lib/MC/MCAssembler.cpp

   1 //===- lib/MC/MCAssembler.cpp - Assembler Backend Implementation ----------===//
   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 #define DEBUG_TYPE "assembler"
  11 #include "llvm/MC/MCAssembler.h"
  12 #include "llvm/MC/MCSectionMachO.h"
  13 #include "llvm/Target/TargetMachOWriterInfo.h"
  14 #include "llvm/ADT/DenseMap.h"
  15 #include "llvm/ADT/SmallString.h"
  16 #include "llvm/ADT/Statistic.h"
  17 #include "llvm/ADT/StringMap.h"
  18 #include "llvm/ADT/Twine.h"
  19 #include "llvm/Support/ErrorHandling.h"
  20 #include "llvm/Support/raw_ostream.h"
  21 #include <vector>
  22 using namespace llvm;
  23
  24 class MachObjectWriter;
  25
  26 STATISTIC(EmittedFragments, "Number of emitted assembler fragments");
  27
  28 // FIXME FIXME FIXME: There are number of places in this file where we convert
  29 // what is a 64-bit assembler value used for computation into a value in the
  30 // object file, which may truncate it. We should detect that truncation where
  31 // invalid and report errors back.
  32
  33 static void WriteFileData(raw_ostream &OS, const MCSectionData &SD,
  34                           MachObjectWriter &MOW);
  35
  36 /// isVirtualSection - Check if this is a section which does not actually exist
  37 /// in the object file.
  38 static bool isVirtualSection(const MCSection &Section) {
  39   // FIXME: Lame.
  40   const MCSectionMachO &SMO = static_cast<const MCSectionMachO&>(Section);
  41   unsigned Type = SMO.getTypeAndAttributes() & MCSectionMachO::SECTION_TYPE;
  42   return (Type == MCSectionMachO::S_ZEROFILL);
  43 }
  44
  45 class MachObjectWriter {
  46   // See <mach-o/loader.h>.
  47   enum {
  48     Header_Magic32 = 0xFEEDFACE,
  49     Header_Magic64 = 0xFEEDFACF
  50   };
  51
  52   static const unsigned Header32Size = 28;
  53   static const unsigned Header64Size = 32;
  54   static const unsigned SegmentLoadCommand32Size = 56;
  55   static const unsigned Section32Size = 68;
  56   static const unsigned SymtabLoadCommandSize = 24;
  57   static const unsigned DysymtabLoadCommandSize = 80;
  58   static const unsigned Nlist32Size = 12;
  59   static const unsigned RelocationInfoSize = 8;
  60
  61   enum HeaderFileType {
  62     HFT_Object = 0x1
  63   };
  64
  65   enum HeaderFlags {
  66     HF_SubsectionsViaSymbols = 0x2000
  67   };
  68
  69   enum LoadCommandType {
  70     LCT_Segment = 0x1,
  71     LCT_Symtab = 0x2,
  72     LCT_Dysymtab = 0xb
  73   };
  74
  75   // See <mach-o/nlist.h>.
  76   enum SymbolTypeType {
  77     STT_Undefined = 0x00,
  78     STT_Absolute  = 0x02,
  79     STT_Section   = 0x0e
  80   };
  81
  82   enum SymbolTypeFlags {
  83     // If any of these bits are set, then the entry is a stab entry number (see
  84     // <mach-o/stab.h>. Otherwise the other masks apply.
  85     STF_StabsEntryMask = 0xe0,
  86
  87     STF_TypeMask       = 0x0e,
  88     STF_External       = 0x01,
  89     STF_PrivateExtern  = 0x10
  90   };
  91
  92   /// IndirectSymbolFlags - Flags for encoding special values in the indirect
  93   /// symbol entry.
  94   enum IndirectSymbolFlags {
  95     ISF_Local    = 0x80000000,
  96     ISF_Absolute = 0x40000000
  97   };
  98
  99   /// RelocationFlags - Special flags for addresses.
 100   enum RelocationFlags {
 101     RF_Scattered = 0x80000000
 102   };
 103
 104   enum RelocationInfoType {
 105     RIT_Vanilla             = 0,
 106     RIT_Pair                = 1,
 107     RIT_Difference          = 2,
 108     RIT_PreboundLazyPointer = 3,
 109     RIT_LocalDifference     = 4
 110   };
 111
 112   /// MachSymbolData - Helper struct for containing some precomputed information
 113   /// on symbols.
 114   struct MachSymbolData {
 115     MCSymbolData *SymbolData;
 116     uint64_t StringIndex;
 117     uint8_t SectionIndex;
 118
 119     // Support lexicographic sorting.
 120     bool operator<(const MachSymbolData &RHS) const {
 121       const std::string &Name = SymbolData->getSymbol().getName();
 122       return Name < RHS.SymbolData->getSymbol().getName();
 123     }
 124   };
 125
 126   raw_ostream &OS;
 127   bool IsLSB;
 128
 129 public:
 130   MachObjectWriter(raw_ostream &_OS, bool _IsLSB = true)
 131     : OS(_OS), IsLSB(_IsLSB) {
 132   }
 133
 134   /// @name Helper Methods
 135   /// @{
 136
 137   void Write8(uint8_t Value) {
 138     OS << char(Value);
 139   }
 140
 141   void Write16(uint16_t Value) {
 142     if (IsLSB) {
 143       Write8(uint8_t(Value >> 0));
 144       Write8(uint8_t(Value >> 8));
 145     } else {
 146       Write8(uint8_t(Value >> 8));
 147       Write8(uint8_t(Value >> 0));
 148     }
 149   }
 150
 151   void Write32(uint32_t Value) {
 152     if (IsLSB) {
 153       Write16(uint16_t(Value >> 0));
 154       Write16(uint16_t(Value >> 16));
 155     } else {
 156       Write16(uint16_t(Value >> 16));
 157       Write16(uint16_t(Value >> 0));
 158     }
 159   }
 160
 161   void Write64(uint64_t Value) {
 162     if (IsLSB) {
 163       Write32(uint32_t(Value >> 0));
 164       Write32(uint32_t(Value >> 32));
 165     } else {
 166       Write32(uint32_t(Value >> 32));
 167       Write32(uint32_t(Value >> 0));
 168     }
 169   }
 170
 171   void WriteZeros(unsigned N) {
 172     const char Zeros[16] = { 0 };
 173
 174     for (unsigned i = 0, e = N / 16; i != e; ++i)
 175       OS << StringRef(Zeros, 16);
 176
 177     OS << StringRef(Zeros, N % 16);
 178   }
 179
 180   void WriteString(const StringRef &Str, unsigned ZeroFillSize = 0) {
 181     OS << Str;
 182     if (ZeroFillSize)
 183       WriteZeros(ZeroFillSize - Str.size());
 184   }
 185
 186   /// @}
 187
 188   void WriteHeader32(unsigned NumLoadCommands, unsigned LoadCommandsSize,
 189                      bool SubsectionsViaSymbols) {
 190     uint32_t Flags = 0;
 191
 192     if (SubsectionsViaSymbols)
 193       Flags |= HF_SubsectionsViaSymbols;
 194
 195     // struct mach_header (28 bytes)
 196
 197     uint64_t Start = OS.tell();
 198     (void) Start;
 199
 200     Write32(Header_Magic32);
 201
 202     // FIXME: Support cputype.
 203     Write32(TargetMachOWriterInfo::HDR_CPU_TYPE_I386);
 204     // FIXME: Support cpusubtype.
 205     Write32(TargetMachOWriterInfo::HDR_CPU_SUBTYPE_I386_ALL);
 206     Write32(HFT_Object);
 207     Write32(NumLoadCommands);    // Object files have a single load command, the
 208                                  // segment.
 209     Write32(LoadCommandsSize);
 210     Write32(Flags);
 211
 212     assert(OS.tell() - Start == Header32Size);
 213   }
 214
 215   /// WriteSegmentLoadCommand32 - Write a 32-bit segment load command.
 216   ///
 217   /// \arg NumSections - The number of sections in this segment.
 218   /// \arg SectionDataSize - The total size of the sections.
 219   void WriteSegmentLoadCommand32(unsigned NumSections,
 220                                  uint64_t VMSize,
 221                                  uint64_t SectionDataStartOffset,
 222                                  uint64_t SectionDataSize) {
 223     // struct segment_command (56 bytes)
 224
 225     uint64_t Start = OS.tell();
 226     (void) Start;
 227
 228     Write32(LCT_Segment);
 229     Write32(SegmentLoadCommand32Size + NumSections * Section32Size);
 230
 231     WriteString("", 16);
 232     Write32(0); // vmaddr
 233     Write32(VMSize); // vmsize
 234     Write32(SectionDataStartOffset); // file offset
 235     Write32(SectionDataSize); // file size
 236     Write32(0x7); // maxprot
 237     Write32(0x7); // initprot
 238     Write32(NumSections);
 239     Write32(0); // flags
 240
 241     assert(OS.tell() - Start == SegmentLoadCommand32Size);
 242   }
 243
 244   void WriteSection32(const MCSectionData &SD, uint64_t FileOffset,
 245                       uint64_t RelocationsStart, unsigned NumRelocations) {
 246     // The offset is unused for virtual sections.
 247     if (isVirtualSection(SD.getSection())) {
 248       assert(SD.getFileSize() == 0 && "Invalid file size!");
 249       FileOffset = 0;
 250     }
 251
 252     // struct section (68 bytes)
 253
 254     uint64_t Start = OS.tell();
 255     (void) Start;
 256
 257     // FIXME: cast<> support!
 258     const MCSectionMachO &Section =
 259       static_cast<const MCSectionMachO&>(SD.getSection());
 260     WriteString(Section.getSectionName(), 16);
 261     WriteString(Section.getSegmentName(), 16);
 262     Write32(SD.getAddress()); // address
 263     Write32(SD.getSize()); // size
 264     Write32(FileOffset);
 265
 266     assert(isPowerOf2_32(SD.getAlignment()) && "Invalid alignment!");
 267     Write32(Log2_32(SD.getAlignment()));
 268     Write32(NumRelocations ? RelocationsStart : 0);
 269     Write32(NumRelocations);
 270     Write32(Section.getTypeAndAttributes());
 271     Write32(0); // reserved1
 272     Write32(Section.getStubSize()); // reserved2
 273
 274     assert(OS.tell() - Start == Section32Size);
 275   }
 276
 277   void WriteSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
 278                               uint32_t StringTableOffset,
 279                               uint32_t StringTableSize) {
 280     // struct symtab_command (24 bytes)
 281
 282     uint64_t Start = OS.tell();
 283     (void) Start;
 284
 285     Write32(LCT_Symtab);
 286     Write32(SymtabLoadCommandSize);
 287     Write32(SymbolOffset);
 288     Write32(NumSymbols);
 289     Write32(StringTableOffset);
 290     Write32(StringTableSize);
 291
 292     assert(OS.tell() - Start == SymtabLoadCommandSize);
 293   }
 294
 295   void WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
 296                                 uint32_t NumLocalSymbols,
 297                                 uint32_t FirstExternalSymbol,
 298                                 uint32_t NumExternalSymbols,
 299                                 uint32_t FirstUndefinedSymbol,
 300                                 uint32_t NumUndefinedSymbols,
 301                                 uint32_t IndirectSymbolOffset,
 302                                 uint32_t NumIndirectSymbols) {
 303     // struct dysymtab_command (80 bytes)
 304
 305     uint64_t Start = OS.tell();
 306     (void) Start;
 307
 308     Write32(LCT_Dysymtab);
 309     Write32(DysymtabLoadCommandSize);
 310     Write32(FirstLocalSymbol);
 311     Write32(NumLocalSymbols);
 312     Write32(FirstExternalSymbol);
 313     Write32(NumExternalSymbols);
 314     Write32(FirstUndefinedSymbol);
 315     Write32(NumUndefinedSymbols);
 316     Write32(0); // tocoff
 317     Write32(0); // ntoc
 318     Write32(0); // modtaboff
 319     Write32(0); // nmodtab
 320     Write32(0); // extrefsymoff
 321     Write32(0); // nextrefsyms
 322     Write32(IndirectSymbolOffset);
 323     Write32(NumIndirectSymbols);
 324     Write32(0); // extreloff
 325     Write32(0); // nextrel
 326     Write32(0); // locreloff
 327     Write32(0); // nlocrel
 328
 329     assert(OS.tell() - Start == DysymtabLoadCommandSize);
 330   }
 331
 332   void WriteNlist32(MachSymbolData &MSD) {
 333     MCSymbolData &Data = *MSD.SymbolData;
 334     const MCSymbol &Symbol = Data.getSymbol();
 335     uint8_t Type = 0;
 336     uint16_t Flags = Data.getFlags();
 337     uint32_t Address = 0;
 338
 339     // Set the N_TYPE bits. See <mach-o/nlist.h>.
 340     //
 341     // FIXME: Are the prebound or indirect fields possible here?
 342     if (Symbol.isUndefined())
 343       Type = STT_Undefined;
 344     else if (Symbol.isAbsolute())
 345       Type = STT_Absolute;
 346     else
 347       Type = STT_Section;
 348
 349     // FIXME: Set STAB bits.
 350
 351     if (Data.isPrivateExtern())
 352       Type |= STF_PrivateExtern;
 353
 354     // Set external bit.
 355     if (Data.isExternal() || Symbol.isUndefined())
 356       Type |= STF_External;
 357
 358     // Compute the symbol address.
 359     if (Symbol.isDefined()) {
 360       if (Symbol.isAbsolute()) {
 361         llvm_unreachable("FIXME: Not yet implemented!");
 362       } else {
 363         Address = Data.getFragment()->getAddress() + Data.getOffset();
 364       }
 365     } else if (Data.isCommon()) {
 366       // Common symbols are encoded with the size in the address
 367       // field, and their alignment in the flags.
 368       Address = Data.getCommonSize();
 369
 370       // Common alignment is packed into the 'desc' bits.
 371       if (unsigned Align = Data.getCommonAlignment()) {
 372         unsigned Log2Size = Log2_32(Align);
 373         assert((1U << Log2Size) == Align && "Invalid 'common' alignment!");
 374         if (Log2Size > 15)
 375           llvm_report_error("invalid 'common' alignment '" +
 376                             Twine(Align) + "'");
 377         // FIXME: Keep this mask with the SymbolFlags enumeration.
 378         Flags = (Flags & 0xF0FF) | (Log2Size << 8);
 379       }
 380     }
 381
 382     // struct nlist (12 bytes)
 383
 384     Write32(MSD.StringIndex);
 385     Write8(Type);
 386     Write8(MSD.SectionIndex);
 387
 388     // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
 389     // value.
 390     Write16(Flags);
 391     Write32(Address);
 392   }
 393
 394   struct MachRelocationEntry {
 395     uint32_t Word0;
 396     uint32_t Word1;
 397   };
 398   void ComputeScatteredRelocationInfo(MCAssembler &Asm,
 399                                       MCSectionData::Fixup &Fixup,
 400                              DenseMap<const MCSymbol*,MCSymbolData*> &SymbolMap,
 401                                      std::vector<MachRelocationEntry> &Relocs) {
 402     uint32_t Address = Fixup.Fragment->getOffset() + Fixup.Offset;
 403     unsigned IsPCRel = 0;
 404     unsigned Type = RIT_Vanilla;
 405
 406     // See <reloc.h>.
 407
 408     const MCSymbol *A = Fixup.Value.getSymA();
 409     MCSymbolData *SD = SymbolMap.lookup(A);
 410     uint32_t Value = SD->getFragment()->getAddress() + SD->getOffset();
 411     uint32_t Value2 = 0;
 412
 413     if (const MCSymbol *B = Fixup.Value.getSymB()) {
 414       Type = RIT_LocalDifference;
 415
 416       MCSymbolData *SD = SymbolMap.lookup(B);
 417       Value2 = SD->getFragment()->getAddress() + SD->getOffset();
 418     }
 419
 420     unsigned Log2Size = Log2_32(Fixup.Size);
 421     assert((1U << Log2Size) == Fixup.Size && "Invalid fixup size!");
 422
 423     // The value which goes in the fixup is current value of the expression.
 424     Fixup.FixedValue = Value - Value2 + Fixup.Value.getConstant();
 425
 426     MachRelocationEntry MRE;
 427     MRE.Word0 = ((Address   <<  0) |
 428                  (Type      << 24) |
 429                  (Log2Size  << 28) |
 430                  (IsPCRel   << 30) |
 431                  RF_Scattered);
 432     MRE.Word1 = Value;
 433     Relocs.push_back(MRE);
 434
 435     if (Type == RIT_LocalDifference) {
 436       Type = RIT_Pair;
 437
 438       MachRelocationEntry MRE;
 439       MRE.Word0 = ((0         <<  0) |
 440                    (Type      << 24) |
 441                    (Log2Size  << 28) |
 442                    (0   << 30) |
 443                    RF_Scattered);
 444       MRE.Word1 = Value2;
 445       Relocs.push_back(MRE);
 446     }
 447   }
 448
 449   void ComputeRelocationInfo(MCAssembler &Asm,
 450                              MCSectionData::Fixup &Fixup,
 451                              DenseMap<const MCSymbol*,MCSymbolData*> &SymbolMap,
 452                              std::vector<MachRelocationEntry> &Relocs) {
 453     // If this is a local symbol plus an offset or a difference, then we need a
 454     // scattered relocation entry.
 455     if (Fixup.Value.getSymB()) // a - b
 456       return ComputeScatteredRelocationInfo(Asm, Fixup, SymbolMap, Relocs);
 457     if (Fixup.Value.getSymA() && Fixup.Value.getConstant())
 458       if (!Fixup.Value.getSymA()->isUndefined())
 459         return ComputeScatteredRelocationInfo(Asm, Fixup, SymbolMap, Relocs);
 460
 461     // See <reloc.h>.
 462     uint32_t Address = Fixup.Fragment->getOffset() + Fixup.Offset;
 463     uint32_t Value = 0;
 464     unsigned Index = 0;
 465     unsigned IsPCRel = 0;
 466     unsigned IsExtern = 0;
 467     unsigned Type = 0;
 468
 469     if (Fixup.Value.isAbsolute()) { // constant
 470       // SymbolNum of 0 indicates the absolute section.
 471       Type = RIT_Vanilla;
 472       Value = 0;
 473       llvm_unreachable("FIXME: Not yet implemented!");
 474     } else {
 475       const MCSymbol *Symbol = Fixup.Value.getSymA();
 476       MCSymbolData *SD = SymbolMap.lookup(Symbol);
 477
 478       if (Symbol->isUndefined()) {
 479         IsExtern = 1;
 480         Index = SD->getIndex();
 481         Value = 0;
 482       } else {
 483         // The index is the section ordinal.
 484         //
 485         // FIXME: O(N)
 486         Index = 1;
 487         for (MCAssembler::iterator it = Asm.begin(),
 488                ie = Asm.end(); it != ie; ++it, ++Index)
 489           if (&*it == SD->getFragment()->getParent())
 490             break;
 491         Value = SD->getFragment()->getAddress() + SD->getOffset();
 492       }
 493
 494       Type = RIT_Vanilla;
 495     }
 496
 497     // The value which goes in the fixup is current value of the expression.
 498     Fixup.FixedValue = Value + Fixup.Value.getConstant();
 499
 500     unsigned Log2Size = Log2_32(Fixup.Size);
 501     assert((1U << Log2Size) == Fixup.Size && "Invalid fixup size!");
 502
 503     // struct relocation_info (8 bytes)
 504     MachRelocationEntry MRE;
 505     MRE.Word0 = Address;
 506     MRE.Word1 = ((Index     <<  0) |
 507                  (IsPCRel   << 24) |
 508                  (Log2Size  << 25) |
 509                  (IsExtern  << 27) |
 510                  (Type      << 28));
 511     Relocs.push_back(MRE);
 512   }
 513
 514   void BindIndirectSymbols(MCAssembler &Asm,
 515                            DenseMap<const MCSymbol*,MCSymbolData*> &SymbolMap) {
 516     // This is the point where 'as' creates actual symbols for indirect symbols
 517     // (in the following two passes). It would be easier for us to do this
 518     // sooner when we see the attribute, but that makes getting the order in the
 519     // symbol table much more complicated than it is worth.
 520     //
 521     // FIXME: Revisit this when the dust settles.
 522
 523     // Bind non lazy symbol pointers first.
 524     for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 525            ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 526       // FIXME: cast<> support!
 527       const MCSectionMachO &Section =
 528         static_cast<const MCSectionMachO&>(it->SectionData->getSection());
 529
 530       unsigned Type =
 531         Section.getTypeAndAttributes() & MCSectionMachO::SECTION_TYPE;
 532       if (Type != MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS)
 533         continue;
 534
 535       MCSymbolData *&Entry = SymbolMap[it->Symbol];
 536       if (!Entry)
 537         Entry = new MCSymbolData(*it->Symbol, 0, 0, &Asm);
 538     }
 539
 540     // Then lazy symbol pointers and symbol stubs.
 541     for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
 542            ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 543       // FIXME: cast<> support!
 544       const MCSectionMachO &Section =
 545         static_cast<const MCSectionMachO&>(it->SectionData->getSection());
 546
 547       unsigned Type =
 548         Section.getTypeAndAttributes() & MCSectionMachO::SECTION_TYPE;
 549       if (Type != MCSectionMachO::S_LAZY_SYMBOL_POINTERS &&
 550           Type != MCSectionMachO::S_SYMBOL_STUBS)
 551         continue;
 552
 553       MCSymbolData *&Entry = SymbolMap[it->Symbol];
 554       if (!Entry) {
 555         Entry = new MCSymbolData(*it->Symbol, 0, 0, &Asm);
 556
 557         // Set the symbol type to undefined lazy, but only on construction.
 558         //
 559         // FIXME: Do not hardcode.
 560         Entry->setFlags(Entry->getFlags() | 0x0001);
 561       }
 562     }
 563   }
 564
 565   /// ComputeSymbolTable - Compute the symbol table data
 566   ///
 567   /// \param StringTable [out] - The string table data.
 568   /// \param StringIndexMap [out] - Map from symbol names to offsets in the
 569   /// string table.
 570   void ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
 571                           std::vector<MachSymbolData> &LocalSymbolData,
 572                           std::vector<MachSymbolData> &ExternalSymbolData,
 573                           std::vector<MachSymbolData> &UndefinedSymbolData) {
 574     // Build section lookup table.
 575     DenseMap<const MCSection*, uint8_t> SectionIndexMap;
 576     unsigned Index = 1;
 577     for (MCAssembler::iterator it = Asm.begin(),
 578            ie = Asm.end(); it != ie; ++it, ++Index)
 579       SectionIndexMap[&it->getSection()] = Index;
 580     assert(Index <= 256 && "Too many sections!");
 581
 582     // Index 0 is always the empty string.
 583     StringMap<uint64_t> StringIndexMap;
 584     StringTable += '\x00';
 585
 586     // Build the symbol arrays and the string table, but only for non-local
 587     // symbols.
 588     //
 589     // The particular order that we collect the symbols and create the string
 590     // table, then sort the symbols is chosen to match 'as'. Even though it
 591     // doesn't matter for correctness, this is important for letting us diff .o
 592     // files.
 593     for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
 594            ie = Asm.symbol_end(); it != ie; ++it) {
 595       const MCSymbol &Symbol = it->getSymbol();
 596
 597       // Ignore assembler temporaries.
 598       if (it->getSymbol().isTemporary())
 599         continue;
 600
 601       if (!it->isExternal() && !Symbol.isUndefined())
 602         continue;
 603
 604       uint64_t &Entry = StringIndexMap[Symbol.getName()];
 605       if (!Entry) {
 606         Entry = StringTable.size();
 607         StringTable += Symbol.getName();
 608         StringTable += '\x00';
 609       }
 610
 611       MachSymbolData MSD;
 612       MSD.SymbolData = it;
 613       MSD.StringIndex = Entry;
 614
 615       if (Symbol.isUndefined()) {
 616         MSD.SectionIndex = 0;
 617         UndefinedSymbolData.push_back(MSD);
 618       } else if (Symbol.isAbsolute()) {
 619         MSD.SectionIndex = 0;
 620         ExternalSymbolData.push_back(MSD);
 621       } else {
 622         MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
 623         assert(MSD.SectionIndex && "Invalid section index!");
 624         ExternalSymbolData.push_back(MSD);
 625       }
 626     }
 627
 628     // Now add the data for local symbols.
 629     for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
 630            ie = Asm.symbol_end(); it != ie; ++it) {
 631       const MCSymbol &Symbol = it->getSymbol();
 632
 633       // Ignore assembler temporaries.
 634       if (it->getSymbol().isTemporary())
 635         continue;
 636
 637       if (it->isExternal() || Symbol.isUndefined())
 638         continue;
 639
 640       uint64_t &Entry = StringIndexMap[Symbol.getName()];
 641       if (!Entry) {
 642         Entry = StringTable.size();
 643         StringTable += Symbol.getName();
 644         StringTable += '\x00';
 645       }
 646
 647       MachSymbolData MSD;
 648       MSD.SymbolData = it;
 649       MSD.StringIndex = Entry;
 650
 651       if (Symbol.isAbsolute()) {
 652         MSD.SectionIndex = 0;
 653         LocalSymbolData.push_back(MSD);
 654       } else {
 655         MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
 656         assert(MSD.SectionIndex && "Invalid section index!");
 657         LocalSymbolData.push_back(MSD);
 658       }
 659     }
 660
 661     // External and undefined symbols are required to be in lexicographic order.
 662     std::sort(ExternalSymbolData.begin(), ExternalSymbolData.end());
 663     std::sort(UndefinedSymbolData.begin(), UndefinedSymbolData.end());
 664
 665     // Set the symbol indices.
 666     Index = 0;
 667     for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
 668       LocalSymbolData[i].SymbolData->setIndex(Index++);
 669     for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
 670       ExternalSymbolData[i].SymbolData->setIndex(Index++);
 671     for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
 672       UndefinedSymbolData[i].SymbolData->setIndex(Index++);
 673
 674     // The string table is padded to a multiple of 4.
 675     //
 676     // FIXME: Check to see if this varies per arch.
 677     while (StringTable.size() % 4)
 678       StringTable += '\x00';
 679   }
 680
 681   void WriteObject(MCAssembler &Asm) {
 682     unsigned NumSections = Asm.size();
 683
 684     // Compute the symbol -> symbol data map.
 685     //
 686     // FIXME: This should not be here.
 687     DenseMap<const MCSymbol*, MCSymbolData *> SymbolMap;
 688     for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
 689            ie = Asm.symbol_end(); it != ie; ++it)
 690       SymbolMap[&it->getSymbol()] = it;
 691
 692     // Create symbol data for any indirect symbols.
 693     BindIndirectSymbols(Asm, SymbolMap);
 694
 695     // Compute symbol table information.
 696     SmallString<256> StringTable;
 697     std::vector<MachSymbolData> LocalSymbolData;
 698     std::vector<MachSymbolData> ExternalSymbolData;
 699     std::vector<MachSymbolData> UndefinedSymbolData;
 700     unsigned NumSymbols = Asm.symbol_size();
 701
 702     // No symbol table command is written if there are no symbols.
 703     if (NumSymbols)
 704       ComputeSymbolTable(Asm, StringTable, LocalSymbolData, ExternalSymbolData,
 705                          UndefinedSymbolData);
 706
 707     // The section data starts after the header, the segment load command (and
 708     // section headers) and the symbol table.
 709     unsigned NumLoadCommands = 1;
 710     uint64_t LoadCommandsSize =
 711       SegmentLoadCommand32Size + NumSections * Section32Size;
 712
 713     // Add the symbol table load command sizes, if used.
 714     if (NumSymbols) {
 715       NumLoadCommands += 2;
 716       LoadCommandsSize += SymtabLoadCommandSize + DysymtabLoadCommandSize;
 717     }
 718
 719     // Compute the total size of the section data, as well as its file size and
 720     // vm size.
 721     uint64_t SectionDataStart = Header32Size + LoadCommandsSize;
 722     uint64_t SectionDataSize = 0;
 723     uint64_t SectionDataFileSize = 0;
 724     uint64_t VMSize = 0;
 725     for (MCAssembler::iterator it = Asm.begin(),
 726            ie = Asm.end(); it != ie; ++it) {
 727       MCSectionData &SD = *it;
 728
 729       VMSize = std::max(VMSize, SD.getAddress() + SD.getSize());
 730
 731       if (isVirtualSection(SD.getSection()))
 732         continue;
 733
 734       SectionDataSize = std::max(SectionDataSize,
 735                                  SD.getAddress() + SD.getSize());
 736       SectionDataFileSize = std::max(SectionDataFileSize,
 737                                      SD.getAddress() + SD.getFileSize());
 738     }
 739
 740     // The section data is passed to 4 bytes.
 741     //
 742     // FIXME: Is this machine dependent?
 743     unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
 744     SectionDataFileSize += SectionDataPadding;
 745
 746     // Write the prolog, starting with the header and load command...
 747     WriteHeader32(NumLoadCommands, LoadCommandsSize,
 748                   Asm.getSubsectionsViaSymbols());
 749     WriteSegmentLoadCommand32(NumSections, VMSize,
 750                               SectionDataStart, SectionDataSize);
 751
 752     // ... and then the section headers.
 753     //
 754     // We also compute the section relocations while we do this. Note that
 755     // compute relocation info will also update the fixup to have the correct
 756     // value; this will be overwrite the appropriate data in the fragment when
 757     // it is written.
 758     std::vector<MachRelocationEntry> RelocInfos;
 759     uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
 760     for (MCAssembler::iterator it = Asm.begin(), ie = Asm.end(); it != ie;
 761          ++it) {
 762       MCSectionData &SD = *it;
 763
 764       // The assembler writes relocations in the reverse order they were seen.
 765       //
 766       // FIXME: It is probably more complicated than this.
 767       unsigned NumRelocsStart = RelocInfos.size();
 768       for (unsigned i = 0, e = SD.fixup_size(); i != e; ++i)
 769         ComputeRelocationInfo(Asm, SD.getFixups()[e - i - 1], SymbolMap,
 770                               RelocInfos);
 771
 772       unsigned NumRelocs = RelocInfos.size() - NumRelocsStart;
 773       uint64_t SectionStart = SectionDataStart + SD.getAddress();
 774       WriteSection32(SD, SectionStart, RelocTableEnd, NumRelocs);
 775       RelocTableEnd += NumRelocs * RelocationInfoSize;
 776     }
 777
 778     // Write the symbol table load command, if used.
 779     if (NumSymbols) {
 780       unsigned FirstLocalSymbol = 0;
 781       unsigned NumLocalSymbols = LocalSymbolData.size();
 782       unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
 783       unsigned NumExternalSymbols = ExternalSymbolData.size();
 784       unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
 785       unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
 786       unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
 787       unsigned NumSymTabSymbols =
 788         NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
 789       uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
 790       uint64_t IndirectSymbolOffset = 0;
 791
 792       // If used, the indirect symbols are written after the section data.
 793       if (NumIndirectSymbols)
 794         IndirectSymbolOffset = RelocTableEnd;
 795
 796       // The symbol table is written after the indirect symbol data.
 797       uint64_t SymbolTableOffset = RelocTableEnd + IndirectSymbolSize;
 798
 799       // The string table is written after symbol table.
 800       uint64_t StringTableOffset =
 801         SymbolTableOffset + NumSymTabSymbols * Nlist32Size;
 802       WriteSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
 803                              StringTableOffset, StringTable.size());
 804
 805       WriteDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
 806                                FirstExternalSymbol, NumExternalSymbols,
 807                                FirstUndefinedSymbol, NumUndefinedSymbols,
 808                                IndirectSymbolOffset, NumIndirectSymbols);
 809     }
 810
 811     // Write the actual section data.
 812     for (MCAssembler::iterator it = Asm.begin(), ie = Asm.end(); it != ie; ++it)
 813       WriteFileData(OS, *it, *this);
 814
 815     // Write the extra padding.
 816     WriteZeros(SectionDataPadding);
 817
 818     // Write the relocation entries.
 819     for (unsigned i = 0, e = RelocInfos.size(); i != e; ++i) {
 820       Write32(RelocInfos[i].Word0);
 821       Write32(RelocInfos[i].Word1);
 822     }
 823
 824     // Write the symbol table data, if used.
 825     if (NumSymbols) {
 826       // Write the indirect symbol entries.
 827       for (MCAssembler::indirect_symbol_iterator
 828              it = Asm.indirect_symbol_begin(),
 829              ie = Asm.indirect_symbol_end(); it != ie; ++it) {
 830         // Indirect symbols in the non lazy symbol pointer section have some
 831         // special handling.
 832         const MCSectionMachO &Section =
 833           static_cast<const MCSectionMachO&>(it->SectionData->getSection());
 834         unsigned Type =
 835           Section.getTypeAndAttributes() & MCSectionMachO::SECTION_TYPE;
 836         if (Type == MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS) {
 837           // If this symbol is defined and internal, mark it as such.
 838           if (it->Symbol->isDefined() &&
 839               !SymbolMap.lookup(it->Symbol)->isExternal()) {
 840             uint32_t Flags = ISF_Local;
 841             if (it->Symbol->isAbsolute())
 842               Flags |= ISF_Absolute;
 843             Write32(Flags);
 844             continue;
 845           }
 846         }
 847
 848         Write32(SymbolMap[it->Symbol]->getIndex());
 849       }
 850
 851       // FIXME: Check that offsets match computed ones.
 852
 853       // Write the symbol table entries.
 854       for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
 855         WriteNlist32(LocalSymbolData[i]);
 856       for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
 857         WriteNlist32(ExternalSymbolData[i]);
 858       for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
 859         WriteNlist32(UndefinedSymbolData[i]);
 860
 861       // Write the string table.
 862       OS << StringTable.str();
 863     }
 864   }
 865 };
 866
 867 /* *** */
 868
 869 MCFragment::MCFragment() : Kind(FragmentType(~0)) {
 870 }
 871
 872 MCFragment::MCFragment(FragmentType _Kind, MCSectionData *_Parent)
 873   : Kind(_Kind),
 874     Parent(_Parent),
 875     FileSize(~UINT64_C(0))
 876 {
 877   if (Parent)
 878     Parent->getFragmentList().push_back(this);
 879 }
 880
 881 MCFragment::~MCFragment() {
 882 }
 883
 884 uint64_t MCFragment::getAddress() const {
 885   assert(getParent() && "Missing Section!");
 886   return getParent()->getAddress() + Offset;
 887 }
 888
 889 /* *** */
 890
 891 MCSectionData::MCSectionData() : Section(0) {}
 892
 893 MCSectionData::MCSectionData(const MCSection &_Section, MCAssembler *A)
 894   : Section(&_Section),
 895     Alignment(1),
 896     Address(~UINT64_C(0)),
 897     Size(~UINT64_C(0)),
 898     FileSize(~UINT64_C(0)),
 899     LastFixupLookup(~0)
 900 {
 901   if (A)
 902     A->getSectionList().push_back(this);
 903 }
 904
 905 const MCSectionData::Fixup *
 906 MCSectionData::LookupFixup(const MCFragment *Fragment, uint64_t Offset) const {
 907   // Use a one level cache to turn the common case of accessing the fixups in
 908   // order into O(1) instead of O(N).
 909   unsigned i = LastFixupLookup, Count = Fixups.size(), End = Fixups.size();
 910   if (i >= End)
 911     i = 0;
 912   while (Count--) {
 913     const Fixup &F = Fixups[i];
 914     if (F.Fragment == Fragment && F.Offset == Offset) {
 915       LastFixupLookup = i;
 916       return &F;
 917     }
 918
 919     ++i;
 920     if (i == End)
 921       i = 0;
 922   }
 923
 924   return 0;
 925 }
 926
 927 /* *** */
 928
 929 MCSymbolData::MCSymbolData() : Symbol(0) {}
 930
 931 MCSymbolData::MCSymbolData(const MCSymbol &_Symbol, MCFragment *_Fragment,
 932                            uint64_t _Offset, MCAssembler *A)
 933   : Symbol(&_Symbol), Fragment(_Fragment), Offset(_Offset),
 934     IsExternal(false), IsPrivateExtern(false),
 935     CommonSize(0), CommonAlign(0), Flags(0), Index(0)
 936 {
 937   if (A)
 938     A->getSymbolList().push_back(this);
 939 }
 940
 941 /* *** */
 942
 943 MCAssembler::MCAssembler(MCContext &_Context, raw_ostream &_OS)
 944   : Context(_Context), OS(_OS), SubsectionsViaSymbols(false)
 945 {
 946 }
 947
 948 MCAssembler::~MCAssembler() {
 949 }
 950
 951 void MCAssembler::LayoutSection(MCSectionData &SD) {
 952   uint64_t Address = SD.getAddress();
 953
 954   for (MCSectionData::iterator it = SD.begin(), ie = SD.end(); it != ie; ++it) {
 955     MCFragment &F = *it;
 956
 957     F.setOffset(Address - SD.getAddress());
 958
 959     // Evaluate fragment size.
 960     switch (F.getKind()) {
 961     case MCFragment::FT_Align: {
 962       MCAlignFragment &AF = cast<MCAlignFragment>(F);
 963
 964       uint64_t Size = OffsetToAlignment(Address, AF.getAlignment());
 965       if (Size > AF.getMaxBytesToEmit())
 966         AF.setFileSize(0);
 967       else
 968         AF.setFileSize(Size);
 969       break;
 970     }
 971
 972     case MCFragment::FT_Data:
 973       F.setFileSize(F.getMaxFileSize());
 974       break;
 975
 976     case MCFragment::FT_Fill: {
 977       MCFillFragment &FF = cast<MCFillFragment>(F);
 978
 979       F.setFileSize(F.getMaxFileSize());
 980
 981       // If the fill value is constant, thats it.
 982       if (FF.getValue().isAbsolute())
 983         break;
 984
 985       // Otherwise, add fixups for the values.
 986       for (uint64_t i = 0, e = FF.getCount(); i != e; ++i) {
 987         MCSectionData::Fixup Fix(F, i * FF.getValueSize(),
 988                                  FF.getValue(),FF.getValueSize());
 989         SD.getFixups().push_back(Fix);
 990       }
 991       break;
 992     }
 993
 994     case MCFragment::FT_Org: {
 995       MCOrgFragment &OF = cast<MCOrgFragment>(F);
 996
 997       if (!OF.getOffset().isAbsolute())
 998         llvm_unreachable("FIXME: Not yet implemented!");
 999       uint64_t OrgOffset = OF.getOffset().getConstant();
1000       uint64_t Offset = Address - SD.getAddress();
1001
1002       // FIXME: We need a way to communicate this error.
1003       if (OrgOffset < Offset)
1004         llvm_report_error("invalid .org offset '" + Twine(OrgOffset) +
1005                           "' (at offset '" + Twine(Offset) + "'");
1006
1007       F.setFileSize(OrgOffset - Offset);
1008       break;
1009     }
1010
1011     case MCFragment::FT_ZeroFill: {
1012       MCZeroFillFragment &ZFF = cast<MCZeroFillFragment>(F);
1013
1014       // Align the fragment offset; it is safe to adjust the offset freely since
1015       // this is only in virtual sections.
1016       uint64_t Aligned = RoundUpToAlignment(Address, ZFF.getAlignment());
1017       F.setOffset(Aligned - SD.getAddress());
1018
1019       // FIXME: This is misnamed.
1020       F.setFileSize(ZFF.getSize());
1021       break;
1022     }
1023     }
1024
1025     Address += F.getFileSize();
1026   }
1027
1028   // Set the section sizes.
1029   SD.setSize(Address - SD.getAddress());
1030   if (isVirtualSection(SD.getSection()))
1031     SD.setFileSize(0);
1032   else
1033     SD.setFileSize(Address - SD.getAddress());
1034 }
1035
1036 /// WriteFileData - Write the \arg F data to the output file.
1037 static void WriteFileData(raw_ostream &OS, const MCFragment &F,
1038                           MachObjectWriter &MOW) {
1039   uint64_t Start = OS.tell();
1040   (void) Start;
1041
1042   ++EmittedFragments;
1043
1044   // FIXME: Embed in fragments instead?
1045   switch (F.getKind()) {
1046   case MCFragment::FT_Align: {
1047     MCAlignFragment &AF = cast<MCAlignFragment>(F);
1048     uint64_t Count = AF.getFileSize() / AF.getValueSize();
1049
1050     // FIXME: This error shouldn't actually occur (the front end should emit
1051     // multiple .align directives to enforce the semantics it wants), but is
1052     // severe enough that we want to report it. How to handle this?
1053     if (Count * AF.getValueSize() != AF.getFileSize())
1054       llvm_report_error("undefined .align directive, value size '" +
1055                         Twine(AF.getValueSize()) +
1056                         "' is not a divisor of padding size '" +
1057                         Twine(AF.getFileSize()) + "'");
1058
1059     for (uint64_t i = 0; i != Count; ++i) {
1060       switch (AF.getValueSize()) {
1061       default:
1062         assert(0 && "Invalid size!");
1063       case 1: MOW.Write8 (uint8_t (AF.getValue())); break;
1064       case 2: MOW.Write16(uint16_t(AF.getValue())); break;
1065       case 4: MOW.Write32(uint32_t(AF.getValue())); break;
1066       case 8: MOW.Write64(uint64_t(AF.getValue())); break;
1067       }
1068     }
1069     break;
1070   }
1071
1072   case MCFragment::FT_Data:
1073     OS << cast<MCDataFragment>(F).getContents().str();
1074     break;
1075
1076   case MCFragment::FT_Fill: {
1077     MCFillFragment &FF = cast<MCFillFragment>(F);
1078
1079     int64_t Value = 0;
1080     if (FF.getValue().isAbsolute())
1081       Value = FF.getValue().getConstant();
1082     for (uint64_t i = 0, e = FF.getCount(); i != e; ++i) {
1083       if (!FF.getValue().isAbsolute()) {
1084         // Find the fixup.
1085         //
1086         // FIXME: Find a better way to write in the fixes.
1087         const MCSectionData::Fixup *Fixup =
1088           F.getParent()->LookupFixup(&F, i * FF.getValueSize());
1089         assert(Fixup && "Missing fixup for fill value!");
1090         Value = Fixup->FixedValue;
1091       }
1092
1093       switch (FF.getValueSize()) {
1094       default:
1095         assert(0 && "Invalid size!");
1096       case 1: MOW.Write8 (uint8_t (Value)); break;
1097       case 2: MOW.Write16(uint16_t(Value)); break;
1098       case 4: MOW.Write32(uint32_t(Value)); break;
1099       case 8: MOW.Write64(uint64_t(Value)); break;
1100       }
1101     }
1102     break;
1103   }
1104
1105   case MCFragment::FT_Org: {
1106     MCOrgFragment &OF = cast<MCOrgFragment>(F);
1107
1108     for (uint64_t i = 0, e = OF.getFileSize(); i != e; ++i)
1109       MOW.Write8(uint8_t(OF.getValue()));
1110
1111     break;
1112   }
1113
1114   case MCFragment::FT_ZeroFill: {
1115     assert(0 && "Invalid zero fill fragment in concrete section!");
1116     break;
1117   }
1118   }
1119
1120   assert(OS.tell() - Start == F.getFileSize());
1121 }
1122
1123 /// WriteFileData - Write the \arg SD data to the output file.
1124 static void WriteFileData(raw_ostream &OS, const MCSectionData &SD,
1125                           MachObjectWriter &MOW) {
1126   // Ignore virtual sections.
1127   if (isVirtualSection(SD.getSection())) {
1128     assert(SD.getFileSize() == 0);
1129     return;
1130   }
1131
1132   uint64_t Start = OS.tell();
1133   (void) Start;
1134
1135   for (MCSectionData::const_iterator it = SD.begin(),
1136          ie = SD.end(); it != ie; ++it)
1137     WriteFileData(OS, *it, MOW);
1138
1139   // Add section padding.
1140   assert(SD.getFileSize() >= SD.getSize() && "Invalid section sizes!");
1141   MOW.WriteZeros(SD.getFileSize() - SD.getSize());
1142
1143   assert(OS.tell() - Start == SD.getFileSize());
1144 }
1145
1146 void MCAssembler::Finish() {
1147   // Layout the concrete sections and fragments.
1148   uint64_t Address = 0;
1149   MCSectionData *Prev = 0;
1150   for (iterator it = begin(), ie = end(); it != ie; ++it) {
1151     MCSectionData &SD = *it;
1152
1153     // Skip virtual sections.
1154     if (isVirtualSection(SD.getSection()))
1155       continue;
1156
1157     // Align this section if necessary by adding padding bytes to the previous
1158     // section.
1159     if (uint64_t Pad = OffsetToAlignment(Address, it->getAlignment())) {
1160       assert(Prev && "Missing prev section!");
1161       Prev->setFileSize(Prev->getFileSize() + Pad);
1162       Address += Pad;
1163     }
1164
1165     // Layout the section fragments and its size.
1166     SD.setAddress(Address);
1167     LayoutSection(SD);
1168     Address += SD.getFileSize();
1169
1170     Prev = &SD;
1171   }
1172
1173   // Layout the virtual sections.
1174   for (iterator it = begin(), ie = end(); it != ie; ++it) {
1175     MCSectionData &SD = *it;
1176
1177     if (!isVirtualSection(SD.getSection()))
1178       continue;
1179
1180     SD.setAddress(Address);
1181     LayoutSection(SD);
1182     Address += SD.getSize();
1183   }
1184
1185   // Write the object file.
1186   MachObjectWriter MOW(OS);
1187   MOW.WriteObject(*this);
1188
1189   OS.flush();
1190 }