Home | History | Annotate | Download | only in MC 
      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/MCAssembler.h"
     14 #include "llvm/MC/MCAsmBackend.h"
     15 #include "llvm/MC/MCAsmLayout.h"
     16 #include "llvm/MC/MCExpr.h"
     17 #include "llvm/MC/MCFixupKindInfo.h"
     18 #include "llvm/MC/MCObjectWriter.h"
     19 #include "llvm/MC/MCSectionMachO.h"
     20 #include "llvm/MC/MCSymbol.h"
     21 #include "llvm/MC/MCMachOSymbolFlags.h"
     22 #include "llvm/MC/MCValue.h"
     23 #include "llvm/Object/MachOFormat.h"
     24 #include "llvm/Support/Debug.h"
     25 #include "llvm/Support/ErrorHandling.h"
     26 
     27 #include <vector>
     28 using namespace llvm;
     29 using namespace llvm::object;
     30 
     31 bool MachObjectWriter::
     32 doesSymbolRequireExternRelocation(const MCSymbolData *SD) {
     33   // Undefined symbols are always extern.
     34   if (SD->Symbol->isUndefined())
     35     return true;
     36 
     37   // References to weak definitions require external relocation entries; the
     38   // definition may not always be the one in the same object file.
     39   if (SD->getFlags() & SF_WeakDefinition)
     40     return true;
     41 
     42   // Otherwise, we can use an internal relocation.
     43   return false;
     44 }
     45 
     46 bool MachObjectWriter::
     47 MachSymbolData::operator<(const MachSymbolData &RHS) const {
     48   return SymbolData->getSymbol().getName() <
     49     RHS.SymbolData->getSymbol().getName();
     50 }
     51 
     52 bool MachObjectWriter::isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) {
     53   const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo(
     54     (MCFixupKind) Kind);
     55 
     56   return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel;
     57 }
     58 
     59 uint64_t MachObjectWriter::getFragmentAddress(const MCFragment *Fragment,
     60                                               const MCAsmLayout &Layout) const {
     61   return getSectionAddress(Fragment->getParent()) +
     62     Layout.getFragmentOffset(Fragment);
     63 }
     64 
     65 uint64_t MachObjectWriter::getSymbolAddress(const MCSymbolData* SD,
     66                                             const MCAsmLayout &Layout) const {
     67   const MCSymbol &S = SD->getSymbol();
     68 
     69   // If this is a variable, then recursively evaluate now.
     70   if (S.isVariable()) {
     71     MCValue Target;
     72     if (!S.getVariableValue()->EvaluateAsRelocatable(Target, Layout))
     73       report_fatal_error("unable to evaluate offset for variable '" +
     74                          S.getName() + "'");
     75 
     76     // Verify that any used symbols are defined.
     77     if (Target.getSymA() && Target.getSymA()->getSymbol().isUndefined())
     78       report_fatal_error("unable to evaluate offset to undefined symbol '" +
     79                          Target.getSymA()->getSymbol().getName() + "'");
     80     if (Target.getSymB() && Target.getSymB()->getSymbol().isUndefined())
     81       report_fatal_error("unable to evaluate offset to undefined symbol '" +
     82                          Target.getSymB()->getSymbol().getName() + "'");
     83 
     84     uint64_t Address = Target.getConstant();
     85     if (Target.getSymA())
     86       Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
     87                                     Target.getSymA()->getSymbol()), Layout);
     88     if (Target.getSymB())
     89       Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
     90                                     Target.getSymB()->getSymbol()), Layout);
     91     return Address;
     92   }
     93 
     94   return getSectionAddress(SD->getFragment()->getParent()) +
     95     Layout.getSymbolOffset(SD);
     96 }
     97 
     98 uint64_t MachObjectWriter::getPaddingSize(const MCSectionData *SD,
     99                                           const MCAsmLayout &Layout) const {
    100   uint64_t EndAddr = getSectionAddress(SD) + Layout.getSectionAddressSize(SD);
    101   unsigned Next = SD->getLayoutOrder() + 1;
    102   if (Next >= Layout.getSectionOrder().size())
    103     return 0;
    104 
    105   const MCSectionData &NextSD = *Layout.getSectionOrder()[Next];
    106   if (NextSD.getSection().isVirtualSection())
    107     return 0;
    108   return OffsetToAlignment(EndAddr, NextSD.getAlignment());
    109 }
    110 
    111 void MachObjectWriter::WriteHeader(unsigned NumLoadCommands,
    112                                    unsigned LoadCommandsSize,
    113                                    bool SubsectionsViaSymbols) {
    114   uint32_t Flags = 0;
    115 
    116   if (SubsectionsViaSymbols)
    117     Flags |= macho::HF_SubsectionsViaSymbols;
    118 
    119   // struct mach_header (28 bytes) or
    120   // struct mach_header_64 (32 bytes)
    121 
    122   uint64_t Start = OS.tell();
    123   (void) Start;
    124 
    125   Write32(is64Bit() ? macho::HM_Object64 : macho::HM_Object32);
    126 
    127   Write32(TargetObjectWriter->getCPUType());
    128   Write32(TargetObjectWriter->getCPUSubtype());
    129 
    130   Write32(macho::HFT_Object);
    131   Write32(NumLoadCommands);
    132   Write32(LoadCommandsSize);
    133   Write32(Flags);
    134   if (is64Bit())
    135     Write32(0); // reserved
    136 
    137   assert(OS.tell() - Start ==
    138          (is64Bit() ? macho::Header64Size : macho::Header32Size));
    139 }
    140 
    141 /// WriteSegmentLoadCommand - Write a segment load command.
    142 ///
    143 /// \arg NumSections - The number of sections in this segment.
    144 /// \arg SectionDataSize - The total size of the sections.
    145 void MachObjectWriter::WriteSegmentLoadCommand(unsigned NumSections,
    146                                                uint64_t VMSize,
    147                                                uint64_t SectionDataStartOffset,
    148                                                uint64_t SectionDataSize) {
    149   // struct segment_command (56 bytes) or
    150   // struct segment_command_64 (72 bytes)
    151 
    152   uint64_t Start = OS.tell();
    153   (void) Start;
    154 
    155   unsigned SegmentLoadCommandSize =
    156     is64Bit() ? macho::SegmentLoadCommand64Size:
    157     macho::SegmentLoadCommand32Size;
    158   Write32(is64Bit() ? macho::LCT_Segment64 : macho::LCT_Segment);
    159   Write32(SegmentLoadCommandSize +
    160           NumSections * (is64Bit() ? macho::Section64Size :
    161                          macho::Section32Size));
    162 
    163   WriteBytes("", 16);
    164   if (is64Bit()) {
    165     Write64(0); // vmaddr
    166     Write64(VMSize); // vmsize
    167     Write64(SectionDataStartOffset); // file offset
    168     Write64(SectionDataSize); // file size
    169   } else {
    170     Write32(0); // vmaddr
    171     Write32(VMSize); // vmsize
    172     Write32(SectionDataStartOffset); // file offset
    173     Write32(SectionDataSize); // file size
    174   }
    175   Write32(0x7); // maxprot
    176   Write32(0x7); // initprot
    177   Write32(NumSections);
    178   Write32(0); // flags
    179 
    180   assert(OS.tell() - Start == SegmentLoadCommandSize);
    181 }
    182 
    183 void MachObjectWriter::WriteSection(const MCAssembler &Asm,
    184                                     const MCAsmLayout &Layout,
    185                                     const MCSectionData &SD,
    186                                     uint64_t FileOffset,
    187                                     uint64_t RelocationsStart,
    188                                     unsigned NumRelocations) {
    189   uint64_t SectionSize = Layout.getSectionAddressSize(&SD);
    190 
    191   // The offset is unused for virtual sections.
    192   if (SD.getSection().isVirtualSection()) {
    193     assert(Layout.getSectionFileSize(&SD) == 0 && "Invalid file size!");
    194     FileOffset = 0;
    195   }
    196 
    197   // struct section (68 bytes) or
    198   // struct section_64 (80 bytes)
    199 
    200   uint64_t Start = OS.tell();
    201   (void) Start;
    202 
    203   const MCSectionMachO &Section = cast<MCSectionMachO>(SD.getSection());
    204   WriteBytes(Section.getSectionName(), 16);
    205   WriteBytes(Section.getSegmentName(), 16);
    206   if (is64Bit()) {
    207     Write64(getSectionAddress(&SD)); // address
    208     Write64(SectionSize); // size
    209   } else {
    210     Write32(getSectionAddress(&SD)); // address
    211     Write32(SectionSize); // size
    212   }
    213   Write32(FileOffset);
    214 
    215   unsigned Flags = Section.getTypeAndAttributes();
    216   if (SD.hasInstructions())
    217     Flags |= MCSectionMachO::S_ATTR_SOME_INSTRUCTIONS;
    218 
    219   assert(isPowerOf2_32(SD.getAlignment()) && "Invalid alignment!");
    220   Write32(Log2_32(SD.getAlignment()));
    221   Write32(NumRelocations ? RelocationsStart : 0);
    222   Write32(NumRelocations);
    223   Write32(Flags);
    224   Write32(IndirectSymBase.lookup(&SD)); // reserved1
    225   Write32(Section.getStubSize()); // reserved2
    226   if (is64Bit())
    227     Write32(0); // reserved3
    228 
    229   assert(OS.tell() - Start == (is64Bit() ? macho::Section64Size :
    230                                macho::Section32Size));
    231 }
    232 
    233 void MachObjectWriter::WriteSymtabLoadCommand(uint32_t SymbolOffset,
    234                                               uint32_t NumSymbols,
    235                                               uint32_t StringTableOffset,
    236                                               uint32_t StringTableSize) {
    237   // struct symtab_command (24 bytes)
    238 
    239   uint64_t Start = OS.tell();
    240   (void) Start;
    241 
    242   Write32(macho::LCT_Symtab);
    243   Write32(macho::SymtabLoadCommandSize);
    244   Write32(SymbolOffset);
    245   Write32(NumSymbols);
    246   Write32(StringTableOffset);
    247   Write32(StringTableSize);
    248 
    249   assert(OS.tell() - Start == macho::SymtabLoadCommandSize);
    250 }
    251 
    252 void MachObjectWriter::WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
    253                                                 uint32_t NumLocalSymbols,
    254                                                 uint32_t FirstExternalSymbol,
    255                                                 uint32_t NumExternalSymbols,
    256                                                 uint32_t FirstUndefinedSymbol,
    257                                                 uint32_t NumUndefinedSymbols,
    258                                                 uint32_t IndirectSymbolOffset,
    259                                                 uint32_t NumIndirectSymbols) {
    260   // struct dysymtab_command (80 bytes)
    261 
    262   uint64_t Start = OS.tell();
    263   (void) Start;
    264 
    265   Write32(macho::LCT_Dysymtab);
    266   Write32(macho::DysymtabLoadCommandSize);
    267   Write32(FirstLocalSymbol);
    268   Write32(NumLocalSymbols);
    269   Write32(FirstExternalSymbol);
    270   Write32(NumExternalSymbols);
    271   Write32(FirstUndefinedSymbol);
    272   Write32(NumUndefinedSymbols);
    273   Write32(0); // tocoff
    274   Write32(0); // ntoc
    275   Write32(0); // modtaboff
    276   Write32(0); // nmodtab
    277   Write32(0); // extrefsymoff
    278   Write32(0); // nextrefsyms
    279   Write32(IndirectSymbolOffset);
    280   Write32(NumIndirectSymbols);
    281   Write32(0); // extreloff
    282   Write32(0); // nextrel
    283   Write32(0); // locreloff
    284   Write32(0); // nlocrel
    285 
    286   assert(OS.tell() - Start == macho::DysymtabLoadCommandSize);
    287 }
    288 
    289 void MachObjectWriter::WriteNlist(MachSymbolData &MSD,
    290                                   const MCAsmLayout &Layout) {
    291   MCSymbolData &Data = *MSD.SymbolData;
    292   const MCSymbol &Symbol = Data.getSymbol();
    293   uint8_t Type = 0;
    294   uint16_t Flags = Data.getFlags();
    295   uint64_t Address = 0;
    296 
    297   // Set the N_TYPE bits. See <mach-o/nlist.h>.
    298   //
    299   // FIXME: Are the prebound or indirect fields possible here?
    300   if (Symbol.isUndefined())
    301     Type = macho::STT_Undefined;
    302   else if (Symbol.isAbsolute())
    303     Type = macho::STT_Absolute;
    304   else
    305     Type = macho::STT_Section;
    306 
    307   // FIXME: Set STAB bits.
    308 
    309   if (Data.isPrivateExtern())
    310     Type |= macho::STF_PrivateExtern;
    311 
    312   // Set external bit.
    313   if (Data.isExternal() || Symbol.isUndefined())
    314     Type |= macho::STF_External;
    315 
    316   // Compute the symbol address.
    317   if (Symbol.isDefined()) {
    318     if (Symbol.isAbsolute()) {
    319       Address = cast<MCConstantExpr>(Symbol.getVariableValue())->getValue();
    320     } else {
    321       Address = getSymbolAddress(&Data, Layout);
    322     }
    323   } else if (Data.isCommon()) {
    324     // Common symbols are encoded with the size in the address
    325     // field, and their alignment in the flags.
    326     Address = Data.getCommonSize();
    327 
    328     // Common alignment is packed into the 'desc' bits.
    329     if (unsigned Align = Data.getCommonAlignment()) {
    330       unsigned Log2Size = Log2_32(Align);
    331       assert((1U << Log2Size) == Align && "Invalid 'common' alignment!");
    332       if (Log2Size > 15)
    333         report_fatal_error("invalid 'common' alignment '" +
    334                            Twine(Align) + "'");
    335       // FIXME: Keep this mask with the SymbolFlags enumeration.
    336       Flags = (Flags & 0xF0FF) | (Log2Size << 8);
    337     }
    338   }
    339 
    340   // struct nlist (12 bytes)
    341 
    342   Write32(MSD.StringIndex);
    343   Write8(Type);
    344   Write8(MSD.SectionIndex);
    345 
    346   // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
    347   // value.
    348   Write16(Flags);
    349   if (is64Bit())
    350     Write64(Address);
    351   else
    352     Write32(Address);
    353 }
    354 
    355 void MachObjectWriter::WriteLinkeditLoadCommand(uint32_t Type,
    356                                                 uint32_t DataOffset,
    357                                                 uint32_t DataSize) {
    358   uint64_t Start = OS.tell();
    359   (void) Start;
    360 
    361   Write32(Type);
    362   Write32(macho::LinkeditLoadCommandSize);
    363   Write32(DataOffset);
    364   Write32(DataSize);
    365 
    366   assert(OS.tell() - Start == macho::LinkeditLoadCommandSize);
    367 }
    368 
    369 
    370 void MachObjectWriter::RecordRelocation(const MCAssembler &Asm,
    371                                         const MCAsmLayout &Layout,
    372                                         const MCFragment *Fragment,
    373                                         const MCFixup &Fixup,
    374                                         MCValue Target,
    375                                         uint64_t &FixedValue) {
    376   TargetObjectWriter->RecordRelocation(this, Asm, Layout, Fragment, Fixup,
    377                                        Target, FixedValue);
    378 }
    379 
    380 void MachObjectWriter::BindIndirectSymbols(MCAssembler &Asm) {
    381   // This is the point where 'as' creates actual symbols for indirect symbols
    382   // (in the following two passes). It would be easier for us to do this sooner
    383   // when we see the attribute, but that makes getting the order in the symbol
    384   // table much more complicated than it is worth.
    385   //
    386   // FIXME: Revisit this when the dust settles.
    387 
    388   // Bind non lazy symbol pointers first.
    389   unsigned IndirectIndex = 0;
    390   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
    391          ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
    392     const MCSectionMachO &Section =
    393       cast<MCSectionMachO>(it->SectionData->getSection());
    394 
    395     if (Section.getType() != MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS)
    396       continue;
    397 
    398     // Initialize the section indirect symbol base, if necessary.
    399     IndirectSymBase.insert(std::make_pair(it->SectionData, IndirectIndex));
    400 
    401     Asm.getOrCreateSymbolData(*it->Symbol);
    402   }
    403 
    404   // Then lazy symbol pointers and symbol stubs.
    405   IndirectIndex = 0;
    406   for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
    407          ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
    408     const MCSectionMachO &Section =
    409       cast<MCSectionMachO>(it->SectionData->getSection());
    410 
    411     if (Section.getType() != MCSectionMachO::S_LAZY_SYMBOL_POINTERS &&
    412         Section.getType() != MCSectionMachO::S_SYMBOL_STUBS)
    413       continue;
    414 
    415     // Initialize the section indirect symbol base, if necessary.
    416     IndirectSymBase.insert(std::make_pair(it->SectionData, IndirectIndex));
    417 
    418     // Set the symbol type to undefined lazy, but only on construction.
    419     //
    420     // FIXME: Do not hardcode.
    421     bool Created;
    422     MCSymbolData &Entry = Asm.getOrCreateSymbolData(*it->Symbol, &Created);
    423     if (Created)
    424       Entry.setFlags(Entry.getFlags() | 0x0001);
    425   }
    426 }
    427 
    428 /// ComputeSymbolTable - Compute the symbol table data
    429 ///
    430 /// \param StringTable [out] - The string table data.
    431 /// \param StringIndexMap [out] - Map from symbol names to offsets in the
    432 /// string table.
    433 void MachObjectWriter::
    434 ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
    435                    std::vector<MachSymbolData> &LocalSymbolData,
    436                    std::vector<MachSymbolData> &ExternalSymbolData,
    437                    std::vector<MachSymbolData> &UndefinedSymbolData) {
    438   // Build section lookup table.
    439   DenseMap<const MCSection*, uint8_t> SectionIndexMap;
    440   unsigned Index = 1;
    441   for (MCAssembler::iterator it = Asm.begin(),
    442          ie = Asm.end(); it != ie; ++it, ++Index)
    443     SectionIndexMap[&it->getSection()] = Index;
    444   assert(Index <= 256 && "Too many sections!");
    445 
    446   // Index 0 is always the empty string.
    447   StringMap<uint64_t> StringIndexMap;
    448   StringTable += '\x00';
    449 
    450   // Build the symbol arrays and the string table, but only for non-local
    451   // symbols.
    452   //
    453   // The particular order that we collect the symbols and create the string
    454   // table, then sort the symbols is chosen to match 'as'. Even though it
    455   // doesn't matter for correctness, this is important for letting us diff .o
    456   // files.
    457   for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
    458          ie = Asm.symbol_end(); it != ie; ++it) {
    459     const MCSymbol &Symbol = it->getSymbol();
    460 
    461     // Ignore non-linker visible symbols.
    462     if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
    463       continue;
    464 
    465     if (!it->isExternal() && !Symbol.isUndefined())
    466       continue;
    467 
    468     uint64_t &Entry = StringIndexMap[Symbol.getName()];
    469     if (!Entry) {
    470       Entry = StringTable.size();
    471       StringTable += Symbol.getName();
    472       StringTable += '\x00';
    473     }
    474 
    475     MachSymbolData MSD;
    476     MSD.SymbolData = it;
    477     MSD.StringIndex = Entry;
    478 
    479     if (Symbol.isUndefined()) {
    480       MSD.SectionIndex = 0;
    481       UndefinedSymbolData.push_back(MSD);
    482     } else if (Symbol.isAbsolute()) {
    483       MSD.SectionIndex = 0;
    484       ExternalSymbolData.push_back(MSD);
    485     } else {
    486       MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
    487       assert(MSD.SectionIndex && "Invalid section index!");
    488       ExternalSymbolData.push_back(MSD);
    489     }
    490   }
    491 
    492   // Now add the data for local symbols.
    493   for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
    494          ie = Asm.symbol_end(); it != ie; ++it) {
    495     const MCSymbol &Symbol = it->getSymbol();
    496 
    497     // Ignore non-linker visible symbols.
    498     if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
    499       continue;
    500 
    501     if (it->isExternal() || Symbol.isUndefined())
    502       continue;
    503 
    504     uint64_t &Entry = StringIndexMap[Symbol.getName()];
    505     if (!Entry) {
    506       Entry = StringTable.size();
    507       StringTable += Symbol.getName();
    508       StringTable += '\x00';
    509     }
    510 
    511     MachSymbolData MSD;
    512     MSD.SymbolData = it;
    513     MSD.StringIndex = Entry;
    514 
    515     if (Symbol.isAbsolute()) {
    516       MSD.SectionIndex = 0;
    517       LocalSymbolData.push_back(MSD);
    518     } else {
    519       MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
    520       assert(MSD.SectionIndex && "Invalid section index!");
    521       LocalSymbolData.push_back(MSD);
    522     }
    523   }
    524 
    525   // External and undefined symbols are required to be in lexicographic order.
    526   std::sort(ExternalSymbolData.begin(), ExternalSymbolData.end());
    527   std::sort(UndefinedSymbolData.begin(), UndefinedSymbolData.end());
    528 
    529   // Set the symbol indices.
    530   Index = 0;
    531   for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
    532     LocalSymbolData[i].SymbolData->setIndex(Index++);
    533   for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
    534     ExternalSymbolData[i].SymbolData->setIndex(Index++);
    535   for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
    536     UndefinedSymbolData[i].SymbolData->setIndex(Index++);
    537 
    538   // The string table is padded to a multiple of 4.
    539   while (StringTable.size() % 4)
    540     StringTable += '\x00';
    541 }
    542 
    543 void MachObjectWriter::computeSectionAddresses(const MCAssembler &Asm,
    544                                                const MCAsmLayout &Layout) {
    545   uint64_t StartAddress = 0;
    546   const SmallVectorImpl<MCSectionData*> &Order = Layout.getSectionOrder();
    547   for (int i = 0, n = Order.size(); i != n ; ++i) {
    548     const MCSectionData *SD = Order[i];
    549     StartAddress = RoundUpToAlignment(StartAddress, SD->getAlignment());
    550     SectionAddress[SD] = StartAddress;
    551     StartAddress += Layout.getSectionAddressSize(SD);
    552 
    553     // Explicitly pad the section to match the alignment requirements of the
    554     // following one. This is for 'gas' compatibility, it shouldn't
    555     /// strictly be necessary.
    556     StartAddress += getPaddingSize(SD, Layout);
    557   }
    558 }
    559 
    560 void MachObjectWriter::ExecutePostLayoutBinding(MCAssembler &Asm,
    561                                                 const MCAsmLayout &Layout) {
    562   computeSectionAddresses(Asm, Layout);
    563 
    564   // Create symbol data for any indirect symbols.
    565   BindIndirectSymbols(Asm);
    566 
    567   // Compute symbol table information and bind symbol indices.
    568   ComputeSymbolTable(Asm, StringTable, LocalSymbolData, ExternalSymbolData,
    569                      UndefinedSymbolData);
    570 }
    571 
    572 bool MachObjectWriter::
    573 IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
    574                                        const MCSymbolData &DataA,
    575                                        const MCFragment &FB,
    576                                        bool InSet,
    577                                        bool IsPCRel) const {
    578   if (InSet)
    579     return true;
    580 
    581   // The effective address is
    582   //     addr(atom(A)) + offset(A)
    583   //   - addr(atom(B)) - offset(B)
    584   // and the offsets are not relocatable, so the fixup is fully resolved when
    585   //  addr(atom(A)) - addr(atom(B)) == 0.
    586   const MCSymbolData *A_Base = 0, *B_Base = 0;
    587 
    588   const MCSymbol &SA = DataA.getSymbol().AliasedSymbol();
    589   const MCSection &SecA = SA.getSection();
    590   const MCSection &SecB = FB.getParent()->getSection();
    591 
    592   if (IsPCRel) {
    593     // The simple (Darwin, except on x86_64) way of dealing with this was to
    594     // assume that any reference to a temporary symbol *must* be a temporary
    595     // symbol in the same atom, unless the sections differ. Therefore, any PCrel
    596     // relocation to a temporary symbol (in the same section) is fully
    597     // resolved. This also works in conjunction with absolutized .set, which
    598     // requires the compiler to use .set to absolutize the differences between
    599     // symbols which the compiler knows to be assembly time constants, so we
    600     // don't need to worry about considering symbol differences fully resolved.
    601     //
    602     // If the file isn't using sub-sections-via-symbols, we can make the
    603     // same assumptions about any symbol that we normally make about
    604     // assembler locals.
    605 
    606     if (!Asm.getBackend().hasReliableSymbolDifference()) {
    607       if (!SA.isInSection() || &SecA != &SecB ||
    608           (!SA.isTemporary() &&
    609            FB.getAtom() != Asm.getSymbolData(SA).getFragment()->getAtom() &&
    610            Asm.getSubsectionsViaSymbols()))
    611         return false;
    612       return true;
    613     }
    614     // For Darwin x86_64, there is one special case when the reference IsPCRel.
    615     // If the fragment with the reference does not have a base symbol but meets
    616     // the simple way of dealing with this, in that it is a temporary symbol in
    617     // the same atom then it is assumed to be fully resolved.  This is needed so
    618     // a relocation entry is not created and so the static linker does not
    619     // mess up the reference later.
    620     else if(!FB.getAtom() &&
    621             SA.isTemporary() && SA.isInSection() && &SecA == &SecB){
    622       return true;
    623     }
    624   } else {
    625     if (!TargetObjectWriter->useAggressiveSymbolFolding())
    626       return false;
    627   }
    628 
    629   const MCFragment *FA = Asm.getSymbolData(SA).getFragment();
    630 
    631   // Bail if the symbol has no fragment.
    632   if (!FA)
    633     return false;
    634 
    635   A_Base = FA->getAtom();
    636   if (!A_Base)
    637     return false;
    638 
    639   B_Base = FB.getAtom();
    640   if (!B_Base)
    641     return false;
    642 
    643   // If the atoms are the same, they are guaranteed to have the same address.
    644   if (A_Base == B_Base)
    645     return true;
    646 
    647   // Otherwise, we can't prove this is fully resolved.
    648   return false;
    649 }
    650 
    651 void MachObjectWriter::WriteObject(MCAssembler &Asm,
    652                                    const MCAsmLayout &Layout) {
    653   unsigned NumSections = Asm.size();
    654 
    655   // The section data starts after the header, the segment load command (and
    656   // section headers) and the symbol table.
    657   unsigned NumLoadCommands = 1;
    658   uint64_t LoadCommandsSize = is64Bit() ?
    659     macho::SegmentLoadCommand64Size + NumSections * macho::Section64Size :
    660     macho::SegmentLoadCommand32Size + NumSections * macho::Section32Size;
    661 
    662   // Add the symbol table load command sizes, if used.
    663   unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() +
    664     UndefinedSymbolData.size();
    665   if (NumSymbols) {
    666     NumLoadCommands += 2;
    667     LoadCommandsSize += (macho::SymtabLoadCommandSize +
    668                          macho::DysymtabLoadCommandSize);
    669   }
    670 
    671   // Add the data-in-code load command size, if used.
    672   unsigned NumDataRegions = Asm.getDataRegions().size();
    673   if (NumDataRegions) {
    674     ++NumLoadCommands;
    675     LoadCommandsSize += macho::LinkeditLoadCommandSize;
    676   }
    677 
    678   // Compute the total size of the section data, as well as its file size and vm
    679   // size.
    680   uint64_t SectionDataStart = (is64Bit() ? macho::Header64Size :
    681                                macho::Header32Size) + LoadCommandsSize;
    682   uint64_t SectionDataSize = 0;
    683   uint64_t SectionDataFileSize = 0;
    684   uint64_t VMSize = 0;
    685   for (MCAssembler::const_iterator it = Asm.begin(),
    686          ie = Asm.end(); it != ie; ++it) {
    687     const MCSectionData &SD = *it;
    688     uint64_t Address = getSectionAddress(&SD);
    689     uint64_t Size = Layout.getSectionAddressSize(&SD);
    690     uint64_t FileSize = Layout.getSectionFileSize(&SD);
    691     FileSize += getPaddingSize(&SD, Layout);
    692 
    693     VMSize = std::max(VMSize, Address + Size);
    694 
    695     if (SD.getSection().isVirtualSection())
    696       continue;
    697 
    698     SectionDataSize = std::max(SectionDataSize, Address + Size);
    699     SectionDataFileSize = std::max(SectionDataFileSize, Address + FileSize);
    700   }
    701 
    702   // The section data is padded to 4 bytes.
    703   //
    704   // FIXME: Is this machine dependent?
    705   unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
    706   SectionDataFileSize += SectionDataPadding;
    707 
    708   // Write the prolog, starting with the header and load command...
    709   WriteHeader(NumLoadCommands, LoadCommandsSize,
    710               Asm.getSubsectionsViaSymbols());
    711   WriteSegmentLoadCommand(NumSections, VMSize,
    712                           SectionDataStart, SectionDataSize);
    713 
    714   // ... and then the section headers.
    715   uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
    716   for (MCAssembler::const_iterator it = Asm.begin(),
    717          ie = Asm.end(); it != ie; ++it) {
    718     std::vector<macho::RelocationEntry> &Relocs = Relocations[it];
    719     unsigned NumRelocs = Relocs.size();
    720     uint64_t SectionStart = SectionDataStart + getSectionAddress(it);
    721     WriteSection(Asm, Layout, *it, SectionStart, RelocTableEnd, NumRelocs);
    722     RelocTableEnd += NumRelocs * macho::RelocationInfoSize;
    723   }
    724 
    725   // Write the data-in-code load command, if used.
    726   uint64_t DataInCodeTableEnd = RelocTableEnd + NumDataRegions * 8;
    727   if (NumDataRegions) {
    728     uint64_t DataRegionsOffset = RelocTableEnd;
    729     uint64_t DataRegionsSize = NumDataRegions * 8;
    730     WriteLinkeditLoadCommand(macho::LCT_DataInCode, DataRegionsOffset,
    731                              DataRegionsSize);
    732   }
    733 
    734   // Write the symbol table load command, if used.
    735   if (NumSymbols) {
    736     unsigned FirstLocalSymbol = 0;
    737     unsigned NumLocalSymbols = LocalSymbolData.size();
    738     unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
    739     unsigned NumExternalSymbols = ExternalSymbolData.size();
    740     unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
    741     unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
    742     unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
    743     unsigned NumSymTabSymbols =
    744       NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
    745     uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
    746     uint64_t IndirectSymbolOffset = 0;
    747 
    748     // If used, the indirect symbols are written after the section data.
    749     if (NumIndirectSymbols)
    750       IndirectSymbolOffset = DataInCodeTableEnd;
    751 
    752     // The symbol table is written after the indirect symbol data.
    753     uint64_t SymbolTableOffset = DataInCodeTableEnd + IndirectSymbolSize;
    754 
    755     // The string table is written after symbol table.
    756     uint64_t StringTableOffset =
    757       SymbolTableOffset + NumSymTabSymbols * (is64Bit() ? macho::Nlist64Size :
    758                                               macho::Nlist32Size);
    759     WriteSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
    760                            StringTableOffset, StringTable.size());
    761 
    762     WriteDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
    763                              FirstExternalSymbol, NumExternalSymbols,
    764                              FirstUndefinedSymbol, NumUndefinedSymbols,
    765                              IndirectSymbolOffset, NumIndirectSymbols);
    766   }
    767 
    768   // Write the actual section data.
    769   for (MCAssembler::const_iterator it = Asm.begin(),
    770          ie = Asm.end(); it != ie; ++it) {
    771     Asm.writeSectionData(it, Layout);
    772 
    773     uint64_t Pad = getPaddingSize(it, Layout);
    774     for (unsigned int i = 0; i < Pad; ++i)
    775       Write8(0);
    776   }
    777 
    778   // Write the extra padding.
    779   WriteZeros(SectionDataPadding);
    780 
    781   // Write the relocation entries.
    782   for (MCAssembler::const_iterator it = Asm.begin(),
    783          ie = Asm.end(); it != ie; ++it) {
    784     // Write the section relocation entries, in reverse order to match 'as'
    785     // (approximately, the exact algorithm is more complicated than this).
    786     std::vector<macho::RelocationEntry> &Relocs = Relocations[it];
    787     for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
    788       Write32(Relocs[e - i - 1].Word0);
    789       Write32(Relocs[e - i - 1].Word1);
    790     }
    791   }
    792 
    793   // Write out the data-in-code region payload, if there is one.
    794   for (MCAssembler::const_data_region_iterator
    795          it = Asm.data_region_begin(), ie = Asm.data_region_end();
    796          it != ie; ++it) {
    797     const DataRegionData *Data = &(*it);
    798     uint64_t Start = getSymbolAddress(&Layout.getAssembler().getSymbolData(*Data->Start), Layout);
    799     uint64_t End = getSymbolAddress(&Layout.getAssembler().getSymbolData(*Data->End), Layout);
    800     DEBUG(dbgs() << "data in code region-- kind: " << Data->Kind
    801                  << "  start: " << Start << "(" << Data->Start->getName() << ")"
    802                  << "  end: " << End << "(" << Data->End->getName() << ")"
    803                  << "  size: " << End - Start
    804                  << "\n");
    805     Write32(Start);
    806     Write16(End - Start);
    807     Write16(Data->Kind);
    808   }
    809 
    810   // Write the symbol table data, if used.
    811   if (NumSymbols) {
    812     // Write the indirect symbol entries.
    813     for (MCAssembler::const_indirect_symbol_iterator
    814            it = Asm.indirect_symbol_begin(),
    815            ie = Asm.indirect_symbol_end(); it != ie; ++it) {
    816       // Indirect symbols in the non lazy symbol pointer section have some
    817       // special handling.
    818       const MCSectionMachO &Section =
    819         static_cast<const MCSectionMachO&>(it->SectionData->getSection());
    820       if (Section.getType() == MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS) {
    821         // If this symbol is defined and internal, mark it as such.
    822         if (it->Symbol->isDefined() &&
    823             !Asm.getSymbolData(*it->Symbol).isExternal()) {
    824           uint32_t Flags = macho::ISF_Local;
    825           if (it->Symbol->isAbsolute())
    826             Flags |= macho::ISF_Absolute;
    827           Write32(Flags);
    828           continue;
    829         }
    830       }
    831 
    832       Write32(Asm.getSymbolData(*it->Symbol).getIndex());
    833     }
    834 
    835     // FIXME: Check that offsets match computed ones.
    836 
    837     // Write the symbol table entries.
    838     for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
    839       WriteNlist(LocalSymbolData[i], Layout);
    840     for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
    841       WriteNlist(ExternalSymbolData[i], Layout);
    842     for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
    843       WriteNlist(UndefinedSymbolData[i], Layout);
    844 
    845     // Write the string table.
    846     OS << StringTable.str();
    847   }
    848 }
    849 
    850 MCObjectWriter *llvm::createMachObjectWriter(MCMachObjectTargetWriter *MOTW,
    851                                              raw_ostream &OS,
    852                                              bool IsLittleEndian) {
    853   return new MachObjectWriter(MOTW, OS, IsLittleEndian);
    854 }
    855