blob: 417a8b8b389e46c7db62e758d76681f1010c6706 [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"
Benjamin Kramer699128e2011-09-21 01:13:19 +000020#include "llvm/DebugInfo/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 Enderby3f0ffab2014-12-03 22:29:40 +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"
37#include "llvm/Support/GraphWriter.h"
Chandler Carruth4d88a1c2012-12-04 10:44:52 +000038#include "llvm/Support/MachO.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000039#include "llvm/Support/MemoryBuffer.h"
Kevin Enderbybf246f52014-09-24 23:08:22 +000040#include "llvm/Support/FormattedStream.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000041#include "llvm/Support/TargetRegistry.h"
42#include "llvm/Support/TargetSelect.h"
43#include "llvm/Support/raw_ostream.h"
Benjamin Kramer43a772e2011-09-19 17:56:04 +000044#include <algorithm>
45#include <cstring>
Rafael Espindolaa6e9c3e2014-06-12 17:38:55 +000046#include <system_error>
Kevin Enderby04bf6932014-10-28 23:39:46 +000047
48#if HAVE_CXXABI_H
49#include <cxxabi.h>
50#endif
51
Benjamin Kramer43a772e2011-09-19 17:56:04 +000052using namespace llvm;
53using namespace object;
54
55static cl::opt<bool>
Kevin Enderbyb28ed012014-10-29 21:28:24 +000056 UseDbg("g",
57 cl::desc("Print line information from debug info if available"));
Benjamin Kramer699128e2011-09-21 01:13:19 +000058
Kevin Enderbyb28ed012014-10-29 21:28:24 +000059static cl::opt<std::string> DSYMFile("dsym",
60 cl::desc("Use .dSYM file for debug info"));
Benjamin Kramer699128e2011-09-21 01:13:19 +000061
Kevin Enderbyb28ed012014-10-29 21:28:24 +000062static cl::opt<bool> FullLeadingAddr("full-leading-addr",
63 cl::desc("Print full leading address"));
Kevin Enderbybf246f52014-09-24 23:08:22 +000064
65static cl::opt<bool>
66 PrintImmHex("print-imm-hex",
67 cl::desc("Use hex format for immediate values"));
68
Kevin Enderby3f0ffab2014-12-03 22:29:40 +000069static cl::list<std::string>
70 ArchFlags("arch", cl::desc("architecture(s) from a Mach-O file to dump"),
71 cl::ZeroOrMore);
72bool ArchAll = false;
73
Kevin Enderbyec5ca032014-08-18 20:21:02 +000074static std::string ThumbTripleName;
75
76static const Target *GetTarget(const MachOObjectFile *MachOObj,
77 const char **McpuDefault,
78 const Target **ThumbTarget) {
Benjamin Kramer43a772e2011-09-19 17:56:04 +000079 // Figure out the target triple.
Cameron Zwarich88cc16a2012-02-03 06:35:22 +000080 if (TripleName.empty()) {
81 llvm::Triple TT("unknown-unknown-unknown");
Kevin Enderbyec5ca032014-08-18 20:21:02 +000082 llvm::Triple ThumbTriple = Triple();
83 TT = MachOObj->getArch(McpuDefault, &ThumbTriple);
Cameron Zwarich88cc16a2012-02-03 06:35:22 +000084 TripleName = TT.str();
Kevin Enderbyec5ca032014-08-18 20:21:02 +000085 ThumbTripleName = ThumbTriple.str();
Benjamin Kramer43a772e2011-09-19 17:56:04 +000086 }
87
Benjamin Kramer43a772e2011-09-19 17:56:04 +000088 // Get the target specific parser.
89 std::string Error;
90 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
Kevin Enderbyec5ca032014-08-18 20:21:02 +000091 if (TheTarget && ThumbTripleName.empty())
Benjamin Kramer43a772e2011-09-19 17:56:04 +000092 return TheTarget;
93
Kevin Enderbyec5ca032014-08-18 20:21:02 +000094 *ThumbTarget = TargetRegistry::lookupTarget(ThumbTripleName, Error);
95 if (*ThumbTarget)
96 return TheTarget;
97
98 errs() << "llvm-objdump: error: unable to get target for '";
99 if (!TheTarget)
100 errs() << TripleName;
101 else
102 errs() << ThumbTripleName;
103 errs() << "', see --version and --triple.\n";
Craig Toppere6cb63e2014-04-25 04:24:47 +0000104 return nullptr;
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000105}
106
Owen Andersond9243c42011-10-17 21:37:35 +0000107struct SymbolSorter {
108 bool operator()(const SymbolRef &A, const SymbolRef &B) {
109 SymbolRef::Type AType, BType;
110 A.getType(AType);
111 B.getType(BType);
112
113 uint64_t AAddr, BAddr;
114 if (AType != SymbolRef::ST_Function)
115 AAddr = 0;
116 else
117 A.getAddress(AAddr);
118 if (BType != SymbolRef::ST_Function)
119 BAddr = 0;
120 else
121 B.getAddress(BAddr);
122 return AAddr < BAddr;
123 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000124};
125
Kevin Enderby273ae012013-06-06 17:20:50 +0000126// Types for the storted data in code table that is built before disassembly
127// and the predicate function to sort them.
128typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
129typedef std::vector<DiceTableEntry> DiceTable;
130typedef DiceTable::iterator dice_table_iterator;
131
Kevin Enderby930fdc72014-11-06 19:00:13 +0000132// This is used to search for a data in code table entry for the PC being
133// disassembled. The j parameter has the PC in j.first. A single data in code
134// table entry can cover many bytes for each of its Kind's. So if the offset,
135// aka the i.first value, of the data in code table entry plus its Length
136// covers the PC being searched for this will return true. If not it will
137// return false.
David Majnemerea9b8ee2014-11-04 08:41:48 +0000138static bool compareDiceTableEntries(const DiceTableEntry &i,
139 const DiceTableEntry &j) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000140 uint16_t Length;
141 i.second.getLength(Length);
142
143 return j.first >= i.first && j.first < i.first + Length;
Kevin Enderby273ae012013-06-06 17:20:50 +0000144}
145
Kevin Enderby930fdc72014-11-06 19:00:13 +0000146static uint64_t DumpDataInCode(const char *bytes, uint64_t Length,
147 unsigned short Kind) {
148 uint32_t Value, Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000149
150 switch (Kind) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000151 default:
Charles Davis8bdfafd2013-09-01 04:28:48 +0000152 case MachO::DICE_KIND_DATA:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000153 if (Length >= 4) {
154 if (!NoShowRawInsn)
155 DumpBytes(StringRef(bytes, 4));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000156 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
Kevin Enderby273ae012013-06-06 17:20:50 +0000157 outs() << "\t.long " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000158 Size = 4;
159 } else if (Length >= 2) {
160 if (!NoShowRawInsn)
161 DumpBytes(StringRef(bytes, 2));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000162 Value = bytes[1] << 8 | bytes[0];
Kevin Enderby273ae012013-06-06 17:20:50 +0000163 outs() << "\t.short " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000164 Size = 2;
165 } else {
166 if (!NoShowRawInsn)
167 DumpBytes(StringRef(bytes, 2));
Kevin Enderby273ae012013-06-06 17:20:50 +0000168 Value = bytes[0];
169 outs() << "\t.byte " << Value;
Kevin Enderby930fdc72014-11-06 19:00:13 +0000170 Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000171 }
Kevin Enderby930fdc72014-11-06 19:00:13 +0000172 if (Kind == MachO::DICE_KIND_DATA)
173 outs() << "\t@ KIND_DATA\n";
174 else
175 outs() << "\t@ data in code kind = " << Kind << "\n";
Kevin Enderby273ae012013-06-06 17:20:50 +0000176 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000177 case MachO::DICE_KIND_JUMP_TABLE8:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000178 if (!NoShowRawInsn)
179 DumpBytes(StringRef(bytes, 1));
Kevin Enderby273ae012013-06-06 17:20:50 +0000180 Value = bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000181 outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n";
182 Size = 1;
Kevin Enderby273ae012013-06-06 17:20:50 +0000183 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000184 case MachO::DICE_KIND_JUMP_TABLE16:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000185 if (!NoShowRawInsn)
186 DumpBytes(StringRef(bytes, 2));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000187 Value = bytes[1] << 8 | bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000188 outs() << "\t.short " << format("%5u", Value & 0xffff)
189 << "\t@ KIND_JUMP_TABLE16\n";
190 Size = 2;
Kevin Enderby273ae012013-06-06 17:20:50 +0000191 break;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000192 case MachO::DICE_KIND_JUMP_TABLE32:
Kevin Enderby930fdc72014-11-06 19:00:13 +0000193 case MachO::DICE_KIND_ABS_JUMP_TABLE32:
194 if (!NoShowRawInsn)
195 DumpBytes(StringRef(bytes, 4));
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000196 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0];
Kevin Enderby930fdc72014-11-06 19:00:13 +0000197 outs() << "\t.long " << Value;
198 if (Kind == MachO::DICE_KIND_JUMP_TABLE32)
199 outs() << "\t@ KIND_JUMP_TABLE32\n";
200 else
201 outs() << "\t@ KIND_ABS_JUMP_TABLE32\n";
202 Size = 4;
Kevin Enderby273ae012013-06-06 17:20:50 +0000203 break;
204 }
Kevin Enderby930fdc72014-11-06 19:00:13 +0000205 return Size;
Kevin Enderby273ae012013-06-06 17:20:50 +0000206}
207
Alexey Samsonov464d2e42014-03-17 07:28:19 +0000208static void getSectionsAndSymbols(const MachO::mach_header Header,
209 MachOObjectFile *MachOObj,
210 std::vector<SectionRef> &Sections,
211 std::vector<SymbolRef> &Symbols,
212 SmallVectorImpl<uint64_t> &FoundFns,
213 uint64_t &BaseSegmentAddress) {
Kevin Enderby3f0ffab2014-12-03 22:29:40 +0000214 for (const SymbolRef &Symbol : MachOObj->symbols()) {
215 StringRef SymName;
216 Symbol.getName(SymName);
217 if (!SymName.startswith("ltmp"))
218 Symbols.push_back(Symbol);
219 }
Owen Andersond9243c42011-10-17 21:37:35 +0000220
Alexey Samsonov48803e52014-03-13 14:37:36 +0000221 for (const SectionRef &Section : MachOObj->sections()) {
Owen Andersond9243c42011-10-17 21:37:35 +0000222 StringRef SectName;
Alexey Samsonov48803e52014-03-13 14:37:36 +0000223 Section.getName(SectName);
224 Sections.push_back(Section);
Owen Andersond9243c42011-10-17 21:37:35 +0000225 }
226
Rafael Espindola56f976f2013-04-18 18:08:55 +0000227 MachOObjectFile::LoadCommandInfo Command =
Alexey Samsonov48803e52014-03-13 14:37:36 +0000228 MachOObj->getFirstLoadCommandInfo();
Kevin Enderby273ae012013-06-06 17:20:50 +0000229 bool BaseSegmentAddressSet = false;
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000230 for (unsigned i = 0;; ++i) {
Charles Davis8bdfafd2013-09-01 04:28:48 +0000231 if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) {
Benjamin Kramer699128e2011-09-21 01:13:19 +0000232 // We found a function starts segment, parse the addresses for later
233 // consumption.
Charles Davis8bdfafd2013-09-01 04:28:48 +0000234 MachO::linkedit_data_command LLC =
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000235 MachOObj->getLinkeditDataLoadCommand(Command);
Benjamin Kramer699128e2011-09-21 01:13:19 +0000236
Charles Davis8bdfafd2013-09-01 04:28:48 +0000237 MachOObj->ReadULEB128s(LLC.dataoff, FoundFns);
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000238 } else if (Command.C.cmd == MachO::LC_SEGMENT) {
239 MachO::segment_command SLC = MachOObj->getSegmentLoadCommand(Command);
Charles Davis8bdfafd2013-09-01 04:28:48 +0000240 StringRef SegName = SLC.segname;
Kevin Enderbyb28ed012014-10-29 21:28:24 +0000241 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
Kevin Enderby273ae012013-06-06 17:20:50 +0000242 BaseSegmentAddressSet = true;
Charles Davis8bdfafd2013-09-01 04:28:48 +0000243 BaseSegmentAddress = SLC.vmaddr;
Kevin Enderby273ae012013-06-06 17:20:50 +0000244 }
245 }
Rafael Espindolafeef8c22013-04-19 11:36:47 +0000246
Charles Davis8bdfafd2013-09-01 04:28:48 +0000247 if (i == Header.ncmds - 1)
Rafael Espindolafeef8c22013-04-19 11:36:47 +0000248 break;
249 else
250 Command = MachOObj->getNextLoadCommandInfo(Command);
Benjamin Kramer8a529dc2011-09-21 22:16:43 +0000251 }
Benjamin Kramer699128e2011-09-21 01:13:19 +0000252}
253
Kevin Enderby3f0ffab2014-12-03 22:29:40 +0000254// checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file
255// and if it is and there is a list of architecture flags is specified then
256// check to make sure this Mach-O file is one of those architectures or all
257// architectures were specified. If not then an error is generated and this
258// routine returns false. Else it returns true.
259static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) {
260 if (isa<MachOObjectFile>(O) && !ArchAll && ArchFlags.size() != 0) {
261 MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(O);
262 bool ArchFound = false;
263 MachO::mach_header H;
264 MachO::mach_header_64 H_64;
265 Triple T;
266 if (MachO->is64Bit()) {
267 H_64 = MachO->MachOObjectFile::getHeader64();
268 T = MachOObjectFile::getArch(H_64.cputype, H_64.cpusubtype);
269 } else {
270 H = MachO->MachOObjectFile::getHeader();
271 T = MachOObjectFile::getArch(H.cputype, H.cpusubtype);
272 }
273 unsigned i;
274 for (i = 0; i < ArchFlags.size(); ++i) {
275 if (ArchFlags[i] == T.getArchName())
276 ArchFound = true;
277 break;
278 }
279 if (!ArchFound) {
280 errs() << "llvm-objdump: file: " + Filename + " does not contain "
281 << "architecture: " + ArchFlags[i] + "\n";
282 return false;
283 }
284 }
285 return true;
286}
287
288static void DisassembleInputMachO2(StringRef Filename, MachOObjectFile *MachOOF,
289 StringRef ArchiveMemberName = StringRef(),
290 StringRef ArchitectureName = StringRef());
Rafael Espindola9b709252013-04-13 01:45:40 +0000291
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000292void llvm::DisassembleInputMachO(StringRef Filename) {
Kevin Enderby3f0ffab2014-12-03 22:29:40 +0000293 // Check for -arch all and verifiy the -arch flags are valid.
294 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
295 if (ArchFlags[i] == "all") {
296 ArchAll = true;
297 } else {
298 if (!MachOObjectFile::isValidArch(ArchFlags[i])) {
299 errs() << "llvm-objdump: Unknown architecture named '" + ArchFlags[i] +
300 "'for the -arch option\n";
301 return;
302 }
303 }
304 }
305
306 // Attempt to open the binary.
307 ErrorOr<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename);
308 if (std::error_code EC = BinaryOrErr.getError()) {
309 errs() << "llvm-objdump: '" << Filename << "': " << EC.message() << ".\n";
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000310 return;
311 }
Kevin Enderby3f0ffab2014-12-03 22:29:40 +0000312 Binary &Bin = *BinaryOrErr.get().getBinary();
Benjamin Kramer43a772e2011-09-19 17:56:04 +0000313
Kevin Enderby3f0ffab2014-12-03 22:29:40 +0000314 if (Archive *A = dyn_cast<Archive>(&Bin)) {
315 outs() << "Archive : " << Filename << "\n";
316 for (Archive::child_iterator I = A->child_begin(), E = A->child_end();
317 I != E; ++I) {
318 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = I->getAsBinary();
319 if (ChildOrErr.getError())
320 continue;
321 if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
322 if (!checkMachOAndArchFlags(O, Filename))
323 return;
324 DisassembleInputMachO2(Filename, O, O->getFileName());
325 }
326 }
327 return;
328 }
329 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) {
330 // If we have a list of architecture flags specified dump only those.
331 if (!ArchAll && ArchFlags.size() != 0) {
332 // Look for a slice in the universal binary that matches each ArchFlag.
333 bool ArchFound;
334 for (unsigned i = 0; i < ArchFlags.size(); ++i) {
335 ArchFound = false;
336 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
337 E = UB->end_objects();
338 I != E; ++I) {
339 if (ArchFlags[i] == I->getArchTypeName()) {
340 ArchFound = true;
341 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr =
342 I->getAsObjectFile();
343 std::unique_ptr<Archive> A;
344 StringRef ArchitectureName = StringRef();
345 if (ArchFlags.size() > 1)
346 ArchitectureName = I->getArchTypeName();
347 if (ObjOrErr) {
348 ObjectFile &O = *ObjOrErr.get();
349 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
350 DisassembleInputMachO2(Filename, MachOOF, "", ArchitectureName);
351 } else if (!I->getAsArchive(A)) {
352 outs() << "Archive : " << Filename;
353 if (!ArchitectureName.empty())
354 outs() << " (architecture " << ArchitectureName << ")";
355 outs() << "\n";
356 for (Archive::child_iterator AI = A->child_begin(),
357 AE = A->child_end();
358 AI != AE; ++AI) {
359 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
360 if (ChildOrErr.getError())
361 continue;
362 if (MachOObjectFile *O =
363 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
364 DisassembleInputMachO2(Filename, O, O->getFileName(),
365 ArchitectureName);
366 }
367 }
368 }
369 }
370 if (!ArchFound) {
371 errs() << "llvm-objdump: file: " + Filename + " does not contain "
372 << "architecture: " + ArchFlags[i] + "\n";
373 return;
374 }
375 }
376 return;
377 }
378 // No architecture flags were specified so if this contains a slice that
379 // matches the host architecture dump only that.
380 if (!ArchAll) {
381 StringRef HostArchName = MachOObjectFile::getHostArch().getArchName();
382 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
383 E = UB->end_objects();
384 I != E; ++I) {
385 if (HostArchName == I->getArchTypeName()) {
386 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
387 std::unique_ptr<Archive> A;
388 std::string ArchiveName;
389 ArchiveName.clear();
390 if (ObjOrErr) {
391 ObjectFile &O = *ObjOrErr.get();
392 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O))
393 DisassembleInputMachO2(Filename, MachOOF);
394 } else if (!I->getAsArchive(A)) {
395 outs() << "Archive : " << Filename << "\n";
396 for (Archive::child_iterator AI = A->child_begin(),
397 AE = A->child_end();
398 AI != AE; ++AI) {
399 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
400 if (ChildOrErr.getError())
401 continue;
402 if (MachOObjectFile *O =
403 dyn_cast<MachOObjectFile>(&*ChildOrErr.get()))
404 DisassembleInputMachO2(Filename, O, O->getFileName());
405 }
406 }
407 return;
408 }
409 }
410 }
411 // Either all architectures have been specified or none have been specified
412 // and this does not contain the host architecture so dump all the slices.
413 bool moreThanOneArch = UB->getNumberOfObjects() > 1;
414 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(),
415 E = UB->end_objects();
416 I != E; ++I) {
417 ErrorOr<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile();
418 std::unique_ptr<Archive> A;
419 StringRef ArchitectureName = StringRef();
420 if (moreThanOneArch)
421 ArchitectureName = I->getArchTypeName();
422 if (ObjOrErr) {
423 ObjectFile &Obj = *ObjOrErr.get();
424 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj))
425 DisassembleInputMachO2(Filename, MachOOF, "", ArchitectureName);
426 } else if (!I->getAsArchive(A)) {
427 outs() << "Archive : " << Filename;
428 if (!ArchitectureName.empty())
429 outs() << " (architecture " << ArchitectureName << ")";
430 outs() << "\n";
431 for (Archive::child_iterator AI = A->child_begin(), AE = A->child_end();
432 AI != AE; ++AI) {
433 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = AI->getAsBinary();
434 if (ChildOrErr.getError())
435 continue;
436 if (MachOObjectFile *O =
437 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) {
438 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O))
439 DisassembleInputMachO2(Filename, MachOOF, MachOOF->getFileName(),
440 ArchitectureName);
441 }
442 }
443 }
444 }
445 return;
446 }
447 if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) {
448 if (!checkMachOAndArchFlags(O, Filename))
449 return;
450 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O)) {
451 DisassembleInputMachO2(Filename, MachOOF);
452 } else
453 errs() << "llvm-objdump: '" << Filename << "': "
454 << "Object is not a Mach-O file type.\n";
455 } else
456 errs() << "llvm-objdump: '" << Filename << "': "
457 << "Unrecognized file type.\n";
Rafael Espindola9b709252013-04-13 01:45:40 +0000458}
459
Kevin Enderbybf246f52014-09-24 23:08:22 +0000460typedef DenseMap<uint64_t, StringRef> SymbolAddressMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +0000461typedef std::pair<uint64_t, const char *> BindInfoEntry;
462typedef std::vector<BindInfoEntry> BindTable;
463typedef BindTable::iterator bind_table_iterator;
Kevin Enderbybf246f52014-09-24 23:08:22 +0000464
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000465// The block of info used by the Symbolizer call backs.
466struct DisassembleInfo {
467 bool verbose;
468 MachOObjectFile *O;
469 SectionRef S;
Kevin Enderbybf246f52014-09-24 23:08:22 +0000470 SymbolAddressMap *AddrMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +0000471 std::vector<SectionRef> *Sections;
472 const char *class_name;
473 const char *selector_name;
474 char *method;
Kevin Enderby04bf6932014-10-28 23:39:46 +0000475 char *demangled_name;
Kevin Enderbyae3c1262014-11-14 21:52:18 +0000476 uint64_t adrp_addr;
477 uint32_t adrp_inst;
Kevin Enderby078be602014-10-23 19:53:12 +0000478 BindTable *bindtable;
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000479};
480
Kevin Enderby9907d0a2014-11-04 00:43:16 +0000481// GuessSymbolName is passed the address of what might be a symbol and a
482// pointer to the DisassembleInfo struct. It returns the name of a symbol
483// with that address or nullptr if no symbol is found with that address.
484static const char *GuessSymbolName(uint64_t value,
485 struct DisassembleInfo *info) {
486 const char *SymbolName = nullptr;
487 // A DenseMap can't lookup up some values.
488 if (value != 0xffffffffffffffffULL && value != 0xfffffffffffffffeULL) {
489 StringRef name = info->AddrMap->lookup(value);
490 if (!name.empty())
491 SymbolName = name.data();
492 }
493 return SymbolName;
494}
495
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000496// SymbolizerGetOpInfo() is the operand information call back function.
497// This is called to get the symbolic information for operand(s) of an
498// instruction when it is being done. This routine does this from
499// the relocation information, symbol table, etc. That block of information
500// is a pointer to the struct DisassembleInfo that was passed when the
501// disassembler context was created and passed to back to here when
502// called back by the disassembler for instruction operands that could have
503// relocation information. The address of the instruction containing operand is
504// at the Pc parameter. The immediate value the operand has is passed in
505// op_info->Value and is at Offset past the start of the instruction and has a
506// byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the
507// LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol
508// names and addends of the symbolic expression to add for the operand. The
509// value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic
510// information is returned then this function returns 1 else it returns 0.
511int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset,
512 uint64_t Size, int TagType, void *TagBuf) {
513 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
514 struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf;
Kevin Enderbyae3c1262014-11-14 21:52:18 +0000515 uint64_t value = op_info->Value;
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000516
517 // Make sure all fields returned are zero if we don't set them.
518 memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1));
519 op_info->Value = value;
520
521 // If the TagType is not the value 1 which it code knows about or if no
522 // verbose symbolic information is wanted then just return 0, indicating no
523 // information is being returned.
524 if (TagType != 1 || info->verbose == false)
525 return 0;
526
527 unsigned int Arch = info->O->getArch();
528 if (Arch == Triple::x86) {
Kevin Enderby9907d0a2014-11-04 00:43:16 +0000529 if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
530 return 0;
531 // First search the section's relocation entries (if any) for an entry
532 // for this section offset.
533 uint32_t sect_addr = info->S.getAddress();
534 uint32_t sect_offset = (Pc + Offset) - sect_addr;
535 bool reloc_found = false;
536 DataRefImpl Rel;
537 MachO::any_relocation_info RE;
538 bool isExtern = false;
539 SymbolRef Symbol;
540 bool r_scattered = false;
541 uint32_t r_value, pair_r_value, r_type;
542 for (const RelocationRef &Reloc : info->S.relocations()) {
543 uint64_t RelocOffset;
544 Reloc.getOffset(RelocOffset);
545 if (RelocOffset == sect_offset) {
546 Rel = Reloc.getRawDataRefImpl();
547 RE = info->O->getRelocation(Rel);
Kevin Enderby3eb73e12014-11-11 19:16:45 +0000548 r_type = info->O->getAnyRelocationType(RE);
Kevin Enderby9907d0a2014-11-04 00:43:16 +0000549 r_scattered = info->O->isRelocationScattered(RE);
550 if (r_scattered) {
551 r_value = info->O->getScatteredRelocationValue(RE);
Kevin Enderby9907d0a2014-11-04 00:43:16 +0000552 if (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
553 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) {
554 DataRefImpl RelNext = Rel;
555 info->O->moveRelocationNext(RelNext);
556 MachO::any_relocation_info RENext;
557 RENext = info->O->getRelocation(RelNext);
558 if (info->O->isRelocationScattered(RENext))
Kevin Enderby930fdc72014-11-06 19:00:13 +0000559 pair_r_value = info->O->getScatteredRelocationValue(RENext);
Kevin Enderby9907d0a2014-11-04 00:43:16 +0000560 else
561 return 0;
562 }
563 } else {
564 isExtern = info->O->getPlainRelocationExternal(RE);
565 if (isExtern) {
566 symbol_iterator RelocSym = Reloc.getSymbol();
567 Symbol = *RelocSym;
568 }
569 }
570 reloc_found = true;
571 break;
572 }
573 }
574 if (reloc_found && isExtern) {
575 StringRef SymName;
576 Symbol.getName(SymName);
577 const char *name = SymName.data();
578 op_info->AddSymbol.Present = 1;
579 op_info->AddSymbol.Name = name;
580 // For i386 extern relocation entries the value in the instruction is
581 // the offset from the symbol, and value is already set in op_info->Value.
582 return 1;
583 }
584 if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF ||
585 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) {
586 const char *add = GuessSymbolName(r_value, info);
587 const char *sub = GuessSymbolName(pair_r_value, info);
588 uint32_t offset = value - (r_value - pair_r_value);
589 op_info->AddSymbol.Present = 1;
590 if (add != nullptr)
591 op_info->AddSymbol.Name = add;
592 else
593 op_info->AddSymbol.Value = r_value;
594 op_info->SubtractSymbol.Present = 1;
595 if (sub != nullptr)
596 op_info->SubtractSymbol.Name = sub;
597 else
598 op_info->SubtractSymbol.Value = pair_r_value;
599 op_info->Value = offset;
600 return 1;
601 }
602 // TODO:
603 // Second search the external relocation entries of a fully linked image
604 // (if any) for an entry that matches this segment offset.
605 // uint32_t seg_offset = (Pc + Offset);
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000606 return 0;
607 } else if (Arch == Triple::x86_64) {
608 if (Size != 1 && Size != 2 && Size != 4 && Size != 0)
609 return 0;
610 // First search the section's relocation entries (if any) for an entry
611 // for this section offset.
Rafael Espindola80291272014-10-08 15:28:58 +0000612 uint64_t sect_addr = info->S.getAddress();
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000613 uint64_t sect_offset = (Pc + Offset) - sect_addr;
614 bool reloc_found = false;
615 DataRefImpl Rel;
616 MachO::any_relocation_info RE;
617 bool isExtern = false;
618 SymbolRef Symbol;
619 for (const RelocationRef &Reloc : info->S.relocations()) {
620 uint64_t RelocOffset;
621 Reloc.getOffset(RelocOffset);
622 if (RelocOffset == sect_offset) {
623 Rel = Reloc.getRawDataRefImpl();
624 RE = info->O->getRelocation(Rel);
625 // NOTE: Scattered relocations don't exist on x86_64.
626 isExtern = info->O->getPlainRelocationExternal(RE);
627 if (isExtern) {
628 symbol_iterator RelocSym = Reloc.getSymbol();
629 Symbol = *RelocSym;
630 }
631 reloc_found = true;
632 break;
633 }
634 }
635 if (reloc_found && isExtern) {
636 // The Value passed in will be adjusted by the Pc if the instruction
637 // adds the Pc. But for x86_64 external relocation entries the Value
638 // is the offset from the external symbol.
639 if (info->O->getAnyRelocationPCRel(RE))
640 op_info->Value -= Pc + Offset + Size;
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000641 StringRef SymName;
642 Symbol.getName(SymName);
643 const char *name = SymName.data();
644 unsigned Type = info->O->getAnyRelocationType(RE);
645 if (Type == MachO::X86_64_RELOC_SUBTRACTOR) {
646 DataRefImpl RelNext = Rel;
647 info->O->moveRelocationNext(RelNext);
648 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
649 unsigned TypeNext = info->O->getAnyRelocationType(RENext);
650 bool isExternNext = info->O->getPlainRelocationExternal(RENext);
651 unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext);
652 if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) {
653 op_info->SubtractSymbol.Present = 1;
654 op_info->SubtractSymbol.Name = name;
655 symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum);
656 Symbol = *RelocSymNext;
657 StringRef SymNameNext;
658 Symbol.getName(SymNameNext);
659 name = SymNameNext.data();
660 }
661 }
662 // TODO: add the VariantKinds to op_info->VariantKind for relocation types
663 // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT.
664 op_info->AddSymbol.Present = 1;
665 op_info->AddSymbol.Name = name;
666 return 1;
667 }
668 // TODO:
669 // Second search the external relocation entries of a fully linked image
670 // (if any) for an entry that matches this segment offset.
Kevin Enderby6f326ce2014-10-23 19:37:31 +0000671 // uint64_t seg_offset = (Pc + Offset);
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000672 return 0;
673 } else if (Arch == Triple::arm) {
Kevin Enderby930fdc72014-11-06 19:00:13 +0000674 if (Offset != 0 || (Size != 4 && Size != 2))
675 return 0;
676 // First search the section's relocation entries (if any) for an entry
677 // for this section offset.
678 uint32_t sect_addr = info->S.getAddress();
679 uint32_t sect_offset = (Pc + Offset) - sect_addr;
680 bool reloc_found = false;
681 DataRefImpl Rel;
682 MachO::any_relocation_info RE;
683 bool isExtern = false;
684 SymbolRef Symbol;
685 bool r_scattered = false;
686 uint32_t r_value, pair_r_value, r_type, r_length, other_half;
687 for (const RelocationRef &Reloc : info->S.relocations()) {
688 uint64_t RelocOffset;
689 Reloc.getOffset(RelocOffset);
690 if (RelocOffset == sect_offset) {
691 Rel = Reloc.getRawDataRefImpl();
692 RE = info->O->getRelocation(Rel);
693 r_length = info->O->getAnyRelocationLength(RE);
694 r_scattered = info->O->isRelocationScattered(RE);
695 if (r_scattered) {
696 r_value = info->O->getScatteredRelocationValue(RE);
697 r_type = info->O->getScatteredRelocationType(RE);
698 } else {
699 r_type = info->O->getAnyRelocationType(RE);
700 isExtern = info->O->getPlainRelocationExternal(RE);
701 if (isExtern) {
702 symbol_iterator RelocSym = Reloc.getSymbol();
703 Symbol = *RelocSym;
704 }
705 }
706 if (r_type == MachO::ARM_RELOC_HALF ||
707 r_type == MachO::ARM_RELOC_SECTDIFF ||
708 r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF ||
709 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
710 DataRefImpl RelNext = Rel;
711 info->O->moveRelocationNext(RelNext);
712 MachO::any_relocation_info RENext;
713 RENext = info->O->getRelocation(RelNext);
714 other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff;
715 if (info->O->isRelocationScattered(RENext))
716 pair_r_value = info->O->getScatteredRelocationValue(RENext);
717 }
718 reloc_found = true;
719 break;
720 }
721 }
722 if (reloc_found && isExtern) {
723 StringRef SymName;
724 Symbol.getName(SymName);
725 const char *name = SymName.data();
726 op_info->AddSymbol.Present = 1;
727 op_info->AddSymbol.Name = name;
728 if (value != 0) {
729 switch (r_type) {
730 case MachO::ARM_RELOC_HALF:
731 if ((r_length & 0x1) == 1) {
732 op_info->Value = value << 16 | other_half;
733 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
734 } else {
735 op_info->Value = other_half << 16 | value;
736 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
737 }
738 break;
739 default:
740 break;
741 }
742 } else {
743 switch (r_type) {
744 case MachO::ARM_RELOC_HALF:
745 if ((r_length & 0x1) == 1) {
746 op_info->Value = value << 16 | other_half;
747 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
748 } else {
749 op_info->Value = other_half << 16 | value;
750 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
751 }
752 break;
753 default:
754 break;
755 }
756 }
757 return 1;
758 }
759 // If we have a branch that is not an external relocation entry then
760 // return 0 so the code in tryAddingSymbolicOperand() can use the
761 // SymbolLookUp call back with the branch target address to look up the
762 // symbol and possiblity add an annotation for a symbol stub.
763 if (reloc_found && isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 ||
764 r_type == MachO::ARM_THUMB_RELOC_BR22))
765 return 0;
766
767 uint32_t offset = 0;
768 if (reloc_found) {
769 if (r_type == MachO::ARM_RELOC_HALF ||
770 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
771 if ((r_length & 0x1) == 1)
772 value = value << 16 | other_half;
773 else
774 value = other_half << 16 | value;
775 }
776 if (r_scattered && (r_type != MachO::ARM_RELOC_HALF &&
777 r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) {
778 offset = value - r_value;
779 value = r_value;
780 }
781 }
782
783 if (reloc_found && r_type == MachO::ARM_RELOC_HALF_SECTDIFF) {
784 if ((r_length & 0x1) == 1)
785 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
786 else
787 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
788 const char *add = GuessSymbolName(r_value, info);
789 const char *sub = GuessSymbolName(pair_r_value, info);
790 int32_t offset = value - (r_value - pair_r_value);
791 op_info->AddSymbol.Present = 1;
792 if (add != nullptr)
793 op_info->AddSymbol.Name = add;
794 else
795 op_info->AddSymbol.Value = r_value;
796 op_info->SubtractSymbol.Present = 1;
797 if (sub != nullptr)
798 op_info->SubtractSymbol.Name = sub;
799 else
800 op_info->SubtractSymbol.Value = pair_r_value;
801 op_info->Value = offset;
802 return 1;
803 }
804
805 if (reloc_found == false)
806 return 0;
807
808 op_info->AddSymbol.Present = 1;
809 op_info->Value = offset;
810 if (reloc_found) {
811 if (r_type == MachO::ARM_RELOC_HALF) {
812 if ((r_length & 0x1) == 1)
813 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16;
814 else
815 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16;
816 }
817 }
818 const char *add = GuessSymbolName(value, info);
819 if (add != nullptr) {
820 op_info->AddSymbol.Name = add;
821 return 1;
822 }
823 op_info->AddSymbol.Value = value;
824 return 1;
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000825 } else if (Arch == Triple::aarch64) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +0000826 if (Offset != 0 || Size != 4)
827 return 0;
828 // First search the section's relocation entries (if any) for an entry
829 // for this section offset.
830 uint64_t sect_addr = info->S.getAddress();
831 uint64_t sect_offset = (Pc + Offset) - sect_addr;
832 bool reloc_found = false;
833 DataRefImpl Rel;
834 MachO::any_relocation_info RE;
835 bool isExtern = false;
836 SymbolRef Symbol;
837 uint32_t r_type = 0;
838 for (const RelocationRef &Reloc : info->S.relocations()) {
839 uint64_t RelocOffset;
840 Reloc.getOffset(RelocOffset);
841 if (RelocOffset == sect_offset) {
842 Rel = Reloc.getRawDataRefImpl();
843 RE = info->O->getRelocation(Rel);
844 r_type = info->O->getAnyRelocationType(RE);
845 if (r_type == MachO::ARM64_RELOC_ADDEND) {
846 DataRefImpl RelNext = Rel;
847 info->O->moveRelocationNext(RelNext);
848 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext);
849 if (value == 0) {
850 value = info->O->getPlainRelocationSymbolNum(RENext);
851 op_info->Value = value;
852 }
853 }
854 // NOTE: Scattered relocations don't exist on arm64.
855 isExtern = info->O->getPlainRelocationExternal(RE);
856 if (isExtern) {
857 symbol_iterator RelocSym = Reloc.getSymbol();
858 Symbol = *RelocSym;
859 }
860 reloc_found = true;
861 break;
862 }
863 }
864 if (reloc_found && isExtern) {
865 StringRef SymName;
866 Symbol.getName(SymName);
867 const char *name = SymName.data();
868 op_info->AddSymbol.Present = 1;
869 op_info->AddSymbol.Name = name;
870
871 switch (r_type) {
872 case MachO::ARM64_RELOC_PAGE21:
873 /* @page */
874 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE;
875 break;
876 case MachO::ARM64_RELOC_PAGEOFF12:
877 /* @pageoff */
878 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF;
879 break;
880 case MachO::ARM64_RELOC_GOT_LOAD_PAGE21:
881 /* @gotpage */
882 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE;
883 break;
884 case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12:
885 /* @gotpageoff */
886 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF;
887 break;
888 case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21:
889 /* @tvlppage is not implemented in llvm-mc */
890 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP;
891 break;
892 case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12:
893 /* @tvlppageoff is not implemented in llvm-mc */
894 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF;
895 break;
896 default:
897 case MachO::ARM64_RELOC_BRANCH26:
898 op_info->VariantKind = LLVMDisassembler_VariantKind_None;
899 break;
900 }
901 return 1;
902 }
Kevin Enderby98c9acc2014-09-16 18:00:57 +0000903 return 0;
904 } else {
905 return 0;
906 }
907}
908
Kevin Enderbybf246f52014-09-24 23:08:22 +0000909// GuessCstringPointer is passed the address of what might be a pointer to a
910// literal string in a cstring section. If that address is in a cstring section
911// it returns a pointer to that string. Else it returns nullptr.
912const char *GuessCstringPointer(uint64_t ReferenceValue,
913 struct DisassembleInfo *info) {
914 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
915 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
916 for (unsigned I = 0;; ++I) {
917 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
918 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
919 for (unsigned J = 0; J < Seg.nsects; ++J) {
920 MachO::section_64 Sec = info->O->getSection64(Load, J);
921 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
922 if (section_type == MachO::S_CSTRING_LITERALS &&
923 ReferenceValue >= Sec.addr &&
924 ReferenceValue < Sec.addr + Sec.size) {
925 uint64_t sect_offset = ReferenceValue - Sec.addr;
926 uint64_t object_offset = Sec.offset + sect_offset;
927 StringRef MachOContents = info->O->getData();
928 uint64_t object_size = MachOContents.size();
929 const char *object_addr = (const char *)MachOContents.data();
930 if (object_offset < object_size) {
931 const char *name = object_addr + object_offset;
932 return name;
933 } else {
934 return nullptr;
935 }
936 }
937 }
938 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
939 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
940 for (unsigned J = 0; J < Seg.nsects; ++J) {
941 MachO::section Sec = info->O->getSection(Load, J);
942 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
943 if (section_type == MachO::S_CSTRING_LITERALS &&
944 ReferenceValue >= Sec.addr &&
945 ReferenceValue < Sec.addr + Sec.size) {
946 uint64_t sect_offset = ReferenceValue - Sec.addr;
947 uint64_t object_offset = Sec.offset + sect_offset;
948 StringRef MachOContents = info->O->getData();
949 uint64_t object_size = MachOContents.size();
950 const char *object_addr = (const char *)MachOContents.data();
951 if (object_offset < object_size) {
952 const char *name = object_addr + object_offset;
953 return name;
954 } else {
955 return nullptr;
956 }
957 }
958 }
959 }
960 if (I == LoadCommandCount - 1)
961 break;
962 else
963 Load = info->O->getNextLoadCommandInfo(Load);
964 }
965 return nullptr;
966}
967
Kevin Enderby85974882014-09-26 22:20:44 +0000968// GuessIndirectSymbol returns the name of the indirect symbol for the
969// ReferenceValue passed in or nullptr. This is used when ReferenceValue maybe
970// an address of a symbol stub or a lazy or non-lazy pointer to associate the
971// symbol name being referenced by the stub or pointer.
972static const char *GuessIndirectSymbol(uint64_t ReferenceValue,
973 struct DisassembleInfo *info) {
974 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
975 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
976 MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand();
977 MachO::symtab_command Symtab = info->O->getSymtabLoadCommand();
978 for (unsigned I = 0;; ++I) {
979 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
980 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
981 for (unsigned J = 0; J < Seg.nsects; ++J) {
982 MachO::section_64 Sec = info->O->getSection64(Load, J);
983 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
984 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
985 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
986 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
987 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
988 section_type == MachO::S_SYMBOL_STUBS) &&
989 ReferenceValue >= Sec.addr &&
990 ReferenceValue < Sec.addr + Sec.size) {
991 uint32_t stride;
992 if (section_type == MachO::S_SYMBOL_STUBS)
993 stride = Sec.reserved2;
994 else
995 stride = 8;
996 if (stride == 0)
997 return nullptr;
998 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
999 if (index < Dysymtab.nindirectsyms) {
1000 uint32_t indirect_symbol =
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001001 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
Kevin Enderby85974882014-09-26 22:20:44 +00001002 if (indirect_symbol < Symtab.nsyms) {
1003 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
1004 SymbolRef Symbol = *Sym;
1005 StringRef SymName;
1006 Symbol.getName(SymName);
1007 const char *name = SymName.data();
1008 return name;
1009 }
1010 }
1011 }
1012 }
1013 } else if (Load.C.cmd == MachO::LC_SEGMENT) {
1014 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load);
1015 for (unsigned J = 0; J < Seg.nsects; ++J) {
1016 MachO::section Sec = info->O->getSection(Load, J);
1017 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE;
1018 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
1019 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
1020 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
1021 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS ||
1022 section_type == MachO::S_SYMBOL_STUBS) &&
1023 ReferenceValue >= Sec.addr &&
1024 ReferenceValue < Sec.addr + Sec.size) {
1025 uint32_t stride;
1026 if (section_type == MachO::S_SYMBOL_STUBS)
1027 stride = Sec.reserved2;
1028 else
1029 stride = 4;
1030 if (stride == 0)
1031 return nullptr;
1032 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride;
1033 if (index < Dysymtab.nindirectsyms) {
1034 uint32_t indirect_symbol =
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001035 info->O->getIndirectSymbolTableEntry(Dysymtab, index);
Kevin Enderby85974882014-09-26 22:20:44 +00001036 if (indirect_symbol < Symtab.nsyms) {
1037 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol);
1038 SymbolRef Symbol = *Sym;
1039 StringRef SymName;
1040 Symbol.getName(SymName);
1041 const char *name = SymName.data();
1042 return name;
1043 }
1044 }
1045 }
1046 }
1047 }
1048 if (I == LoadCommandCount - 1)
1049 break;
1050 else
1051 Load = info->O->getNextLoadCommandInfo(Load);
1052 }
1053 return nullptr;
1054}
1055
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001056// method_reference() is called passing it the ReferenceName that might be
1057// a reference it to an Objective-C method call. If so then it allocates and
1058// assembles a method call string with the values last seen and saved in
1059// the DisassembleInfo's class_name and selector_name fields. This is saved
1060// into the method field of the info and any previous string is free'ed.
1061// Then the class_name field in the info is set to nullptr. The method call
1062// string is set into ReferenceName and ReferenceType is set to
1063// LLVMDisassembler_ReferenceType_Out_Objc_Message. If this not a method call
1064// then both ReferenceType and ReferenceName are left unchanged.
1065static void method_reference(struct DisassembleInfo *info,
1066 uint64_t *ReferenceType,
1067 const char **ReferenceName) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001068 unsigned int Arch = info->O->getArch();
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001069 if (*ReferenceName != nullptr) {
1070 if (strcmp(*ReferenceName, "_objc_msgSend") == 0) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001071 if (info->selector_name != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001072 if (info->method != nullptr)
1073 free(info->method);
1074 if (info->class_name != nullptr) {
1075 info->method = (char *)malloc(5 + strlen(info->class_name) +
1076 strlen(info->selector_name));
1077 if (info->method != nullptr) {
1078 strcpy(info->method, "+[");
1079 strcat(info->method, info->class_name);
1080 strcat(info->method, " ");
1081 strcat(info->method, info->selector_name);
1082 strcat(info->method, "]");
1083 *ReferenceName = info->method;
1084 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
1085 }
1086 } else {
1087 info->method = (char *)malloc(9 + strlen(info->selector_name));
1088 if (info->method != nullptr) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001089 if (Arch == Triple::x86_64)
1090 strcpy(info->method, "-[%rdi ");
1091 else if (Arch == Triple::aarch64)
1092 strcpy(info->method, "-[x0 ");
1093 else
1094 strcpy(info->method, "-[r? ");
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001095 strcat(info->method, info->selector_name);
1096 strcat(info->method, "]");
1097 *ReferenceName = info->method;
1098 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
1099 }
1100 }
1101 info->class_name = nullptr;
1102 }
1103 } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001104 if (info->selector_name != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001105 if (info->method != nullptr)
1106 free(info->method);
1107 info->method = (char *)malloc(17 + strlen(info->selector_name));
1108 if (info->method != nullptr) {
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001109 if (Arch == Triple::x86_64)
1110 strcpy(info->method, "-[[%rdi super] ");
1111 else if (Arch == Triple::aarch64)
1112 strcpy(info->method, "-[[x0 super] ");
1113 else
1114 strcpy(info->method, "-[[r? super] ");
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001115 strcat(info->method, info->selector_name);
1116 strcat(info->method, "]");
1117 *ReferenceName = info->method;
1118 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message;
1119 }
1120 info->class_name = nullptr;
1121 }
1122 }
1123 }
1124}
1125
1126// GuessPointerPointer() is passed the address of what might be a pointer to
1127// a reference to an Objective-C class, selector, message ref or cfstring.
1128// If so the value of the pointer is returned and one of the booleans are set
1129// to true. If not zero is returned and all the booleans are set to false.
1130static uint64_t GuessPointerPointer(uint64_t ReferenceValue,
1131 struct DisassembleInfo *info,
1132 bool &classref, bool &selref, bool &msgref,
1133 bool &cfstring) {
1134 classref = false;
1135 selref = false;
1136 msgref = false;
1137 cfstring = false;
1138 uint32_t LoadCommandCount = info->O->getHeader().ncmds;
1139 MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo();
1140 for (unsigned I = 0;; ++I) {
1141 if (Load.C.cmd == MachO::LC_SEGMENT_64) {
1142 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load);
1143 for (unsigned J = 0; J < Seg.nsects; ++J) {
1144 MachO::section_64 Sec = info->O->getSection64(Load, J);
1145 if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 ||
1146 strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
1147 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 ||
1148 strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 ||
1149 strncmp(Sec.sectname, "__cfstring", 16) == 0) &&
1150 ReferenceValue >= Sec.addr &&
1151 ReferenceValue < Sec.addr + Sec.size) {
1152 uint64_t sect_offset = ReferenceValue - Sec.addr;
1153 uint64_t object_offset = Sec.offset + sect_offset;
1154 StringRef MachOContents = info->O->getData();
1155 uint64_t object_size = MachOContents.size();
1156 const char *object_addr = (const char *)MachOContents.data();
1157 if (object_offset < object_size) {
1158 uint64_t pointer_value;
1159 memcpy(&pointer_value, object_addr + object_offset,
1160 sizeof(uint64_t));
1161 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
1162 sys::swapByteOrder(pointer_value);
1163 if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0)
1164 selref = true;
1165 else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 ||
1166 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0)
1167 classref = true;
1168 else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 &&
1169 ReferenceValue + 8 < Sec.addr + Sec.size) {
1170 msgref = true;
1171 memcpy(&pointer_value, object_addr + object_offset + 8,
1172 sizeof(uint64_t));
1173 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
1174 sys::swapByteOrder(pointer_value);
1175 } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0)
1176 cfstring = true;
1177 return pointer_value;
1178 } else {
1179 return 0;
1180 }
1181 }
1182 }
1183 }
1184 // TODO: Look for LC_SEGMENT for 32-bit Mach-O files.
1185 if (I == LoadCommandCount - 1)
1186 break;
1187 else
1188 Load = info->O->getNextLoadCommandInfo(Load);
1189 }
1190 return 0;
1191}
1192
1193// get_pointer_64 returns a pointer to the bytes in the object file at the
1194// Address from a section in the Mach-O file. And indirectly returns the
1195// offset into the section, number of bytes left in the section past the offset
1196// and which section is was being referenced. If the Address is not in a
1197// section nullptr is returned.
1198const char *get_pointer_64(uint64_t Address, uint32_t &offset, uint32_t &left,
1199 SectionRef &S, DisassembleInfo *info) {
1200 offset = 0;
1201 left = 0;
1202 S = SectionRef();
1203 for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) {
1204 uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress();
1205 uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize();
1206 if (Address >= SectAddress && Address < SectAddress + SectSize) {
1207 S = (*(info->Sections))[SectIdx];
1208 offset = Address - SectAddress;
1209 left = SectSize - offset;
1210 StringRef SectContents;
1211 ((*(info->Sections))[SectIdx]).getContents(SectContents);
1212 return SectContents.data() + offset;
1213 }
1214 }
1215 return nullptr;
1216}
1217
1218// get_symbol_64() returns the name of a symbol (or nullptr) and the address of
1219// the symbol indirectly through n_value. Based on the relocation information
1220// for the specified section offset in the specified section reference.
1221const char *get_symbol_64(uint32_t sect_offset, SectionRef S,
1222 DisassembleInfo *info, uint64_t &n_value) {
1223 n_value = 0;
1224 if (info->verbose == false)
1225 return nullptr;
1226
1227 // See if there is an external relocation entry at the sect_offset.
1228 bool reloc_found = false;
1229 DataRefImpl Rel;
1230 MachO::any_relocation_info RE;
1231 bool isExtern = false;
1232 SymbolRef Symbol;
1233 for (const RelocationRef &Reloc : S.relocations()) {
1234 uint64_t RelocOffset;
1235 Reloc.getOffset(RelocOffset);
1236 if (RelocOffset == sect_offset) {
1237 Rel = Reloc.getRawDataRefImpl();
1238 RE = info->O->getRelocation(Rel);
1239 if (info->O->isRelocationScattered(RE))
1240 continue;
1241 isExtern = info->O->getPlainRelocationExternal(RE);
1242 if (isExtern) {
1243 symbol_iterator RelocSym = Reloc.getSymbol();
1244 Symbol = *RelocSym;
1245 }
1246 reloc_found = true;
1247 break;
1248 }
1249 }
1250 // If there is an external relocation entry for a symbol in this section
1251 // at this section_offset then use that symbol's value for the n_value
1252 // and return its name.
1253 const char *SymbolName = nullptr;
1254 if (reloc_found && isExtern) {
1255 Symbol.getAddress(n_value);
1256 StringRef name;
1257 Symbol.getName(name);
1258 if (!name.empty()) {
1259 SymbolName = name.data();
1260 return SymbolName;
1261 }
1262 }
1263
1264 // TODO: For fully linked images, look through the external relocation
1265 // entries off the dynamic symtab command. For these the r_offset is from the
1266 // start of the first writeable segment in the Mach-O file. So the offset
1267 // to this section from that segment is passed to this routine by the caller,
1268 // as the database_offset. Which is the difference of the section's starting
1269 // address and the first writable segment.
1270 //
1271 // NOTE: need add passing the database_offset to this routine.
1272
1273 // TODO: We did not find an external relocation entry so look up the
1274 // ReferenceValue as an address of a symbol and if found return that symbol's
1275 // name.
1276 //
1277 // NOTE: need add passing the ReferenceValue to this routine. Then that code
1278 // would simply be this:
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001279 // SymbolName = GuessSymbolName(ReferenceValue, info);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001280
1281 return SymbolName;
1282}
1283
1284// These are structs in the Objective-C meta data and read to produce the
1285// comments for disassembly. While these are part of the ABI they are no
1286// public defintions. So the are here not in include/llvm/Support/MachO.h .
1287
1288// The cfstring object in a 64-bit Mach-O file.
1289struct cfstring64_t {
1290 uint64_t isa; // class64_t * (64-bit pointer)
1291 uint64_t flags; // flag bits
1292 uint64_t characters; // char * (64-bit pointer)
1293 uint64_t length; // number of non-NULL characters in above
1294};
1295
1296// The class object in a 64-bit Mach-O file.
1297struct class64_t {
1298 uint64_t isa; // class64_t * (64-bit pointer)
1299 uint64_t superclass; // class64_t * (64-bit pointer)
1300 uint64_t cache; // Cache (64-bit pointer)
1301 uint64_t vtable; // IMP * (64-bit pointer)
1302 uint64_t data; // class_ro64_t * (64-bit pointer)
1303};
1304
1305struct class_ro64_t {
1306 uint32_t flags;
1307 uint32_t instanceStart;
1308 uint32_t instanceSize;
1309 uint32_t reserved;
1310 uint64_t ivarLayout; // const uint8_t * (64-bit pointer)
1311 uint64_t name; // const char * (64-bit pointer)
1312 uint64_t baseMethods; // const method_list_t * (64-bit pointer)
1313 uint64_t baseProtocols; // const protocol_list_t * (64-bit pointer)
1314 uint64_t ivars; // const ivar_list_t * (64-bit pointer)
1315 uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer)
1316 uint64_t baseProperties; // const struct objc_property_list (64-bit pointer)
1317};
1318
1319inline void swapStruct(struct cfstring64_t &cfs) {
1320 sys::swapByteOrder(cfs.isa);
1321 sys::swapByteOrder(cfs.flags);
1322 sys::swapByteOrder(cfs.characters);
1323 sys::swapByteOrder(cfs.length);
1324}
1325
1326inline void swapStruct(struct class64_t &c) {
1327 sys::swapByteOrder(c.isa);
1328 sys::swapByteOrder(c.superclass);
1329 sys::swapByteOrder(c.cache);
1330 sys::swapByteOrder(c.vtable);
1331 sys::swapByteOrder(c.data);
1332}
1333
1334inline void swapStruct(struct class_ro64_t &cro) {
1335 sys::swapByteOrder(cro.flags);
1336 sys::swapByteOrder(cro.instanceStart);
1337 sys::swapByteOrder(cro.instanceSize);
1338 sys::swapByteOrder(cro.reserved);
1339 sys::swapByteOrder(cro.ivarLayout);
1340 sys::swapByteOrder(cro.name);
1341 sys::swapByteOrder(cro.baseMethods);
1342 sys::swapByteOrder(cro.baseProtocols);
1343 sys::swapByteOrder(cro.ivars);
1344 sys::swapByteOrder(cro.weakIvarLayout);
1345 sys::swapByteOrder(cro.baseProperties);
1346}
1347
1348static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
1349 struct DisassembleInfo *info);
1350
1351// get_objc2_64bit_class_name() is used for disassembly and is passed a pointer
1352// to an Objective-C class and returns the class name. It is also passed the
1353// address of the pointer, so when the pointer is zero as it can be in an .o
1354// file, that is used to look for an external relocation entry with a symbol
1355// name.
1356const char *get_objc2_64bit_class_name(uint64_t pointer_value,
1357 uint64_t ReferenceValue,
1358 struct DisassembleInfo *info) {
1359 const char *r;
1360 uint32_t offset, left;
1361 SectionRef S;
1362
1363 // The pointer_value can be 0 in an object file and have a relocation
1364 // entry for the class symbol at the ReferenceValue (the address of the
1365 // pointer).
1366 if (pointer_value == 0) {
1367 r = get_pointer_64(ReferenceValue, offset, left, S, info);
1368 if (r == nullptr || left < sizeof(uint64_t))
1369 return nullptr;
1370 uint64_t n_value;
1371 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
1372 if (symbol_name == nullptr)
1373 return nullptr;
Hans Wennborgdb53e302014-10-23 21:59:17 +00001374 const char *class_name = strrchr(symbol_name, '$');
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001375 if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0')
1376 return class_name + 2;
1377 else
1378 return nullptr;
1379 }
1380
1381 // The case were the pointer_value is non-zero and points to a class defined
1382 // in this Mach-O file.
1383 r = get_pointer_64(pointer_value, offset, left, S, info);
1384 if (r == nullptr || left < sizeof(struct class64_t))
1385 return nullptr;
1386 struct class64_t c;
1387 memcpy(&c, r, sizeof(struct class64_t));
1388 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
1389 swapStruct(c);
1390 if (c.data == 0)
1391 return nullptr;
1392 r = get_pointer_64(c.data, offset, left, S, info);
1393 if (r == nullptr || left < sizeof(struct class_ro64_t))
1394 return nullptr;
1395 struct class_ro64_t cro;
1396 memcpy(&cro, r, sizeof(struct class_ro64_t));
1397 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
1398 swapStruct(cro);
1399 if (cro.name == 0)
1400 return nullptr;
1401 const char *name = get_pointer_64(cro.name, offset, left, S, info);
1402 return name;
1403}
1404
1405// get_objc2_64bit_cfstring_name is used for disassembly and is passed a
1406// pointer to a cfstring and returns its name or nullptr.
1407const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue,
1408 struct DisassembleInfo *info) {
1409 const char *r, *name;
1410 uint32_t offset, left;
1411 SectionRef S;
1412 struct cfstring64_t cfs;
1413 uint64_t cfs_characters;
1414
1415 r = get_pointer_64(ReferenceValue, offset, left, S, info);
1416 if (r == nullptr || left < sizeof(struct cfstring64_t))
1417 return nullptr;
1418 memcpy(&cfs, r, sizeof(struct cfstring64_t));
1419 if (info->O->isLittleEndian() != sys::IsLittleEndianHost)
1420 swapStruct(cfs);
1421 if (cfs.characters == 0) {
1422 uint64_t n_value;
1423 const char *symbol_name = get_symbol_64(
1424 offset + offsetof(struct cfstring64_t, characters), S, info, n_value);
1425 if (symbol_name == nullptr)
1426 return nullptr;
1427 cfs_characters = n_value;
1428 } else
1429 cfs_characters = cfs.characters;
1430 name = get_pointer_64(cfs_characters, offset, left, S, info);
1431
1432 return name;
1433}
1434
1435// get_objc2_64bit_selref() is used for disassembly and is passed a the address
1436// of a pointer to an Objective-C selector reference when the pointer value is
1437// zero as in a .o file and is likely to have a external relocation entry with
1438// who's symbol's n_value is the real pointer to the selector name. If that is
1439// the case the real pointer to the selector name is returned else 0 is
1440// returned
1441uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue,
1442 struct DisassembleInfo *info) {
1443 uint32_t offset, left;
1444 SectionRef S;
1445
1446 const char *r = get_pointer_64(ReferenceValue, offset, left, S, info);
1447 if (r == nullptr || left < sizeof(uint64_t))
1448 return 0;
1449 uint64_t n_value;
1450 const char *symbol_name = get_symbol_64(offset, S, info, n_value);
1451 if (symbol_name == nullptr)
1452 return 0;
1453 return n_value;
1454}
1455
Kevin Enderbybf246f52014-09-24 23:08:22 +00001456// GuessLiteralPointer returns a string which for the item in the Mach-O file
1457// for the address passed in as ReferenceValue for printing as a comment with
1458// the instruction and also returns the corresponding type of that item
1459// indirectly through ReferenceType.
1460//
1461// If ReferenceValue is an address of literal cstring then a pointer to the
1462// cstring is returned and ReferenceType is set to
1463// LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr .
1464//
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001465// If ReferenceValue is an address of an Objective-C CFString, Selector ref or
1466// Class ref that name is returned and the ReferenceType is set accordingly.
1467//
1468// Lastly, literals which are Symbol address in a literal pool are looked for
1469// and if found the symbol name is returned and ReferenceType is set to
1470// LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr .
1471//
1472// If there is no item in the Mach-O file for the address passed in as
1473// ReferenceValue nullptr is returned and ReferenceType is unchanged.
Kevin Enderbybf246f52014-09-24 23:08:22 +00001474const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC,
1475 uint64_t *ReferenceType,
1476 struct DisassembleInfo *info) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00001477 // First see if there is an external relocation entry at the ReferencePC.
Rafael Espindola80291272014-10-08 15:28:58 +00001478 uint64_t sect_addr = info->S.getAddress();
Kevin Enderbybf246f52014-09-24 23:08:22 +00001479 uint64_t sect_offset = ReferencePC - sect_addr;
1480 bool reloc_found = false;
1481 DataRefImpl Rel;
1482 MachO::any_relocation_info RE;
1483 bool isExtern = false;
1484 SymbolRef Symbol;
1485 for (const RelocationRef &Reloc : info->S.relocations()) {
1486 uint64_t RelocOffset;
1487 Reloc.getOffset(RelocOffset);
1488 if (RelocOffset == sect_offset) {
1489 Rel = Reloc.getRawDataRefImpl();
1490 RE = info->O->getRelocation(Rel);
1491 if (info->O->isRelocationScattered(RE))
1492 continue;
1493 isExtern = info->O->getPlainRelocationExternal(RE);
1494 if (isExtern) {
1495 symbol_iterator RelocSym = Reloc.getSymbol();
1496 Symbol = *RelocSym;
1497 }
1498 reloc_found = true;
1499 break;
1500 }
1501 }
1502 // If there is an external relocation entry for a symbol in a section
1503 // then used that symbol's value for the value of the reference.
1504 if (reloc_found && isExtern) {
1505 if (info->O->getAnyRelocationPCRel(RE)) {
1506 unsigned Type = info->O->getAnyRelocationType(RE);
1507 if (Type == MachO::X86_64_RELOC_SIGNED) {
1508 Symbol.getAddress(ReferenceValue);
1509 }
1510 }
1511 }
1512
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001513 // Look for literals such as Objective-C CFStrings refs, Selector refs,
1514 // Message refs and Class refs.
1515 bool classref, selref, msgref, cfstring;
1516 uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref,
1517 selref, msgref, cfstring);
1518 if (classref == true && pointer_value == 0) {
1519 // Note the ReferenceValue is a pointer into the __objc_classrefs section.
1520 // And the pointer_value in that section is typically zero as it will be
1521 // set by dyld as part of the "bind information".
1522 const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info);
1523 if (name != nullptr) {
1524 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
Hans Wennborgdb53e302014-10-23 21:59:17 +00001525 const char *class_name = strrchr(name, '$');
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001526 if (class_name != nullptr && class_name[1] == '_' &&
1527 class_name[2] != '\0') {
1528 info->class_name = class_name + 2;
1529 return name;
1530 }
1531 }
1532 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00001533
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001534 if (classref == true) {
1535 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref;
1536 const char *name =
1537 get_objc2_64bit_class_name(pointer_value, ReferenceValue, info);
1538 if (name != nullptr)
1539 info->class_name = name;
1540 else
1541 name = "bad class ref";
Kevin Enderbybf246f52014-09-24 23:08:22 +00001542 return name;
1543 }
1544
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001545 if (cfstring == true) {
1546 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref;
1547 const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info);
1548 return name;
1549 }
1550
1551 if (selref == true && pointer_value == 0)
1552 pointer_value = get_objc2_64bit_selref(ReferenceValue, info);
1553
1554 if (pointer_value != 0)
1555 ReferenceValue = pointer_value;
1556
1557 const char *name = GuessCstringPointer(ReferenceValue, info);
1558 if (name) {
1559 if (pointer_value != 0 && selref == true) {
1560 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref;
1561 info->selector_name = name;
1562 } else if (pointer_value != 0 && msgref == true) {
1563 info->class_name = nullptr;
1564 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref;
1565 info->selector_name = name;
1566 } else
1567 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr;
1568 return name;
1569 }
1570
1571 // Lastly look for an indirect symbol with this ReferenceValue which is in
1572 // a literal pool. If found return that symbol name.
1573 name = GuessIndirectSymbol(ReferenceValue, info);
1574 if (name) {
1575 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr;
1576 return name;
1577 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00001578
1579 return nullptr;
1580}
1581
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001582// SymbolizerSymbolLookUp is the symbol lookup function passed when creating
Kevin Enderbybf246f52014-09-24 23:08:22 +00001583// the Symbolizer. It looks up the ReferenceValue using the info passed via the
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001584// pointer to the struct DisassembleInfo that was passed when MCSymbolizer
1585// is created and returns the symbol name that matches the ReferenceValue or
1586// nullptr if none. The ReferenceType is passed in for the IN type of
1587// reference the instruction is making from the values in defined in the header
1588// "llvm-c/Disassembler.h". On return the ReferenceType can set to a specific
1589// Out type and the ReferenceName will also be set which is added as a comment
1590// to the disassembled instruction.
1591//
Kevin Enderby04bf6932014-10-28 23:39:46 +00001592#if HAVE_CXXABI_H
1593// If the symbol name is a C++ mangled name then the demangled name is
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001594// returned through ReferenceName and ReferenceType is set to
1595// LLVMDisassembler_ReferenceType_DeMangled_Name .
Kevin Enderby04bf6932014-10-28 23:39:46 +00001596#endif
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001597//
1598// When this is called to get a symbol name for a branch target then the
1599// ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then
1600// SymbolValue will be looked for in the indirect symbol table to determine if
1601// it is an address for a symbol stub. If so then the symbol name for that
1602// stub is returned indirectly through ReferenceName and then ReferenceType is
1603// set to LLVMDisassembler_ReferenceType_Out_SymbolStub.
1604//
Kevin Enderbybf246f52014-09-24 23:08:22 +00001605// When this is called with an value loaded via a PC relative load then
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001606// ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the
1607// SymbolValue is checked to be an address of literal pointer, symbol pointer,
1608// or an Objective-C meta data reference. If so the output ReferenceType is
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001609// set to correspond to that as well as setting the ReferenceName.
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001610const char *SymbolizerSymbolLookUp(void *DisInfo, uint64_t ReferenceValue,
1611 uint64_t *ReferenceType,
1612 uint64_t ReferencePC,
1613 const char **ReferenceName) {
1614 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo;
Kevin Enderbybf246f52014-09-24 23:08:22 +00001615 // If no verbose symbolic information is wanted then just return nullptr.
1616 if (info->verbose == false) {
1617 *ReferenceName = nullptr;
1618 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001619 return nullptr;
1620 }
Kevin Enderbybf246f52014-09-24 23:08:22 +00001621
Kevin Enderby9907d0a2014-11-04 00:43:16 +00001622 const char *SymbolName = GuessSymbolName(ReferenceValue, info);
Kevin Enderbybf246f52014-09-24 23:08:22 +00001623
Kevin Enderby85974882014-09-26 22:20:44 +00001624 if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) {
1625 *ReferenceName = GuessIndirectSymbol(ReferenceValue, info);
Kevin Enderby04bf6932014-10-28 23:39:46 +00001626 if (*ReferenceName != nullptr) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001627 method_reference(info, ReferenceType, ReferenceName);
1628 if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message)
1629 *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub;
1630 } else
Kevin Enderby04bf6932014-10-28 23:39:46 +00001631#if HAVE_CXXABI_H
Kevin Enderbyb28ed012014-10-29 21:28:24 +00001632 if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
Kevin Enderby04bf6932014-10-28 23:39:46 +00001633 if (info->demangled_name != nullptr)
1634 free(info->demangled_name);
1635 int status;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00001636 info->demangled_name =
1637 abi::__cxa_demangle(SymbolName + 1, nullptr, nullptr, &status);
Kevin Enderby04bf6932014-10-28 23:39:46 +00001638 if (info->demangled_name != nullptr) {
1639 *ReferenceName = info->demangled_name;
1640 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
1641 } else
1642 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
1643 } else
1644#endif
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001645 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
1646 } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) {
1647 *ReferenceName =
1648 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
Kevin Enderby85974882014-09-26 22:20:44 +00001649 if (*ReferenceName)
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001650 method_reference(info, ReferenceType, ReferenceName);
Kevin Enderby85974882014-09-26 22:20:44 +00001651 else
1652 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001653 // If this is arm64 and the reference is an adrp instruction save the
1654 // instruction, passed in ReferenceValue and the address of the instruction
1655 // for use later if we see and add immediate instruction.
1656 } else if (info->O->getArch() == Triple::aarch64 &&
1657 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
1658 info->adrp_inst = ReferenceValue;
1659 info->adrp_addr = ReferencePC;
1660 SymbolName = nullptr;
1661 *ReferenceName = nullptr;
1662 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
1663 // If this is arm64 and reference is an add immediate instruction and we
1664 // have
1665 // seen an adrp instruction just before it and the adrp's Xd register
1666 // matches
1667 // this add's Xn register reconstruct the value being referenced and look to
1668 // see if it is a literal pointer. Note the add immediate instruction is
1669 // passed in ReferenceValue.
1670 } else if (info->O->getArch() == Triple::aarch64 &&
1671 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri &&
1672 ReferencePC - 4 == info->adrp_addr &&
1673 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
1674 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
1675 uint32_t addxri_inst;
1676 uint64_t adrp_imm, addxri_imm;
1677
1678 adrp_imm =
1679 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
1680 if (info->adrp_inst & 0x0200000)
1681 adrp_imm |= 0xfffffffffc000000LL;
1682
1683 addxri_inst = ReferenceValue;
1684 addxri_imm = (addxri_inst >> 10) & 0xfff;
1685 if (((addxri_inst >> 22) & 0x3) == 1)
1686 addxri_imm <<= 12;
1687
1688 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
1689 (adrp_imm << 12) + addxri_imm;
1690
1691 *ReferenceName =
1692 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
1693 if (*ReferenceName == nullptr)
1694 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
1695 // If this is arm64 and the reference is a load register instruction and we
1696 // have seen an adrp instruction just before it and the adrp's Xd register
1697 // matches this add's Xn register reconstruct the value being referenced and
1698 // look to see if it is a literal pointer. Note the load register
1699 // instruction is passed in ReferenceValue.
1700 } else if (info->O->getArch() == Triple::aarch64 &&
1701 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui &&
1702 ReferencePC - 4 == info->adrp_addr &&
1703 (info->adrp_inst & 0x9f000000) == 0x90000000 &&
1704 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) {
1705 uint32_t ldrxui_inst;
1706 uint64_t adrp_imm, ldrxui_imm;
1707
1708 adrp_imm =
1709 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3);
1710 if (info->adrp_inst & 0x0200000)
1711 adrp_imm |= 0xfffffffffc000000LL;
1712
1713 ldrxui_inst = ReferenceValue;
1714 ldrxui_imm = (ldrxui_inst >> 10) & 0xfff;
1715
1716 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) +
1717 (adrp_imm << 12) + (ldrxui_imm << 3);
1718
1719 *ReferenceName =
1720 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
1721 if (*ReferenceName == nullptr)
1722 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
1723 }
1724 // If this arm64 and is an load register (PC-relative) instruction the
1725 // ReferenceValue is the PC plus the immediate value.
1726 else if (info->O->getArch() == Triple::aarch64 &&
1727 (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl ||
1728 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) {
1729 *ReferenceName =
1730 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info);
1731 if (*ReferenceName == nullptr)
1732 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
Kevin Enderby85974882014-09-26 22:20:44 +00001733 }
Kevin Enderby04bf6932014-10-28 23:39:46 +00001734#if HAVE_CXXABI_H
1735 else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) {
1736 if (info->demangled_name != nullptr)
1737 free(info->demangled_name);
1738 int status;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00001739 info->demangled_name =
1740 abi::__cxa_demangle(SymbolName + 1, nullptr, nullptr, &status);
Kevin Enderby04bf6932014-10-28 23:39:46 +00001741 if (info->demangled_name != nullptr) {
1742 *ReferenceName = info->demangled_name;
1743 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name;
1744 }
1745 }
1746#endif
Kevin Enderby6f326ce2014-10-23 19:37:31 +00001747 else {
Kevin Enderbybf246f52014-09-24 23:08:22 +00001748 *ReferenceName = nullptr;
1749 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None;
1750 }
1751
1752 return SymbolName;
1753}
1754
Kevin Enderbybf246f52014-09-24 23:08:22 +00001755/// \brief Emits the comments that are stored in the CommentStream.
1756/// Each comment in the CommentStream must end with a newline.
1757static void emitComments(raw_svector_ostream &CommentStream,
1758 SmallString<128> &CommentsToEmit,
1759 formatted_raw_ostream &FormattedOS,
1760 const MCAsmInfo &MAI) {
1761 // Flush the stream before taking its content.
1762 CommentStream.flush();
1763 StringRef Comments = CommentsToEmit.str();
1764 // Get the default information for printing a comment.
1765 const char *CommentBegin = MAI.getCommentString();
1766 unsigned CommentColumn = MAI.getCommentColumn();
1767 bool IsFirst = true;
1768 while (!Comments.empty()) {
1769 if (!IsFirst)
1770 FormattedOS << '\n';
1771 // Emit a line of comments.
1772 FormattedOS.PadToColumn(CommentColumn);
1773 size_t Position = Comments.find('\n');
1774 FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position);
1775 // Move after the newline character.
1776 Comments = Comments.substr(Position + 1);
1777 IsFirst = false;
1778 }
1779 FormattedOS.flush();
1780
1781 // Tell the comment stream that the vector changed underneath it.
1782 CommentsToEmit.clear();
1783 CommentStream.resync();
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001784}
1785
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001786static void DisassembleInputMachO2(StringRef Filename, MachOObjectFile *MachOOF,
1787 StringRef ArchiveMemberName,
1788 StringRef ArchitectureName) {
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001789 const char *McpuDefault = nullptr;
1790 const Target *ThumbTarget = nullptr;
1791 const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget);
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001792 if (!TheTarget) {
1793 // GetTarget prints out stuff.
1794 return;
1795 }
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001796 if (MCPU.empty() && McpuDefault)
1797 MCPU = McpuDefault;
1798
Ahmed Charles56440fd2014-03-06 05:51:42 +00001799 std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001800 std::unique_ptr<const MCInstrInfo> ThumbInstrInfo;
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001801 if (ThumbTarget)
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001802 ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo());
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001803
Kevin Enderbyc9595622014-08-06 23:24:41 +00001804 // Package up features to be passed to target/subtarget
1805 std::string FeaturesStr;
1806 if (MAttrs.size()) {
1807 SubtargetFeatures Features;
1808 for (unsigned i = 0; i != MAttrs.size(); ++i)
1809 Features.AddFeature(MAttrs[i]);
1810 FeaturesStr = Features.getString();
1811 }
1812
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001813 // Set up disassembler.
Ahmed Charles56440fd2014-03-06 05:51:42 +00001814 std::unique_ptr<const MCRegisterInfo> MRI(
1815 TheTarget->createMCRegInfo(TripleName));
1816 std::unique_ptr<const MCAsmInfo> AsmInfo(
Rafael Espindola227144c2013-05-13 01:16:13 +00001817 TheTarget->createMCAsmInfo(*MRI, TripleName));
Ahmed Charles56440fd2014-03-06 05:51:42 +00001818 std::unique_ptr<const MCSubtargetInfo> STI(
Kevin Enderbyc9595622014-08-06 23:24:41 +00001819 TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
Craig Toppere6cb63e2014-04-25 04:24:47 +00001820 MCContext Ctx(AsmInfo.get(), MRI.get(), nullptr);
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001821 std::unique_ptr<MCDisassembler> DisAsm(
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001822 TheTarget->createMCDisassembler(*STI, Ctx));
Kevin Enderby98c9acc2014-09-16 18:00:57 +00001823 std::unique_ptr<MCSymbolizer> Symbolizer;
1824 struct DisassembleInfo SymbolizerInfo;
1825 std::unique_ptr<MCRelocationInfo> RelInfo(
1826 TheTarget->createMCRelocationInfo(TripleName, Ctx));
1827 if (RelInfo) {
1828 Symbolizer.reset(TheTarget->createMCSymbolizer(
1829 TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
1830 &SymbolizerInfo, &Ctx, RelInfo.release()));
1831 DisAsm->setSymbolizer(std::move(Symbolizer));
1832 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001833 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
Ahmed Charles56440fd2014-03-06 05:51:42 +00001834 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
1835 AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI, *STI));
Kevin Enderbybf246f52014-09-24 23:08:22 +00001836 // Set the display preference for hex vs. decimal immediates.
1837 IP->setPrintImmHex(PrintImmHex);
1838 // Comment stream and backing vector.
1839 SmallString<128> CommentsToEmit;
1840 raw_svector_ostream CommentStream(CommentsToEmit);
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001841 // FIXME: Setting the CommentStream in the InstPrinter is problematic in that
1842 // if it is done then arm64 comments for string literals don't get printed
1843 // and some constant get printed instead and not setting it causes intel
1844 // (32-bit and 64-bit) comments printed with different spacing before the
1845 // comment causing different diffs with the 'C' disassembler library API.
1846 // IP->setCommentStream(CommentStream);
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00001847
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001848 if (!AsmInfo || !STI || !DisAsm || !IP) {
Michael J. Spencerc1363cf2011-10-07 19:25:47 +00001849 errs() << "error: couldn't initialize disassembler for target "
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00001850 << TripleName << '\n';
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001851 return;
1852 }
1853
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001854 // Set up thumb disassembler.
1855 std::unique_ptr<const MCRegisterInfo> ThumbMRI;
1856 std::unique_ptr<const MCAsmInfo> ThumbAsmInfo;
1857 std::unique_ptr<const MCSubtargetInfo> ThumbSTI;
Kevin Enderby930fdc72014-11-06 19:00:13 +00001858 std::unique_ptr<MCDisassembler> ThumbDisAsm;
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001859 std::unique_ptr<MCInstPrinter> ThumbIP;
1860 std::unique_ptr<MCContext> ThumbCtx;
Kevin Enderby930fdc72014-11-06 19:00:13 +00001861 std::unique_ptr<MCSymbolizer> ThumbSymbolizer;
1862 struct DisassembleInfo ThumbSymbolizerInfo;
1863 std::unique_ptr<MCRelocationInfo> ThumbRelInfo;
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001864 if (ThumbTarget) {
1865 ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName));
1866 ThumbAsmInfo.reset(
1867 ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName));
1868 ThumbSTI.reset(
1869 ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MCPU, FeaturesStr));
1870 ThumbCtx.reset(new MCContext(ThumbAsmInfo.get(), ThumbMRI.get(), nullptr));
1871 ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx));
Kevin Enderby930fdc72014-11-06 19:00:13 +00001872 MCContext *PtrThumbCtx = ThumbCtx.get();
1873 ThumbRelInfo.reset(
1874 ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx));
1875 if (ThumbRelInfo) {
1876 ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer(
1877 ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp,
1878 &ThumbSymbolizerInfo, PtrThumbCtx, ThumbRelInfo.release()));
1879 ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer));
1880 }
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001881 int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect();
1882 ThumbIP.reset(ThumbTarget->createMCInstPrinter(
1883 ThumbAsmPrinterVariant, *ThumbAsmInfo, *ThumbInstrInfo, *ThumbMRI,
1884 *ThumbSTI));
Kevin Enderbybf246f52014-09-24 23:08:22 +00001885 // Set the display preference for hex vs. decimal immediates.
1886 ThumbIP->setPrintImmHex(PrintImmHex);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001887 }
1888
Kevin Enderbyae3c1262014-11-14 21:52:18 +00001889 if (ThumbTarget && (!ThumbAsmInfo || !ThumbSTI || !ThumbDisAsm || !ThumbIP)) {
Kevin Enderbyec5ca032014-08-18 20:21:02 +00001890 errs() << "error: couldn't initialize disassembler for target "
1891 << ThumbTripleName << '\n';
1892 return;
1893 }
1894
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001895 outs() << Filename;
1896 if (!ArchiveMemberName.empty())
1897 outs() << '(' << ArchiveMemberName << ')';
1898 if (!ArchitectureName.empty())
1899 outs() << " (architecture " << ArchitectureName << ")";
1900 outs() << ":\n";
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001901
Charles Davis8bdfafd2013-09-01 04:28:48 +00001902 MachO::mach_header Header = MachOOF->getHeader();
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001903
Ahmed Bougachaaa790682013-05-24 01:07:04 +00001904 // FIXME: Using the -cfg command line option, this code used to be able to
1905 // annotate relocations with the referenced symbol's name, and if this was
1906 // inside a __[cf]string section, the data it points to. This is now replaced
1907 // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
Owen Andersond9243c42011-10-17 21:37:35 +00001908 std::vector<SectionRef> Sections;
1909 std::vector<SymbolRef> Symbols;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001910 SmallVector<uint64_t, 8> FoundFns;
Kevin Enderby273ae012013-06-06 17:20:50 +00001911 uint64_t BaseSegmentAddress;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001912
Kevin Enderby273ae012013-06-06 17:20:50 +00001913 getSectionsAndSymbols(Header, MachOOF, Sections, Symbols, FoundFns,
1914 BaseSegmentAddress);
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001915
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001916 // Sort the symbols by address, just in case they didn't come in that way.
Owen Andersond9243c42011-10-17 21:37:35 +00001917 std::sort(Symbols.begin(), Symbols.end(), SymbolSorter());
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001918
Kevin Enderby273ae012013-06-06 17:20:50 +00001919 // Build a data in code table that is sorted on by the address of each entry.
1920 uint64_t BaseAddress = 0;
Charles Davis8bdfafd2013-09-01 04:28:48 +00001921 if (Header.filetype == MachO::MH_OBJECT)
Rafael Espindola80291272014-10-08 15:28:58 +00001922 BaseAddress = Sections[0].getAddress();
Kevin Enderby273ae012013-06-06 17:20:50 +00001923 else
1924 BaseAddress = BaseSegmentAddress;
1925 DiceTable Dices;
Kevin Enderby273ae012013-06-06 17:20:50 +00001926 for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
Rafael Espindola5e812af2014-01-30 02:49:50 +00001927 DI != DE; ++DI) {
Kevin Enderby273ae012013-06-06 17:20:50 +00001928 uint32_t Offset;
1929 DI->getOffset(Offset);
1930 Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
1931 }
1932 array_pod_sort(Dices.begin(), Dices.end());
1933
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001934#ifndef NDEBUG
1935 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
1936#else
1937 raw_ostream &DebugOut = nulls();
1938#endif
1939
Ahmed Charles56440fd2014-03-06 05:51:42 +00001940 std::unique_ptr<DIContext> diContext;
Rafael Espindola9b709252013-04-13 01:45:40 +00001941 ObjectFile *DbgObj = MachOOF;
Benjamin Kramer699128e2011-09-21 01:13:19 +00001942 // Try to find debug info and set up the DIContext for it.
1943 if (UseDbg) {
Benjamin Kramer699128e2011-09-21 01:13:19 +00001944 // A separate DSym file path was specified, parse it as a macho file,
1945 // get the sections and supply it to the section name parsing machinery.
1946 if (!DSYMFile.empty()) {
Rafael Espindola48af1c22014-08-19 18:44:46 +00001947 ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr =
Rafael Espindolaadf21f22014-07-06 17:43:13 +00001948 MemoryBuffer::getFileOrSTDIN(DSYMFile);
Rafael Espindola48af1c22014-08-19 18:44:46 +00001949 if (std::error_code EC = BufOrErr.getError()) {
Rafael Espindolaadf21f22014-07-06 17:43:13 +00001950 errs() << "llvm-objdump: " << Filename << ": " << EC.message() << '\n';
Benjamin Kramer699128e2011-09-21 01:13:19 +00001951 return;
1952 }
Rafael Espindola48af1c22014-08-19 18:44:46 +00001953 DbgObj =
1954 ObjectFile::createMachOObjectFile(BufOrErr.get()->getMemBufferRef())
1955 .get()
1956 .release();
Benjamin Kramer699128e2011-09-21 01:13:19 +00001957 }
1958
Eric Christopher7370b552012-11-12 21:40:38 +00001959 // Setup the DIContext
Rafael Espindolaa04bb5b2014-07-31 20:19:36 +00001960 diContext.reset(DIContext::getDWARFContext(*DbgObj));
Benjamin Kramer699128e2011-09-21 01:13:19 +00001961 }
1962
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00001963 // TODO: For now this only disassembles the (__TEXT,__text) section (see the
1964 // checks in the code below at the top of this loop). It should allow a
1965 // darwin otool(1) like -s option to disassemble any named segment & section
1966 // that is marked as containing instructions with the attributes
1967 // S_ATTR_PURE_INSTRUCTIONS or S_ATTR_SOME_INSTRUCTIONS in the flags field of
1968 // the section structure.
1969 outs() << "(__TEXT,__text) section\n";
1970
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001971 for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
Ahmed Bougachaaa790682013-05-24 01:07:04 +00001972
Rafael Espindola80291272014-10-08 15:28:58 +00001973 bool SectIsText = Sections[SectIdx].isText();
Ahmed Bougachaaa790682013-05-24 01:07:04 +00001974 if (SectIsText == false)
1975 continue;
1976
Owen Andersond9243c42011-10-17 21:37:35 +00001977 StringRef SectName;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00001978 if (Sections[SectIdx].getName(SectName) || SectName != "__text")
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00001979 continue; // Skip non-text sections
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001980
Rafael Espindolaa9f810b2012-12-21 03:47:03 +00001981 DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
Ahmed Bougachaaa790682013-05-24 01:07:04 +00001982
Rafael Espindolab0f76a42013-04-05 15:15:22 +00001983 StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
1984 if (SegmentName != "__TEXT")
Rafael Espindolaa9f810b2012-12-21 03:47:03 +00001985 continue;
1986
Rafael Espindola7fc5b872014-11-12 02:04:27 +00001987 StringRef BytesStr;
1988 Sections[SectIdx].getContents(BytesStr);
Aaron Ballman106fd7b2014-11-12 14:01:17 +00001989 ArrayRef<uint8_t> Bytes(reinterpret_cast<const uint8_t *>(BytesStr.data()),
1990 BytesStr.size());
Rafael Espindola80291272014-10-08 15:28:58 +00001991 uint64_t SectAddress = Sections[SectIdx].getAddress();
Rafael Espindolabd604f22014-11-07 00:52:15 +00001992
Benjamin Kramer43a772e2011-09-19 17:56:04 +00001993 bool symbolTableWorked = false;
1994
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00001995 // Parse relocations.
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00001996 std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
1997 for (const RelocationRef &Reloc : Sections[SectIdx].relocations()) {
Rafael Espindola80291272014-10-08 15:28:58 +00001998 uint64_t RelocOffset;
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00001999 Reloc.getOffset(RelocOffset);
Rafael Espindola80291272014-10-08 15:28:58 +00002000 uint64_t SectionAddress = Sections[SectIdx].getAddress();
Owen Andersond9243c42011-10-17 21:37:35 +00002001 RelocOffset -= SectionAddress;
2002
Alexey Samsonovaa4d2952014-03-14 14:22:49 +00002003 symbol_iterator RelocSym = Reloc.getSymbol();
Owen Andersond9243c42011-10-17 21:37:35 +00002004
Rafael Espindola806f0062013-06-05 01:33:53 +00002005 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002006 }
2007 array_pod_sort(Relocs.begin(), Relocs.end());
2008
Kevin Enderbybf246f52014-09-24 23:08:22 +00002009 // Create a map of symbol addresses to symbol names for use by
2010 // the SymbolizerSymbolLookUp() routine.
2011 SymbolAddressMap AddrMap;
2012 for (const SymbolRef &Symbol : MachOOF->symbols()) {
2013 SymbolRef::Type ST;
2014 Symbol.getType(ST);
2015 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data ||
2016 ST == SymbolRef::ST_Other) {
2017 uint64_t Address;
2018 Symbol.getAddress(Address);
2019 StringRef SymName;
2020 Symbol.getName(SymName);
2021 AddrMap[Address] = SymName;
2022 }
2023 }
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002024 // Set up the block of info used by the Symbolizer call backs.
2025 SymbolizerInfo.verbose = true;
2026 SymbolizerInfo.O = MachOOF;
2027 SymbolizerInfo.S = Sections[SectIdx];
Kevin Enderbybf246f52014-09-24 23:08:22 +00002028 SymbolizerInfo.AddrMap = &AddrMap;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002029 SymbolizerInfo.Sections = &Sections;
2030 SymbolizerInfo.class_name = nullptr;
2031 SymbolizerInfo.selector_name = nullptr;
2032 SymbolizerInfo.method = nullptr;
Kevin Enderby04bf6932014-10-28 23:39:46 +00002033 SymbolizerInfo.demangled_name = nullptr;
Kevin Enderby078be602014-10-23 19:53:12 +00002034 SymbolizerInfo.bindtable = nullptr;
Kevin Enderby10738222014-11-19 20:20:16 +00002035 SymbolizerInfo.adrp_addr = 0;
2036 SymbolizerInfo.adrp_inst = 0;
Kevin Enderby930fdc72014-11-06 19:00:13 +00002037 // Same for the ThumbSymbolizer
2038 ThumbSymbolizerInfo.verbose = true;
2039 ThumbSymbolizerInfo.O = MachOOF;
2040 ThumbSymbolizerInfo.S = Sections[SectIdx];
2041 ThumbSymbolizerInfo.AddrMap = &AddrMap;
2042 ThumbSymbolizerInfo.Sections = &Sections;
2043 ThumbSymbolizerInfo.class_name = nullptr;
2044 ThumbSymbolizerInfo.selector_name = nullptr;
2045 ThumbSymbolizerInfo.method = nullptr;
2046 ThumbSymbolizerInfo.demangled_name = nullptr;
2047 ThumbSymbolizerInfo.bindtable = nullptr;
Kevin Enderby10738222014-11-19 20:20:16 +00002048 ThumbSymbolizerInfo.adrp_addr = 0;
2049 ThumbSymbolizerInfo.adrp_inst = 0;
Kevin Enderby98c9acc2014-09-16 18:00:57 +00002050
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00002051 // Disassemble symbol by symbol.
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002052 for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
Owen Andersond9243c42011-10-17 21:37:35 +00002053 StringRef SymName;
2054 Symbols[SymIdx].getName(SymName);
2055
2056 SymbolRef::Type ST;
2057 Symbols[SymIdx].getType(ST);
2058 if (ST != SymbolRef::ST_Function)
2059 continue;
2060
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00002061 // Make sure the symbol is defined in this section.
Rafael Espindola80291272014-10-08 15:28:58 +00002062 bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]);
Owen Andersond9243c42011-10-17 21:37:35 +00002063 if (!containsSym)
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002064 continue;
2065
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00002066 // Start at the address of the symbol relative to the section's address.
Owen Andersond9243c42011-10-17 21:37:35 +00002067 uint64_t Start = 0;
Rafael Espindola80291272014-10-08 15:28:58 +00002068 uint64_t SectionAddress = Sections[SectIdx].getAddress();
Danil Malyshevcbe72fc2011-11-29 17:40:10 +00002069 Symbols[SymIdx].getAddress(Start);
Cameron Zwarich54478a52012-02-03 05:42:17 +00002070 Start -= SectionAddress;
Owen Andersond9243c42011-10-17 21:37:35 +00002071
Benjamin Kramer2ad2eb52011-09-20 17:53:01 +00002072 // Stop disassembling either at the beginning of the next symbol or at
2073 // the end of the section.
Kevin Enderbyedd58722012-05-15 18:57:14 +00002074 bool containsNextSym = false;
Owen Andersond9243c42011-10-17 21:37:35 +00002075 uint64_t NextSym = 0;
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002076 uint64_t NextSymIdx = SymIdx + 1;
Owen Andersond9243c42011-10-17 21:37:35 +00002077 while (Symbols.size() > NextSymIdx) {
2078 SymbolRef::Type NextSymType;
2079 Symbols[NextSymIdx].getType(NextSymType);
2080 if (NextSymType == SymbolRef::ST_Function) {
Rafael Espindola80291272014-10-08 15:28:58 +00002081 containsNextSym =
2082 Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]);
Danil Malyshevcbe72fc2011-11-29 17:40:10 +00002083 Symbols[NextSymIdx].getAddress(NextSym);
Cameron Zwarich54478a52012-02-03 05:42:17 +00002084 NextSym -= SectionAddress;
Owen Andersond9243c42011-10-17 21:37:35 +00002085 break;
2086 }
2087 ++NextSymIdx;
2088 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002089
Rafael Espindola80291272014-10-08 15:28:58 +00002090 uint64_t SectSize = Sections[SectIdx].getSize();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002091 uint64_t End = containsNextSym ? NextSym : SectSize;
Owen Andersond9243c42011-10-17 21:37:35 +00002092 uint64_t Size;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002093
2094 symbolTableWorked = true;
Rafael Espindolabd604f22014-11-07 00:52:15 +00002095
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002096 DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl();
2097 bool isThumb =
2098 (MachOOF->getSymbolFlags(Symb) & SymbolRef::SF_Thumb) && ThumbTarget;
2099
Ahmed Bougachaaa790682013-05-24 01:07:04 +00002100 outs() << SymName << ":\n";
2101 DILineInfo lastLine;
2102 for (uint64_t Index = Start; Index < End; Index += Size) {
2103 MCInst Inst;
Owen Andersond9243c42011-10-17 21:37:35 +00002104
Kevin Enderbybf246f52014-09-24 23:08:22 +00002105 uint64_t PC = SectAddress + Index;
2106 if (FullLeadingAddr) {
2107 if (MachOOF->is64Bit())
2108 outs() << format("%016" PRIx64, PC);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002109 else
Kevin Enderbybf246f52014-09-24 23:08:22 +00002110 outs() << format("%08" PRIx64, PC);
2111 } else {
2112 outs() << format("%8" PRIx64 ":", PC);
2113 }
2114 if (!NoShowRawInsn)
2115 outs() << "\t";
Kevin Enderby273ae012013-06-06 17:20:50 +00002116
2117 // Check the data in code table here to see if this is data not an
2118 // instruction to be disassembled.
2119 DiceTable Dice;
Kevin Enderbybf246f52014-09-24 23:08:22 +00002120 Dice.push_back(std::make_pair(PC, DiceRef()));
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002121 dice_table_iterator DTI =
2122 std::search(Dices.begin(), Dices.end(), Dice.begin(), Dice.end(),
2123 compareDiceTableEntries);
2124 if (DTI != Dices.end()) {
Kevin Enderby273ae012013-06-06 17:20:50 +00002125 uint16_t Length;
2126 DTI->second.getLength(Length);
Kevin Enderby273ae012013-06-06 17:20:50 +00002127 uint16_t Kind;
2128 DTI->second.getKind(Kind);
Aaron Ballman106fd7b2014-11-12 14:01:17 +00002129 Size = DumpDataInCode(reinterpret_cast<const char *>(Bytes.data()) +
2130 Index,
2131 Length, Kind);
Kevin Enderby930fdc72014-11-06 19:00:13 +00002132 if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) &&
2133 (PC == (DTI->first + Length - 1)) && (Length & 1))
2134 Size++;
Kevin Enderby273ae012013-06-06 17:20:50 +00002135 continue;
2136 }
2137
Kevin Enderbybf246f52014-09-24 23:08:22 +00002138 SmallVector<char, 64> AnnotationsBytes;
2139 raw_svector_ostream Annotations(AnnotationsBytes);
2140
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002141 bool gotInst;
2142 if (isThumb)
Rafael Espindola7fc5b872014-11-12 02:04:27 +00002143 gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index),
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002144 PC, DebugOut, Annotations);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002145 else
Rafael Espindola7fc5b872014-11-12 02:04:27 +00002146 gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC,
Kevin Enderbybf246f52014-09-24 23:08:22 +00002147 DebugOut, Annotations);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002148 if (gotInst) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002149 if (!NoShowRawInsn) {
Aaron Ballman106fd7b2014-11-12 14:01:17 +00002150 DumpBytes(StringRef(
2151 reinterpret_cast<const char *>(Bytes.data()) + Index, Size));
Kevin Enderbybf246f52014-09-24 23:08:22 +00002152 }
2153 formatted_raw_ostream FormattedOS(outs());
2154 Annotations.flush();
2155 StringRef AnnotationsStr = Annotations.str();
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002156 if (isThumb)
Kevin Enderbybf246f52014-09-24 23:08:22 +00002157 ThumbIP->printInst(&Inst, FormattedOS, AnnotationsStr);
Kevin Enderbyec5ca032014-08-18 20:21:02 +00002158 else
Kevin Enderbybf246f52014-09-24 23:08:22 +00002159 IP->printInst(&Inst, FormattedOS, AnnotationsStr);
2160 emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo);
Owen Andersond9243c42011-10-17 21:37:35 +00002161
Ahmed Bougachaaa790682013-05-24 01:07:04 +00002162 // Print debug info.
2163 if (diContext) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002164 DILineInfo dli = diContext->getLineInfoForAddress(PC);
Ahmed Bougachaaa790682013-05-24 01:07:04 +00002165 // Print valid line info if it changed.
Alexey Samsonovd0109992014-04-18 21:36:39 +00002166 if (dli != lastLine && dli.Line != 0)
2167 outs() << "\t## " << dli.FileName << ':' << dli.Line << ':'
2168 << dli.Column;
Ahmed Bougachaaa790682013-05-24 01:07:04 +00002169 lastLine = dli;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002170 }
Ahmed Bougachaaa790682013-05-24 01:07:04 +00002171 outs() << "\n";
2172 } else {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002173 unsigned int Arch = MachOOF->getArch();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002174 if (Arch == Triple::x86_64 || Arch == Triple::x86) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002175 outs() << format("\t.byte 0x%02x #bad opcode\n",
2176 *(Bytes.data() + Index) & 0xff);
2177 Size = 1; // skip exactly one illegible byte and move on.
Kevin Enderbyae3c1262014-11-14 21:52:18 +00002178 } else if (Arch == Triple::aarch64) {
2179 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) |
2180 (*(Bytes.data() + Index + 1) & 0xff) << 8 |
2181 (*(Bytes.data() + Index + 2) & 0xff) << 16 |
2182 (*(Bytes.data() + Index + 3) & 0xff) << 24;
2183 outs() << format("\t.long\t0x%08x\n", opcode);
2184 Size = 4;
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002185 } else {
2186 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
2187 if (Size == 0)
2188 Size = 1; // skip illegible bytes
2189 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002190 }
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002191 }
2192 }
Ahmed Bougachaaa790682013-05-24 01:07:04 +00002193 if (!symbolTableWorked) {
Rafael Espindola80291272014-10-08 15:28:58 +00002194 // Reading the symbol table didn't work, disassemble the whole section.
2195 uint64_t SectAddress = Sections[SectIdx].getAddress();
2196 uint64_t SectSize = Sections[SectIdx].getSize();
Kevin Enderbybadd1002012-05-18 00:13:56 +00002197 uint64_t InstSize;
2198 for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
Bill Wendling4e68e062012-07-19 00:17:40 +00002199 MCInst Inst;
Kevin Enderbybadd1002012-05-18 00:13:56 +00002200
Kevin Enderbybf246f52014-09-24 23:08:22 +00002201 uint64_t PC = SectAddress + Index;
Rafael Espindola7fc5b872014-11-12 02:04:27 +00002202 if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC,
2203 DebugOut, nulls())) {
Kevin Enderbybf246f52014-09-24 23:08:22 +00002204 if (FullLeadingAddr) {
2205 if (MachOOF->is64Bit())
2206 outs() << format("%016" PRIx64, PC);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002207 else
Kevin Enderbybf246f52014-09-24 23:08:22 +00002208 outs() << format("%08" PRIx64, PC);
2209 } else {
2210 outs() << format("%8" PRIx64 ":", PC);
2211 }
2212 if (!NoShowRawInsn) {
2213 outs() << "\t";
Aaron Ballman106fd7b2014-11-12 14:01:17 +00002214 DumpBytes(
2215 StringRef(reinterpret_cast<const char *>(Bytes.data()) + Index,
2216 InstSize));
Kevin Enderbybf246f52014-09-24 23:08:22 +00002217 }
Bill Wendling4e68e062012-07-19 00:17:40 +00002218 IP->printInst(&Inst, outs(), "");
2219 outs() << "\n";
2220 } else {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002221 unsigned int Arch = MachOOF->getArch();
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002222 if (Arch == Triple::x86_64 || Arch == Triple::x86) {
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002223 outs() << format("\t.byte 0x%02x #bad opcode\n",
2224 *(Bytes.data() + Index) & 0xff);
2225 InstSize = 1; // skip exactly one illegible byte and move on.
2226 } else {
2227 errs() << "llvm-objdump: warning: invalid instruction encoding\n";
2228 if (InstSize == 0)
2229 InstSize = 1; // skip illegible bytes
2230 }
Bill Wendling4e68e062012-07-19 00:17:40 +00002231 }
Kevin Enderbybadd1002012-05-18 00:13:56 +00002232 }
2233 }
Kevin Enderby3f0ffab2014-12-03 22:29:40 +00002234 // The TripleName's need to be reset if we are called again for a different
2235 // archtecture.
2236 TripleName = "";
2237 ThumbTripleName = "";
2238
Kevin Enderby6f326ce2014-10-23 19:37:31 +00002239 if (SymbolizerInfo.method != nullptr)
2240 free(SymbolizerInfo.method);
Kevin Enderby04bf6932014-10-28 23:39:46 +00002241 if (SymbolizerInfo.demangled_name != nullptr)
2242 free(SymbolizerInfo.demangled_name);
Kevin Enderby078be602014-10-23 19:53:12 +00002243 if (SymbolizerInfo.bindtable != nullptr)
2244 delete SymbolizerInfo.bindtable;
Kevin Enderby930fdc72014-11-06 19:00:13 +00002245 if (ThumbSymbolizerInfo.method != nullptr)
2246 free(ThumbSymbolizerInfo.method);
2247 if (ThumbSymbolizerInfo.demangled_name != nullptr)
2248 free(ThumbSymbolizerInfo.demangled_name);
2249 if (ThumbSymbolizerInfo.bindtable != nullptr)
2250 delete ThumbSymbolizerInfo.bindtable;
Benjamin Kramer43a772e2011-09-19 17:56:04 +00002251 }
2252}
Tim Northover4bd286a2014-08-01 13:07:19 +00002253
Tim Northover39c70bb2014-08-12 11:52:59 +00002254//===----------------------------------------------------------------------===//
2255// __compact_unwind section dumping
2256//===----------------------------------------------------------------------===//
2257
Tim Northover4bd286a2014-08-01 13:07:19 +00002258namespace {
Tim Northover39c70bb2014-08-12 11:52:59 +00002259
2260template <typename T> static uint64_t readNext(const char *&Buf) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002261 using llvm::support::little;
2262 using llvm::support::unaligned;
Tim Northover39c70bb2014-08-12 11:52:59 +00002263
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002264 uint64_t Val = support::endian::read<T, little, unaligned>(Buf);
2265 Buf += sizeof(T);
2266 return Val;
2267}
Tim Northover39c70bb2014-08-12 11:52:59 +00002268
Tim Northover4bd286a2014-08-01 13:07:19 +00002269struct CompactUnwindEntry {
2270 uint32_t OffsetInSection;
2271
2272 uint64_t FunctionAddr;
2273 uint32_t Length;
2274 uint32_t CompactEncoding;
2275 uint64_t PersonalityAddr;
2276 uint64_t LSDAAddr;
2277
2278 RelocationRef FunctionReloc;
2279 RelocationRef PersonalityReloc;
2280 RelocationRef LSDAReloc;
2281
2282 CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64)
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002283 : OffsetInSection(Offset) {
Tim Northover4bd286a2014-08-01 13:07:19 +00002284 if (Is64)
2285 read<uint64_t>(Contents.data() + Offset);
2286 else
2287 read<uint32_t>(Contents.data() + Offset);
2288 }
2289
2290private:
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002291 template <typename UIntPtr> void read(const char *Buf) {
Tim Northover4bd286a2014-08-01 13:07:19 +00002292 FunctionAddr = readNext<UIntPtr>(Buf);
2293 Length = readNext<uint32_t>(Buf);
2294 CompactEncoding = readNext<uint32_t>(Buf);
2295 PersonalityAddr = readNext<UIntPtr>(Buf);
2296 LSDAAddr = readNext<UIntPtr>(Buf);
2297 }
2298};
2299}
2300
2301/// Given a relocation from __compact_unwind, consisting of the RelocationRef
2302/// and data being relocated, determine the best base Name and Addend to use for
2303/// display purposes.
2304///
2305/// 1. An Extern relocation will directly reference a symbol (and the data is
2306/// then already an addend), so use that.
2307/// 2. Otherwise the data is an offset in the object file's layout; try to find
2308// a symbol before it in the same section, and use the offset from there.
2309/// 3. Finally, if all that fails, fall back to an offset from the start of the
2310/// referenced section.
2311static void findUnwindRelocNameAddend(const MachOObjectFile *Obj,
2312 std::map<uint64_t, SymbolRef> &Symbols,
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002313 const RelocationRef &Reloc, uint64_t Addr,
Tim Northover4bd286a2014-08-01 13:07:19 +00002314 StringRef &Name, uint64_t &Addend) {
2315 if (Reloc.getSymbol() != Obj->symbol_end()) {
2316 Reloc.getSymbol()->getName(Name);
2317 Addend = Addr;
2318 return;
2319 }
2320
2321 auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl());
2322 SectionRef RelocSection = Obj->getRelocationSection(RE);
2323
Rafael Espindola80291272014-10-08 15:28:58 +00002324 uint64_t SectionAddr = RelocSection.getAddress();
Tim Northover4bd286a2014-08-01 13:07:19 +00002325
2326 auto Sym = Symbols.upper_bound(Addr);
2327 if (Sym == Symbols.begin()) {
2328 // The first symbol in the object is after this reference, the best we can
2329 // do is section-relative notation.
2330 RelocSection.getName(Name);
2331 Addend = Addr - SectionAddr;
2332 return;
2333 }
2334
2335 // Go back one so that SymbolAddress <= Addr.
2336 --Sym;
2337
2338 section_iterator SymSection = Obj->section_end();
2339 Sym->second.getSection(SymSection);
2340 if (RelocSection == *SymSection) {
2341 // There's a valid symbol in the same section before this reference.
2342 Sym->second.getName(Name);
2343 Addend = Addr - Sym->first;
2344 return;
2345 }
2346
2347 // There is a symbol before this reference, but it's in a different
2348 // section. Probably not helpful to mention it, so use the section name.
2349 RelocSection.getName(Name);
2350 Addend = Addr - SectionAddr;
2351}
2352
2353static void printUnwindRelocDest(const MachOObjectFile *Obj,
2354 std::map<uint64_t, SymbolRef> &Symbols,
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002355 const RelocationRef &Reloc, uint64_t Addr) {
Tim Northover4bd286a2014-08-01 13:07:19 +00002356 StringRef Name;
2357 uint64_t Addend;
2358
Tim Northover0b0add52014-09-09 10:45:06 +00002359 if (!Reloc.getObjectFile())
2360 return;
2361
Tim Northover4bd286a2014-08-01 13:07:19 +00002362 findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend);
2363
2364 outs() << Name;
2365 if (Addend)
Tim Northover63a25622014-08-11 09:14:06 +00002366 outs() << " + " << format("0x%" PRIx64, Addend);
Tim Northover4bd286a2014-08-01 13:07:19 +00002367}
2368
2369static void
2370printMachOCompactUnwindSection(const MachOObjectFile *Obj,
2371 std::map<uint64_t, SymbolRef> &Symbols,
2372 const SectionRef &CompactUnwind) {
2373
2374 assert(Obj->isLittleEndian() &&
2375 "There should not be a big-endian .o with __compact_unwind");
2376
2377 bool Is64 = Obj->is64Bit();
2378 uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t);
2379 uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t);
2380
2381 StringRef Contents;
2382 CompactUnwind.getContents(Contents);
2383
2384 SmallVector<CompactUnwindEntry, 4> CompactUnwinds;
2385
2386 // First populate the initial raw offsets, encodings and so on from the entry.
2387 for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) {
2388 CompactUnwindEntry Entry(Contents.data(), Offset, Is64);
2389 CompactUnwinds.push_back(Entry);
2390 }
2391
2392 // Next we need to look at the relocations to find out what objects are
2393 // actually being referred to.
2394 for (const RelocationRef &Reloc : CompactUnwind.relocations()) {
2395 uint64_t RelocAddress;
2396 Reloc.getOffset(RelocAddress);
2397
2398 uint32_t EntryIdx = RelocAddress / EntrySize;
2399 uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize;
2400 CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx];
2401
2402 if (OffsetInEntry == 0)
2403 Entry.FunctionReloc = Reloc;
2404 else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t))
2405 Entry.PersonalityReloc = Reloc;
2406 else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t))
2407 Entry.LSDAReloc = Reloc;
2408 else
2409 llvm_unreachable("Unexpected relocation in __compact_unwind section");
2410 }
2411
2412 // Finally, we're ready to print the data we've gathered.
2413 outs() << "Contents of __compact_unwind section:\n";
2414 for (auto &Entry : CompactUnwinds) {
Tim Northover06af2602014-08-08 12:08:51 +00002415 outs() << " Entry at offset "
2416 << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n";
Tim Northover4bd286a2014-08-01 13:07:19 +00002417
2418 // 1. Start of the region this entry applies to.
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002419 outs() << " start: " << format("0x%" PRIx64,
2420 Entry.FunctionAddr) << ' ';
2421 printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr);
Tim Northover4bd286a2014-08-01 13:07:19 +00002422 outs() << '\n';
2423
2424 // 2. Length of the region this entry applies to.
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002425 outs() << " length: " << format("0x%" PRIx32, Entry.Length)
2426 << '\n';
Tim Northover4bd286a2014-08-01 13:07:19 +00002427 // 3. The 32-bit compact encoding.
2428 outs() << " compact encoding: "
Tim Northoverb911bf82014-08-08 12:00:09 +00002429 << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n';
Tim Northover4bd286a2014-08-01 13:07:19 +00002430
2431 // 4. The personality function, if present.
2432 if (Entry.PersonalityReloc.getObjectFile()) {
2433 outs() << " personality function: "
Tim Northoverb911bf82014-08-08 12:00:09 +00002434 << format("0x%" PRIx64, Entry.PersonalityAddr) << ' ';
Tim Northover4bd286a2014-08-01 13:07:19 +00002435 printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc,
2436 Entry.PersonalityAddr);
2437 outs() << '\n';
2438 }
2439
2440 // 5. This entry's language-specific data area.
2441 if (Entry.LSDAReloc.getObjectFile()) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002442 outs() << " LSDA: " << format("0x%" PRIx64,
2443 Entry.LSDAAddr) << ' ';
Tim Northover4bd286a2014-08-01 13:07:19 +00002444 printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr);
2445 outs() << '\n';
2446 }
2447 }
2448}
2449
Tim Northover39c70bb2014-08-12 11:52:59 +00002450//===----------------------------------------------------------------------===//
2451// __unwind_info section dumping
2452//===----------------------------------------------------------------------===//
2453
2454static void printRegularSecondLevelUnwindPage(const char *PageStart) {
2455 const char *Pos = PageStart;
2456 uint32_t Kind = readNext<uint32_t>(Pos);
2457 (void)Kind;
2458 assert(Kind == 2 && "kind for a regular 2nd level index should be 2");
2459
2460 uint16_t EntriesStart = readNext<uint16_t>(Pos);
2461 uint16_t NumEntries = readNext<uint16_t>(Pos);
2462
2463 Pos = PageStart + EntriesStart;
2464 for (unsigned i = 0; i < NumEntries; ++i) {
2465 uint32_t FunctionOffset = readNext<uint32_t>(Pos);
2466 uint32_t Encoding = readNext<uint32_t>(Pos);
2467
2468 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002469 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
2470 << ", "
2471 << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00002472 }
2473}
2474
2475static void printCompressedSecondLevelUnwindPage(
2476 const char *PageStart, uint32_t FunctionBase,
2477 const SmallVectorImpl<uint32_t> &CommonEncodings) {
2478 const char *Pos = PageStart;
2479 uint32_t Kind = readNext<uint32_t>(Pos);
2480 (void)Kind;
2481 assert(Kind == 3 && "kind for a compressed 2nd level index should be 3");
2482
2483 uint16_t EntriesStart = readNext<uint16_t>(Pos);
2484 uint16_t NumEntries = readNext<uint16_t>(Pos);
2485
2486 uint16_t EncodingsStart = readNext<uint16_t>(Pos);
2487 readNext<uint16_t>(Pos);
Aaron Ballman80930af2014-08-14 13:53:19 +00002488 const auto *PageEncodings = reinterpret_cast<const support::ulittle32_t *>(
2489 PageStart + EncodingsStart);
Tim Northover39c70bb2014-08-12 11:52:59 +00002490
2491 Pos = PageStart + EntriesStart;
2492 for (unsigned i = 0; i < NumEntries; ++i) {
2493 uint32_t Entry = readNext<uint32_t>(Pos);
2494 uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff);
2495 uint32_t EncodingIdx = Entry >> 24;
2496
2497 uint32_t Encoding;
2498 if (EncodingIdx < CommonEncodings.size())
2499 Encoding = CommonEncodings[EncodingIdx];
2500 else
2501 Encoding = PageEncodings[EncodingIdx - CommonEncodings.size()];
2502
2503 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002504 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
2505 << ", "
2506 << "encoding[" << EncodingIdx
2507 << "]=" << format("0x%08" PRIx32, Encoding) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00002508 }
2509}
2510
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002511static void printMachOUnwindInfoSection(const MachOObjectFile *Obj,
2512 std::map<uint64_t, SymbolRef> &Symbols,
2513 const SectionRef &UnwindInfo) {
Tim Northover39c70bb2014-08-12 11:52:59 +00002514
2515 assert(Obj->isLittleEndian() &&
2516 "There should not be a big-endian .o with __unwind_info");
2517
2518 outs() << "Contents of __unwind_info section:\n";
2519
2520 StringRef Contents;
2521 UnwindInfo.getContents(Contents);
2522 const char *Pos = Contents.data();
2523
2524 //===----------------------------------
2525 // Section header
2526 //===----------------------------------
2527
2528 uint32_t Version = readNext<uint32_t>(Pos);
2529 outs() << " Version: "
2530 << format("0x%" PRIx32, Version) << '\n';
2531 assert(Version == 1 && "only understand version 1");
2532
2533 uint32_t CommonEncodingsStart = readNext<uint32_t>(Pos);
2534 outs() << " Common encodings array section offset: "
2535 << format("0x%" PRIx32, CommonEncodingsStart) << '\n';
2536 uint32_t NumCommonEncodings = readNext<uint32_t>(Pos);
2537 outs() << " Number of common encodings in array: "
2538 << format("0x%" PRIx32, NumCommonEncodings) << '\n';
2539
2540 uint32_t PersonalitiesStart = readNext<uint32_t>(Pos);
2541 outs() << " Personality function array section offset: "
2542 << format("0x%" PRIx32, PersonalitiesStart) << '\n';
2543 uint32_t NumPersonalities = readNext<uint32_t>(Pos);
2544 outs() << " Number of personality functions in array: "
2545 << format("0x%" PRIx32, NumPersonalities) << '\n';
2546
2547 uint32_t IndicesStart = readNext<uint32_t>(Pos);
2548 outs() << " Index array section offset: "
2549 << format("0x%" PRIx32, IndicesStart) << '\n';
2550 uint32_t NumIndices = readNext<uint32_t>(Pos);
2551 outs() << " Number of indices in array: "
2552 << format("0x%" PRIx32, NumIndices) << '\n';
2553
2554 //===----------------------------------
2555 // A shared list of common encodings
2556 //===----------------------------------
2557
2558 // These occupy indices in the range [0, N] whenever an encoding is referenced
2559 // from a compressed 2nd level index table. In practice the linker only
2560 // creates ~128 of these, so that indices are available to embed encodings in
2561 // the 2nd level index.
2562
2563 SmallVector<uint32_t, 64> CommonEncodings;
2564 outs() << " Common encodings: (count = " << NumCommonEncodings << ")\n";
2565 Pos = Contents.data() + CommonEncodingsStart;
2566 for (unsigned i = 0; i < NumCommonEncodings; ++i) {
2567 uint32_t Encoding = readNext<uint32_t>(Pos);
2568 CommonEncodings.push_back(Encoding);
2569
2570 outs() << " encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding)
2571 << '\n';
2572 }
2573
Tim Northover39c70bb2014-08-12 11:52:59 +00002574 //===----------------------------------
2575 // Personality functions used in this executable
2576 //===----------------------------------
2577
2578 // There should be only a handful of these (one per source language,
2579 // roughly). Particularly since they only get 2 bits in the compact encoding.
2580
2581 outs() << " Personality functions: (count = " << NumPersonalities << ")\n";
2582 Pos = Contents.data() + PersonalitiesStart;
2583 for (unsigned i = 0; i < NumPersonalities; ++i) {
2584 uint32_t PersonalityFn = readNext<uint32_t>(Pos);
2585 outs() << " personality[" << i + 1
2586 << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n';
2587 }
2588
2589 //===----------------------------------
2590 // The level 1 index entries
2591 //===----------------------------------
2592
2593 // These specify an approximate place to start searching for the more detailed
2594 // information, sorted by PC.
2595
2596 struct IndexEntry {
2597 uint32_t FunctionOffset;
2598 uint32_t SecondLevelPageStart;
2599 uint32_t LSDAStart;
2600 };
2601
2602 SmallVector<IndexEntry, 4> IndexEntries;
2603
2604 outs() << " Top level indices: (count = " << NumIndices << ")\n";
2605 Pos = Contents.data() + IndicesStart;
2606 for (unsigned i = 0; i < NumIndices; ++i) {
2607 IndexEntry Entry;
2608
2609 Entry.FunctionOffset = readNext<uint32_t>(Pos);
2610 Entry.SecondLevelPageStart = readNext<uint32_t>(Pos);
2611 Entry.LSDAStart = readNext<uint32_t>(Pos);
2612 IndexEntries.push_back(Entry);
2613
2614 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002615 << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset)
2616 << ", "
Tim Northover39c70bb2014-08-12 11:52:59 +00002617 << "2nd level page offset="
2618 << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002619 << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00002620 }
2621
Tim Northover39c70bb2014-08-12 11:52:59 +00002622 //===----------------------------------
2623 // Next come the LSDA tables
2624 //===----------------------------------
2625
2626 // The LSDA layout is rather implicit: it's a contiguous array of entries from
2627 // the first top-level index's LSDAOffset to the last (sentinel).
2628
2629 outs() << " LSDA descriptors:\n";
2630 Pos = Contents.data() + IndexEntries[0].LSDAStart;
2631 int NumLSDAs = (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) /
2632 (2 * sizeof(uint32_t));
2633 for (int i = 0; i < NumLSDAs; ++i) {
2634 uint32_t FunctionOffset = readNext<uint32_t>(Pos);
2635 uint32_t LSDAOffset = readNext<uint32_t>(Pos);
2636 outs() << " [" << i << "]: "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00002637 << "function offset=" << format("0x%08" PRIx32, FunctionOffset)
2638 << ", "
2639 << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n';
Tim Northover39c70bb2014-08-12 11:52:59 +00002640 }
2641
2642 //===----------------------------------
2643 // Finally, the 2nd level indices
2644 //===----------------------------------
2645
2646 // Generally these are 4K in size, and have 2 possible forms:
2647 // + Regular stores up to 511 entries with disparate encodings
2648 // + Compressed stores up to 1021 entries if few enough compact encoding
2649 // values are used.
2650 outs() << " Second level indices:\n";
2651 for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) {
2652 // The final sentinel top-level index has no associated 2nd level page
2653 if (IndexEntries[i].SecondLevelPageStart == 0)
2654 break;
2655
2656 outs() << " Second level index[" << i << "]: "
2657 << "offset in section="
2658 << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart)
2659 << ", "
2660 << "base function offset="
2661 << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n';
2662
2663 Pos = Contents.data() + IndexEntries[i].SecondLevelPageStart;
Aaron Ballman80930af2014-08-14 13:53:19 +00002664 uint32_t Kind = *reinterpret_cast<const support::ulittle32_t *>(Pos);
Tim Northover39c70bb2014-08-12 11:52:59 +00002665 if (Kind == 2)
2666 printRegularSecondLevelUnwindPage(Pos);
2667 else if (Kind == 3)
2668 printCompressedSecondLevelUnwindPage(Pos, IndexEntries[i].FunctionOffset,
2669 CommonEncodings);
2670 else
2671 llvm_unreachable("Do not know how to print this kind of 2nd level page");
Tim Northover39c70bb2014-08-12 11:52:59 +00002672 }
2673}
2674
Tim Northover4bd286a2014-08-01 13:07:19 +00002675void llvm::printMachOUnwindInfo(const MachOObjectFile *Obj) {
2676 std::map<uint64_t, SymbolRef> Symbols;
2677 for (const SymbolRef &SymRef : Obj->symbols()) {
2678 // Discard any undefined or absolute symbols. They're not going to take part
2679 // in the convenience lookup for unwind info and just take up resources.
2680 section_iterator Section = Obj->section_end();
2681 SymRef.getSection(Section);
2682 if (Section == Obj->section_end())
2683 continue;
2684
2685 uint64_t Addr;
2686 SymRef.getAddress(Addr);
2687 Symbols.insert(std::make_pair(Addr, SymRef));
2688 }
2689
2690 for (const SectionRef &Section : Obj->sections()) {
2691 StringRef SectName;
2692 Section.getName(SectName);
2693 if (SectName == "__compact_unwind")
2694 printMachOCompactUnwindSection(Obj, Symbols, Section);
2695 else if (SectName == "__unwind_info")
Tim Northover39c70bb2014-08-12 11:52:59 +00002696 printMachOUnwindInfoSection(Obj, Symbols, Section);
Tim Northover4bd286a2014-08-01 13:07:19 +00002697 else if (SectName == "__eh_frame")
2698 outs() << "llvm-objdump: warning: unhandled __eh_frame section\n";
Tim Northover4bd286a2014-08-01 13:07:19 +00002699 }
2700}
Kevin Enderbyb76d3862014-08-22 20:35:18 +00002701
2702static void PrintMachHeader(uint32_t magic, uint32_t cputype,
2703 uint32_t cpusubtype, uint32_t filetype,
2704 uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags,
2705 bool verbose) {
2706 outs() << "Mach header\n";
2707 outs() << " magic cputype cpusubtype caps filetype ncmds "
2708 "sizeofcmds flags\n";
2709 if (verbose) {
2710 if (magic == MachO::MH_MAGIC)
2711 outs() << " MH_MAGIC";
2712 else if (magic == MachO::MH_MAGIC_64)
2713 outs() << "MH_MAGIC_64";
2714 else
2715 outs() << format(" 0x%08" PRIx32, magic);
2716 switch (cputype) {
2717 case MachO::CPU_TYPE_I386:
2718 outs() << " I386";
2719 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2720 case MachO::CPU_SUBTYPE_I386_ALL:
2721 outs() << " ALL";
2722 break;
2723 default:
2724 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2725 break;
2726 }
2727 break;
2728 case MachO::CPU_TYPE_X86_64:
2729 outs() << " X86_64";
2730 case MachO::CPU_SUBTYPE_X86_64_ALL:
2731 outs() << " ALL";
2732 break;
2733 case MachO::CPU_SUBTYPE_X86_64_H:
2734 outs() << " Haswell";
Aaron Ballman9d515ff2014-08-24 13:25:16 +00002735 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
Kevin Enderbyb76d3862014-08-22 20:35:18 +00002736 break;
2737 case MachO::CPU_TYPE_ARM:
2738 outs() << " ARM";
2739 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2740 case MachO::CPU_SUBTYPE_ARM_ALL:
2741 outs() << " ALL";
2742 break;
2743 case MachO::CPU_SUBTYPE_ARM_V4T:
2744 outs() << " V4T";
2745 break;
2746 case MachO::CPU_SUBTYPE_ARM_V5TEJ:
2747 outs() << " V5TEJ";
2748 break;
2749 case MachO::CPU_SUBTYPE_ARM_XSCALE:
2750 outs() << " XSCALE";
2751 break;
2752 case MachO::CPU_SUBTYPE_ARM_V6:
2753 outs() << " V6";
2754 break;
2755 case MachO::CPU_SUBTYPE_ARM_V6M:
2756 outs() << " V6M";
2757 break;
2758 case MachO::CPU_SUBTYPE_ARM_V7:
2759 outs() << " V7";
2760 break;
2761 case MachO::CPU_SUBTYPE_ARM_V7EM:
2762 outs() << " V7EM";
2763 break;
2764 case MachO::CPU_SUBTYPE_ARM_V7K:
2765 outs() << " V7K";
2766 break;
2767 case MachO::CPU_SUBTYPE_ARM_V7M:
2768 outs() << " V7M";
2769 break;
2770 case MachO::CPU_SUBTYPE_ARM_V7S:
2771 outs() << " V7S";
2772 break;
2773 default:
2774 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2775 break;
2776 }
2777 break;
2778 case MachO::CPU_TYPE_ARM64:
2779 outs() << " ARM64";
2780 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2781 case MachO::CPU_SUBTYPE_ARM64_ALL:
2782 outs() << " ALL";
2783 break;
2784 default:
2785 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2786 break;
2787 }
2788 break;
2789 case MachO::CPU_TYPE_POWERPC:
2790 outs() << " PPC";
2791 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2792 case MachO::CPU_SUBTYPE_POWERPC_ALL:
2793 outs() << " ALL";
2794 break;
2795 default:
2796 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2797 break;
2798 }
2799 break;
2800 case MachO::CPU_TYPE_POWERPC64:
2801 outs() << " PPC64";
2802 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) {
2803 case MachO::CPU_SUBTYPE_POWERPC_ALL:
2804 outs() << " ALL";
2805 break;
2806 default:
2807 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2808 break;
2809 }
2810 break;
2811 }
2812 if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00002813 outs() << " LIB64";
Kevin Enderbyb76d3862014-08-22 20:35:18 +00002814 } else {
2815 outs() << format(" 0x%02" PRIx32,
2816 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
2817 }
2818 switch (filetype) {
2819 case MachO::MH_OBJECT:
2820 outs() << " OBJECT";
2821 break;
2822 case MachO::MH_EXECUTE:
2823 outs() << " EXECUTE";
2824 break;
2825 case MachO::MH_FVMLIB:
2826 outs() << " FVMLIB";
2827 break;
2828 case MachO::MH_CORE:
2829 outs() << " CORE";
2830 break;
2831 case MachO::MH_PRELOAD:
2832 outs() << " PRELOAD";
2833 break;
2834 case MachO::MH_DYLIB:
2835 outs() << " DYLIB";
2836 break;
2837 case MachO::MH_DYLIB_STUB:
2838 outs() << " DYLIB_STUB";
2839 break;
2840 case MachO::MH_DYLINKER:
2841 outs() << " DYLINKER";
2842 break;
2843 case MachO::MH_BUNDLE:
2844 outs() << " BUNDLE";
2845 break;
2846 case MachO::MH_DSYM:
2847 outs() << " DSYM";
2848 break;
2849 case MachO::MH_KEXT_BUNDLE:
2850 outs() << " KEXTBUNDLE";
2851 break;
2852 default:
2853 outs() << format(" %10u", filetype);
2854 break;
2855 }
2856 outs() << format(" %5u", ncmds);
2857 outs() << format(" %10u", sizeofcmds);
2858 uint32_t f = flags;
2859 if (f & MachO::MH_NOUNDEFS) {
2860 outs() << " NOUNDEFS";
2861 f &= ~MachO::MH_NOUNDEFS;
2862 }
2863 if (f & MachO::MH_INCRLINK) {
2864 outs() << " INCRLINK";
2865 f &= ~MachO::MH_INCRLINK;
2866 }
2867 if (f & MachO::MH_DYLDLINK) {
2868 outs() << " DYLDLINK";
2869 f &= ~MachO::MH_DYLDLINK;
2870 }
2871 if (f & MachO::MH_BINDATLOAD) {
2872 outs() << " BINDATLOAD";
2873 f &= ~MachO::MH_BINDATLOAD;
2874 }
2875 if (f & MachO::MH_PREBOUND) {
2876 outs() << " PREBOUND";
2877 f &= ~MachO::MH_PREBOUND;
2878 }
2879 if (f & MachO::MH_SPLIT_SEGS) {
2880 outs() << " SPLIT_SEGS";
2881 f &= ~MachO::MH_SPLIT_SEGS;
2882 }
2883 if (f & MachO::MH_LAZY_INIT) {
2884 outs() << " LAZY_INIT";
2885 f &= ~MachO::MH_LAZY_INIT;
2886 }
2887 if (f & MachO::MH_TWOLEVEL) {
2888 outs() << " TWOLEVEL";
2889 f &= ~MachO::MH_TWOLEVEL;
2890 }
2891 if (f & MachO::MH_FORCE_FLAT) {
2892 outs() << " FORCE_FLAT";
2893 f &= ~MachO::MH_FORCE_FLAT;
2894 }
2895 if (f & MachO::MH_NOMULTIDEFS) {
2896 outs() << " NOMULTIDEFS";
2897 f &= ~MachO::MH_NOMULTIDEFS;
2898 }
2899 if (f & MachO::MH_NOFIXPREBINDING) {
2900 outs() << " NOFIXPREBINDING";
2901 f &= ~MachO::MH_NOFIXPREBINDING;
2902 }
2903 if (f & MachO::MH_PREBINDABLE) {
2904 outs() << " PREBINDABLE";
2905 f &= ~MachO::MH_PREBINDABLE;
2906 }
2907 if (f & MachO::MH_ALLMODSBOUND) {
2908 outs() << " ALLMODSBOUND";
2909 f &= ~MachO::MH_ALLMODSBOUND;
2910 }
2911 if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) {
2912 outs() << " SUBSECTIONS_VIA_SYMBOLS";
2913 f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
2914 }
2915 if (f & MachO::MH_CANONICAL) {
2916 outs() << " CANONICAL";
2917 f &= ~MachO::MH_CANONICAL;
2918 }
2919 if (f & MachO::MH_WEAK_DEFINES) {
2920 outs() << " WEAK_DEFINES";
2921 f &= ~MachO::MH_WEAK_DEFINES;
2922 }
2923 if (f & MachO::MH_BINDS_TO_WEAK) {
2924 outs() << " BINDS_TO_WEAK";
2925 f &= ~MachO::MH_BINDS_TO_WEAK;
2926 }
2927 if (f & MachO::MH_ALLOW_STACK_EXECUTION) {
2928 outs() << " ALLOW_STACK_EXECUTION";
2929 f &= ~MachO::MH_ALLOW_STACK_EXECUTION;
2930 }
2931 if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) {
2932 outs() << " DEAD_STRIPPABLE_DYLIB";
2933 f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB;
2934 }
2935 if (f & MachO::MH_PIE) {
2936 outs() << " PIE";
2937 f &= ~MachO::MH_PIE;
2938 }
2939 if (f & MachO::MH_NO_REEXPORTED_DYLIBS) {
2940 outs() << " NO_REEXPORTED_DYLIBS";
2941 f &= ~MachO::MH_NO_REEXPORTED_DYLIBS;
2942 }
2943 if (f & MachO::MH_HAS_TLV_DESCRIPTORS) {
2944 outs() << " MH_HAS_TLV_DESCRIPTORS";
2945 f &= ~MachO::MH_HAS_TLV_DESCRIPTORS;
2946 }
2947 if (f & MachO::MH_NO_HEAP_EXECUTION) {
2948 outs() << " MH_NO_HEAP_EXECUTION";
2949 f &= ~MachO::MH_NO_HEAP_EXECUTION;
2950 }
2951 if (f & MachO::MH_APP_EXTENSION_SAFE) {
2952 outs() << " APP_EXTENSION_SAFE";
2953 f &= ~MachO::MH_APP_EXTENSION_SAFE;
2954 }
2955 if (f != 0 || flags == 0)
2956 outs() << format(" 0x%08" PRIx32, f);
2957 } else {
2958 outs() << format(" 0x%08" PRIx32, magic);
2959 outs() << format(" %7d", cputype);
2960 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK);
2961 outs() << format(" 0x%02" PRIx32,
2962 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24);
2963 outs() << format(" %10u", filetype);
2964 outs() << format(" %5u", ncmds);
2965 outs() << format(" %10u", sizeofcmds);
2966 outs() << format(" 0x%08" PRIx32, flags);
2967 }
2968 outs() << "\n";
2969}
2970
Kevin Enderby956366c2014-08-29 22:30:52 +00002971static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize,
2972 StringRef SegName, uint64_t vmaddr,
2973 uint64_t vmsize, uint64_t fileoff,
2974 uint64_t filesize, uint32_t maxprot,
2975 uint32_t initprot, uint32_t nsects,
2976 uint32_t flags, uint32_t object_size,
2977 bool verbose) {
2978 uint64_t expected_cmdsize;
2979 if (cmd == MachO::LC_SEGMENT) {
2980 outs() << " cmd LC_SEGMENT\n";
2981 expected_cmdsize = nsects;
2982 expected_cmdsize *= sizeof(struct MachO::section);
2983 expected_cmdsize += sizeof(struct MachO::segment_command);
2984 } else {
2985 outs() << " cmd LC_SEGMENT_64\n";
2986 expected_cmdsize = nsects;
2987 expected_cmdsize *= sizeof(struct MachO::section_64);
2988 expected_cmdsize += sizeof(struct MachO::segment_command_64);
2989 }
2990 outs() << " cmdsize " << cmdsize;
2991 if (cmdsize != expected_cmdsize)
2992 outs() << " Inconsistent size\n";
2993 else
2994 outs() << "\n";
2995 outs() << " segname " << SegName << "\n";
2996 if (cmd == MachO::LC_SEGMENT_64) {
2997 outs() << " vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n";
2998 outs() << " vmsize " << format("0x%016" PRIx64, vmsize) << "\n";
2999 } else {
3000 outs() << " vmaddr " << format("0x%08" PRIx32, vmaddr) << "\n";
3001 outs() << " vmsize " << format("0x%08" PRIx32, vmsize) << "\n";
3002 }
3003 outs() << " fileoff " << fileoff;
3004 if (fileoff > object_size)
3005 outs() << " (past end of file)\n";
3006 else
3007 outs() << "\n";
3008 outs() << " filesize " << filesize;
3009 if (fileoff + filesize > object_size)
3010 outs() << " (past end of file)\n";
3011 else
3012 outs() << "\n";
3013 if (verbose) {
3014 if ((maxprot &
3015 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
3016 MachO::VM_PROT_EXECUTE)) != 0)
3017 outs() << " maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n";
3018 else {
3019 if (maxprot & MachO::VM_PROT_READ)
3020 outs() << " maxprot r";
3021 else
3022 outs() << " maxprot -";
3023 if (maxprot & MachO::VM_PROT_WRITE)
3024 outs() << "w";
3025 else
3026 outs() << "-";
3027 if (maxprot & MachO::VM_PROT_EXECUTE)
3028 outs() << "x\n";
3029 else
3030 outs() << "-\n";
3031 }
3032 if ((initprot &
3033 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE |
3034 MachO::VM_PROT_EXECUTE)) != 0)
3035 outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n";
3036 else {
3037 if (initprot & MachO::VM_PROT_READ)
3038 outs() << " initprot r";
3039 else
3040 outs() << " initprot -";
3041 if (initprot & MachO::VM_PROT_WRITE)
3042 outs() << "w";
3043 else
3044 outs() << "-";
3045 if (initprot & MachO::VM_PROT_EXECUTE)
3046 outs() << "x\n";
3047 else
3048 outs() << "-\n";
3049 }
3050 } else {
3051 outs() << " maxprot " << format("0x%08" PRIx32, maxprot) << "\n";
3052 outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n";
3053 }
3054 outs() << " nsects " << nsects << "\n";
3055 if (verbose) {
3056 outs() << " flags";
3057 if (flags == 0)
3058 outs() << " (none)\n";
3059 else {
3060 if (flags & MachO::SG_HIGHVM) {
3061 outs() << " HIGHVM";
3062 flags &= ~MachO::SG_HIGHVM;
3063 }
3064 if (flags & MachO::SG_FVMLIB) {
3065 outs() << " FVMLIB";
3066 flags &= ~MachO::SG_FVMLIB;
3067 }
3068 if (flags & MachO::SG_NORELOC) {
3069 outs() << " NORELOC";
3070 flags &= ~MachO::SG_NORELOC;
3071 }
3072 if (flags & MachO::SG_PROTECTED_VERSION_1) {
3073 outs() << " PROTECTED_VERSION_1";
3074 flags &= ~MachO::SG_PROTECTED_VERSION_1;
3075 }
3076 if (flags)
3077 outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n";
3078 else
3079 outs() << "\n";
3080 }
3081 } else {
3082 outs() << " flags " << format("0x%" PRIx32, flags) << "\n";
3083 }
3084}
3085
3086static void PrintSection(const char *sectname, const char *segname,
3087 uint64_t addr, uint64_t size, uint32_t offset,
3088 uint32_t align, uint32_t reloff, uint32_t nreloc,
3089 uint32_t flags, uint32_t reserved1, uint32_t reserved2,
3090 uint32_t cmd, const char *sg_segname,
3091 uint32_t filetype, uint32_t object_size,
3092 bool verbose) {
3093 outs() << "Section\n";
3094 outs() << " sectname " << format("%.16s\n", sectname);
3095 outs() << " segname " << format("%.16s", segname);
3096 if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0)
3097 outs() << " (does not match segment)\n";
3098 else
3099 outs() << "\n";
3100 if (cmd == MachO::LC_SEGMENT_64) {
3101 outs() << " addr " << format("0x%016" PRIx64, addr) << "\n";
3102 outs() << " size " << format("0x%016" PRIx64, size);
3103 } else {
3104 outs() << " addr " << format("0x%08" PRIx32, addr) << "\n";
3105 outs() << " size " << format("0x%08" PRIx32, size);
3106 }
3107 if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size)
3108 outs() << " (past end of file)\n";
3109 else
3110 outs() << "\n";
3111 outs() << " offset " << offset;
3112 if (offset > object_size)
3113 outs() << " (past end of file)\n";
3114 else
3115 outs() << "\n";
3116 uint32_t align_shifted = 1 << align;
3117 outs() << " align 2^" << align << " (" << align_shifted << ")\n";
3118 outs() << " reloff " << reloff;
3119 if (reloff > object_size)
3120 outs() << " (past end of file)\n";
3121 else
3122 outs() << "\n";
3123 outs() << " nreloc " << nreloc;
3124 if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size)
3125 outs() << " (past end of file)\n";
3126 else
3127 outs() << "\n";
3128 uint32_t section_type = flags & MachO::SECTION_TYPE;
3129 if (verbose) {
3130 outs() << " type";
3131 if (section_type == MachO::S_REGULAR)
3132 outs() << " S_REGULAR\n";
3133 else if (section_type == MachO::S_ZEROFILL)
3134 outs() << " S_ZEROFILL\n";
3135 else if (section_type == MachO::S_CSTRING_LITERALS)
3136 outs() << " S_CSTRING_LITERALS\n";
3137 else if (section_type == MachO::S_4BYTE_LITERALS)
3138 outs() << " S_4BYTE_LITERALS\n";
3139 else if (section_type == MachO::S_8BYTE_LITERALS)
3140 outs() << " S_8BYTE_LITERALS\n";
3141 else if (section_type == MachO::S_16BYTE_LITERALS)
3142 outs() << " S_16BYTE_LITERALS\n";
3143 else if (section_type == MachO::S_LITERAL_POINTERS)
3144 outs() << " S_LITERAL_POINTERS\n";
3145 else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS)
3146 outs() << " S_NON_LAZY_SYMBOL_POINTERS\n";
3147 else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS)
3148 outs() << " S_LAZY_SYMBOL_POINTERS\n";
3149 else if (section_type == MachO::S_SYMBOL_STUBS)
3150 outs() << " S_SYMBOL_STUBS\n";
3151 else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS)
3152 outs() << " S_MOD_INIT_FUNC_POINTERS\n";
3153 else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS)
3154 outs() << " S_MOD_TERM_FUNC_POINTERS\n";
3155 else if (section_type == MachO::S_COALESCED)
3156 outs() << " S_COALESCED\n";
3157 else if (section_type == MachO::S_INTERPOSING)
3158 outs() << " S_INTERPOSING\n";
3159 else if (section_type == MachO::S_DTRACE_DOF)
3160 outs() << " S_DTRACE_DOF\n";
3161 else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS)
3162 outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n";
3163 else if (section_type == MachO::S_THREAD_LOCAL_REGULAR)
3164 outs() << " S_THREAD_LOCAL_REGULAR\n";
3165 else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL)
3166 outs() << " S_THREAD_LOCAL_ZEROFILL\n";
3167 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES)
3168 outs() << " S_THREAD_LOCAL_VARIABLES\n";
3169 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
3170 outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n";
3171 else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS)
3172 outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n";
3173 else
3174 outs() << format("0x%08" PRIx32, section_type) << "\n";
3175 outs() << "attributes";
3176 uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES;
3177 if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS)
3178 outs() << " PURE_INSTRUCTIONS";
3179 if (section_attributes & MachO::S_ATTR_NO_TOC)
3180 outs() << " NO_TOC";
3181 if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS)
3182 outs() << " STRIP_STATIC_SYMS";
3183 if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP)
3184 outs() << " NO_DEAD_STRIP";
3185 if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT)
3186 outs() << " LIVE_SUPPORT";
3187 if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE)
3188 outs() << " SELF_MODIFYING_CODE";
3189 if (section_attributes & MachO::S_ATTR_DEBUG)
3190 outs() << " DEBUG";
3191 if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS)
3192 outs() << " SOME_INSTRUCTIONS";
3193 if (section_attributes & MachO::S_ATTR_EXT_RELOC)
3194 outs() << " EXT_RELOC";
3195 if (section_attributes & MachO::S_ATTR_LOC_RELOC)
3196 outs() << " LOC_RELOC";
3197 if (section_attributes == 0)
3198 outs() << " (none)";
3199 outs() << "\n";
3200 } else
3201 outs() << " flags " << format("0x%08" PRIx32, flags) << "\n";
3202 outs() << " reserved1 " << reserved1;
3203 if (section_type == MachO::S_SYMBOL_STUBS ||
3204 section_type == MachO::S_LAZY_SYMBOL_POINTERS ||
3205 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS ||
3206 section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS ||
3207 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
3208 outs() << " (index into indirect symbol table)\n";
3209 else
3210 outs() << "\n";
3211 outs() << " reserved2 " << reserved2;
3212 if (section_type == MachO::S_SYMBOL_STUBS)
3213 outs() << " (size of stubs)\n";
3214 else
3215 outs() << "\n";
3216}
3217
David Majnemer73cc6ff2014-11-13 19:48:56 +00003218static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit,
Kevin Enderby956366c2014-08-29 22:30:52 +00003219 uint32_t object_size) {
3220 outs() << " cmd LC_SYMTAB\n";
3221 outs() << " cmdsize " << st.cmdsize;
3222 if (st.cmdsize != sizeof(struct MachO::symtab_command))
3223 outs() << " Incorrect size\n";
3224 else
3225 outs() << "\n";
3226 outs() << " symoff " << st.symoff;
3227 if (st.symoff > object_size)
3228 outs() << " (past end of file)\n";
3229 else
3230 outs() << "\n";
3231 outs() << " nsyms " << st.nsyms;
3232 uint64_t big_size;
David Majnemer73cc6ff2014-11-13 19:48:56 +00003233 if (Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00003234 big_size = st.nsyms;
3235 big_size *= sizeof(struct MachO::nlist_64);
3236 big_size += st.symoff;
3237 if (big_size > object_size)
3238 outs() << " (past end of file)\n";
3239 else
3240 outs() << "\n";
3241 } else {
3242 big_size = st.nsyms;
3243 big_size *= sizeof(struct MachO::nlist);
3244 big_size += st.symoff;
3245 if (big_size > object_size)
3246 outs() << " (past end of file)\n";
3247 else
3248 outs() << "\n";
3249 }
3250 outs() << " stroff " << st.stroff;
3251 if (st.stroff > object_size)
3252 outs() << " (past end of file)\n";
3253 else
3254 outs() << "\n";
3255 outs() << " strsize " << st.strsize;
3256 big_size = st.stroff;
3257 big_size += st.strsize;
3258 if (big_size > object_size)
3259 outs() << " (past end of file)\n";
3260 else
3261 outs() << "\n";
3262}
3263
3264static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst,
3265 uint32_t nsyms, uint32_t object_size,
David Majnemer73cc6ff2014-11-13 19:48:56 +00003266 bool Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00003267 outs() << " cmd LC_DYSYMTAB\n";
3268 outs() << " cmdsize " << dyst.cmdsize;
3269 if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command))
3270 outs() << " Incorrect size\n";
3271 else
3272 outs() << "\n";
3273 outs() << " ilocalsym " << dyst.ilocalsym;
3274 if (dyst.ilocalsym > nsyms)
3275 outs() << " (greater than the number of symbols)\n";
3276 else
3277 outs() << "\n";
3278 outs() << " nlocalsym " << dyst.nlocalsym;
3279 uint64_t big_size;
3280 big_size = dyst.ilocalsym;
3281 big_size += dyst.nlocalsym;
3282 if (big_size > nsyms)
3283 outs() << " (past the end of the symbol table)\n";
3284 else
3285 outs() << "\n";
3286 outs() << " iextdefsym " << dyst.iextdefsym;
3287 if (dyst.iextdefsym > nsyms)
3288 outs() << " (greater than the number of symbols)\n";
3289 else
3290 outs() << "\n";
3291 outs() << " nextdefsym " << dyst.nextdefsym;
3292 big_size = dyst.iextdefsym;
3293 big_size += dyst.nextdefsym;
3294 if (big_size > nsyms)
3295 outs() << " (past the end of the symbol table)\n";
3296 else
3297 outs() << "\n";
3298 outs() << " iundefsym " << dyst.iundefsym;
3299 if (dyst.iundefsym > nsyms)
3300 outs() << " (greater than the number of symbols)\n";
3301 else
3302 outs() << "\n";
3303 outs() << " nundefsym " << dyst.nundefsym;
3304 big_size = dyst.iundefsym;
3305 big_size += dyst.nundefsym;
3306 if (big_size > nsyms)
3307 outs() << " (past the end of the symbol table)\n";
3308 else
3309 outs() << "\n";
3310 outs() << " tocoff " << dyst.tocoff;
3311 if (dyst.tocoff > object_size)
3312 outs() << " (past end of file)\n";
3313 else
3314 outs() << "\n";
3315 outs() << " ntoc " << dyst.ntoc;
3316 big_size = dyst.ntoc;
3317 big_size *= sizeof(struct MachO::dylib_table_of_contents);
3318 big_size += dyst.tocoff;
3319 if (big_size > object_size)
3320 outs() << " (past end of file)\n";
3321 else
3322 outs() << "\n";
3323 outs() << " modtaboff " << dyst.modtaboff;
3324 if (dyst.modtaboff > object_size)
3325 outs() << " (past end of file)\n";
3326 else
3327 outs() << "\n";
3328 outs() << " nmodtab " << dyst.nmodtab;
3329 uint64_t modtabend;
David Majnemer73cc6ff2014-11-13 19:48:56 +00003330 if (Is64Bit) {
Kevin Enderby956366c2014-08-29 22:30:52 +00003331 modtabend = dyst.nmodtab;
3332 modtabend *= sizeof(struct MachO::dylib_module_64);
3333 modtabend += dyst.modtaboff;
3334 } else {
3335 modtabend = dyst.nmodtab;
3336 modtabend *= sizeof(struct MachO::dylib_module);
3337 modtabend += dyst.modtaboff;
3338 }
3339 if (modtabend > object_size)
3340 outs() << " (past end of file)\n";
3341 else
3342 outs() << "\n";
3343 outs() << " extrefsymoff " << dyst.extrefsymoff;
3344 if (dyst.extrefsymoff > object_size)
3345 outs() << " (past end of file)\n";
3346 else
3347 outs() << "\n";
3348 outs() << " nextrefsyms " << dyst.nextrefsyms;
3349 big_size = dyst.nextrefsyms;
3350 big_size *= sizeof(struct MachO::dylib_reference);
3351 big_size += dyst.extrefsymoff;
3352 if (big_size > object_size)
3353 outs() << " (past end of file)\n";
3354 else
3355 outs() << "\n";
3356 outs() << " indirectsymoff " << dyst.indirectsymoff;
3357 if (dyst.indirectsymoff > object_size)
3358 outs() << " (past end of file)\n";
3359 else
3360 outs() << "\n";
3361 outs() << " nindirectsyms " << dyst.nindirectsyms;
3362 big_size = dyst.nindirectsyms;
3363 big_size *= sizeof(uint32_t);
3364 big_size += dyst.indirectsymoff;
3365 if (big_size > object_size)
3366 outs() << " (past end of file)\n";
3367 else
3368 outs() << "\n";
3369 outs() << " extreloff " << dyst.extreloff;
3370 if (dyst.extreloff > object_size)
3371 outs() << " (past end of file)\n";
3372 else
3373 outs() << "\n";
3374 outs() << " nextrel " << dyst.nextrel;
3375 big_size = dyst.nextrel;
3376 big_size *= sizeof(struct MachO::relocation_info);
3377 big_size += dyst.extreloff;
3378 if (big_size > object_size)
3379 outs() << " (past end of file)\n";
3380 else
3381 outs() << "\n";
3382 outs() << " locreloff " << dyst.locreloff;
3383 if (dyst.locreloff > object_size)
3384 outs() << " (past end of file)\n";
3385 else
3386 outs() << "\n";
3387 outs() << " nlocrel " << dyst.nlocrel;
3388 big_size = dyst.nlocrel;
3389 big_size *= sizeof(struct MachO::relocation_info);
3390 big_size += dyst.locreloff;
3391 if (big_size > object_size)
3392 outs() << " (past end of file)\n";
3393 else
3394 outs() << "\n";
3395}
3396
Kevin Enderby8ae63c12014-09-04 16:54:47 +00003397static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc,
3398 uint32_t object_size) {
3399 if (dc.cmd == MachO::LC_DYLD_INFO)
3400 outs() << " cmd LC_DYLD_INFO\n";
3401 else
3402 outs() << " cmd LC_DYLD_INFO_ONLY\n";
3403 outs() << " cmdsize " << dc.cmdsize;
3404 if (dc.cmdsize != sizeof(struct MachO::dyld_info_command))
3405 outs() << " Incorrect size\n";
3406 else
3407 outs() << "\n";
3408 outs() << " rebase_off " << dc.rebase_off;
3409 if (dc.rebase_off > object_size)
3410 outs() << " (past end of file)\n";
3411 else
3412 outs() << "\n";
3413 outs() << " rebase_size " << dc.rebase_size;
3414 uint64_t big_size;
3415 big_size = dc.rebase_off;
3416 big_size += dc.rebase_size;
3417 if (big_size > object_size)
3418 outs() << " (past end of file)\n";
3419 else
3420 outs() << "\n";
3421 outs() << " bind_off " << dc.bind_off;
3422 if (dc.bind_off > object_size)
3423 outs() << " (past end of file)\n";
3424 else
3425 outs() << "\n";
3426 outs() << " bind_size " << dc.bind_size;
3427 big_size = dc.bind_off;
3428 big_size += dc.bind_size;
3429 if (big_size > object_size)
3430 outs() << " (past end of file)\n";
3431 else
3432 outs() << "\n";
3433 outs() << " weak_bind_off " << dc.weak_bind_off;
3434 if (dc.weak_bind_off > object_size)
3435 outs() << " (past end of file)\n";
3436 else
3437 outs() << "\n";
3438 outs() << " weak_bind_size " << dc.weak_bind_size;
3439 big_size = dc.weak_bind_off;
3440 big_size += dc.weak_bind_size;
3441 if (big_size > object_size)
3442 outs() << " (past end of file)\n";
3443 else
3444 outs() << "\n";
3445 outs() << " lazy_bind_off " << dc.lazy_bind_off;
3446 if (dc.lazy_bind_off > object_size)
3447 outs() << " (past end of file)\n";
3448 else
3449 outs() << "\n";
3450 outs() << " lazy_bind_size " << dc.lazy_bind_size;
3451 big_size = dc.lazy_bind_off;
3452 big_size += dc.lazy_bind_size;
3453 if (big_size > object_size)
3454 outs() << " (past end of file)\n";
3455 else
3456 outs() << "\n";
3457 outs() << " export_off " << dc.export_off;
3458 if (dc.export_off > object_size)
3459 outs() << " (past end of file)\n";
3460 else
3461 outs() << "\n";
3462 outs() << " export_size " << dc.export_size;
3463 big_size = dc.export_off;
3464 big_size += dc.export_size;
3465 if (big_size > object_size)
3466 outs() << " (past end of file)\n";
3467 else
3468 outs() << "\n";
3469}
3470
3471static void PrintDyldLoadCommand(MachO::dylinker_command dyld,
3472 const char *Ptr) {
3473 if (dyld.cmd == MachO::LC_ID_DYLINKER)
3474 outs() << " cmd LC_ID_DYLINKER\n";
3475 else if (dyld.cmd == MachO::LC_LOAD_DYLINKER)
3476 outs() << " cmd LC_LOAD_DYLINKER\n";
3477 else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT)
3478 outs() << " cmd LC_DYLD_ENVIRONMENT\n";
3479 else
3480 outs() << " cmd ?(" << dyld.cmd << ")\n";
3481 outs() << " cmdsize " << dyld.cmdsize;
3482 if (dyld.cmdsize < sizeof(struct MachO::dylinker_command))
3483 outs() << " Incorrect size\n";
3484 else
3485 outs() << "\n";
3486 if (dyld.name >= dyld.cmdsize)
3487 outs() << " name ?(bad offset " << dyld.name << ")\n";
3488 else {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003489 const char *P = (const char *)(Ptr) + dyld.name;
Kevin Enderby8ae63c12014-09-04 16:54:47 +00003490 outs() << " name " << P << " (offset " << dyld.name << ")\n";
3491 }
3492}
3493
3494static void PrintUuidLoadCommand(MachO::uuid_command uuid) {
3495 outs() << " cmd LC_UUID\n";
3496 outs() << " cmdsize " << uuid.cmdsize;
3497 if (uuid.cmdsize != sizeof(struct MachO::uuid_command))
3498 outs() << " Incorrect size\n";
3499 else
3500 outs() << "\n";
3501 outs() << " uuid ";
3502 outs() << format("%02" PRIX32, uuid.uuid[0]);
3503 outs() << format("%02" PRIX32, uuid.uuid[1]);
3504 outs() << format("%02" PRIX32, uuid.uuid[2]);
3505 outs() << format("%02" PRIX32, uuid.uuid[3]);
3506 outs() << "-";
3507 outs() << format("%02" PRIX32, uuid.uuid[4]);
3508 outs() << format("%02" PRIX32, uuid.uuid[5]);
3509 outs() << "-";
3510 outs() << format("%02" PRIX32, uuid.uuid[6]);
3511 outs() << format("%02" PRIX32, uuid.uuid[7]);
3512 outs() << "-";
3513 outs() << format("%02" PRIX32, uuid.uuid[8]);
3514 outs() << format("%02" PRIX32, uuid.uuid[9]);
3515 outs() << "-";
3516 outs() << format("%02" PRIX32, uuid.uuid[10]);
3517 outs() << format("%02" PRIX32, uuid.uuid[11]);
3518 outs() << format("%02" PRIX32, uuid.uuid[12]);
3519 outs() << format("%02" PRIX32, uuid.uuid[13]);
3520 outs() << format("%02" PRIX32, uuid.uuid[14]);
3521 outs() << format("%02" PRIX32, uuid.uuid[15]);
3522 outs() << "\n";
3523}
3524
3525static void PrintVersionMinLoadCommand(MachO::version_min_command vd) {
3526 if (vd.cmd == MachO::LC_VERSION_MIN_MACOSX)
3527 outs() << " cmd LC_VERSION_MIN_MACOSX\n";
3528 else if (vd.cmd == MachO::LC_VERSION_MIN_IPHONEOS)
3529 outs() << " cmd LC_VERSION_MIN_IPHONEOS\n";
3530 else
3531 outs() << " cmd " << vd.cmd << " (?)\n";
3532 outs() << " cmdsize " << vd.cmdsize;
3533 if (vd.cmdsize != sizeof(struct MachO::version_min_command))
3534 outs() << " Incorrect size\n";
3535 else
3536 outs() << "\n";
3537 outs() << " version " << ((vd.version >> 16) & 0xffff) << "."
3538 << ((vd.version >> 8) & 0xff);
3539 if ((vd.version & 0xff) != 0)
3540 outs() << "." << (vd.version & 0xff);
3541 outs() << "\n";
3542 if (vd.sdk == 0)
3543 outs() << " sdk n/a\n";
3544 else {
3545 outs() << " sdk " << ((vd.sdk >> 16) & 0xffff) << "."
3546 << ((vd.sdk >> 8) & 0xff);
3547 }
3548 if ((vd.sdk & 0xff) != 0)
3549 outs() << "." << (vd.sdk & 0xff);
3550 outs() << "\n";
3551}
3552
3553static void PrintSourceVersionCommand(MachO::source_version_command sd) {
3554 outs() << " cmd LC_SOURCE_VERSION\n";
3555 outs() << " cmdsize " << sd.cmdsize;
3556 if (sd.cmdsize != sizeof(struct MachO::source_version_command))
3557 outs() << " Incorrect size\n";
3558 else
3559 outs() << "\n";
3560 uint64_t a = (sd.version >> 40) & 0xffffff;
3561 uint64_t b = (sd.version >> 30) & 0x3ff;
3562 uint64_t c = (sd.version >> 20) & 0x3ff;
3563 uint64_t d = (sd.version >> 10) & 0x3ff;
3564 uint64_t e = sd.version & 0x3ff;
3565 outs() << " version " << a << "." << b;
3566 if (e != 0)
3567 outs() << "." << c << "." << d << "." << e;
3568 else if (d != 0)
3569 outs() << "." << c << "." << d;
3570 else if (c != 0)
3571 outs() << "." << c;
3572 outs() << "\n";
3573}
3574
3575static void PrintEntryPointCommand(MachO::entry_point_command ep) {
3576 outs() << " cmd LC_MAIN\n";
3577 outs() << " cmdsize " << ep.cmdsize;
3578 if (ep.cmdsize != sizeof(struct MachO::entry_point_command))
3579 outs() << " Incorrect size\n";
3580 else
3581 outs() << "\n";
3582 outs() << " entryoff " << ep.entryoff << "\n";
3583 outs() << " stacksize " << ep.stacksize << "\n";
3584}
3585
3586static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) {
3587 if (dl.cmd == MachO::LC_ID_DYLIB)
3588 outs() << " cmd LC_ID_DYLIB\n";
3589 else if (dl.cmd == MachO::LC_LOAD_DYLIB)
3590 outs() << " cmd LC_LOAD_DYLIB\n";
3591 else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB)
3592 outs() << " cmd LC_LOAD_WEAK_DYLIB\n";
3593 else if (dl.cmd == MachO::LC_REEXPORT_DYLIB)
3594 outs() << " cmd LC_REEXPORT_DYLIB\n";
3595 else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB)
3596 outs() << " cmd LC_LAZY_LOAD_DYLIB\n";
3597 else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB)
3598 outs() << " cmd LC_LOAD_UPWARD_DYLIB\n";
3599 else
3600 outs() << " cmd " << dl.cmd << " (unknown)\n";
3601 outs() << " cmdsize " << dl.cmdsize;
3602 if (dl.cmdsize < sizeof(struct MachO::dylib_command))
3603 outs() << " Incorrect size\n";
3604 else
3605 outs() << "\n";
3606 if (dl.dylib.name < dl.cmdsize) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003607 const char *P = (const char *)(Ptr) + dl.dylib.name;
Kevin Enderby8ae63c12014-09-04 16:54:47 +00003608 outs() << " name " << P << " (offset " << dl.dylib.name << ")\n";
3609 } else {
3610 outs() << " name ?(bad offset " << dl.dylib.name << ")\n";
3611 }
3612 outs() << " time stamp " << dl.dylib.timestamp << " ";
3613 time_t t = dl.dylib.timestamp;
3614 outs() << ctime(&t);
3615 outs() << " current version ";
3616 if (dl.dylib.current_version == 0xffffffff)
3617 outs() << "n/a\n";
3618 else
3619 outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "."
3620 << ((dl.dylib.current_version >> 8) & 0xff) << "."
3621 << (dl.dylib.current_version & 0xff) << "\n";
3622 outs() << "compatibility version ";
3623 if (dl.dylib.compatibility_version == 0xffffffff)
3624 outs() << "n/a\n";
3625 else
3626 outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "."
3627 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "."
3628 << (dl.dylib.compatibility_version & 0xff) << "\n";
3629}
3630
3631static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld,
3632 uint32_t object_size) {
3633 if (ld.cmd == MachO::LC_CODE_SIGNATURE)
3634 outs() << " cmd LC_FUNCTION_STARTS\n";
3635 else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO)
3636 outs() << " cmd LC_SEGMENT_SPLIT_INFO\n";
3637 else if (ld.cmd == MachO::LC_FUNCTION_STARTS)
3638 outs() << " cmd LC_FUNCTION_STARTS\n";
3639 else if (ld.cmd == MachO::LC_DATA_IN_CODE)
3640 outs() << " cmd LC_DATA_IN_CODE\n";
3641 else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS)
3642 outs() << " cmd LC_DYLIB_CODE_SIGN_DRS\n";
3643 else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT)
3644 outs() << " cmd LC_LINKER_OPTIMIZATION_HINT\n";
3645 else
3646 outs() << " cmd " << ld.cmd << " (?)\n";
3647 outs() << " cmdsize " << ld.cmdsize;
3648 if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command))
3649 outs() << " Incorrect size\n";
3650 else
3651 outs() << "\n";
3652 outs() << " dataoff " << ld.dataoff;
3653 if (ld.dataoff > object_size)
3654 outs() << " (past end of file)\n";
3655 else
3656 outs() << "\n";
3657 outs() << " datasize " << ld.datasize;
3658 uint64_t big_size = ld.dataoff;
3659 big_size += ld.datasize;
3660 if (big_size > object_size)
3661 outs() << " (past end of file)\n";
3662 else
3663 outs() << "\n";
3664}
3665
Kevin Enderby956366c2014-08-29 22:30:52 +00003666static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t ncmds,
3667 uint32_t filetype, uint32_t cputype,
3668 bool verbose) {
3669 StringRef Buf = Obj->getData();
3670 MachOObjectFile::LoadCommandInfo Command = Obj->getFirstLoadCommandInfo();
3671 for (unsigned i = 0;; ++i) {
3672 outs() << "Load command " << i << "\n";
3673 if (Command.C.cmd == MachO::LC_SEGMENT) {
3674 MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command);
3675 const char *sg_segname = SLC.segname;
3676 PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr,
3677 SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot,
3678 SLC.initprot, SLC.nsects, SLC.flags, Buf.size(),
3679 verbose);
3680 for (unsigned j = 0; j < SLC.nsects; j++) {
3681 MachO::section_64 S = Obj->getSection64(Command, j);
3682 PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align,
3683 S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2,
3684 SLC.cmd, sg_segname, filetype, Buf.size(), verbose);
3685 }
3686 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) {
3687 MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command);
3688 const char *sg_segname = SLC_64.segname;
3689 PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname,
3690 SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff,
3691 SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot,
3692 SLC_64.nsects, SLC_64.flags, Buf.size(), verbose);
3693 for (unsigned j = 0; j < SLC_64.nsects; j++) {
3694 MachO::section_64 S_64 = Obj->getSection64(Command, j);
3695 PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size,
3696 S_64.offset, S_64.align, S_64.reloff, S_64.nreloc,
3697 S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd,
3698 sg_segname, filetype, Buf.size(), verbose);
3699 }
3700 } else if (Command.C.cmd == MachO::LC_SYMTAB) {
3701 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
David Majnemer73cc6ff2014-11-13 19:48:56 +00003702 PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size());
Kevin Enderby956366c2014-08-29 22:30:52 +00003703 } else if (Command.C.cmd == MachO::LC_DYSYMTAB) {
3704 MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand();
3705 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand();
David Majnemer73cc6ff2014-11-13 19:48:56 +00003706 PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(),
3707 Obj->is64Bit());
Kevin Enderby8ae63c12014-09-04 16:54:47 +00003708 } else if (Command.C.cmd == MachO::LC_DYLD_INFO ||
3709 Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) {
3710 MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command);
3711 PrintDyldInfoLoadCommand(DyldInfo, Buf.size());
3712 } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER ||
3713 Command.C.cmd == MachO::LC_ID_DYLINKER ||
3714 Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) {
3715 MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command);
3716 PrintDyldLoadCommand(Dyld, Command.Ptr);
3717 } else if (Command.C.cmd == MachO::LC_UUID) {
3718 MachO::uuid_command Uuid = Obj->getUuidCommand(Command);
3719 PrintUuidLoadCommand(Uuid);
3720 } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX) {
3721 MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command);
3722 PrintVersionMinLoadCommand(Vd);
3723 } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) {
3724 MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command);
3725 PrintSourceVersionCommand(Sd);
3726 } else if (Command.C.cmd == MachO::LC_MAIN) {
3727 MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command);
3728 PrintEntryPointCommand(Ep);
Nick Kledzik15558912014-10-16 18:58:20 +00003729 } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB ||
3730 Command.C.cmd == MachO::LC_ID_DYLIB ||
3731 Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
3732 Command.C.cmd == MachO::LC_REEXPORT_DYLIB ||
3733 Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
3734 Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
Kevin Enderby8ae63c12014-09-04 16:54:47 +00003735 MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command);
3736 PrintDylibCommand(Dl, Command.Ptr);
3737 } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE ||
3738 Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO ||
3739 Command.C.cmd == MachO::LC_FUNCTION_STARTS ||
3740 Command.C.cmd == MachO::LC_DATA_IN_CODE ||
3741 Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS ||
3742 Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) {
3743 MachO::linkedit_data_command Ld =
3744 Obj->getLinkeditDataLoadCommand(Command);
3745 PrintLinkEditDataCommand(Ld, Buf.size());
Kevin Enderby956366c2014-08-29 22:30:52 +00003746 } else {
3747 outs() << " cmd ?(" << format("0x%08" PRIx32, Command.C.cmd)
3748 << ")\n";
3749 outs() << " cmdsize " << Command.C.cmdsize << "\n";
3750 // TODO: get and print the raw bytes of the load command.
3751 }
3752 // TODO: print all the other kinds of load commands.
3753 if (i == ncmds - 1)
3754 break;
3755 else
3756 Command = Obj->getNextLoadCommandInfo(Command);
3757 }
3758}
3759
3760static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds,
3761 uint32_t &filetype, uint32_t &cputype,
3762 bool verbose) {
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003763 if (Obj->is64Bit()) {
3764 MachO::mach_header_64 H_64;
3765 H_64 = Obj->getHeader64();
3766 PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype,
3767 H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose);
Kevin Enderby956366c2014-08-29 22:30:52 +00003768 ncmds = H_64.ncmds;
3769 filetype = H_64.filetype;
3770 cputype = H_64.cputype;
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003771 } else {
3772 MachO::mach_header H;
3773 H = Obj->getHeader();
3774 PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds,
3775 H.sizeofcmds, H.flags, verbose);
Kevin Enderby956366c2014-08-29 22:30:52 +00003776 ncmds = H.ncmds;
3777 filetype = H.filetype;
3778 cputype = H.cputype;
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003779 }
3780}
3781
3782void llvm::printMachOFileHeader(const object::ObjectFile *Obj) {
3783 const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj);
Kevin Enderby956366c2014-08-29 22:30:52 +00003784 uint32_t ncmds = 0;
3785 uint32_t filetype = 0;
3786 uint32_t cputype = 0;
3787 getAndPrintMachHeader(file, ncmds, filetype, cputype, true);
3788 PrintLoadCommands(file, ncmds, filetype, cputype, true);
Kevin Enderbyb76d3862014-08-22 20:35:18 +00003789}
Nick Kledzikd04bc352014-08-30 00:20:14 +00003790
3791//===----------------------------------------------------------------------===//
3792// export trie dumping
3793//===----------------------------------------------------------------------===//
3794
3795void llvm::printMachOExportsTrie(const object::MachOObjectFile *Obj) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003796 for (const llvm::object::ExportEntry &Entry : Obj->exports()) {
3797 uint64_t Flags = Entry.flags();
Nick Kledzikd04bc352014-08-30 00:20:14 +00003798 bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT);
3799 bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION);
3800 bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
3801 MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL);
3802 bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) ==
3803 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE);
3804 bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER);
3805 if (ReExport)
3806 outs() << "[re-export] ";
3807 else
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003808 outs() << format("0x%08llX ",
3809 Entry.address()); // FIXME:add in base address
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003810 outs() << Entry.name();
Nick Kledzikd04bc352014-08-30 00:20:14 +00003811 if (WeakDef || ThreadLocal || Resolver || Abs) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003812 bool NeedsComma = false;
Nick Kledzik1d1ac4b2014-09-03 01:12:52 +00003813 outs() << " [";
Nick Kledzikd04bc352014-08-30 00:20:14 +00003814 if (WeakDef) {
3815 outs() << "weak_def";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003816 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00003817 }
3818 if (ThreadLocal) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003819 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00003820 outs() << ", ";
3821 outs() << "per-thread";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003822 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00003823 }
3824 if (Abs) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003825 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00003826 outs() << ", ";
3827 outs() << "absolute";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003828 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00003829 }
3830 if (Resolver) {
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003831 if (NeedsComma)
Nick Kledzikd04bc352014-08-30 00:20:14 +00003832 outs() << ", ";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003833 outs() << format("resolver=0x%08llX", Entry.other());
3834 NeedsComma = true;
Nick Kledzikd04bc352014-08-30 00:20:14 +00003835 }
3836 outs() << "]";
3837 }
3838 if (ReExport) {
3839 StringRef DylibName = "unknown";
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003840 int Ordinal = Entry.other() - 1;
3841 Obj->getLibraryShortNameByIndex(Ordinal, DylibName);
3842 if (Entry.otherName().empty())
Nick Kledzikd04bc352014-08-30 00:20:14 +00003843 outs() << " (from " << DylibName << ")";
3844 else
Nick Kledzikac7cbdc2014-09-02 18:50:24 +00003845 outs() << " (" << Entry.otherName() << " from " << DylibName << ")";
Nick Kledzikd04bc352014-08-30 00:20:14 +00003846 }
3847 outs() << "\n";
3848 }
3849}
Nick Kledzikac431442014-09-12 21:34:15 +00003850
Nick Kledzikac431442014-09-12 21:34:15 +00003851//===----------------------------------------------------------------------===//
3852// rebase table dumping
3853//===----------------------------------------------------------------------===//
3854
3855namespace {
3856class SegInfo {
3857public:
3858 SegInfo(const object::MachOObjectFile *Obj);
3859
3860 StringRef segmentName(uint32_t SegIndex);
3861 StringRef sectionName(uint32_t SegIndex, uint64_t SegOffset);
3862 uint64_t address(uint32_t SegIndex, uint64_t SegOffset);
3863
3864private:
3865 struct SectionInfo {
3866 uint64_t Address;
3867 uint64_t Size;
3868 StringRef SectionName;
3869 StringRef SegmentName;
3870 uint64_t OffsetInSegment;
3871 uint64_t SegmentStartAddress;
3872 uint32_t SegmentIndex;
3873 };
3874 const SectionInfo &findSection(uint32_t SegIndex, uint64_t SegOffset);
3875 SmallVector<SectionInfo, 32> Sections;
3876};
3877}
3878
3879SegInfo::SegInfo(const object::MachOObjectFile *Obj) {
3880 // Build table of sections so segIndex/offset pairs can be translated.
Nick Kledzik56ebef42014-09-16 01:41:51 +00003881 uint32_t CurSegIndex = Obj->hasPageZeroSegment() ? 1 : 0;
Nick Kledzikac431442014-09-12 21:34:15 +00003882 StringRef CurSegName;
3883 uint64_t CurSegAddress;
3884 for (const SectionRef &Section : Obj->sections()) {
3885 SectionInfo Info;
3886 if (error(Section.getName(Info.SectionName)))
3887 return;
Rafael Espindola80291272014-10-08 15:28:58 +00003888 Info.Address = Section.getAddress();
3889 Info.Size = Section.getSize();
Nick Kledzikac431442014-09-12 21:34:15 +00003890 Info.SegmentName =
3891 Obj->getSectionFinalSegmentName(Section.getRawDataRefImpl());
3892 if (!Info.SegmentName.equals(CurSegName)) {
3893 ++CurSegIndex;
3894 CurSegName = Info.SegmentName;
3895 CurSegAddress = Info.Address;
3896 }
3897 Info.SegmentIndex = CurSegIndex - 1;
3898 Info.OffsetInSegment = Info.Address - CurSegAddress;
3899 Info.SegmentStartAddress = CurSegAddress;
3900 Sections.push_back(Info);
3901 }
3902}
3903
3904StringRef SegInfo::segmentName(uint32_t SegIndex) {
3905 for (const SectionInfo &SI : Sections) {
3906 if (SI.SegmentIndex == SegIndex)
3907 return SI.SegmentName;
3908 }
3909 llvm_unreachable("invalid segIndex");
3910}
3911
3912const SegInfo::SectionInfo &SegInfo::findSection(uint32_t SegIndex,
3913 uint64_t OffsetInSeg) {
3914 for (const SectionInfo &SI : Sections) {
3915 if (SI.SegmentIndex != SegIndex)
3916 continue;
3917 if (SI.OffsetInSegment > OffsetInSeg)
3918 continue;
3919 if (OffsetInSeg >= (SI.OffsetInSegment + SI.Size))
3920 continue;
3921 return SI;
3922 }
3923 llvm_unreachable("segIndex and offset not in any section");
3924}
3925
3926StringRef SegInfo::sectionName(uint32_t SegIndex, uint64_t OffsetInSeg) {
3927 return findSection(SegIndex, OffsetInSeg).SectionName;
3928}
3929
3930uint64_t SegInfo::address(uint32_t SegIndex, uint64_t OffsetInSeg) {
3931 const SectionInfo &SI = findSection(SegIndex, OffsetInSeg);
3932 return SI.SegmentStartAddress + OffsetInSeg;
3933}
3934
3935void llvm::printMachORebaseTable(const object::MachOObjectFile *Obj) {
3936 // Build table of sections so names can used in final output.
3937 SegInfo sectionTable(Obj);
3938
3939 outs() << "segment section address type\n";
3940 for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable()) {
3941 uint32_t SegIndex = Entry.segmentIndex();
3942 uint64_t OffsetInSeg = Entry.segmentOffset();
3943 StringRef SegmentName = sectionTable.segmentName(SegIndex);
3944 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
3945 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
3946
3947 // Table lines look like: __DATA __nl_symbol_ptr 0x0000F00C pointer
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003948 outs() << format("%-8s %-18s 0x%08" PRIX64 " %s\n",
3949 SegmentName.str().c_str(), SectionName.str().c_str(),
3950 Address, Entry.typeName().str().c_str());
Nick Kledzikac431442014-09-12 21:34:15 +00003951 }
3952}
Nick Kledzik56ebef42014-09-16 01:41:51 +00003953
3954static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) {
3955 StringRef DylibName;
3956 switch (Ordinal) {
3957 case MachO::BIND_SPECIAL_DYLIB_SELF:
3958 return "this-image";
3959 case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE:
3960 return "main-executable";
3961 case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP:
3962 return "flat-namespace";
3963 default:
Nick Kledzikabd29872014-09-16 22:03:13 +00003964 if (Ordinal > 0) {
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003965 std::error_code EC =
3966 Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName);
Nick Kledzikabd29872014-09-16 22:03:13 +00003967 if (EC)
Nick Kledzik51d2c2b2014-10-14 23:29:38 +00003968 return "<<bad library ordinal>>";
Nick Kledzikabd29872014-09-16 22:03:13 +00003969 return DylibName;
3970 }
Nick Kledzik56ebef42014-09-16 01:41:51 +00003971 }
Nick Kledzikabd29872014-09-16 22:03:13 +00003972 return "<<unknown special ordinal>>";
Nick Kledzik56ebef42014-09-16 01:41:51 +00003973}
3974
3975//===----------------------------------------------------------------------===//
3976// bind table dumping
3977//===----------------------------------------------------------------------===//
3978
3979void llvm::printMachOBindTable(const object::MachOObjectFile *Obj) {
3980 // Build table of sections so names can used in final output.
3981 SegInfo sectionTable(Obj);
3982
Nick Kledzik5ffacc12014-09-30 00:19:58 +00003983 outs() << "segment section address type "
3984 "addend dylib symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00003985 for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable()) {
3986 uint32_t SegIndex = Entry.segmentIndex();
3987 uint64_t OffsetInSeg = Entry.segmentOffset();
3988 StringRef SegmentName = sectionTable.segmentName(SegIndex);
3989 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
3990 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
3991
3992 // Table lines look like:
3993 // __DATA __got 0x00012010 pointer 0 libSystem ___stack_chk_guard
Nick Kledzik5ffacc12014-09-30 00:19:58 +00003994 StringRef Attr;
Nick Kledzik56ebef42014-09-16 01:41:51 +00003995 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT)
Nick Kledzik5ffacc12014-09-30 00:19:58 +00003996 Attr = " (weak_import)";
Kevin Enderbyb28ed012014-10-29 21:28:24 +00003997 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00003998 << left_justify(SectionName, 18) << " "
3999 << format_hex(Address, 10, true) << " "
4000 << left_justify(Entry.typeName(), 8) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004001 << format_decimal(Entry.addend(), 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00004002 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004003 << Entry.symbolName() << Attr << "\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00004004 }
4005}
4006
4007//===----------------------------------------------------------------------===//
4008// lazy bind table dumping
4009//===----------------------------------------------------------------------===//
4010
4011void llvm::printMachOLazyBindTable(const object::MachOObjectFile *Obj) {
4012 // Build table of sections so names can used in final output.
4013 SegInfo sectionTable(Obj);
4014
Nick Kledzik5ffacc12014-09-30 00:19:58 +00004015 outs() << "segment section address "
4016 "dylib symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00004017 for (const llvm::object::MachOBindEntry &Entry : Obj->lazyBindTable()) {
4018 uint32_t SegIndex = Entry.segmentIndex();
4019 uint64_t OffsetInSeg = Entry.segmentOffset();
4020 StringRef SegmentName = sectionTable.segmentName(SegIndex);
4021 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
4022 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
4023
4024 // Table lines look like:
4025 // __DATA __got 0x00012010 libSystem ___stack_chk_guard
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004026 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00004027 << left_justify(SectionName, 18) << " "
4028 << format_hex(Address, 10, true) << " "
4029 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " "
Nick Kledzik56ebef42014-09-16 01:41:51 +00004030 << Entry.symbolName() << "\n";
4031 }
4032}
4033
Nick Kledzik56ebef42014-09-16 01:41:51 +00004034//===----------------------------------------------------------------------===//
4035// weak bind table dumping
4036//===----------------------------------------------------------------------===//
4037
4038void llvm::printMachOWeakBindTable(const object::MachOObjectFile *Obj) {
4039 // Build table of sections so names can used in final output.
4040 SegInfo sectionTable(Obj);
4041
Nick Kledzik5ffacc12014-09-30 00:19:58 +00004042 outs() << "segment section address "
4043 "type addend symbol\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00004044 for (const llvm::object::MachOBindEntry &Entry : Obj->weakBindTable()) {
4045 // Strong symbols don't have a location to update.
4046 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) {
Nick Kledzik5ffacc12014-09-30 00:19:58 +00004047 outs() << " strong "
Nick Kledzik56ebef42014-09-16 01:41:51 +00004048 << Entry.symbolName() << "\n";
4049 continue;
4050 }
4051 uint32_t SegIndex = Entry.segmentIndex();
4052 uint64_t OffsetInSeg = Entry.segmentOffset();
4053 StringRef SegmentName = sectionTable.segmentName(SegIndex);
4054 StringRef SectionName = sectionTable.sectionName(SegIndex, OffsetInSeg);
4055 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
4056
4057 // Table lines look like:
4058 // __DATA __data 0x00001000 pointer 0 _foo
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004059 outs() << left_justify(SegmentName, 8) << " "
Nick Kledzik5ffacc12014-09-30 00:19:58 +00004060 << left_justify(SectionName, 18) << " "
4061 << format_hex(Address, 10, true) << " "
4062 << left_justify(Entry.typeName(), 8) << " "
Kevin Enderbyb28ed012014-10-29 21:28:24 +00004063 << format_decimal(Entry.addend(), 8) << " " << Entry.symbolName()
4064 << "\n";
Nick Kledzik56ebef42014-09-16 01:41:51 +00004065 }
4066}
4067
Kevin Enderby6f326ce2014-10-23 19:37:31 +00004068// get_dyld_bind_info_symbolname() is used for disassembly and passed an
4069// address, ReferenceValue, in the Mach-O file and looks in the dyld bind
4070// information for that address. If the address is found its binding symbol
4071// name is returned. If not nullptr is returned.
4072static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue,
4073 struct DisassembleInfo *info) {
Kevin Enderby078be602014-10-23 19:53:12 +00004074 if (info->bindtable == nullptr) {
4075 info->bindtable = new (BindTable);
Kevin Enderby6f326ce2014-10-23 19:37:31 +00004076 SegInfo sectionTable(info->O);
4077 for (const llvm::object::MachOBindEntry &Entry : info->O->bindTable()) {
4078 uint32_t SegIndex = Entry.segmentIndex();
4079 uint64_t OffsetInSeg = Entry.segmentOffset();
4080 uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg);
4081 const char *SymbolName = nullptr;
4082 StringRef name = Entry.symbolName();
4083 if (!name.empty())
4084 SymbolName = name.data();
Kevin Enderby078be602014-10-23 19:53:12 +00004085 info->bindtable->push_back(std::make_pair(Address, SymbolName));
Kevin Enderby6f326ce2014-10-23 19:37:31 +00004086 }
4087 }
Kevin Enderby078be602014-10-23 19:53:12 +00004088 for (bind_table_iterator BI = info->bindtable->begin(),
4089 BE = info->bindtable->end();
Kevin Enderby6f326ce2014-10-23 19:37:31 +00004090 BI != BE; ++BI) {
4091 uint64_t Address = BI->first;
4092 if (ReferenceValue == Address) {
4093 const char *SymbolName = BI->second;
4094 return SymbolName;
4095 }
4096 }
4097 return nullptr;
4098}