
5243 outs() << "\t fsh.flavor x86_FLOAT_STATE64 ";
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005244 if (fs.fsh.count == MachO::x86_FLOAT_STATE64_COUNT)
Kevin Enderby48ef5342014-12-23 22:56:39 +00005245 outs() << "fsh.count x86_FLOAT_STATE64_COUNT\n";
5246 else
5247 outs() << "fsh.count " << fs.fsh.count
5248 << " (not x86_FLOAT_STATE64_COUNT\n";
5249 Print_x86_float_state_t(fs.ufs.fs64);
5250 } else {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005251 outs() << "\t fsh.flavor " << fs.fsh.flavor << " fsh.count "
5252 << fs.fsh.count << "\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005253 }
5254 } else if (flavor == MachO::x86_EXCEPTION_STATE) {
5255 outs() << " flavor x86_EXCEPTION_STATE\n";
5256 if (count == MachO::x86_EXCEPTION_STATE_COUNT)
5257 outs() << " count x86_EXCEPTION_STATE_COUNT\n";
5258 else
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005259 outs() << " count " << count
Kevin Enderby48ef5342014-12-23 22:56:39 +00005260 << " (not x86_EXCEPTION_STATE_COUNT)\n";
5261 struct MachO::x86_exception_state_t es;
5262 left = end - begin;
5263 if (left >= sizeof(MachO::x86_exception_state_t)) {
5264 memcpy(&es, begin, sizeof(MachO::x86_exception_state_t));
5265 begin += sizeof(MachO::x86_exception_state_t);
5266 } else {
5267 memset(&es, '\0', sizeof(MachO::x86_exception_state_t));
5268 memcpy(&es, begin, left);
5269 begin += left;
5270 }
5271 if (isLittleEndian != sys::IsLittleEndianHost)
5272 swapStruct(es);
5273 if (es.esh.flavor == MachO::x86_EXCEPTION_STATE64) {
5274 outs() << "\t esh.flavor x86_EXCEPTION_STATE64\n";
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005275 if (es.esh.count == MachO::x86_EXCEPTION_STATE64_COUNT)
Kevin Enderby48ef5342014-12-23 22:56:39 +00005276 outs() << "\t esh.count x86_EXCEPTION_STATE64_COUNT\n";
5277 else
5278 outs() << "\t esh.count " << es.esh.count
5279 << " (not x86_EXCEPTION_STATE64_COUNT\n";
5280 Print_x86_exception_state_t(es.ues.es64);
5281 } else {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005282 outs() << "\t esh.flavor " << es.esh.flavor << " esh.count "
5283 << es.esh.count << "\n";
Kevin Enderby48ef5342014-12-23 22:56:39 +00005284 }
5285 } else {
5286 outs() << " flavor " << flavor << " (unknown)\n";
5287 outs() << " count " << count << "\n";
5288 outs() << " state (unknown)\n";
5289 begin += count * sizeof(uint32_t);
5290 }
5291 }
5292 } else {
5293 while (begin < end) {
5294 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5295 memcpy((char *)&flavor, begin, sizeof(uint32_t));
5296 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005297 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005298 flavor = 0;
5299 begin = end;
5300 }
5301 if (isLittleEndian != sys::IsLittleEndianHost)
5302 sys::swapByteOrder(flavor);
5303 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) {
5304 memcpy((char *)&count, begin, sizeof(uint32_t));
5305 begin += sizeof(uint32_t);
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005306 } else {
Kevin Enderby48ef5342014-12-23 22:56:39 +00005307 count = 0;
5308 begin = end;
5309 }
5310 if (isLittleEndian != sys::IsLittleEndianHost)
5311 sys::swapByteOrder(count);
5312 outs() << " flavor " << flavor << "\n";
5313 outs() << " count " << count << "\n";
5314 outs() << " state (Unknown cputype/cpusubtype)\n";
5315 begin += count * sizeof(uint32_t);
5316 }
5317 }
5318}
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005319
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005320static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) {
5321 if (dl.cmd == MachO::LC_ID_DYLIB)
5322 outs() << " cmd LC_ID_DYLIB\n";
5323 else if (dl.cmd == MachO::LC_LOAD_DYLIB)
5324 outs() << " cmd LC_LOAD_DYLIB\n";
5325 else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB)
5326 outs() << " cmd LC_LOAD_WEAK_DYLIB\n";
5327 else if (dl.cmd == MachO::LC_REEXPORT_DYLIB)
5328 outs() << " cmd LC_REEXPORT_DYLIB\n";
5329 else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB)
5330 outs() << " cmd LC_LAZY_LOAD_DYLIB\n";
5331 else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
5332 outs() << " cmd LC_LOAD_UPWARD_DYLIB\n";
5333 else
5334 outs() << " cmd " << dl.cmd << " (unknown)\n";
5335 outs() << " cmdsize " << dl.cmdsize;
5336 if (dl.cmdsize < sizeof(struct MachO::dylib_command))
5337 outs() << " Incorrect size\n";
5338 else
5339 outs() << "\n";
5340 if (dl.dylib.name < dl.cmdsize) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005341 const char *P = (const char *)(Ptr) + dl.dylib.name;
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005342 outs() << " name " << P << " (offset " << dl.dylib.name << ")\n";
5343 } else {
5344 outs() << " name ?(bad offset " << dl.dylib.name << ")\n";
5345 }
5346 outs() << " time stamp " << dl.dylib.timestamp << " ";
5347 time_t t = dl.dylib.timestamp;
5348 outs() << ctime(&t);
5349 outs() << " current version ";
5350 if (dl.dylib.current_version == 0xffffffff)
5351 outs() << "n/a\n";
5352 else
5353 outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "."
5354 << ((dl.dylib.current_version >> 8) & 0xff) << "."
5355 << (dl.dylib.current_version & 0xff) << "\n";
5356 outs() << "compatibility version ";
5357 if (dl.dylib.compatibility_version == 0xffffffff)
5358 outs() << "n/a\n";
5359 else
5360 outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
5361 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
5362 << (dl.dylib.compatibility_version & 0xff) << "\n";
5363}
5364
5365static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld,
5366 uint32_t object_size) {
5367 if (ld.cmd == MachO::LC_CODE_SIGNATURE)
5368 outs() << " cmd LC_FUNCTION_STARTS\n";
5369 else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO)
5370 outs() << " cmd LC_SEGMENT_SPLIT_INFO\n";
5371 else if (ld.cmd == MachO::LC_FUNCTION_STARTS)
5372 outs() << " cmd LC_FUNCTION_STARTS\n";
5373 else if (ld.cmd == MachO::LC_DATA_IN_CODE)
5374 outs() << " cmd LC_DATA_IN_CODE\n";
5375 else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS)
5376 outs() << " cmd LC_DYLIB_CODE_SIGN_DRS\n";
5377 else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT)
5378 outs() << " cmd LC_LINKER_OPTIMIZATION_HINT\n";
5379 else
5380 outs() << " cmd " << ld.cmd << " (?)\n";
5381 outs() << " cmdsize " << ld.cmdsize;
5382 if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command))
5383 outs() << " Incorrect size\n";
5384 else
5385 outs() << "\n";
5386 outs() << " dataoff " << ld.dataoff;
5387 if (ld.dataoff > object_size)
5388 outs() << " (past end of file)\n";
5389 else
5390 outs() << "\n";
5391 outs() << " datasize " << ld.datasize;
5392 uint64_t big_size = ld.dataoff;
5393 big_size += ld.datasize;
5394 if (big_size > object_size)
5395 outs() << " (past end of file)\n";
5396 else
5397 outs() << "\n";
5398}
5399
Kevin Enderby956366c2014-08-29 22:30:52 +00005400static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t ncmds,
5401 uint32_t filetype, uint32_t cputype,
5402 bool verbose) {
Filipe Cabecinhase71bd0c2015-01-06 17:08:26 +00005403 if (ncmds == 0)
5404 return;
Kevin Enderby956366c2014-08-29 22:30:52 +00005405 StringRef Buf = Obj->getData();
5406 MachOObjectFile::LoadCommandInfo Command = Obj->getFirstLoadCommandInfo();
5407 for (unsigned i = 0;; ++i) {
5408 outs() << "Load command " << i << "\n";
5409 if (Command.C.cmd == MachO::LC_SEGMENT) {
5410 MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command);
5411 const char *sg_segname = SLC.segname;
5412 PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr,
5413 SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot,
5414 SLC.initprot, SLC.nsects, SLC.flags, Buf.size(),
5415 verbose);
5416 for (unsigned j = 0; j < SLC.nsects; j++) {
Kevin Enderbyc9713382014-12-16 01:14:45 +00005417 MachO::section S = Obj->getSection(Command, j);
Kevin Enderby956366c2014-08-29 22:30:52 +00005418 PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align,
5419 S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2,
5420 SLC.cmd, sg_segname, filetype, Buf.size(), verbose);
5421 }
5422 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
5423 MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command);
5424 const char *sg_segname = SLC_64.segname;
5425 PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname,
5426 SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff,
5427 SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot,
5428 SLC_64.nsects, SLC_64.flags, Buf.size(), verbose);
5429 for (unsigned j = 0; j < SLC_64.nsects; j++) {
5430 MachO::section_64 S_64 = Obj->getSection64(Command, j);
5431 PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size,
5432 S_64.offset, S_64.align, S_64.reloff, S_64.nreloc,
5433 S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd,
5434 sg_segname, filetype, Buf.size(), verbose);
5435 }
5436 } else if (Command.C.cmd == MachO::LC_SYMTAB) {
5437 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
David Majnemer73cc6ff2014-11-13 19:48:56 +00005438 PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size());
Kevin Enderby956366c2014-08-29 22:30:52 +00005439 } else if (Command.C.cmd == MachO::LC_DYSYMTAB) {
5440 MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand();
5441 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
David Majnemer73cc6ff2014-11-13 19:48:56 +00005442 PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(),
5443 Obj->is64Bit());
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005444 } else if (Command.C.cmd == MachO::LC_DYLD_INFO ||
5445 Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) {
5446 MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command);
5447 PrintDyldInfoLoadCommand(DyldInfo, Buf.size());
5448 } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER ||
5449 Command.C.cmd == MachO::LC_ID_DYLINKER ||
5450 Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) {
5451 MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command);
5452 PrintDyldLoadCommand(Dyld, Command.Ptr);
5453 } else if (Command.C.cmd == MachO::LC_UUID) {
5454 MachO::uuid_command Uuid = Obj->getUuidCommand(Command);
5455 PrintUuidLoadCommand(Uuid);
Jean-Daniel Dupas00cc1f52014-12-04 07:37:02 +00005456 } else if (Command.C.cmd == MachO::LC_RPATH) {
5457 MachO::rpath_command Rpath = Obj->getRpathCommand(Command);
5458 PrintRpathLoadCommand(Rpath, Command.Ptr);
Kevin Enderby1ff0ecc2014-12-16 21:48:27 +00005459 } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX ||
5460 Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005461 MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command);
5462 PrintVersionMinLoadCommand(Vd);
5463 } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) {
5464 MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command);
5465 PrintSourceVersionCommand(Sd);
5466 } else if (Command.C.cmd == MachO::LC_MAIN) {
5467 MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command);
5468 PrintEntryPointCommand(Ep);
Kevin Enderby0804f4672014-12-16 23:25:52 +00005469 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO) {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005470 MachO::encryption_info_command Ei =
5471 Obj->getEncryptionInfoCommand(Command);
Kevin Enderby0804f4672014-12-16 23:25:52 +00005472 PrintEncryptionInfoCommand(Ei, Buf.size());
Kevin Enderby57538292014-12-17 01:01:30 +00005473 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO_64) {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005474 MachO::encryption_info_command_64 Ei =
5475 Obj->getEncryptionInfoCommand64(Command);
Kevin Enderby57538292014-12-17 01:01:30 +00005476 PrintEncryptionInfoCommand64(Ei, Buf.size());
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00005477 } else if (Command.C.cmd == MachO::LC_LINKER_OPTION) {
Kevin Enderby66d51fc2015-01-08 00:25:24 +00005478 MachO::linker_option_command Lo =
5479 Obj->getLinkerOptionLoadCommand(Command);
Kevin Enderbyd0b6b7f2014-12-18 00:53:40 +00005480 PrintLinkerOptionCommand(Lo, Command.Ptr);
Kevin Enderbyb4b79312014-12-18 19:24:35 +00005481 } else if (Command.C.cmd == MachO::LC_SUB_FRAMEWORK) {
5482 MachO::sub_framework_command Sf = Obj->getSubFrameworkCommand(Command);
5483 PrintSubFrameworkCommand(Sf, Command.Ptr);
Kevin Enderbya2bd8d92014-12-18 23:13:26 +00005484 } else if (Command.C.cmd == MachO::LC_SUB_UMBRELLA) {
5485 MachO::sub_umbrella_command Sf = Obj->getSubUmbrellaCommand(Command);
5486 PrintSubUmbrellaCommand(Sf, Command.Ptr);
Kevin Enderby36c8d3a2014-12-19 19:48:16 +00005487 } else if (Command.C.cmd == MachO::LC_SUB_LIBRARY) {
5488 MachO::sub_library_command Sl = Obj->getSubLibraryCommand(Command);
5489 PrintSubLibraryCommand(Sl, Command.Ptr);
Kevin Enderby186eac32014-12-19 21:06:24 +00005490 } else if (Command.C.cmd == MachO::LC_SUB_CLIENT) {
5491 MachO::sub_client_command Sc = Obj->getSubClientCommand(Command);
5492 PrintSubClientCommand(Sc, Command.Ptr);
Kevin Enderby52e4ce42014-12-19 22:25:22 +00005493 } else if (Command.C.cmd == MachO::LC_ROUTINES) {
5494 MachO::routines_command Rc = Obj->getRoutinesCommand(Command);
5495 PrintRoutinesCommand(Rc);
5496 } else if (Command.C.cmd == MachO::LC_ROUTINES_64) {
5497 MachO::routines_command_64 Rc = Obj->getRoutinesCommand64(Command);
5498 PrintRoutinesCommand64(Rc);
Kevin Enderby48ef5342014-12-23 22:56:39 +00005499 } else if (Command.C.cmd == MachO::LC_THREAD ||
5500 Command.C.cmd == MachO::LC_UNIXTHREAD) {
5501 MachO::thread_command Tc = Obj->getThreadCommand(Command);
5502 PrintThreadCommand(Tc, Command.Ptr, Obj->isLittleEndian(), cputype);
Nick Kledzik15558912014-10-16 18:58:20 +00005503 } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB ||
5504 Command.C.cmd == MachO::LC_ID_DYLIB ||
5505 Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
5506 Command.C.cmd == MachO::LC_REEXPORT_DYLIB ||
5507 Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
5508 Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00005509 MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command);
5510 PrintDylibCommand(Dl, Command.Ptr);
5511 } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE ||
5512 Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO ||
5513 Command.C.cmd == MachO::LC_FUNCTION_STARTS ||
5514 Command.C.cmd == MachO::LC_DATA_IN_CODE ||
5515 Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS ||
5516 Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) {
5517 MachO::linkedit_data_command Ld =
5518 Obj->getLinkeditDataLoadCommand(Command);
5519 PrintLinkEditDataCommand(Ld, Buf.size());
Kevin Enderby956366c2014-08-29 22:30:52 +00005520 } else {
5521 outs() << " cmd ?(" << format("0x%08" PRIx32, Command.C.cmd)
5522 << ")\n";
5523 outs() << " cmdsize " << Command.C.cmdsize << "\n";
5524 // TODO: get and print the raw bytes of the load command.
5525 }
5526 // TODO: print all the other kinds of load commands.
5527 if (i == ncmds - 1)
5528 break;
5529 else
5530 Command = Obj->getNextLoadCommandInfo(Command);
5531 }
5532}
5533
5534static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds,
5535 uint32_t &filetype, uint32_t &cputype,
5536 bool verbose) {
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005537 if (Obj->is64Bit()) {
5538 MachO::mach_header_64 H_64;
5539 H_64 = Obj->getHeader64();
5540 PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype,
5541 H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose);
Kevin Enderby956366c2014-08-29 22:30:52 +00005542 ncmds = H_64.ncmds;
5543 filetype = H_64.filetype;
5544 cputype = H_64.cputype;
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005545 } else {
5546 MachO::mach_header H;
5547 H = Obj->getHeader();
5548 PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds,
5549 H.sizeofcmds, H.flags, verbose);
Kevin Enderby956366c2014-08-29 22:30:52 +00005550 ncmds = H.ncmds;
5551 filetype = H.filetype;
5552 cputype = H.cputype;
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005553 }
5554}
5555
5556void llvm::printMachOFileHeader(const object::ObjectFile *Obj) {
5557 const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
Kevin Enderby956366c2014-08-29 22:30:52 +00005558 uint32_t ncmds = 0;
5559 uint32_t filetype = 0;
5560 uint32_t cputype = 0;
Kevin Enderbyf0640752015-03-13 17:56:32 +00005561 getAndPrintMachHeader(file, ncmds, filetype, cputype, !NonVerbose);
5562 PrintLoadCommands(file, ncmds, filetype, cputype, !NonVerbose);
Kevin Enderbyb76d3862014-08-22 20:35:18 +00005563}
Nick Kledzikd04bc352014-08-30 00:20:14 +00005564
5565//===----------------------------------------------------------------------===//
5566// export trie dumping
5567//===----------------------------------------------------------------------===//
5568
5569void llvm::printMachOExportsTrie(const object::MachOObjectFile *Obj) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005570 for (const llvm::object::ExportEntry &Entry : Obj->exports()) {
5571 uint64_t Flags = Entry.flags();
Nick Kledzikd04bc352014-08-30 00:20:14 +00005572 bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT);
5573 bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION);
5574 bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
5575 MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL);
5576 bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
5577 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE);
5578 bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER);
5579 if (ReExport)
5580 outs() << "[re-export] ";
5581 else
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005582 outs() << format("0x%08llX ",
5583 Entry.address()); // FIXME:add in base address
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005584 outs() << Entry.name();
Nick Kledzikd04bc352014-08-30 00:20:14 +00005585 if (WeakDef || ThreadLocal || Resolver || Abs) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005586 bool NeedsComma = false;
Nick Kledzik1d1ac4b2014-09-03 01:12:52 +00005587 outs() << " [";
Nick Kledzikd04bc352014-08-30 00:20:14 +00005588 if (WeakDef) {
5589 outs() << "weak_def";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005590 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005591 }
5592 if (ThreadLocal) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005593 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00005594 outs() << ", ";
5595 outs() << "per-thread";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005596 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005597 }
5598 if (Abs) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005599 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00005600 outs() << ", ";
5601 outs() << "absolute";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005602 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005603 }
5604 if (Resolver) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005605 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00005606 outs() << ", ";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005607 outs() << format("resolver=0x%08llX", Entry.other());
5608 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00005609 }
5610 outs() << "]";
5611 }
5612 if (ReExport) {
5613 StringRef DylibName = "unknown";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005614 int Ordinal = Entry.other() - 1;
5615 Obj->getLibraryShortNameByIndex(Ordinal, DylibName);
5616 if (Entry.otherName().empty())
Nick Kledzikd04bc352014-08-30 00:20:14 +00005617 outs() << " (from " << DylibName << ")";
5618 else
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00005619 outs() << " (" << Entry.otherName() << " from " << DylibName << ")";
Nick Kledzikd04bc352014-08-30 00:20:14 +00005620 }
5621 outs() << "\n";
5622 }
5623}
Nick Kledzikac431442014-09-12 21:34:15 +00005624
Nick Kledzikac431442014-09-12 21:34:15 +00005625//===----------------------------------------------------------------------===//
5626// rebase table dumping
5627//===----------------------------------------------------------------------===//
5628
5629namespace {
5630class SegInfo {
5631public:
5632 SegInfo(const object::MachOObjectFile *Obj);
5633
5634 StringRef segmentName(uint32_t SegIndex);
5635 StringRef sectionName(uint32_t SegIndex, uint64_t SegOffset);
5636 uint64_t address(uint32_t SegIndex, uint64_t SegOffset);
5637
5638private:
5639 struct SectionInfo {
5640 uint64_t Address;
5641 uint64_t Size;
5642 StringRef SectionName;
5643 StringRef SegmentName;
5644 uint64_t OffsetInSegment;
5645 uint64_t SegmentStartAddress;
5646 uint32_t SegmentIndex;
5647 };
5648 const SectionInfo &findSection(uint32_t SegIndex, uint64_t SegOffset);
5649 SmallVector<SectionInfo, 32> Sections;
5650};
5651}
5652
5653SegInfo::SegInfo(const object::MachOObjectFile *Obj) {
5654 // Build table of sections so segIndex/offset pairs can be translated.
Nick Kledzik56ebef42014-09-16 01:41:51 +00005655 uint32_t CurSegIndex = Obj->hasPageZeroSegment() ? 1 : 0;
Nick Kledzikac431442014-09-12 21:34:15 +00005656 StringRef CurSegName;
5657 uint64_t CurSegAddress;
5658 for (const SectionRef &Section : Obj->sections()) {
5659 SectionInfo Info;
5660 if (error(Section.getName(Info.SectionName)))
5661 return;
Rafael Espindola80291272014-10-08 15:28:58 +00005662 Info.Address = Section.getAddress();
5663 Info.Size = Section.getSize();
Nick Kledzikac431442014-09-12 21:34:15 +00005664 Info.SegmentName =
5665 Obj->getSectionFinalSegmentName(Section.getRawDataRefImpl());
5666 if (!Info.SegmentName.equals(CurSegName)) {
5667 ++CurSegIndex;
5668 CurSegName = Info.SegmentName;
5669 CurSegAddress = Info.Address;
5670 }
5671 Info.SegmentIndex = CurSegIndex - 1;
5672 Info.OffsetInSegment = Info.Address - CurSegAddress;
5673 Info.SegmentStartAddress = CurSegAddress;
5674 Sections.push_back(Info);
5675 }
5676}
5677
5678StringRef SegInfo::segmentName(uint32_t SegIndex) {
5679 for (const SectionInfo &SI : Sections) {
5680 if (SI.SegmentIndex == SegIndex)
5681 return SI.SegmentName;
5682 }
5683 llvm_unreachable("invalid segIndex");
5684}
5685
5686const SegInfo::SectionInfo &SegInfo::findSection(uint32_t SegIndex,
5687 uint64_t OffsetInSeg) {
5688 for (const SectionInfo &SI : Sections) {
5689 if (SI.SegmentIndex != SegIndex)
5690 continue;
5691 if (SI.OffsetInSegment > OffsetInSeg)
5692 continue;
5693 if (OffsetInSeg >= (SI.OffsetInSegment + SI.Size))
5694 continue;
5695 return SI;
5696 }
5697 llvm_unreachable("segIndex and offset not in any section");
5698}
5699
5700StringRef SegInfo::sectionName(uint32_t SegIndex, uint64_t OffsetInSeg) {
5701 return findSection(SegIndex, OffsetInSeg).SectionName;
5702}
5703
5704uint64_t SegInfo::address(uint32_t SegIndex, uint64_t OffsetInSeg) {
5705 const SectionInfo &SI = findSection(SegIndex, OffsetInSeg);
5706 return SI.SegmentStartAddress + OffsetInSeg;
5707}
5708
5709void llvm::printMachORebaseTable(const object::MachOObjectFile *Obj) {
5710 // Build table of sections so names can used in final output.
5711 SegInfo sectionTable(Obj);
5712
5713 outs() << "segment section address type\n";
5714 for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable()) {
5715 uint32_t SegIndex = Entry.segmentIndex();
5716 uint64_t OffsetInSeg = Entry.segmentOffset();
5717 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5718 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5719 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5720
5721 // Table lines look like: __DATA __nl_symbol_ptr 0x0000F00C pointer
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005722 outs() << format("%-8s %-18s 0x%08" PRIX64 " %s\n",
5723 SegmentName.str().c_str(), SectionName.str().c_str(),
5724 Address, Entry.typeName().str().c_str());
Nick Kledzikac431442014-09-12 21:34:15 +00005725 }
5726}
Nick Kledzik56ebef42014-09-16 01:41:51 +00005727
5728static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) {
5729 StringRef DylibName;
5730 switch (Ordinal) {
5731 case MachO::BIND_SPECIAL_DYLIB_SELF:
5732 return "this-image";
5733 case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE:
5734 return "main-executable";
5735 case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP:
5736 return "flat-namespace";
5737 default:
Nick Kledzikabd29872014-09-16 22:03:13 +00005738 if (Ordinal > 0) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005739 std::error_code EC =
5740 Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName);
Nick Kledzikabd29872014-09-16 22:03:13 +00005741 if (EC)
Nick Kledzik51d2c2b2014-10-14 23:29:38 +00005742 return "<<bad library ordinal>>";
Nick Kledzikabd29872014-09-16 22:03:13 +00005743 return DylibName;
5744 }
Nick Kledzik56ebef42014-09-16 01:41:51 +00005745 }
Nick Kledzikabd29872014-09-16 22:03:13 +00005746 return "<<unknown special ordinal>>";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005747}
5748
5749//===----------------------------------------------------------------------===//
5750// bind table dumping
5751//===----------------------------------------------------------------------===//
5752
5753void llvm::printMachOBindTable(const object::MachOObjectFile *Obj) {
5754 // Build table of sections so names can used in final output.
5755 SegInfo sectionTable(Obj);
5756
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005757 outs() << "segment section address type "
5758 "addend dylib symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005759 for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable()) {
5760 uint32_t SegIndex = Entry.segmentIndex();
5761 uint64_t OffsetInSeg = Entry.segmentOffset();
5762 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5763 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5764 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5765
5766 // Table lines look like:
5767 // __DATA __got 0x00012010 pointer 0 libSystem ___stack_chk_guard
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005768 StringRef Attr;
Nick Kledzik56ebef42014-09-16 01:41:51 +00005769 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT)
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005770 Attr = " (weak_import)";
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005771 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005772 << left_justify(SectionName, 18) << " "
5773 << format_hex(Address, 10, true) << " "
5774 << left_justify(Entry.typeName(), 8) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005775 << format_decimal(Entry.addend(), 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005776 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005777 << Entry.symbolName() << Attr << "\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005778 }
5779}
5780
5781//===----------------------------------------------------------------------===//
5782// lazy bind table dumping
5783//===----------------------------------------------------------------------===//
5784
5785void llvm::printMachOLazyBindTable(const object::MachOObjectFile *Obj) {
5786 // Build table of sections so names can used in final output.
5787 SegInfo sectionTable(Obj);
5788
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005789 outs() << "segment section address "
5790 "dylib symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005791 for (const llvm::object::MachOBindEntry &Entry : Obj->lazyBindTable()) {
5792 uint32_t SegIndex = Entry.segmentIndex();
5793 uint64_t OffsetInSeg = Entry.segmentOffset();
5794 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5795 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5796 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5797
5798 // Table lines look like:
5799 // __DATA __got 0x00012010 libSystem ___stack_chk_guard
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005800 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005801 << left_justify(SectionName, 18) << " "
5802 << format_hex(Address, 10, true) << " "
5803 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
Nick Kledzik56ebef42014-09-16 01:41:51 +00005804 << Entry.symbolName() << "\n";
5805 }
5806}
5807
Nick Kledzik56ebef42014-09-16 01:41:51 +00005808//===----------------------------------------------------------------------===//
5809// weak bind table dumping
5810//===----------------------------------------------------------------------===//
5811
5812void llvm::printMachOWeakBindTable(const object::MachOObjectFile *Obj) {
5813 // Build table of sections so names can used in final output.
5814 SegInfo sectionTable(Obj);
5815
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005816 outs() << "segment section address "
5817 "type addend symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005818 for (const llvm::object::MachOBindEntry &Entry : Obj->weakBindTable()) {
5819 // Strong symbols don't have a location to update.
5820 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) {
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005821 outs() << " strong "
Nick Kledzik56ebef42014-09-16 01:41:51 +00005822 << Entry.symbolName() << "\n";
5823 continue;
5824 }
5825 uint32_t SegIndex = Entry.segmentIndex();
5826 uint64_t OffsetInSeg = Entry.segmentOffset();
5827 StringRef SegmentName = sectionTable.segmentName(SegIndex);
5828 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
5829 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5830
5831 // Table lines look like:
5832 // __DATA __data 0x00001000 pointer 0 _foo
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005833 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00005834 << left_justify(SectionName, 18) << " "
5835 << format_hex(Address, 10, true) << " "
5836 << left_justify(Entry.typeName(), 8) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00005837 << format_decimal(Entry.addend(), 8) << " " << Entry.symbolName()
5838 << "\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00005839 }
5840}
5841
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005842// get_dyld_bind_info_symbolname() is used for disassembly and passed an
5843// address, ReferenceValue, in the Mach-O file and looks in the dyld bind
5844// information for that address. If the address is found its binding symbol
5845// name is returned. If not nullptr is returned.
5846static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
5847 struct DisassembleInfo *info) {
Kevin Enderby078be602014-10-23 19:53:12 +00005848 if (info->bindtable == nullptr) {
5849 info->bindtable = new (BindTable);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005850 SegInfo sectionTable(info->O);
5851 for (const llvm::object::MachOBindEntry &Entry : info->O->bindTable()) {
5852 uint32_t SegIndex = Entry.segmentIndex();
5853 uint64_t OffsetInSeg = Entry.segmentOffset();
5854 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
5855 const char *SymbolName = nullptr;
5856 StringRef name = Entry.symbolName();
5857 if (!name.empty())
5858 SymbolName = name.data();
Kevin Enderby078be602014-10-23 19:53:12 +00005859 info->bindtable->push_back(std::make_pair(Address, SymbolName));
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005860 }
5861 }
Kevin Enderby078be602014-10-23 19:53:12 +00005862 for (bind_table_iterator BI = info->bindtable->begin(),
5863 BE = info->bindtable->end();
Kevin Enderby6f326ce2014-10-23 19:37:31 +00005864 BI != BE; ++BI) {
5865 uint64_t Address = BI->first;
5866 if (ReferenceValue == Address) {
5867 const char *SymbolName = BI->second;
5868 return SymbolName;
5869 }
5870 }
5871 return nullptr;
5872}