src/llvm/lib/MC/MachObjectWriter.cpp

   1 //===- lib/MC/MachObjectWriter.cpp - Mach-O File Writer -------------------===//
   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 #include "llvm/MC/MCMachObjectWriter.h"
  11 #include "llvm/ADT/StringMap.h"
  12 #include "llvm/ADT/Twine.h"
  13 #include "llvm/MC/MCAsmBackend.h"
  14 #include "llvm/MC/MCAsmLayout.h"
  15 #include "llvm/MC/MCAssembler.h"
  16 #include "llvm/MC/MCExpr.h"
  17 #include "llvm/MC/MCFixupKindInfo.h"
  18 #include "llvm/MC/MCMachOSymbolFlags.h"
  19 #include "llvm/MC/MCObjectWriter.h"
  20 #include "llvm/MC/MCSectionMachO.h"
  21 #include "llvm/MC/MCSymbol.h"
  22 #include "llvm/MC/MCValue.h"
  23 #include "llvm/Support/Debug.h"
  24 #include "llvm/Support/ErrorHandling.h"
  25 #include "llvm/Support/MachO.h"
  26 #include <vector>
  27 using namespace llvm;
  28
  29 #define DEBUG_TYPE "mc"
  30
  31 void MachObjectWriter::reset() {
  32   Relocations.clear();
  33   IndirectSymBase.clear();
  34   StringTable.clear();
  35   LocalSymbolData.clear();
  36   ExternalSymbolData.clear();
  37   UndefinedSymbolData.clear();
  38   MCObjectWriter::reset();
  39 }
  40
  41 bool MachObjectWriter::
  42 doesSymbolRequireExternRelocation(const MCSymbolData *SD) {
  43   // Undefined symbols are always extern.
  44   if (SD->getSymbol().isUndefined())
  45     return true;
  46
  47   // References to weak definitions require external relocation entries; the
  48   // definition may not always be the one in the same object file.
  49   if (SD->getFlags() & SF_WeakDefinition)
  50     return true;
  51
  52   // Otherwise, we can use an internal relocation.
  53   return false;
  54 }
  55
  56 bool MachObjectWriter::
  57 MachSymbolData::operator<(const MachSymbolData &RHS) const {
  58   return SymbolData->getSymbol().getName() <
  59     RHS.SymbolData->getSymbol().getName();
  60 }
  61
  62 bool MachObjectWriter::isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) {
  63   const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo(
  64     (MCFixupKind) Kind);
  65
  66   return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel;
  67 }
  68
  69 uint64_t MachObjectWriter::getFragmentAddress(const MCFragment *Fragment,
  70                                               const MCAsmLayout &Layout) const {
  71   return getSectionAddress(Fragment->getParent()) +
  72     Layout.getFragmentOffset(Fragment);
  73 }
  74
  75 uint64_t MachObjectWriter::getSymbolAddress(const MCSymbolData* SD,
  76                                             const MCAsmLayout &Layout) const {
  77   const MCSymbol &S = SD->getSymbol();
  78
  79   // If this is a variable, then recursively evaluate now.
  80   if (S.isVariable()) {
  81     if (const MCConstantExpr *C =
  82           dyn_cast<const MCConstantExpr>(S.getVariableValue()))
  83       return C->getValue();
  84
  85
  86     MCValue Target;
  87     if (!S.getVariableValue()->EvaluateAsRelocatable(Target, &Layout, nullptr))
  88       report_fatal_error("unable to evaluate offset for variable '" +
  89                          S.getName() + "'");
  90
  91     // Verify that any used symbols are defined.
  92     if (Target.getSymA() && Target.getSymA()->getSymbol().isUndefined())
  93       report_fatal_error("unable to evaluate offset to undefined symbol '" +
  94                          Target.getSymA()->getSymbol().getName() + "'");
  95     if (Target.getSymB() && Target.getSymB()->getSymbol().isUndefined())
  96       report_fatal_error("unable to evaluate offset to undefined symbol '" +
  97                          Target.getSymB()->getSymbol().getName() + "'");
  98
  99     uint64_t Address = Target.getConstant();
 100     if (Target.getSymA())
 101       Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
 102                                     Target.getSymA()->getSymbol()), Layout);
 103     if (Target.getSymB())
 104       Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
 105                                     Target.getSymB()->getSymbol()), Layout);
 106     return Address;
 107   }
 108
 109   return getSectionAddress(SD->getFragment()->getParent()) +
 110     Layout.getSymbolOffset(SD);
 111 }
 112
 113 uint64_t MachObjectWriter::getPaddingSize(const MCSectionData *SD,
 114                                           const MCAsmLayout &Layout) const {
 115   uint64_t EndAddr = getSectionAddress(SD) + Layout.getSectionAddressSize(SD);
 116   unsigned Next = SD->getLayoutOrder() + 1;
 117   if (Next >= Layout.getSectionOrder().size())
 118     return 0;
 119
 120   const MCSectionData &NextSD = *Layout.getSectionOrder()[Next];
 121   if (NextSD.getSection().isVirtualSection())
 122     return 0;
 123   return OffsetToAlignment(EndAddr, NextSD.getAlignment());
 124 }
 125
 126 void MachObjectWriter::WriteHeader(unsigned NumLoadCommands,
 127                                    unsigned LoadCommandsSize,
 128                                    bool SubsectionsViaSymbols) {
 129   uint32_t Flags = 0;
 130
 131   if (SubsectionsViaSymbols)
 132     Flags |= MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
 133
 134   // struct mach_header (28 bytes) or
 135   // struct mach_header_64 (32 bytes)
 136
 137   uint64_t Start = OS.tell();
 138   (void) Start;
 139
 140   Write32(is64Bit() ? MachO::MH_MAGIC_64 : MachO::MH_MAGIC);
 141
 142   Write32(TargetObjectWriter->getCPUType());
 143   Write32(TargetObjectWriter->getCPUSubtype());
 144
 145   Write32(MachO::MH_OBJECT);
 146   Write32(NumLoadCommands);
 147   Write32(LoadCommandsSize);
 148   Write32(Flags);
 149   if (is64Bit())
 150     Write32(0); // reserved
 151
 152   assert(OS.tell() - Start ==
 153          (is64Bit()?sizeof(MachO::mach_header_64): sizeof(MachO::mach_header)));
 154 }
 155
 156 /// WriteSegmentLoadCommand - Write a segment load command.
 157 ///
 158 /// \param NumSections The number of sections in this segment.
 159 /// \param SectionDataSize The total size of the sections.
 160 void MachObjectWriter::WriteSegmentLoadCommand(unsigned NumSections,
 161                                                uint64_t VMSize,
 162                                                uint64_t SectionDataStartOffset,
 163                                                uint64_t SectionDataSize) {
 164   // struct segment_command (56 bytes) or
 165   // struct segment_command_64 (72 bytes)
 166
 167   uint64_t Start = OS.tell();
 168   (void) Start;
 169
 170   unsigned SegmentLoadCommandSize =
 171     is64Bit() ? sizeof(MachO::segment_command_64):
 172     sizeof(MachO::segment_command);
 173   Write32(is64Bit() ? MachO::LC_SEGMENT_64 : MachO::LC_SEGMENT);
 174   Write32(SegmentLoadCommandSize +
 175           NumSections * (is64Bit() ? sizeof(MachO::section_64) :
 176                          sizeof(MachO::section)));
 177
 178   WriteBytes("", 16);
 179   if (is64Bit()) {
 180     Write64(0); // vmaddr
 181     Write64(VMSize); // vmsize
 182     Write64(SectionDataStartOffset); // file offset
 183     Write64(SectionDataSize); // file size
 184   } else {
 185     Write32(0); // vmaddr
 186     Write32(VMSize); // vmsize
 187     Write32(SectionDataStartOffset); // file offset
 188     Write32(SectionDataSize); // file size
 189   }
 190   // maxprot
 191   Write32(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE | MachO::VM_PROT_EXECUTE);
 192   // initprot
 193   Write32(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE | MachO::VM_PROT_EXECUTE);
 194   Write32(NumSections);
 195   Write32(0); // flags
 196
 197   assert(OS.tell() - Start == SegmentLoadCommandSize);
 198 }
 199
 200 void MachObjectWriter::WriteSection(const MCAssembler &Asm,
 201                                     const MCAsmLayout &Layout,
 202                                     const MCSectionData &SD,
 203                                     uint64_t FileOffset,
 204                                     uint64_t RelocationsStart,
 205                                     unsigned NumRelocations) {
 206   uint64_t SectionSize = Layout.getSectionAddressSize(&SD);
 207
 208   // The offset is unused for virtual sections.
 209   if (SD.getSection().isVirtualSection()) {
 210     assert(Layout.getSectionFileSize(&SD) == 0 && "Invalid file size!");
 211     FileOffset = 0;
 212   }
 213
 214   // struct section (68 bytes) or
 215   // struct section_64 (80 bytes)
 216
 217   uint64_t Start = OS.tell();
 218   (void) Start;
 219
 220   const MCSectionMachO &Section = cast<MCSectionMachO>(SD.getSection());
 221   WriteBytes(Section.getSectionName(), 16);
 222   WriteBytes(Section.getSegmentName(), 16);
 223   if (is64Bit()) {
 224     Write64(getSectionAddress(&SD)); // address
 225     Write64(SectionSize); // size
 226   } else {
 227     Write32(getSectionAddress(&SD)); // address
 228     Write32(SectionSize); // size
 229   }
 230   Write32(FileOffset);
 231
 232   unsigned Flags = Section.getTypeAndAttributes();
 233   if (SD.hasInstructions())
 234     Flags |= MachO::S_ATTR_SOME_INSTRUCTIONS;
 235
 236   assert(isPowerOf2_32(SD.getAlignment()) && "Invalid alignment!");
 237   Write32(Log2_32(SD.getAlignment()));
 238   Write32(NumRelocations ? RelocationsStart : 0);
 239   Write32(NumRelocations);
 240   Write32(Flags);
 241   Write32(IndirectSymBase.lookup(&SD)); // reserved1
 242   Write32(Section.getStubSize()); // reserved2
 243   if (is64Bit())
 244     Write32(0); // reserved3
 245
 246   assert(OS.tell() - Start == (is64Bit() ? sizeof(MachO::section_64) :
 247                                sizeof(MachO::section)));
 248 }
 249
 250 void MachObjectWriter::WriteSymtabLoadCommand(uint32_t SymbolOffset,
 251                                               uint32_t NumSymbols,
 252                                               uint32_t StringTableOffset,
 253                                               uint32_t StringTableSize) {
 254   // struct symtab_command (24 bytes)
 255
 256   uint64_t Start = OS.tell();
 257   (void) Start;
 258
 259   Write32(MachO::LC_SYMTAB);
 260   Write32(sizeof(MachO::symtab_command));
 261   Write32(SymbolOffset);
 262   Write32(NumSymbols);
 263   Write32(StringTableOffset);
 264   Write32(StringTableSize);
 265
 266   assert(OS.tell() - Start == sizeof(MachO::symtab_command));
 267 }
 268
 269 void MachObjectWriter::WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
 270                                                 uint32_t NumLocalSymbols,
 271                                                 uint32_t FirstExternalSymbol,
 272                                                 uint32_t NumExternalSymbols,
 273                                                 uint32_t FirstUndefinedSymbol,
 274                                                 uint32_t NumUndefinedSymbols,
 275                                                 uint32_t IndirectSymbolOffset,
 276                                                 uint32_t NumIndirectSymbols) {
 277   // struct dysymtab_command (80 bytes)
 278
 279   uint64_t Start = OS.tell();
 280   (void) Start;
 281
 282   Write32(MachO::LC_DYSYMTAB);
 283   Write32(sizeof(MachO::dysymtab_command));
 284   Write32(FirstLocalSymbol);
 285   Write32(NumLocalSymbols);
 286   Write32(FirstExternalSymbol);
 287   Write32(NumExternalSymbols);
 288   Write32(FirstUndefinedSymbol);
 289   Write32(NumUndefinedSymbols);
 290   Write32(0); // tocoff
 291   Write32(0); // ntoc
 292   Write32(0); // modtaboff
 293   Write32(0); // nmodtab
 294   Write32(0); // extrefsymoff
 295   Write32(0); // nextrefsyms
 296   Write32(IndirectSymbolOffset);
 297   Write32(NumIndirectSymbols);
 298   Write32(0); // extreloff
 299   Write32(0); // nextrel
 300   Write32(0); // locreloff
 301   Write32(0); // nlocrel
 302
 303   assert(OS.tell() - Start == sizeof(MachO::dysymtab_command));
 304 }
 305
 306 MachObjectWriter::MachSymbolData *
 307 MachObjectWriter::findSymbolData(const MCSymbol &Sym) {
 308   for (auto &Entry : LocalSymbolData)
 309     if (&Entry.SymbolData->getSymbol() == &Sym)
 310       return &Entry;
 311
 312   for (auto &Entry : ExternalSymbolData)
 313     if (&Entry.SymbolData->getSymbol() == &Sym)
 314       return &Entry;
 315
 316   for (auto &Entry : UndefinedSymbolData)
 317     if (&Entry.SymbolData->getSymbol() == &Sym)
 318       return &Entry;
 319
 320   return nullptr;
 321 }
 322
 323 void MachObjectWriter::WriteNlist(MachSymbolData &MSD,
 324                                   const MCAsmLayout &Layout) {
 325   MCSymbolData &Data = *MSD.SymbolData;
 326   const MCSymbol *Symbol = &Data.getSymbol();
 327   const MCSymbol *AliasedSymbol = &Symbol->AliasedSymbol();
 328   uint8_t SectionIndex = MSD.SectionIndex;
 329   uint8_t Type = 0;
 330   uint16_t Flags = Data.getFlags();
 331   uint64_t Address = 0;
 332   bool IsAlias = Symbol != AliasedSymbol;
 333
 334   MachSymbolData *AliaseeInfo;
 335   if (IsAlias) {
 336     AliaseeInfo = findSymbolData(*AliasedSymbol);
 337     if (AliaseeInfo)
 338       SectionIndex = AliaseeInfo->SectionIndex;
 339     Symbol = AliasedSymbol;
 340   }
 341
 342   // Set the N_TYPE bits. See <mach-o/nlist.h>.
 343   //
 344   // FIXME: Are the prebound or indirect fields possible here?
 345   if (IsAlias && Symbol->isUndefined())
 346     Type = MachO::N_INDR;
 347   else if (Symbol->isUndefined())
 348     Type = MachO::N_UNDF;
 349   else if (Symbol->isAbsolute())
 350     Type = MachO::N_ABS;
 351   else
 352     Type = MachO::N_SECT;
 353
 354   // FIXME: Set STAB bits.
 355
 356   if (Data.isPrivateExtern())
 357     Type |= MachO::N_PEXT;
 358
 359   // Set external bit.
 360   if (Data.isExternal() || (!IsAlias && Symbol->isUndefined()))
 361     Type |= MachO::N_EXT;
 362
 363   // Compute the symbol address.
 364   if (IsAlias && Symbol->isUndefined())
 365     Address = AliaseeInfo->StringIndex;
 366   else if (Symbol->isDefined())
 367     Address = getSymbolAddress(&Data, Layout);
 368   else if (Data.isCommon()) {
 369     // Common symbols are encoded with the size in the address
 370     // field, and their alignment in the flags.
 371     Address = Data.getCommonSize();
 372
 373     // Common alignment is packed into the 'desc' bits.
 374     if (unsigned Align = Data.getCommonAlignment()) {
 375       unsigned Log2Size = Log2_32(Align);
 376       assert((1U << Log2Size) == Align && "Invalid 'common' alignment!");
 377       if (Log2Size > 15)
 378         report_fatal_error("invalid 'common' alignment '" +
 379                            Twine(Align) + "' for '" + Symbol->getName() + "'",
 380                            false);
 381       // FIXME: Keep this mask with the SymbolFlags enumeration.
 382       Flags = (Flags & 0xF0FF) | (Log2Size << 8);
 383     }
 384   }
 385
 386   if (Layout.getAssembler().isThumbFunc(Symbol))
 387     Flags |= SF_ThumbFunc;
 388
 389   // struct nlist (12 bytes)
 390
 391   Write32(MSD.StringIndex);
 392   Write8(Type);
 393   Write8(SectionIndex);
 394
 395   // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
 396   // value.
 397   Write16(Flags);
 398   if (is64Bit())
 399     Write64(Address);
 400   else
 401     Write32(Address);
 402 }
 403
 404 void MachObjectWriter::WriteLinkeditLoadCommand(uint32_t Type,
 405                                                 uint32_t DataOffset,
 406                                                 uint32_t DataSize) {
 407   uint64_t Start = OS.tell();
 408   (void) Start;
 409
 410   Write32(Type);
 411   Write32(sizeof(MachO::linkedit_data_command));
 412   Write32(DataOffset);
 413   Write32(DataSize);
 414
 415   assert(OS.tell() - Start == sizeof(MachO::linkedit_data_command));
 416 }
 417
 418 static unsigned ComputeLinkerOptionsLoadCommandSize(
 419   const std::vector<std::string> &Options, bool is64Bit)
 420 {
 421   unsigned Size = sizeof(MachO::linker_option_command);
 422   for (unsigned i = 0, e = Options.size(); i != e; ++i)
 423     Size += Options[i].size() + 1;
 424   return RoundUpToAlignment(Size, is64Bit ? 8 : 4);
 425 }
 426
 427 void MachObjectWriter::WriteLinkerOptionsLoadCommand(
 428   const std::vector<std::string> &Options)
 429 {
 430   unsigned Size = ComputeLinkerOptionsLoadCommandSize(Options, is64Bit());
 431   uint64_t Start = OS.tell();
 432   (void) Start;
 433
 434   Write32(MachO::LC_LINKER_OPTION);
 435   Write32(Size);
 436   Write32(Options.size());
 437   uint64_t BytesWritten = sizeof(MachO::linker_option_command);
 438   for (unsigned i = 0, e = Options.size(); i != e; ++i) {
 439     // Write each string, including the null byte.
 440     const std::string &Option = Options[i];
 441     WriteBytes(Option.c_str(), Option.size() + 1);
 442     BytesWritten += Option.size() + 1;
 443   }
 444
 445   // Pad to a multiple of the pointer size.
 446   WriteBytes("", OffsetToAlignment(BytesWritten, is64Bit() ? 8 : 4));
 447
 448   assert(OS.tell() - Start == Size);
 449 }
 450
 451
 452 void MachObjectWriter::RecordRelocation(const MCAssembler &Asm,
 453                                         const MCAsmLayout &Layout,
 454                                         const MCFragment *Fragment,
 455                                         const MCFixup &Fixup,
 456                                         MCValue Target,
 457                                         bool &IsPCRel,
 458                                         uint64_t &FixedValue) {
 459   TargetObjectWriter->RecordRelocation(this, Asm, Layout, Fragment, Fixup,
 460                                        Target, FixedValue);
 461 }
 462
 463 void MachObjectWriter::BindIndirectSymbols(MCAssembler &Asm) {
 464   // This is the point where 'as' creates actual symbols for indirect symbols
 465   // (in the following two passes). It would be easier for us to do this sooner
 466   // when we see the attribute, but that makes getting the order in the symbol
 467   // table much more complicated than it is worth.
 468   //
 469   // FIXME: Revisit this when the dust settles.
 470
 471   // Report errors for use of .indirect_symbol not in a symbol pointer section
 472   // or stub section.
 473   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 474          ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 475     const MCSectionMachO &Section =
 476       cast<MCSectionMachO>(it->SectionData->getSection());
 477
 478     if (Section.getType() != MachO::S_NON_LAZY_SYMBOL_POINTERS &&
 479         Section.getType() != MachO::S_LAZY_SYMBOL_POINTERS &&
 480         Section.getType() != MachO::S_SYMBOL_STUBS) {
 481         MCSymbol &Symbol = *it->Symbol;
 482         report_fatal_error("indirect symbol '" + Symbol.getName() +
 483                            "' not in a symbol pointer or stub section");
 484     }
 485   }
 486
 487   // Bind non-lazy symbol pointers first.
 488   unsigned IndirectIndex = 0;
 489   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 490          ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
 491     const MCSectionMachO &Section =
 492       cast<MCSectionMachO>(it->SectionData->getSection());
 493
 494     if (Section.getType() != MachO::S_NON_LAZY_SYMBOL_POINTERS)
 495       continue;
 496
 497     // Initialize the section indirect symbol base, if necessary.
 498     IndirectSymBase.insert(std::make_pair(it->SectionData, IndirectIndex));
 499
 500     Asm.getOrCreateSymbolData(*it->Symbol);
 501   }
 502
 503   // Then lazy symbol pointers and symbol stubs.
 504   IndirectIndex = 0;
 505   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 506          ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
 507     const MCSectionMachO &Section =
 508       cast<MCSectionMachO>(it->SectionData->getSection());
 509
 510     if (Section.getType() != MachO::S_LAZY_SYMBOL_POINTERS &&
 511         Section.getType() != MachO::S_SYMBOL_STUBS)
 512       continue;
 513
 514     // Initialize the section indirect symbol base, if necessary.
 515     IndirectSymBase.insert(std::make_pair(it->SectionData, IndirectIndex));
 516
 517     // Set the symbol type to undefined lazy, but only on construction.
 518     //
 519     // FIXME: Do not hardcode.
 520     bool Created;
 521     MCSymbolData &Entry = Asm.getOrCreateSymbolData(*it->Symbol, &Created);
 522     if (Created)
 523       Entry.setFlags(Entry.getFlags() | 0x0001);
 524   }
 525 }
 526
 527 /// ComputeSymbolTable - Compute the symbol table data
 528 void MachObjectWriter::ComputeSymbolTable(
 529     MCAssembler &Asm, std::vector<MachSymbolData> &LocalSymbolData,
 530     std::vector<MachSymbolData> &ExternalSymbolData,
 531     std::vector<MachSymbolData> &UndefinedSymbolData) {
 532   // Build section lookup table.
 533   DenseMap<const MCSection*, uint8_t> SectionIndexMap;
 534   unsigned Index = 1;
 535   for (MCAssembler::iterator it = Asm.begin(),
 536          ie = Asm.end(); it != ie; ++it, ++Index)
 537     SectionIndexMap[&it->getSection()] = Index;
 538   assert(Index <= 256 && "Too many sections!");
 539
 540   // Build the string table.
 541   for (MCSymbolData &SD : Asm.symbols()) {
 542     const MCSymbol &Symbol = SD.getSymbol();
 543     if (!Asm.isSymbolLinkerVisible(Symbol))
 544       continue;
 545
 546     StringTable.add(Symbol.getName());
 547   }
 548   StringTable.finalize(StringTableBuilder::MachO);
 549
 550   // Build the symbol arrays but only for non-local symbols.
 551   //
 552   // The particular order that we collect and then sort the symbols is chosen to
 553   // match 'as'. Even though it doesn't matter for correctness, this is
 554   // important for letting us diff .o files.
 555   for (MCSymbolData &SD : Asm.symbols()) {
 556     const MCSymbol &Symbol = SD.getSymbol();
 557
 558     // Ignore non-linker visible symbols.
 559     if (!Asm.isSymbolLinkerVisible(Symbol))
 560       continue;
 561
 562     if (!SD.isExternal() && !Symbol.isUndefined())
 563       continue;
 564
 565     MachSymbolData MSD;
 566     MSD.SymbolData = &SD;
 567     MSD.StringIndex = StringTable.getOffset(Symbol.getName());
 568
 569     if (Symbol.isUndefined()) {
 570       MSD.SectionIndex = 0;
 571       UndefinedSymbolData.push_back(MSD);
 572     } else if (Symbol.isAbsolute()) {
 573       MSD.SectionIndex = 0;
 574       ExternalSymbolData.push_back(MSD);
 575     } else {
 576       MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
 577       assert(MSD.SectionIndex && "Invalid section index!");
 578       ExternalSymbolData.push_back(MSD);
 579     }
 580   }
 581
 582   // Now add the data for local symbols.
 583   for (MCSymbolData &SD : Asm.symbols()) {
 584     const MCSymbol &Symbol = SD.getSymbol();
 585
 586     // Ignore non-linker visible symbols.
 587     if (!Asm.isSymbolLinkerVisible(Symbol))
 588       continue;
 589
 590     if (SD.isExternal() || Symbol.isUndefined())
 591       continue;
 592
 593     MachSymbolData MSD;
 594     MSD.SymbolData = &SD;
 595     MSD.StringIndex = StringTable.getOffset(Symbol.getName());
 596
 597     if (Symbol.isAbsolute()) {
 598       MSD.SectionIndex = 0;
 599       LocalSymbolData.push_back(MSD);
 600     } else {
 601       MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
 602       assert(MSD.SectionIndex && "Invalid section index!");
 603       LocalSymbolData.push_back(MSD);
 604     }
 605   }
 606
 607   // External and undefined symbols are required to be in lexicographic order.
 608   std::sort(ExternalSymbolData.begin(), ExternalSymbolData.end());
 609   std::sort(UndefinedSymbolData.begin(), UndefinedSymbolData.end());
 610
 611   // Set the symbol indices.
 612   Index = 0;
 613   for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
 614     LocalSymbolData[i].SymbolData->setIndex(Index++);
 615   for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
 616     ExternalSymbolData[i].SymbolData->setIndex(Index++);
 617   for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
 618     UndefinedSymbolData[i].SymbolData->setIndex(Index++);
 619 }
 620
 621 void MachObjectWriter::computeSectionAddresses(const MCAssembler &Asm,
 622                                                const MCAsmLayout &Layout) {
 623   uint64_t StartAddress = 0;
 624   const SmallVectorImpl<MCSectionData*> &Order = Layout.getSectionOrder();
 625   for (int i = 0, n = Order.size(); i != n ; ++i) {
 626     const MCSectionData *SD = Order[i];
 627     StartAddress = RoundUpToAlignment(StartAddress, SD->getAlignment());
 628     SectionAddress[SD] = StartAddress;
 629     StartAddress += Layout.getSectionAddressSize(SD);
 630
 631     // Explicitly pad the section to match the alignment requirements of the
 632     // following one. This is for 'gas' compatibility, it shouldn't
 633     /// strictly be necessary.
 634     StartAddress += getPaddingSize(SD, Layout);
 635   }
 636 }
 637
 638 void MachObjectWriter::markAbsoluteVariableSymbols(MCAssembler &Asm,
 639                                                    const MCAsmLayout &Layout) {
 640   for (MCSymbolData &SD : Asm.symbols()) {
 641     if (!SD.getSymbol().isVariable())
 642       continue;
 643
 644     // Is the variable is a symbol difference (SA - SB + C) expression,
 645     // and neither symbol is external, mark the variable as absolute.
 646     const MCExpr *Expr = SD.getSymbol().getVariableValue();
 647     MCValue Value;
 648     if (Expr->EvaluateAsRelocatable(Value, &Layout, nullptr)) {
 649       if (Value.getSymA() && Value.getSymB())
 650         const_cast<MCSymbol*>(&SD.getSymbol())->setAbsolute();
 651     }
 652   }
 653 }
 654
 655 void MachObjectWriter::ExecutePostLayoutBinding(MCAssembler &Asm,
 656                                                 const MCAsmLayout &Layout) {
 657   computeSectionAddresses(Asm, Layout);
 658
 659   // Create symbol data for any indirect symbols.
 660   BindIndirectSymbols(Asm);
 661
 662   // Mark symbol difference expressions in variables (from .set or = directives)
 663   // as absolute.
 664   markAbsoluteVariableSymbols(Asm, Layout);
 665
 666   // Compute symbol table information and bind symbol indices.
 667   ComputeSymbolTable(Asm, LocalSymbolData, ExternalSymbolData,
 668                      UndefinedSymbolData);
 669 }
 670
 671 bool MachObjectWriter::
 672 IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
 673                                        const MCSymbolData &DataA,
 674                                        const MCFragment &FB,
 675                                        bool InSet,
 676                                        bool IsPCRel) const {
 677   if (InSet)
 678     return true;
 679
 680   // The effective address is
 681   //     addr(atom(A)) + offset(A)
 682   //   - addr(atom(B)) - offset(B)
 683   // and the offsets are not relocatable, so the fixup is fully resolved when
 684   //  addr(atom(A)) - addr(atom(B)) == 0.
 685   const MCSymbolData *A_Base = nullptr, *B_Base = nullptr;
 686
 687   const MCSymbol &SA = DataA.getSymbol().AliasedSymbol();
 688   const MCSection &SecA = SA.getSection();
 689   const MCSection &SecB = FB.getParent()->getSection();
 690
 691   if (IsPCRel) {
 692     // The simple (Darwin, except on x86_64) way of dealing with this was to
 693     // assume that any reference to a temporary symbol *must* be a temporary
 694     // symbol in the same atom, unless the sections differ. Therefore, any PCrel
 695     // relocation to a temporary symbol (in the same section) is fully
 696     // resolved. This also works in conjunction with absolutized .set, which
 697     // requires the compiler to use .set to absolutize the differences between
 698     // symbols which the compiler knows to be assembly time constants, so we
 699     // don't need to worry about considering symbol differences fully resolved.
 700     //
 701     // If the file isn't using sub-sections-via-symbols, we can make the
 702     // same assumptions about any symbol that we normally make about
 703     // assembler locals.
 704
 705     bool hasReliableSymbolDifference = isX86_64();
 706     if (!hasReliableSymbolDifference) {
 707       if (!SA.isInSection() || &SecA != &SecB ||
 708           (!SA.isTemporary() &&
 709            FB.getAtom() != Asm.getSymbolData(SA).getFragment()->getAtom() &&
 710            Asm.getSubsectionsViaSymbols()))
 711         return false;
 712       return true;
 713     }
 714     // For Darwin x86_64, there is one special case when the reference IsPCRel.
 715     // If the fragment with the reference does not have a base symbol but meets
 716     // the simple way of dealing with this, in that it is a temporary symbol in
 717     // the same atom then it is assumed to be fully resolved.  This is needed so
 718     // a relocation entry is not created and so the static linker does not
 719     // mess up the reference later.
 720     else if(!FB.getAtom() &&
 721             SA.isTemporary() && SA.isInSection() && &SecA == &SecB){
 722       return true;
 723     }
 724   } else {
 725     if (!TargetObjectWriter->useAggressiveSymbolFolding())
 726       return false;
 727   }
 728
 729   // If they are not in the same section, we can't compute the diff.
 730   if (&SecA != &SecB)
 731     return false;
 732
 733   const MCFragment *FA = Asm.getSymbolData(SA).getFragment();
 734
 735   // Bail if the symbol has no fragment.
 736   if (!FA)
 737     return false;
 738
 739   A_Base = FA->getAtom();
 740   B_Base = FB.getAtom();
 741
 742   // If the atoms are the same, they are guaranteed to have the same address.
 743   if (A_Base == B_Base)
 744     return true;
 745
 746   // Otherwise, we can't prove this is fully resolved.
 747   return false;
 748 }
 749
 750 void MachObjectWriter::WriteObject(MCAssembler &Asm,
 751                                    const MCAsmLayout &Layout) {
 752   unsigned NumSections = Asm.size();
 753   const MCAssembler::VersionMinInfoType &VersionInfo =
 754     Layout.getAssembler().getVersionMinInfo();
 755
 756   // The section data starts after the header, the segment load command (and
 757   // section headers) and the symbol table.
 758   unsigned NumLoadCommands = 1;
 759   uint64_t LoadCommandsSize = is64Bit() ?
 760     sizeof(MachO::segment_command_64) + NumSections * sizeof(MachO::section_64):
 761     sizeof(MachO::segment_command) + NumSections * sizeof(MachO::section);
 762
 763   // Add the deployment target version info load command size, if used.
 764   if (VersionInfo.Major != 0) {
 765     ++NumLoadCommands;
 766     LoadCommandsSize += sizeof(MachO::version_min_command);
 767   }
 768
 769   // Add the data-in-code load command size, if used.
 770   unsigned NumDataRegions = Asm.getDataRegions().size();
 771   if (NumDataRegions) {
 772     ++NumLoadCommands;
 773     LoadCommandsSize += sizeof(MachO::linkedit_data_command);
 774   }
 775
 776   // Add the loh load command size, if used.
 777   uint64_t LOHRawSize = Asm.getLOHContainer().getEmitSize(*this, Layout);
 778   uint64_t LOHSize = RoundUpToAlignment(LOHRawSize, is64Bit() ? 8 : 4);
 779   if (LOHSize) {
 780     ++NumLoadCommands;
 781     LoadCommandsSize += sizeof(MachO::linkedit_data_command);
 782   }
 783
 784   // Add the symbol table load command sizes, if used.
 785   unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() +
 786     UndefinedSymbolData.size();
 787   if (NumSymbols) {
 788     NumLoadCommands += 2;
 789     LoadCommandsSize += (sizeof(MachO::symtab_command) +
 790                          sizeof(MachO::dysymtab_command));
 791   }
 792
 793   // Add the linker option load commands sizes.
 794   const std::vector<std::vector<std::string> > &LinkerOptions =
 795     Asm.getLinkerOptions();
 796   for (unsigned i = 0, e = LinkerOptions.size(); i != e; ++i) {
 797     ++NumLoadCommands;
 798     LoadCommandsSize += ComputeLinkerOptionsLoadCommandSize(LinkerOptions[i],
 799                                                             is64Bit());
 800   }
 801
 802   // Compute the total size of the section data, as well as its file size and vm
 803   // size.
 804   uint64_t SectionDataStart = (is64Bit() ? sizeof(MachO::mach_header_64) :
 805                                sizeof(MachO::mach_header)) + LoadCommandsSize;
 806   uint64_t SectionDataSize = 0;
 807   uint64_t SectionDataFileSize = 0;
 808   uint64_t VMSize = 0;
 809   for (MCAssembler::const_iterator it = Asm.begin(),
 810          ie = Asm.end(); it != ie; ++it) {
 811     const MCSectionData &SD = *it;
 812     uint64_t Address = getSectionAddress(&SD);
 813     uint64_t Size = Layout.getSectionAddressSize(&SD);
 814     uint64_t FileSize = Layout.getSectionFileSize(&SD);
 815     FileSize += getPaddingSize(&SD, Layout);
 816
 817     VMSize = std::max(VMSize, Address + Size);
 818
 819     if (SD.getSection().isVirtualSection())
 820       continue;
 821
 822     SectionDataSize = std::max(SectionDataSize, Address + Size);
 823     SectionDataFileSize = std::max(SectionDataFileSize, Address + FileSize);
 824   }
 825
 826   // The section data is padded to 4 bytes.
 827   //
 828   // FIXME: Is this machine dependent?
 829   unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
 830   SectionDataFileSize += SectionDataPadding;
 831
 832   // Write the prolog, starting with the header and load command...
 833   WriteHeader(NumLoadCommands, LoadCommandsSize,
 834               Asm.getSubsectionsViaSymbols());
 835   WriteSegmentLoadCommand(NumSections, VMSize,
 836                           SectionDataStart, SectionDataSize);
 837
 838   // ... and then the section headers.
 839   uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
 840   for (MCAssembler::const_iterator it = Asm.begin(),
 841          ie = Asm.end(); it != ie; ++it) {
 842     std::vector<MachO::any_relocation_info> &Relocs = Relocations[it];
 843     unsigned NumRelocs = Relocs.size();
 844     uint64_t SectionStart = SectionDataStart + getSectionAddress(it);
 845     WriteSection(Asm, Layout, *it, SectionStart, RelocTableEnd, NumRelocs);
 846     RelocTableEnd += NumRelocs * sizeof(MachO::any_relocation_info);
 847   }
 848
 849   // Write out the deployment target information, if it's available.
 850   if (VersionInfo.Major != 0) {
 851     assert(VersionInfo.Update < 256 && "unencodable update target version");
 852     assert(VersionInfo.Minor < 256 && "unencodable minor target version");
 853     assert(VersionInfo.Major < 65536 && "unencodable major target version");
 854     uint32_t EncodedVersion = VersionInfo.Update | (VersionInfo.Minor << 8) |
 855       (VersionInfo.Major << 16);
 856     Write32(VersionInfo.Kind == MCVM_OSXVersionMin ? MachO::LC_VERSION_MIN_MACOSX :
 857             MachO::LC_VERSION_MIN_IPHONEOS);
 858     Write32(sizeof(MachO::version_min_command));
 859     Write32(EncodedVersion);
 860     Write32(0);         // reserved.
 861   }
 862
 863   // Write the data-in-code load command, if used.
 864   uint64_t DataInCodeTableEnd = RelocTableEnd + NumDataRegions * 8;
 865   if (NumDataRegions) {
 866     uint64_t DataRegionsOffset = RelocTableEnd;
 867     uint64_t DataRegionsSize = NumDataRegions * 8;
 868     WriteLinkeditLoadCommand(MachO::LC_DATA_IN_CODE, DataRegionsOffset,
 869                              DataRegionsSize);
 870   }
 871
 872   // Write the loh load command, if used.
 873   uint64_t LOHTableEnd = DataInCodeTableEnd + LOHSize;
 874   if (LOHSize)
 875     WriteLinkeditLoadCommand(MachO::LC_LINKER_OPTIMIZATION_HINT,
 876                              DataInCodeTableEnd, LOHSize);
 877
 878   // Write the symbol table load command, if used.
 879   if (NumSymbols) {
 880     unsigned FirstLocalSymbol = 0;
 881     unsigned NumLocalSymbols = LocalSymbolData.size();
 882     unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
 883     unsigned NumExternalSymbols = ExternalSymbolData.size();
 884     unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
 885     unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
 886     unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
 887     unsigned NumSymTabSymbols =
 888       NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
 889     uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
 890     uint64_t IndirectSymbolOffset = 0;
 891
 892     // If used, the indirect symbols are written after the section data.
 893     if (NumIndirectSymbols)
 894       IndirectSymbolOffset = LOHTableEnd;
 895
 896     // The symbol table is written after the indirect symbol data.
 897     uint64_t SymbolTableOffset = LOHTableEnd + IndirectSymbolSize;
 898
 899     // The string table is written after symbol table.
 900     uint64_t StringTableOffset =
 901       SymbolTableOffset + NumSymTabSymbols * (is64Bit() ?
 902                                               sizeof(MachO::nlist_64) :
 903                                               sizeof(MachO::nlist));
 904     WriteSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
 905                            StringTableOffset, StringTable.data().size());
 906
 907     WriteDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
 908                              FirstExternalSymbol, NumExternalSymbols,
 909                              FirstUndefinedSymbol, NumUndefinedSymbols,
 910                              IndirectSymbolOffset, NumIndirectSymbols);
 911   }
 912
 913   // Write the linker options load commands.
 914   for (unsigned i = 0, e = LinkerOptions.size(); i != e; ++i) {
 915     WriteLinkerOptionsLoadCommand(LinkerOptions[i]);
 916   }
 917
 918   // Write the actual section data.
 919   for (MCAssembler::const_iterator it = Asm.begin(),
 920          ie = Asm.end(); it != ie; ++it) {
 921     Asm.writeSectionData(it, Layout);
 922
 923     uint64_t Pad = getPaddingSize(it, Layout);
 924     for (unsigned int i = 0; i < Pad; ++i)
 925       Write8(0);
 926   }
 927
 928   // Write the extra padding.
 929   WriteZeros(SectionDataPadding);
 930
 931   // Write the relocation entries.
 932   for (MCAssembler::const_iterator it = Asm.begin(),
 933          ie = Asm.end(); it != ie; ++it) {
 934     // Write the section relocation entries, in reverse order to match 'as'
 935     // (approximately, the exact algorithm is more complicated than this).
 936     std::vector<MachO::any_relocation_info> &Relocs = Relocations[it];
 937     for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
 938       Write32(Relocs[e - i - 1].r_word0);
 939       Write32(Relocs[e - i - 1].r_word1);
 940     }
 941   }
 942
 943   // Write out the data-in-code region payload, if there is one.
 944   for (MCAssembler::const_data_region_iterator
 945          it = Asm.data_region_begin(), ie = Asm.data_region_end();
 946          it != ie; ++it) {
 947     const DataRegionData *Data = &(*it);
 948     uint64_t Start =
 949       getSymbolAddress(&Layout.getAssembler().getSymbolData(*Data->Start),
 950                        Layout);
 951     uint64_t End =
 952       getSymbolAddress(&Layout.getAssembler().getSymbolData(*Data->End),
 953                        Layout);
 954     DEBUG(dbgs() << "data in code region-- kind: " << Data->Kind
 955                  << "  start: " << Start << "(" << Data->Start->getName() << ")"
 956                  << "  end: " << End << "(" << Data->End->getName() << ")"
 957                  << "  size: " << End - Start
 958                  << "\n");
 959     Write32(Start);
 960     Write16(End - Start);
 961     Write16(Data->Kind);
 962   }
 963
 964   // Write out the loh commands, if there is one.
 965   if (LOHSize) {
 966 #ifndef NDEBUG
 967     unsigned Start = OS.tell();
 968 #endif
 969     Asm.getLOHContainer().Emit(*this, Layout);
 970     // Pad to a multiple of the pointer size.
 971     WriteBytes("", OffsetToAlignment(LOHRawSize, is64Bit() ? 8 : 4));
 972     assert(OS.tell() - Start == LOHSize);
 973   }
 974
 975   // Write the symbol table data, if used.
 976   if (NumSymbols) {
 977     // Write the indirect symbol entries.
 978     for (MCAssembler::const_indirect_symbol_iterator
 979            it = Asm.indirect_symbol_begin(),
 980            ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 981       // Indirect symbols in the non-lazy symbol pointer section have some
 982       // special handling.
 983       const MCSectionMachO &Section =
 984         static_cast<const MCSectionMachO&>(it->SectionData->getSection());
 985       if (Section.getType() == MachO::S_NON_LAZY_SYMBOL_POINTERS) {
 986         // If this symbol is defined and internal, mark it as such.
 987         if (it->Symbol->isDefined() &&
 988             !Asm.getSymbolData(*it->Symbol).isExternal()) {
 989           uint32_t Flags = MachO::INDIRECT_SYMBOL_LOCAL;
 990           if (it->Symbol->isAbsolute())
 991             Flags |= MachO::INDIRECT_SYMBOL_ABS;
 992           Write32(Flags);
 993           continue;
 994         }
 995       }
 996
 997       Write32(Asm.getSymbolData(*it->Symbol).getIndex());
 998     }
 999
1000     // FIXME: Check that offsets match computed ones.
1001
1002     // Write the symbol table entries.
1003     for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
1004       WriteNlist(LocalSymbolData[i], Layout);
1005     for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
1006       WriteNlist(ExternalSymbolData[i], Layout);
1007     for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
1008       WriteNlist(UndefinedSymbolData[i], Layout);
1009
1010     // Write the string table.
1011     OS << StringTable.data();
1012   }
1013 }
1014
1015 MCObjectWriter *llvm::createMachObjectWriter(MCMachObjectTargetWriter *MOTW,
1016                                              raw_ostream &OS,
1017                                              bool IsLittleEndian) {
1018   return new MachObjectWriter(MOTW, OS, IsLittleEndian);
1019 }