lib/MC/MachObjectWriter.cpp

   1 //===- lib/MC/MachObjectWriter.cpp - Mach-O File Writer -------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #include "llvm/ADT/DenseMap.h"
  10 #include "llvm/ADT/Twine.h"
  11 #include "llvm/ADT/iterator_range.h"
  12 #include "llvm/BinaryFormat/MachO.h"
  13 #include "llvm/MC/MCAsmBackend.h"
  14 #include "llvm/MC/MCAsmLayout.h"
  15 #include "llvm/MC/MCAssembler.h"
  16 #include "llvm/MC/MCDirectives.h"
  17 #include "llvm/MC/MCExpr.h"
  18 #include "llvm/MC/MCFixupKindInfo.h"
  19 #include "llvm/MC/MCFragment.h"
  20 #include "llvm/MC/MCMachObjectWriter.h"
  21 #include "llvm/MC/MCObjectWriter.h"
  22 #include "llvm/MC/MCSection.h"
  23 #include "llvm/MC/MCSectionMachO.h"
  24 #include "llvm/MC/MCSymbol.h"
  25 #include "llvm/MC/MCSymbolMachO.h"
  26 #include "llvm/MC/MCValue.h"
  27 #include "llvm/Support/Casting.h"
  28 #include "llvm/Support/Debug.h"
  29 #include "llvm/Support/ErrorHandling.h"
  30 #include "llvm/Support/MathExtras.h"
  31 #include "llvm/Support/raw_ostream.h"
  32 #include <algorithm>
  33 #include <cassert>
  34 #include <cstdint>
  35 #include <string>
  36 #include <utility>
  37 #include <vector>
  38
  39 using namespace llvm;
  40
  41 #define DEBUG_TYPE "mc"
  42
  43 void MachObjectWriter::reset() {
  44   Relocations.clear();
  45   IndirectSymBase.clear();
  46   StringTable.clear();
  47   LocalSymbolData.clear();
  48   ExternalSymbolData.clear();
  49   UndefinedSymbolData.clear();
  50   MCObjectWriter::reset();
  51 }
  52
  53 bool MachObjectWriter::doesSymbolRequireExternRelocation(const MCSymbol &S) {
  54   // Undefined symbols are always extern.
  55   if (S.isUndefined())
  56     return true;
  57
  58   // References to weak definitions require external relocation entries; the
  59   // definition may not always be the one in the same object file.
  60   if (cast<MCSymbolMachO>(S).isWeakDefinition())
  61     return true;
  62
  63   // Otherwise, we can use an internal relocation.
  64   return false;
  65 }
  66
  67 bool MachObjectWriter::
  68 MachSymbolData::operator<(const MachSymbolData &RHS) const {
  69   return Symbol->getName() < RHS.Symbol->getName();
  70 }
  71
  72 bool MachObjectWriter::isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) {
  73   const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo(
  74     (MCFixupKind) Kind);
  75
  76   return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel;
  77 }
  78
  79 uint64_t MachObjectWriter::getFragmentAddress(const MCFragment *Fragment,
  80                                               const MCAsmLayout &Layout) const {
  81   return getSectionAddress(Fragment->getParent()) +
  82          Layout.getFragmentOffset(Fragment);
  83 }
  84
  85 uint64_t MachObjectWriter::getSymbolAddress(const MCSymbol &S,
  86                                             const MCAsmLayout &Layout) const {
  87   // If this is a variable, then recursively evaluate now.
  88   if (S.isVariable()) {
  89     if (const MCConstantExpr *C =
  90           dyn_cast<const MCConstantExpr>(S.getVariableValue()))
  91       return C->getValue();
  92
  93     MCValue Target;
  94     if (!S.getVariableValue()->evaluateAsRelocatable(Target, &Layout, nullptr))
  95       report_fatal_error("unable to evaluate offset for variable '" +
  96                          S.getName() + "'");
  97
  98     // Verify that any used symbols are defined.
  99     if (Target.getSymA() && Target.getSymA()->getSymbol().isUndefined())
 100       report_fatal_error("unable to evaluate offset to undefined symbol '" +
 101                          Target.getSymA()->getSymbol().getName() + "'");
 102     if (Target.getSymB() && Target.getSymB()->getSymbol().isUndefined())
 103       report_fatal_error("unable to evaluate offset to undefined symbol '" +
 104                          Target.getSymB()->getSymbol().getName() + "'");
 105
 106     uint64_t Address = Target.getConstant();
 107     if (Target.getSymA())
 108       Address += getSymbolAddress(Target.getSymA()->getSymbol(), Layout);
 109     if (Target.getSymB())
 110       Address += getSymbolAddress(Target.getSymB()->getSymbol(), Layout);
 111     return Address;
 112   }
 113
 114   return getSectionAddress(S.getFragment()->getParent()) +
 115          Layout.getSymbolOffset(S);
 116 }
 117
 118 uint64_t MachObjectWriter::getPaddingSize(const MCSection *Sec,
 119                                           const MCAsmLayout &Layout) const {
 120   uint64_t EndAddr = getSectionAddress(Sec) + Layout.getSectionAddressSize(Sec);
 121   unsigned Next = Sec->getLayoutOrder() + 1;
 122   if (Next >= Layout.getSectionOrder().size())
 123     return 0;
 124
 125   const MCSection &NextSec = *Layout.getSectionOrder()[Next];
 126   if (NextSec.isVirtualSection())
 127     return 0;
 128   return OffsetToAlignment(EndAddr, NextSec.getAlignment());
 129 }
 130
 131 void MachObjectWriter::writeHeader(MachO::HeaderFileType Type,
 132                                    unsigned NumLoadCommands,
 133                                    unsigned LoadCommandsSize,
 134                                    bool SubsectionsViaSymbols) {
 135   uint32_t Flags = 0;
 136
 137   if (SubsectionsViaSymbols)
 138     Flags |= MachO::MH_SUBSECTIONS_VIA_SYMBOLS;
 139
 140   // struct mach_header (28 bytes) or
 141   // struct mach_header_64 (32 bytes)
 142
 143   uint64_t Start = W.OS.tell();
 144   (void) Start;
 145
 146   W.write<uint32_t>(is64Bit() ? MachO::MH_MAGIC_64 : MachO::MH_MAGIC);
 147
 148   W.write<uint32_t>(TargetObjectWriter->getCPUType());
 149   W.write<uint32_t>(TargetObjectWriter->getCPUSubtype());
 150
 151   W.write<uint32_t>(Type);
 152   W.write<uint32_t>(NumLoadCommands);
 153   W.write<uint32_t>(LoadCommandsSize);
 154   W.write<uint32_t>(Flags);
 155   if (is64Bit())
 156     W.write<uint32_t>(0); // reserved
 157
 158   assert(W.OS.tell() - Start == (is64Bit() ? sizeof(MachO::mach_header_64)
 159                                            : sizeof(MachO::mach_header)));
 160 }
 161
 162 void MachObjectWriter::writeWithPadding(StringRef Str, uint64_t Size) {
 163   assert(Size >= Str.size());
 164   W.OS << Str;
 165   W.OS.write_zeros(Size - Str.size());
 166 }
 167
 168 /// writeSegmentLoadCommand - Write a segment load command.
 169 ///
 170 /// \param NumSections The number of sections in this segment.
 171 /// \param SectionDataSize The total size of the sections.
 172 void MachObjectWriter::writeSegmentLoadCommand(
 173     StringRef Name, unsigned NumSections, uint64_t VMAddr, uint64_t VMSize,
 174     uint64_t SectionDataStartOffset, uint64_t SectionDataSize, uint32_t MaxProt,
 175     uint32_t InitProt) {
 176   // struct segment_command (56 bytes) or
 177   // struct segment_command_64 (72 bytes)
 178
 179   uint64_t Start = W.OS.tell();
 180   (void) Start;
 181
 182   unsigned SegmentLoadCommandSize =
 183     is64Bit() ? sizeof(MachO::segment_command_64):
 184     sizeof(MachO::segment_command);
 185   W.write<uint32_t>(is64Bit() ? MachO::LC_SEGMENT_64 : MachO::LC_SEGMENT);
 186   W.write<uint32_t>(SegmentLoadCommandSize +
 187           NumSections * (is64Bit() ? sizeof(MachO::section_64) :
 188                          sizeof(MachO::section)));
 189
 190   writeWithPadding(Name, 16);
 191   if (is64Bit()) {
 192     W.write<uint64_t>(VMAddr);                 // vmaddr
 193     W.write<uint64_t>(VMSize); // vmsize
 194     W.write<uint64_t>(SectionDataStartOffset); // file offset
 195     W.write<uint64_t>(SectionDataSize); // file size
 196   } else {
 197     W.write<uint32_t>(VMAddr);                 // vmaddr
 198     W.write<uint32_t>(VMSize); // vmsize
 199     W.write<uint32_t>(SectionDataStartOffset); // file offset
 200     W.write<uint32_t>(SectionDataSize); // file size
 201   }
 202   // maxprot
 203   W.write<uint32_t>(MaxProt);
 204   // initprot
 205   W.write<uint32_t>(InitProt);
 206   W.write<uint32_t>(NumSections);
 207   W.write<uint32_t>(0); // flags
 208
 209   assert(W.OS.tell() - Start == SegmentLoadCommandSize);
 210 }
 211
 212 void MachObjectWriter::writeSection(const MCAsmLayout &Layout,
 213                                     const MCSection &Sec, uint64_t VMAddr,
 214                                     uint64_t FileOffset, unsigned Flags,
 215                                     uint64_t RelocationsStart,
 216                                     unsigned NumRelocations) {
 217   uint64_t SectionSize = Layout.getSectionAddressSize(&Sec);
 218   const MCSectionMachO &Section = cast<MCSectionMachO>(Sec);
 219
 220   // The offset is unused for virtual sections.
 221   if (Section.isVirtualSection()) {
 222     assert(Layout.getSectionFileSize(&Sec) == 0 && "Invalid file size!");
 223     FileOffset = 0;
 224   }
 225
 226   // struct section (68 bytes) or
 227   // struct section_64 (80 bytes)
 228
 229   uint64_t Start = W.OS.tell();
 230   (void) Start;
 231
 232   writeWithPadding(Section.getSectionName(), 16);
 233   writeWithPadding(Section.getSegmentName(), 16);
 234   if (is64Bit()) {
 235     W.write<uint64_t>(VMAddr);      // address
 236     W.write<uint64_t>(SectionSize); // size
 237   } else {
 238     W.write<uint32_t>(VMAddr);      // address
 239     W.write<uint32_t>(SectionSize); // size
 240   }
 241   W.write<uint32_t>(FileOffset);
 242
 243   assert(isPowerOf2_32(Section.getAlignment()) && "Invalid alignment!");
 244   W.write<uint32_t>(Log2_32(Section.getAlignment()));
 245   W.write<uint32_t>(NumRelocations ? RelocationsStart : 0);
 246   W.write<uint32_t>(NumRelocations);
 247   W.write<uint32_t>(Flags);
 248   W.write<uint32_t>(IndirectSymBase.lookup(&Sec)); // reserved1
 249   W.write<uint32_t>(Section.getStubSize()); // reserved2
 250   if (is64Bit())
 251     W.write<uint32_t>(0); // reserved3
 252
 253   assert(W.OS.tell() - Start ==
 254          (is64Bit() ? sizeof(MachO::section_64) : sizeof(MachO::section)));
 255 }
 256
 257 void MachObjectWriter::writeSymtabLoadCommand(uint32_t SymbolOffset,
 258                                               uint32_t NumSymbols,
 259                                               uint32_t StringTableOffset,
 260                                               uint32_t StringTableSize) {
 261   // struct symtab_command (24 bytes)
 262
 263   uint64_t Start = W.OS.tell();
 264   (void) Start;
 265
 266   W.write<uint32_t>(MachO::LC_SYMTAB);
 267   W.write<uint32_t>(sizeof(MachO::symtab_command));
 268   W.write<uint32_t>(SymbolOffset);
 269   W.write<uint32_t>(NumSymbols);
 270   W.write<uint32_t>(StringTableOffset);
 271   W.write<uint32_t>(StringTableSize);
 272
 273   assert(W.OS.tell() - Start == sizeof(MachO::symtab_command));
 274 }
 275
 276 void MachObjectWriter::writeDysymtabLoadCommand(uint32_t FirstLocalSymbol,
 277                                                 uint32_t NumLocalSymbols,
 278                                                 uint32_t FirstExternalSymbol,
 279                                                 uint32_t NumExternalSymbols,
 280                                                 uint32_t FirstUndefinedSymbol,
 281                                                 uint32_t NumUndefinedSymbols,
 282                                                 uint32_t IndirectSymbolOffset,
 283                                                 uint32_t NumIndirectSymbols) {
 284   // struct dysymtab_command (80 bytes)
 285
 286   uint64_t Start = W.OS.tell();
 287   (void) Start;
 288
 289   W.write<uint32_t>(MachO::LC_DYSYMTAB);
 290   W.write<uint32_t>(sizeof(MachO::dysymtab_command));
 291   W.write<uint32_t>(FirstLocalSymbol);
 292   W.write<uint32_t>(NumLocalSymbols);
 293   W.write<uint32_t>(FirstExternalSymbol);
 294   W.write<uint32_t>(NumExternalSymbols);
 295   W.write<uint32_t>(FirstUndefinedSymbol);
 296   W.write<uint32_t>(NumUndefinedSymbols);
 297   W.write<uint32_t>(0); // tocoff
 298   W.write<uint32_t>(0); // ntoc
 299   W.write<uint32_t>(0); // modtaboff
 300   W.write<uint32_t>(0); // nmodtab
 301   W.write<uint32_t>(0); // extrefsymoff
 302   W.write<uint32_t>(0); // nextrefsyms
 303   W.write<uint32_t>(IndirectSymbolOffset);
 304   W.write<uint32_t>(NumIndirectSymbols);
 305   W.write<uint32_t>(0); // extreloff
 306   W.write<uint32_t>(0); // nextrel
 307   W.write<uint32_t>(0); // locreloff
 308   W.write<uint32_t>(0); // nlocrel
 309
 310   assert(W.OS.tell() - Start == sizeof(MachO::dysymtab_command));
 311 }
 312
 313 MachObjectWriter::MachSymbolData *
 314 MachObjectWriter::findSymbolData(const MCSymbol &Sym) {
 315   for (auto *SymbolData :
 316        {&LocalSymbolData, &ExternalSymbolData, &UndefinedSymbolData})
 317     for (MachSymbolData &Entry : *SymbolData)
 318       if (Entry.Symbol == &Sym)
 319         return &Entry;
 320
 321   return nullptr;
 322 }
 323
 324 const MCSymbol &MachObjectWriter::findAliasedSymbol(const MCSymbol &Sym) const {
 325   const MCSymbol *S = &Sym;
 326   while (S->isVariable()) {
 327     const MCExpr *Value = S->getVariableValue();
 328     const auto *Ref = dyn_cast<MCSymbolRefExpr>(Value);
 329     if (!Ref)
 330       return *S;
 331     S = &Ref->getSymbol();
 332   }
 333   return *S;
 334 }
 335
 336 void MachObjectWriter::writeNlist(MachSymbolData &MSD,
 337                                   const MCAsmLayout &Layout) {
 338   const MCSymbol *Symbol = MSD.Symbol;
 339   const MCSymbol &Data = *Symbol;
 340   const MCSymbol *AliasedSymbol = &findAliasedSymbol(*Symbol);
 341   uint8_t SectionIndex = MSD.SectionIndex;
 342   uint8_t Type = 0;
 343   uint64_t Address = 0;
 344   bool IsAlias = Symbol != AliasedSymbol;
 345
 346   const MCSymbol &OrigSymbol = *Symbol;
 347   MachSymbolData *AliaseeInfo;
 348   if (IsAlias) {
 349     AliaseeInfo = findSymbolData(*AliasedSymbol);
 350     if (AliaseeInfo)
 351       SectionIndex = AliaseeInfo->SectionIndex;
 352     Symbol = AliasedSymbol;
 353     // FIXME: Should this update Data as well?
 354   }
 355
 356   // Set the N_TYPE bits. See <mach-o/nlist.h>.
 357   //
 358   // FIXME: Are the prebound or indirect fields possible here?
 359   if (IsAlias && Symbol->isUndefined())
 360     Type = MachO::N_INDR;
 361   else if (Symbol->isUndefined())
 362     Type = MachO::N_UNDF;
 363   else if (Symbol->isAbsolute())
 364     Type = MachO::N_ABS;
 365   else
 366     Type = MachO::N_SECT;
 367
 368   // FIXME: Set STAB bits.
 369
 370   if (Data.isPrivateExtern())
 371     Type |= MachO::N_PEXT;
 372
 373   // Set external bit.
 374   if (Data.isExternal() || (!IsAlias && Symbol->isUndefined()))
 375     Type |= MachO::N_EXT;
 376
 377   // Compute the symbol address.
 378   if (IsAlias && Symbol->isUndefined())
 379     Address = AliaseeInfo->StringIndex;
 380   else if (Symbol->isDefined())
 381     Address = getSymbolAddress(OrigSymbol, Layout);
 382   else if (Symbol->isCommon()) {
 383     // Common symbols are encoded with the size in the address
 384     // field, and their alignment in the flags.
 385     Address = Symbol->getCommonSize();
 386   }
 387
 388   // struct nlist (12 bytes)
 389
 390   W.write<uint32_t>(MSD.StringIndex);
 391   W.OS << char(Type);
 392   W.OS << char(SectionIndex);
 393
 394   // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
 395   // value.
 396   bool EncodeAsAltEntry =
 397     IsAlias && cast<MCSymbolMachO>(OrigSymbol).isAltEntry();
 398   W.write<uint16_t>(cast<MCSymbolMachO>(Symbol)->getEncodedFlags(EncodeAsAltEntry));
 399   if (is64Bit())
 400     W.write<uint64_t>(Address);
 401   else
 402     W.write<uint32_t>(Address);
 403 }
 404
 405 void MachObjectWriter::writeLinkeditLoadCommand(uint32_t Type,
 406                                                 uint32_t DataOffset,
 407                                                 uint32_t DataSize) {
 408   uint64_t Start = W.OS.tell();
 409   (void) Start;
 410
 411   W.write<uint32_t>(Type);
 412   W.write<uint32_t>(sizeof(MachO::linkedit_data_command));
 413   W.write<uint32_t>(DataOffset);
 414   W.write<uint32_t>(DataSize);
 415
 416   assert(W.OS.tell() - Start == sizeof(MachO::linkedit_data_command));
 417 }
 418
 419 static unsigned ComputeLinkerOptionsLoadCommandSize(
 420   const std::vector<std::string> &Options, bool is64Bit)
 421 {
 422   unsigned Size = sizeof(MachO::linker_option_command);
 423   for (const std::string &Option : Options)
 424     Size += Option.size() + 1;
 425   return alignTo(Size, is64Bit ? 8 : 4);
 426 }
 427
 428 void MachObjectWriter::writeLinkerOptionsLoadCommand(
 429   const std::vector<std::string> &Options)
 430 {
 431   unsigned Size = ComputeLinkerOptionsLoadCommandSize(Options, is64Bit());
 432   uint64_t Start = W.OS.tell();
 433   (void) Start;
 434
 435   W.write<uint32_t>(MachO::LC_LINKER_OPTION);
 436   W.write<uint32_t>(Size);
 437   W.write<uint32_t>(Options.size());
 438   uint64_t BytesWritten = sizeof(MachO::linker_option_command);
 439   for (const std::string &Option : Options) {
 440     // Write each string, including the null byte.
 441     W.OS << Option << '\0';
 442     BytesWritten += Option.size() + 1;
 443   }
 444
 445   // Pad to a multiple of the pointer size.
 446   W.OS.write_zeros(OffsetToAlignment(BytesWritten, is64Bit() ? 8 : 4));
 447
 448   assert(W.OS.tell() - Start == Size);
 449 }
 450
 451 void MachObjectWriter::recordRelocation(MCAssembler &Asm,
 452                                         const MCAsmLayout &Layout,
 453                                         const MCFragment *Fragment,
 454                                         const MCFixup &Fixup, MCValue Target,
 455                                         uint64_t &FixedValue) {
 456   TargetObjectWriter->recordRelocation(this, Asm, Layout, Fragment, Fixup,
 457                                        Target, FixedValue);
 458 }
 459
 460 void MachObjectWriter::bindIndirectSymbols(MCAssembler &Asm) {
 461   // This is the point where 'as' creates actual symbols for indirect symbols
 462   // (in the following two passes). It would be easier for us to do this sooner
 463   // when we see the attribute, but that makes getting the order in the symbol
 464   // table much more complicated than it is worth.
 465   //
 466   // FIXME: Revisit this when the dust settles.
 467
 468   // Report errors for use of .indirect_symbol not in a symbol pointer section
 469   // or stub section.
 470   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 471          ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 472     const MCSectionMachO &Section = cast<MCSectionMachO>(*it->Section);
 473
 474     if (Section.getType() != MachO::S_NON_LAZY_SYMBOL_POINTERS &&
 475         Section.getType() != MachO::S_LAZY_SYMBOL_POINTERS &&
 476         Section.getType() != MachO::S_THREAD_LOCAL_VARIABLE_POINTERS &&
 477         Section.getType() != MachO::S_SYMBOL_STUBS) {
 478       MCSymbol &Symbol = *it->Symbol;
 479       report_fatal_error("indirect symbol '" + Symbol.getName() +
 480                          "' not in a symbol pointer or stub section");
 481     }
 482   }
 483
 484   // Bind non-lazy symbol pointers first.
 485   unsigned IndirectIndex = 0;
 486   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 487          ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
 488     const MCSectionMachO &Section = cast<MCSectionMachO>(*it->Section);
 489
 490     if (Section.getType() != MachO::S_NON_LAZY_SYMBOL_POINTERS &&
 491         Section.getType() !=  MachO::S_THREAD_LOCAL_VARIABLE_POINTERS)
 492       continue;
 493
 494     // Initialize the section indirect symbol base, if necessary.
 495     IndirectSymBase.insert(std::make_pair(it->Section, IndirectIndex));
 496
 497     Asm.registerSymbol(*it->Symbol);
 498   }
 499
 500   // Then lazy symbol pointers and symbol stubs.
 501   IndirectIndex = 0;
 502   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 503          ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
 504     const MCSectionMachO &Section = cast<MCSectionMachO>(*it->Section);
 505
 506     if (Section.getType() != MachO::S_LAZY_SYMBOL_POINTERS &&
 507         Section.getType() != MachO::S_SYMBOL_STUBS)
 508       continue;
 509
 510     // Initialize the section indirect symbol base, if necessary.
 511     IndirectSymBase.insert(std::make_pair(it->Section, IndirectIndex));
 512
 513     // Set the symbol type to undefined lazy, but only on construction.
 514     //
 515     // FIXME: Do not hardcode.
 516     bool Created;
 517     Asm.registerSymbol(*it->Symbol, &Created);
 518     if (Created)
 519       cast<MCSymbolMachO>(it->Symbol)->setReferenceTypeUndefinedLazy(true);
 520   }
 521 }
 522
 523 /// computeSymbolTable - Compute the symbol table data
 524 void MachObjectWriter::computeSymbolTable(
 525     MCAssembler &Asm, std::vector<MachSymbolData> &LocalSymbolData,
 526     std::vector<MachSymbolData> &ExternalSymbolData,
 527     std::vector<MachSymbolData> &UndefinedSymbolData) {
 528   // Build section lookup table.
 529   DenseMap<const MCSection*, uint8_t> SectionIndexMap;
 530   unsigned Index = 1;
 531   for (MCAssembler::iterator it = Asm.begin(),
 532          ie = Asm.end(); it != ie; ++it, ++Index)
 533     SectionIndexMap[&*it] = Index;
 534   assert(Index <= 256 && "Too many sections!");
 535
 536   // Build the string table.
 537   for (const MCSymbol &Symbol : Asm.symbols()) {
 538     if (!Asm.isSymbolLinkerVisible(Symbol))
 539       continue;
 540
 541     StringTable.add(Symbol.getName());
 542   }
 543   StringTable.finalize();
 544
 545   // Build the symbol arrays but only for non-local symbols.
 546   //
 547   // The particular order that we collect and then sort the symbols is chosen to
 548   // match 'as'. Even though it doesn't matter for correctness, this is
 549   // important for letting us diff .o files.
 550   for (const MCSymbol &Symbol : Asm.symbols()) {
 551     // Ignore non-linker visible symbols.
 552     if (!Asm.isSymbolLinkerVisible(Symbol))
 553       continue;
 554
 555     if (!Symbol.isExternal() && !Symbol.isUndefined())
 556       continue;
 557
 558     MachSymbolData MSD;
 559     MSD.Symbol = &Symbol;
 560     MSD.StringIndex = StringTable.getOffset(Symbol.getName());
 561
 562     if (Symbol.isUndefined()) {
 563       MSD.SectionIndex = 0;
 564       UndefinedSymbolData.push_back(MSD);
 565     } else if (Symbol.isAbsolute()) {
 566       MSD.SectionIndex = 0;
 567       ExternalSymbolData.push_back(MSD);
 568     } else {
 569       MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
 570       assert(MSD.SectionIndex && "Invalid section index!");
 571       ExternalSymbolData.push_back(MSD);
 572     }
 573   }
 574
 575   // Now add the data for local symbols.
 576   for (const MCSymbol &Symbol : Asm.symbols()) {
 577     // Ignore non-linker visible symbols.
 578     if (!Asm.isSymbolLinkerVisible(Symbol))
 579       continue;
 580
 581     if (Symbol.isExternal() || Symbol.isUndefined())
 582       continue;
 583
 584     MachSymbolData MSD;
 585     MSD.Symbol = &Symbol;
 586     MSD.StringIndex = StringTable.getOffset(Symbol.getName());
 587
 588     if (Symbol.isAbsolute()) {
 589       MSD.SectionIndex = 0;
 590       LocalSymbolData.push_back(MSD);
 591     } else {
 592       MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
 593       assert(MSD.SectionIndex && "Invalid section index!");
 594       LocalSymbolData.push_back(MSD);
 595     }
 596   }
 597
 598   // External and undefined symbols are required to be in lexicographic order.
 599   llvm::sort(ExternalSymbolData);
 600   llvm::sort(UndefinedSymbolData);
 601
 602   // Set the symbol indices.
 603   Index = 0;
 604   for (auto *SymbolData :
 605        {&LocalSymbolData, &ExternalSymbolData, &UndefinedSymbolData})
 606     for (MachSymbolData &Entry : *SymbolData)
 607       Entry.Symbol->setIndex(Index++);
 608
 609   for (const MCSection &Section : Asm) {
 610     for (RelAndSymbol &Rel : Relocations[&Section]) {
 611       if (!Rel.Sym)
 612         continue;
 613
 614       // Set the Index and the IsExtern bit.
 615       unsigned Index = Rel.Sym->getIndex();
 616       assert(isInt<24>(Index));
 617       if (W.Endian == support::little)
 618         Rel.MRE.r_word1 = (Rel.MRE.r_word1 & (~0U << 24)) | Index | (1 << 27);
 619       else
 620         Rel.MRE.r_word1 = (Rel.MRE.r_word1 & 0xff) | Index << 8 | (1 << 4);
 621     }
 622   }
 623 }
 624
 625 void MachObjectWriter::computeSectionAddresses(const MCAssembler &Asm,
 626                                                const MCAsmLayout &Layout) {
 627   uint64_t StartAddress = 0;
 628   for (const MCSection *Sec : Layout.getSectionOrder()) {
 629     StartAddress = alignTo(StartAddress, Sec->getAlignment());
 630     SectionAddress[Sec] = StartAddress;
 631     StartAddress += Layout.getSectionAddressSize(Sec);
 632
 633     // Explicitly pad the section to match the alignment requirements of the
 634     // following one. This is for 'gas' compatibility, it shouldn't
 635     /// strictly be necessary.
 636     StartAddress += getPaddingSize(Sec, Layout);
 637   }
 638 }
 639
 640 void MachObjectWriter::executePostLayoutBinding(MCAssembler &Asm,
 641                                                 const MCAsmLayout &Layout) {
 642   computeSectionAddresses(Asm, Layout);
 643
 644   // Create symbol data for any indirect symbols.
 645   bindIndirectSymbols(Asm);
 646 }
 647
 648 bool MachObjectWriter::isSymbolRefDifferenceFullyResolvedImpl(
 649     const MCAssembler &Asm, const MCSymbol &A, const MCSymbol &B,
 650     bool InSet) const {
 651   // FIXME: We don't handle things like
 652   // foo = .
 653   // creating atoms.
 654   if (A.isVariable() || B.isVariable())
 655     return false;
 656   return MCObjectWriter::isSymbolRefDifferenceFullyResolvedImpl(Asm, A, B,
 657                                                                 InSet);
 658 }
 659
 660 bool MachObjectWriter::isSymbolRefDifferenceFullyResolvedImpl(
 661     const MCAssembler &Asm, const MCSymbol &SymA, const MCFragment &FB,
 662     bool InSet, bool IsPCRel) const {
 663   if (InSet)
 664     return true;
 665
 666   // The effective address is
 667   //     addr(atom(A)) + offset(A)
 668   //   - addr(atom(B)) - offset(B)
 669   // and the offsets are not relocatable, so the fixup is fully resolved when
 670   //  addr(atom(A)) - addr(atom(B)) == 0.
 671   const MCSymbol &SA = findAliasedSymbol(SymA);
 672   const MCSection &SecA = SA.getSection();
 673   const MCSection &SecB = *FB.getParent();
 674
 675   if (IsPCRel) {
 676     // The simple (Darwin, except on x86_64) way of dealing with this was to
 677     // assume that any reference to a temporary symbol *must* be a temporary
 678     // symbol in the same atom, unless the sections differ. Therefore, any PCrel
 679     // relocation to a temporary symbol (in the same section) is fully
 680     // resolved. This also works in conjunction with absolutized .set, which
 681     // requires the compiler to use .set to absolutize the differences between
 682     // symbols which the compiler knows to be assembly time constants, so we
 683     // don't need to worry about considering symbol differences fully resolved.
 684     //
 685     // If the file isn't using sub-sections-via-symbols, we can make the
 686     // same assumptions about any symbol that we normally make about
 687     // assembler locals.
 688
 689     bool hasReliableSymbolDifference = isX86_64();
 690     if (!hasReliableSymbolDifference) {
 691       if (!SA.isInSection() || &SecA != &SecB ||
 692           (!SA.isTemporary() && FB.getAtom() != SA.getFragment()->getAtom() &&
 693            Asm.getSubsectionsViaSymbols()))
 694         return false;
 695       return true;
 696     }
 697     // For Darwin x86_64, there is one special case when the reference IsPCRel.
 698     // If the fragment with the reference does not have a base symbol but meets
 699     // the simple way of dealing with this, in that it is a temporary symbol in
 700     // the same atom then it is assumed to be fully resolved.  This is needed so
 701     // a relocation entry is not created and so the static linker does not
 702     // mess up the reference later.
 703     else if(!FB.getAtom() &&
 704             SA.isTemporary() && SA.isInSection() && &SecA == &SecB){
 705       return true;
 706     }
 707   }
 708
 709   // If they are not in the same section, we can't compute the diff.
 710   if (&SecA != &SecB)
 711     return false;
 712
 713   const MCFragment *FA = SA.getFragment();
 714
 715   // Bail if the symbol has no fragment.
 716   if (!FA)
 717     return false;
 718
 719   // If the atoms are the same, they are guaranteed to have the same address.
 720   if (FA->getAtom() == FB.getAtom())
 721     return true;
 722
 723   // Otherwise, we can't prove this is fully resolved.
 724   return false;
 725 }
 726
 727 static MachO::LoadCommandType getLCFromMCVM(MCVersionMinType Type) {
 728   switch (Type) {
 729   case MCVM_OSXVersionMin:     return MachO::LC_VERSION_MIN_MACOSX;
 730   case MCVM_IOSVersionMin:     return MachO::LC_VERSION_MIN_IPHONEOS;
 731   case MCVM_TvOSVersionMin:    return MachO::LC_VERSION_MIN_TVOS;
 732   case MCVM_WatchOSVersionMin: return MachO::LC_VERSION_MIN_WATCHOS;
 733   }
 734   llvm_unreachable("Invalid mc version min type");
 735 }
 736
 737 uint64_t MachObjectWriter::writeObject(MCAssembler &Asm,
 738                                        const MCAsmLayout &Layout) {
 739   uint64_t StartOffset = W.OS.tell();
 740
 741   // Compute symbol table information and bind symbol indices.
 742   computeSymbolTable(Asm, LocalSymbolData, ExternalSymbolData,
 743                      UndefinedSymbolData);
 744
 745   unsigned NumSections = Asm.size();
 746   const MCAssembler::VersionInfoType &VersionInfo =
 747     Layout.getAssembler().getVersionInfo();
 748
 749   // The section data starts after the header, the segment load command (and
 750   // section headers) and the symbol table.
 751   unsigned NumLoadCommands = 1;
 752   uint64_t LoadCommandsSize = is64Bit() ?
 753     sizeof(MachO::segment_command_64) + NumSections * sizeof(MachO::section_64):
 754     sizeof(MachO::segment_command) + NumSections * sizeof(MachO::section);
 755
 756   // Add the deployment target version info load command size, if used.
 757   if (VersionInfo.Major != 0) {
 758     ++NumLoadCommands;
 759     if (VersionInfo.EmitBuildVersion)
 760       LoadCommandsSize += sizeof(MachO::build_version_command);
 761     else
 762       LoadCommandsSize += sizeof(MachO::version_min_command);
 763   }
 764
 765   // Add the data-in-code load command size, if used.
 766   unsigned NumDataRegions = Asm.getDataRegions().size();
 767   if (NumDataRegions) {
 768     ++NumLoadCommands;
 769     LoadCommandsSize += sizeof(MachO::linkedit_data_command);
 770   }
 771
 772   // Add the loh load command size, if used.
 773   uint64_t LOHRawSize = Asm.getLOHContainer().getEmitSize(*this, Layout);
 774   uint64_t LOHSize = alignTo(LOHRawSize, is64Bit() ? 8 : 4);
 775   if (LOHSize) {
 776     ++NumLoadCommands;
 777     LoadCommandsSize += sizeof(MachO::linkedit_data_command);
 778   }
 779
 780   // Add the symbol table load command sizes, if used.
 781   unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() +
 782     UndefinedSymbolData.size();
 783   if (NumSymbols) {
 784     NumLoadCommands += 2;
 785     LoadCommandsSize += (sizeof(MachO::symtab_command) +
 786                          sizeof(MachO::dysymtab_command));
 787   }
 788
 789   // Add the linker option load commands sizes.
 790   for (const auto &Option : Asm.getLinkerOptions()) {
 791     ++NumLoadCommands;
 792     LoadCommandsSize += ComputeLinkerOptionsLoadCommandSize(Option, is64Bit());
 793   }
 794
 795   // Compute the total size of the section data, as well as its file size and vm
 796   // size.
 797   uint64_t SectionDataStart = (is64Bit() ? sizeof(MachO::mach_header_64) :
 798                                sizeof(MachO::mach_header)) + LoadCommandsSize;
 799   uint64_t SectionDataSize = 0;
 800   uint64_t SectionDataFileSize = 0;
 801   uint64_t VMSize = 0;
 802   for (const MCSection &Sec : Asm) {
 803     uint64_t Address = getSectionAddress(&Sec);
 804     uint64_t Size = Layout.getSectionAddressSize(&Sec);
 805     uint64_t FileSize = Layout.getSectionFileSize(&Sec);
 806     FileSize += getPaddingSize(&Sec, Layout);
 807
 808     VMSize = std::max(VMSize, Address + Size);
 809
 810     if (Sec.isVirtualSection())
 811       continue;
 812
 813     SectionDataSize = std::max(SectionDataSize, Address + Size);
 814     SectionDataFileSize = std::max(SectionDataFileSize, Address + FileSize);
 815   }
 816
 817   // The section data is padded to 4 bytes.
 818   //
 819   // FIXME: Is this machine dependent?
 820   unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
 821   SectionDataFileSize += SectionDataPadding;
 822
 823   // Write the prolog, starting with the header and load command...
 824   writeHeader(MachO::MH_OBJECT, NumLoadCommands, LoadCommandsSize,
 825               Asm.getSubsectionsViaSymbols());
 826   uint32_t Prot =
 827       MachO::VM_PROT_READ | MachO::VM_PROT_WRITE | MachO::VM_PROT_EXECUTE;
 828   writeSegmentLoadCommand("", NumSections, 0, VMSize, SectionDataStart,
 829                           SectionDataSize, Prot, Prot);
 830
 831   // ... and then the section headers.
 832   uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
 833   for (const MCSection &Section : Asm) {
 834     const auto &Sec = cast<MCSectionMachO>(Section);
 835     std::vector<RelAndSymbol> &Relocs = Relocations[&Sec];
 836     unsigned NumRelocs = Relocs.size();
 837     uint64_t SectionStart = SectionDataStart + getSectionAddress(&Sec);
 838     unsigned Flags = Sec.getTypeAndAttributes();
 839     if (Sec.hasInstructions())
 840       Flags |= MachO::S_ATTR_SOME_INSTRUCTIONS;
 841     writeSection(Layout, Sec, getSectionAddress(&Sec), SectionStart, Flags,
 842                  RelocTableEnd, NumRelocs);
 843     RelocTableEnd += NumRelocs * sizeof(MachO::any_relocation_info);
 844   }
 845
 846   // Write out the deployment target information, if it's available.
 847   if (VersionInfo.Major != 0) {
 848     auto EncodeVersion = [](VersionTuple V) -> uint32_t {
 849       assert(!V.empty() && "empty version");
 850       unsigned Update = V.getSubminor() ? *V.getSubminor() : 0;
 851       unsigned Minor = V.getMinor() ? *V.getMinor() : 0;
 852       assert(Update < 256 && "unencodable update target version");
 853       assert(Minor < 256 && "unencodable minor target version");
 854       assert(V.getMajor() < 65536 && "unencodable major target version");
 855       return Update | (Minor << 8) | (V.getMajor() << 16);
 856     };
 857     uint32_t EncodedVersion = EncodeVersion(
 858         VersionTuple(VersionInfo.Major, VersionInfo.Minor, VersionInfo.Update));
 859     uint32_t SDKVersion = !VersionInfo.SDKVersion.empty()
 860                               ? EncodeVersion(VersionInfo.SDKVersion)
 861                               : 0;
 862     if (VersionInfo.EmitBuildVersion) {
 863       // FIXME: Currently empty tools. Add clang version in the future.
 864       W.write<uint32_t>(MachO::LC_BUILD_VERSION);
 865       W.write<uint32_t>(sizeof(MachO::build_version_command));
 866       W.write<uint32_t>(VersionInfo.TypeOrPlatform.Platform);
 867       W.write<uint32_t>(EncodedVersion);
 868       W.write<uint32_t>(SDKVersion);
 869       W.write<uint32_t>(0);         // Empty tools list.
 870     } else {
 871       MachO::LoadCommandType LCType
 872         = getLCFromMCVM(VersionInfo.TypeOrPlatform.Type);
 873       W.write<uint32_t>(LCType);
 874       W.write<uint32_t>(sizeof(MachO::version_min_command));
 875       W.write<uint32_t>(EncodedVersion);
 876       W.write<uint32_t>(SDKVersion);
 877     }
 878   }
 879
 880   // Write the data-in-code load command, if used.
 881   uint64_t DataInCodeTableEnd = RelocTableEnd + NumDataRegions * 8;
 882   if (NumDataRegions) {
 883     uint64_t DataRegionsOffset = RelocTableEnd;
 884     uint64_t DataRegionsSize = NumDataRegions * 8;
 885     writeLinkeditLoadCommand(MachO::LC_DATA_IN_CODE, DataRegionsOffset,
 886                              DataRegionsSize);
 887   }
 888
 889   // Write the loh load command, if used.
 890   uint64_t LOHTableEnd = DataInCodeTableEnd + LOHSize;
 891   if (LOHSize)
 892     writeLinkeditLoadCommand(MachO::LC_LINKER_OPTIMIZATION_HINT,
 893                              DataInCodeTableEnd, LOHSize);
 894
 895   // Write the symbol table load command, if used.
 896   if (NumSymbols) {
 897     unsigned FirstLocalSymbol = 0;
 898     unsigned NumLocalSymbols = LocalSymbolData.size();
 899     unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
 900     unsigned NumExternalSymbols = ExternalSymbolData.size();
 901     unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
 902     unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
 903     unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
 904     unsigned NumSymTabSymbols =
 905       NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
 906     uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
 907     uint64_t IndirectSymbolOffset = 0;
 908
 909     // If used, the indirect symbols are written after the section data.
 910     if (NumIndirectSymbols)
 911       IndirectSymbolOffset = LOHTableEnd;
 912
 913     // The symbol table is written after the indirect symbol data.
 914     uint64_t SymbolTableOffset = LOHTableEnd + IndirectSymbolSize;
 915
 916     // The string table is written after symbol table.
 917     uint64_t StringTableOffset =
 918       SymbolTableOffset + NumSymTabSymbols * (is64Bit() ?
 919                                               sizeof(MachO::nlist_64) :
 920                                               sizeof(MachO::nlist));
 921     writeSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
 922                            StringTableOffset, StringTable.getSize());
 923
 924     writeDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
 925                              FirstExternalSymbol, NumExternalSymbols,
 926                              FirstUndefinedSymbol, NumUndefinedSymbols,
 927                              IndirectSymbolOffset, NumIndirectSymbols);
 928   }
 929
 930   // Write the linker options load commands.
 931   for (const auto &Option : Asm.getLinkerOptions())
 932     writeLinkerOptionsLoadCommand(Option);
 933
 934   // Write the actual section data.
 935   for (const MCSection &Sec : Asm) {
 936     Asm.writeSectionData(W.OS, &Sec, Layout);
 937
 938     uint64_t Pad = getPaddingSize(&Sec, Layout);
 939     W.OS.write_zeros(Pad);
 940   }
 941
 942   // Write the extra padding.
 943   W.OS.write_zeros(SectionDataPadding);
 944
 945   // Write the relocation entries.
 946   for (const MCSection &Sec : Asm) {
 947     // Write the section relocation entries, in reverse order to match 'as'
 948     // (approximately, the exact algorithm is more complicated than this).
 949     std::vector<RelAndSymbol> &Relocs = Relocations[&Sec];
 950     for (const RelAndSymbol &Rel : make_range(Relocs.rbegin(), Relocs.rend())) {
 951       W.write<uint32_t>(Rel.MRE.r_word0);
 952       W.write<uint32_t>(Rel.MRE.r_word1);
 953     }
 954   }
 955
 956   // Write out the data-in-code region payload, if there is one.
 957   for (MCAssembler::const_data_region_iterator
 958          it = Asm.data_region_begin(), ie = Asm.data_region_end();
 959          it != ie; ++it) {
 960     const DataRegionData *Data = &(*it);
 961     uint64_t Start = getSymbolAddress(*Data->Start, Layout);
 962     uint64_t End;
 963     if (Data->End)
 964       End = getSymbolAddress(*Data->End, Layout);
 965     else
 966       report_fatal_error("Data region not terminated");
 967
 968     LLVM_DEBUG(dbgs() << "data in code region-- kind: " << Data->Kind
 969                       << "  start: " << Start << "(" << Data->Start->getName()
 970                       << ")"
 971                       << "  end: " << End << "(" << Data->End->getName() << ")"
 972                       << "  size: " << End - Start << "\n");
 973     W.write<uint32_t>(Start);
 974     W.write<uint16_t>(End - Start);
 975     W.write<uint16_t>(Data->Kind);
 976   }
 977
 978   // Write out the loh commands, if there is one.
 979   if (LOHSize) {
 980 #ifndef NDEBUG
 981     unsigned Start = W.OS.tell();
 982 #endif
 983     Asm.getLOHContainer().emit(*this, Layout);
 984     // Pad to a multiple of the pointer size.
 985     W.OS.write_zeros(OffsetToAlignment(LOHRawSize, is64Bit() ? 8 : 4));
 986     assert(W.OS.tell() - Start == LOHSize);
 987   }
 988
 989   // Write the symbol table data, if used.
 990   if (NumSymbols) {
 991     // Write the indirect symbol entries.
 992     for (MCAssembler::const_indirect_symbol_iterator
 993            it = Asm.indirect_symbol_begin(),
 994            ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 995       // Indirect symbols in the non-lazy symbol pointer section have some
 996       // special handling.
 997       const MCSectionMachO &Section =
 998           static_cast<const MCSectionMachO &>(*it->Section);
 999       if (Section.getType() == MachO::S_NON_LAZY_SYMBOL_POINTERS) {
1000         // If this symbol is defined and internal, mark it as such.
1001         if (it->Symbol->isDefined() && !it->Symbol->isExternal()) {
1002           uint32_t Flags = MachO::INDIRECT_SYMBOL_LOCAL;
1003           if (it->Symbol->isAbsolute())
1004             Flags |= MachO::INDIRECT_SYMBOL_ABS;
1005           W.write<uint32_t>(Flags);
1006           continue;
1007         }
1008       }
1009
1010       W.write<uint32_t>(it->Symbol->getIndex());
1011     }
1012
1013     // FIXME: Check that offsets match computed ones.
1014
1015     // Write the symbol table entries.
1016     for (auto *SymbolData :
1017          {&LocalSymbolData, &ExternalSymbolData, &UndefinedSymbolData})
1018       for (MachSymbolData &Entry : *SymbolData)
1019         writeNlist(Entry, Layout);
1020
1021     // Write the string table.
1022     StringTable.write(W.OS);
1023   }
1024
1025   return W.OS.tell() - StartOffset;
1026 }
1027
1028 std::unique_ptr<MCObjectWriter>
1029 llvm::createMachObjectWriter(std::unique_ptr<MCMachObjectTargetWriter> MOTW,
1030                              raw_pwrite_stream &OS, bool IsLittleEndian) {
1031   return llvm::make_unique<MachObjectWriter>(std::move(MOTW), OS,
1032                                              IsLittleEndian);
1033 }