X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FMC%2FMachObjectWriter.cpp;h=5820a224c527b57f060d3212fc36512de3649079;hb=fef904d0e824a2c587f8c1063b6c4fbf47fec898;hp=e11b9df82b714ea3b8e83a1612a4e7a8c2c744cd;hpb=32e7abd0804de228c50d90e9ee43568c4316b624;p=oota-llvm.git diff --git a/lib/MC/MachObjectWriter.cpp b/lib/MC/MachObjectWriter.cpp index e11b9df82b7..5820a224c52 100644 --- a/lib/MC/MachObjectWriter.cpp +++ b/lib/MC/MachObjectWriter.cpp @@ -8,49 +8,28 @@ //===----------------------------------------------------------------------===// #include "llvm/MC/MCMachObjectWriter.h" -#include "llvm/ADT/OwningPtr.h" #include "llvm/ADT/StringMap.h" #include "llvm/ADT/Twine.h" #include "llvm/MC/MCAssembler.h" +#include "llvm/MC/MCAsmBackend.h" #include "llvm/MC/MCAsmLayout.h" #include "llvm/MC/MCExpr.h" +#include "llvm/MC/MCFixupKindInfo.h" #include "llvm/MC/MCObjectWriter.h" #include "llvm/MC/MCSectionMachO.h" #include "llvm/MC/MCSymbol.h" #include "llvm/MC/MCMachOSymbolFlags.h" #include "llvm/MC/MCValue.h" #include "llvm/Object/MachOFormat.h" +#include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" -#include "llvm/Target/TargetAsmBackend.h" - -// FIXME: Gross. -#include "../Target/ARM/ARMFixupKinds.h" -#include "../Target/X86/X86FixupKinds.h" #include using namespace llvm; using namespace llvm::object; -// FIXME: this has been copied from (or to) X86AsmBackend.cpp -static unsigned getFixupKindLog2Size(unsigned Kind) { - switch (Kind) { - default: - llvm_unreachable("invalid fixup kind!"); - case FK_PCRel_1: - case FK_Data_1: return 0; - case FK_PCRel_2: - case FK_Data_2: return 1; - case FK_PCRel_4: - // FIXME: Remove these!!! - case X86::reloc_riprel_4byte: - case X86::reloc_riprel_4byte_movq_load: - case X86::reloc_signed_4byte: - case FK_Data_4: return 2; - case FK_Data_8: return 3; - } -} - -static bool doesSymbolRequireExternRelocation(MCSymbolData *SD) { +bool MachObjectWriter:: +doesSymbolRequireExternRelocation(const MCSymbolData *SD) { // Undefined symbols are always extern. if (SD->Symbol->isUndefined()) return true; @@ -64,207 +43,24 @@ static bool doesSymbolRequireExternRelocation(MCSymbolData *SD) { return false; } -namespace { - -class MachObjectWriter : public MCObjectWriter { - /// MachSymbolData - Helper struct for containing some precomputed information - /// on symbols. - struct MachSymbolData { - MCSymbolData *SymbolData; - uint64_t StringIndex; - uint8_t SectionIndex; - - // Support lexicographic sorting. - bool operator<(const MachSymbolData &RHS) const { - return SymbolData->getSymbol().getName() < - RHS.SymbolData->getSymbol().getName(); - } - }; - - /// The target specific Mach-O writer instance. - llvm::OwningPtr TargetObjectWriter; - - /// @name Relocation Data - /// @{ - - llvm::DenseMap > Relocations; - llvm::DenseMap IndirectSymBase; - - /// @} - /// @name Symbol Table Data - /// @{ - - SmallString<256> StringTable; - std::vector LocalSymbolData; - std::vector ExternalSymbolData; - std::vector UndefinedSymbolData; - - /// @} - -private: - /// @name Utility Methods - /// @{ - - bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) { - const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo( - (MCFixupKind) Kind); - - return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel; - } - - /// @} - - SectionAddrMap SectionAddress; - uint64_t getSectionAddress(const MCSectionData* SD) const { - return SectionAddress.lookup(SD); - } - uint64_t getSymbolAddress(const MCSymbolData* SD, - const MCAsmLayout &Layout) const; - - uint64_t getFragmentAddress(const MCFragment *Fragment, - const MCAsmLayout &Layout) const { - return getSectionAddress(Fragment->getParent()) + - Layout.getFragmentOffset(Fragment); - } - - uint64_t getPaddingSize(const MCSectionData *SD, - const MCAsmLayout &Layout) const; -public: - MachObjectWriter(MCMachObjectTargetWriter *MOTW, raw_ostream &_OS, - bool _IsLittleEndian) - : MCObjectWriter(_OS, _IsLittleEndian), TargetObjectWriter(MOTW) { - } - - /// @name Target Writer Proxy Accessors - /// @{ +bool MachObjectWriter:: +MachSymbolData::operator<(const MachSymbolData &RHS) const { + return SymbolData->getSymbol().getName() < + RHS.SymbolData->getSymbol().getName(); +} - bool is64Bit() const { return TargetObjectWriter->is64Bit(); } - bool isARM() const { - uint32_t CPUType = TargetObjectWriter->getCPUType() & ~mach::CTFM_ArchMask; - return CPUType == mach::CTM_ARM; - } +bool MachObjectWriter::isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) { + const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo( + (MCFixupKind) Kind); - /// @} - - void WriteHeader(unsigned NumLoadCommands, unsigned LoadCommandsSize, - bool SubsectionsViaSymbols); - - /// WriteSegmentLoadCommand - Write a segment load command. - /// - /// \arg NumSections - The number of sections in this segment. - /// \arg SectionDataSize - The total size of the sections. - void WriteSegmentLoadCommand(unsigned NumSections, - uint64_t VMSize, - uint64_t SectionDataStartOffset, - uint64_t SectionDataSize); - - void WriteSection(const MCAssembler &Asm, const MCAsmLayout &Layout, - const MCSectionData &SD, uint64_t FileOffset, - uint64_t RelocationsStart, unsigned NumRelocations); - - void WriteSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols, - uint32_t StringTableOffset, - uint32_t StringTableSize); - - void WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol, - uint32_t NumLocalSymbols, - uint32_t FirstExternalSymbol, - uint32_t NumExternalSymbols, - uint32_t FirstUndefinedSymbol, - uint32_t NumUndefinedSymbols, - uint32_t IndirectSymbolOffset, - uint32_t NumIndirectSymbols); - - void WriteNlist(MachSymbolData &MSD, const MCAsmLayout &Layout); - - // FIXME: We really need to improve the relocation validation. Basically, we - // want to implement a separate computation which evaluates the relocation - // entry as the linker would, and verifies that the resultant fixup value is - // exactly what the encoder wanted. This will catch several classes of - // problems: - // - // - Relocation entry bugs, the two algorithms are unlikely to have the same - // exact bug. - // - // - Relaxation issues, where we forget to relax something. - // - // - Input errors, where something cannot be correctly encoded. 'as' allows - // these through in many cases. + return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel; +} - static bool isFixupKindRIPRel(unsigned Kind) { - return Kind == X86::reloc_riprel_4byte || - Kind == X86::reloc_riprel_4byte_movq_load; - } - void RecordX86_64Relocation(const MCAssembler &Asm, const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, MCValue Target, - uint64_t &FixedValue); - - void RecordScatteredRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, MCValue Target, - unsigned Log2Size, - uint64_t &FixedValue); - - void RecordARMScatteredRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, MCValue Target, - unsigned Log2Size, - uint64_t &FixedValue); - - void RecordARMMovwMovtRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, MCValue Target, - uint64_t &FixedValue); - - void RecordTLVPRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, MCValue Target, - uint64_t &FixedValue); - - static bool getARMFixupKindMachOInfo(unsigned Kind, unsigned &RelocType, - unsigned &Log2Size); - - void RecordARMRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout, - const MCFragment *Fragment, const MCFixup &Fixup, - MCValue Target, uint64_t &FixedValue); - - void RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout, - const MCFragment *Fragment, const MCFixup &Fixup, - MCValue Target, uint64_t &FixedValue); - - void BindIndirectSymbols(MCAssembler &Asm); - - /// ComputeSymbolTable - Compute the symbol table data - /// - /// \param StringTable [out] - The string table data. - /// \param StringIndexMap [out] - Map from symbol names to offsets in the - /// string table. - void ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable, - std::vector &LocalSymbolData, - std::vector &ExternalSymbolData, - std::vector &UndefinedSymbolData); - - void computeSectionAddresses(const MCAssembler &Asm, - const MCAsmLayout &Layout); - - void ExecutePostLayoutBinding(MCAssembler &Asm, const MCAsmLayout &Layout); - - virtual bool IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm, - const MCSymbolData &DataA, - const MCFragment &FB, - bool InSet, - bool IsPCRel) const; - - void WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout); -}; - -} // end anonymous namespace +uint64_t MachObjectWriter::getFragmentAddress(const MCFragment *Fragment, + const MCAsmLayout &Layout) const { + return getSectionAddress(Fragment->getParent()) + + Layout.getFragmentOffset(Fragment); +} uint64_t MachObjectWriter::getSymbolAddress(const MCSymbolData* SD, const MCAsmLayout &Layout) const { @@ -496,7 +292,7 @@ void MachObjectWriter::WriteNlist(MachSymbolData &MSD, const MCSymbol &Symbol = Data.getSymbol(); uint8_t Type = 0; uint16_t Flags = Data.getFlags(); - uint32_t Address = 0; + uint64_t Address = 0; // Set the N_TYPE bits. See . // @@ -556,673 +352,20 @@ void MachObjectWriter::WriteNlist(MachSymbolData &MSD, Write32(Address); } -void MachObjectWriter::RecordX86_64Relocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, - MCValue Target, - uint64_t &FixedValue) { - unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind()); - unsigned IsRIPRel = isFixupKindRIPRel(Fixup.getKind()); - unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind()); - - // See . - uint32_t FixupOffset = - Layout.getFragmentOffset(Fragment) + Fixup.getOffset(); - uint32_t FixupAddress = - getFragmentAddress(Fragment, Layout) + Fixup.getOffset(); - int64_t Value = 0; - unsigned Index = 0; - unsigned IsExtern = 0; - unsigned Type = 0; - - Value = Target.getConstant(); - - if (IsPCRel) { - // Compensate for the relocation offset, Darwin x86_64 relocations only have - // the addend and appear to have attempted to define it to be the actual - // expression addend without the PCrel bias. However, instructions with data - // following the relocation are not accommodated for (see comment below - // regarding SIGNED{1,2,4}), so it isn't exactly that either. - Value += 1LL << Log2Size; - } - - if (Target.isAbsolute()) { // constant - // SymbolNum of 0 indicates the absolute section. - Type = macho::RIT_X86_64_Unsigned; - Index = 0; - - // FIXME: I believe this is broken, I don't think the linker can understand - // it. I think it would require a local relocation, but I'm not sure if that - // would work either. The official way to get an absolute PCrel relocation - // is to use an absolute symbol (which we don't support yet). - if (IsPCRel) { - IsExtern = 1; - Type = macho::RIT_X86_64_Branch; - } - } else if (Target.getSymB()) { // A - B + constant - const MCSymbol *A = &Target.getSymA()->getSymbol(); - MCSymbolData &A_SD = Asm.getSymbolData(*A); - const MCSymbolData *A_Base = Asm.getAtom(&A_SD); - - const MCSymbol *B = &Target.getSymB()->getSymbol(); - MCSymbolData &B_SD = Asm.getSymbolData(*B); - const MCSymbolData *B_Base = Asm.getAtom(&B_SD); - - // Neither symbol can be modified. - if (Target.getSymA()->getKind() != MCSymbolRefExpr::VK_None || - Target.getSymB()->getKind() != MCSymbolRefExpr::VK_None) - report_fatal_error("unsupported relocation of modified symbol"); - - // We don't support PCrel relocations of differences. Darwin 'as' doesn't - // implement most of these correctly. - if (IsPCRel) - report_fatal_error("unsupported pc-relative relocation of difference"); - - // The support for the situation where one or both of the symbols would - // require a local relocation is handled just like if the symbols were - // external. This is certainly used in the case of debug sections where the - // section has only temporary symbols and thus the symbols don't have base - // symbols. This is encoded using the section ordinal and non-extern - // relocation entries. - - // Darwin 'as' doesn't emit correct relocations for this (it ends up with a - // single SIGNED relocation); reject it for now. Except the case where both - // symbols don't have a base, equal but both NULL. - if (A_Base == B_Base && A_Base) - report_fatal_error("unsupported relocation with identical base"); - - Value += getSymbolAddress(&A_SD, Layout) - - (A_Base == NULL ? 0 : getSymbolAddress(A_Base, Layout)); - Value -= getSymbolAddress(&B_SD, Layout) - - (B_Base == NULL ? 0 : getSymbolAddress(B_Base, Layout)); - - if (A_Base) { - Index = A_Base->getIndex(); - IsExtern = 1; - } - else { - Index = A_SD.getFragment()->getParent()->getOrdinal() + 1; - IsExtern = 0; - } - Type = macho::RIT_X86_64_Unsigned; - - macho::RelocationEntry MRE; - MRE.Word0 = FixupOffset; - MRE.Word1 = ((Index << 0) | - (IsPCRel << 24) | - (Log2Size << 25) | - (IsExtern << 27) | - (Type << 28)); - Relocations[Fragment->getParent()].push_back(MRE); - - if (B_Base) { - Index = B_Base->getIndex(); - IsExtern = 1; - } - else { - Index = B_SD.getFragment()->getParent()->getOrdinal() + 1; - IsExtern = 0; - } - Type = macho::RIT_X86_64_Subtractor; - } else { - const MCSymbol *Symbol = &Target.getSymA()->getSymbol(); - MCSymbolData &SD = Asm.getSymbolData(*Symbol); - const MCSymbolData *Base = Asm.getAtom(&SD); - - // Relocations inside debug sections always use local relocations when - // possible. This seems to be done because the debugger doesn't fully - // understand x86_64 relocation entries, and expects to find values that - // have already been fixed up. - if (Symbol->isInSection()) { - const MCSectionMachO &Section = static_cast( - Fragment->getParent()->getSection()); - if (Section.hasAttribute(MCSectionMachO::S_ATTR_DEBUG)) - Base = 0; - } - - // x86_64 almost always uses external relocations, except when there is no - // symbol to use as a base address (a local symbol with no preceding - // non-local symbol). - if (Base) { - Index = Base->getIndex(); - IsExtern = 1; - - // Add the local offset, if needed. - if (Base != &SD) - Value += Layout.getSymbolOffset(&SD) - Layout.getSymbolOffset(Base); - } else if (Symbol->isInSection() && !Symbol->isVariable()) { - // The index is the section ordinal (1-based). - Index = SD.getFragment()->getParent()->getOrdinal() + 1; - IsExtern = 0; - Value += getSymbolAddress(&SD, Layout); - - if (IsPCRel) - Value -= FixupAddress + (1 << Log2Size); - } else if (Symbol->isVariable()) { - const MCExpr *Value = Symbol->getVariableValue(); - int64_t Res; - bool isAbs = Value->EvaluateAsAbsolute(Res, Layout, SectionAddress); - if (isAbs) { - FixedValue = Res; - return; - } else { - report_fatal_error("unsupported relocation of variable '" + - Symbol->getName() + "'"); - } - } else { - report_fatal_error("unsupported relocation of undefined symbol '" + - Symbol->getName() + "'"); - } - - MCSymbolRefExpr::VariantKind Modifier = Target.getSymA()->getKind(); - if (IsPCRel) { - if (IsRIPRel) { - if (Modifier == MCSymbolRefExpr::VK_GOTPCREL) { - // x86_64 distinguishes movq foo@GOTPCREL so that the linker can - // rewrite the movq to an leaq at link time if the symbol ends up in - // the same linkage unit. - if (unsigned(Fixup.getKind()) == X86::reloc_riprel_4byte_movq_load) - Type = macho::RIT_X86_64_GOTLoad; - else - Type = macho::RIT_X86_64_GOT; - } else if (Modifier == MCSymbolRefExpr::VK_TLVP) { - Type = macho::RIT_X86_64_TLV; - } else if (Modifier != MCSymbolRefExpr::VK_None) { - report_fatal_error("unsupported symbol modifier in relocation"); - } else { - Type = macho::RIT_X86_64_Signed; - - // The Darwin x86_64 relocation format has a problem where it cannot - // encode an address (L + ) which is outside the atom - // containing L. Generally, this shouldn't occur but it does - // happen when we have a RIPrel instruction with data following the - // relocation entry (e.g., movb $012, L0(%rip)). Even with the PCrel - // adjustment Darwin x86_64 uses, the offset is still negative and the - // linker has no way to recognize this. - // - // To work around this, Darwin uses several special relocation types - // to indicate the offsets. However, the specification or - // implementation of these seems to also be incomplete; they should - // adjust the addend as well based on the actual encoded instruction - // (the additional bias), but instead appear to just look at the final - // offset. - switch (-(Target.getConstant() + (1LL << Log2Size))) { - case 1: Type = macho::RIT_X86_64_Signed1; break; - case 2: Type = macho::RIT_X86_64_Signed2; break; - case 4: Type = macho::RIT_X86_64_Signed4; break; - } - } - } else { - if (Modifier != MCSymbolRefExpr::VK_None) - report_fatal_error("unsupported symbol modifier in branch " - "relocation"); - - Type = macho::RIT_X86_64_Branch; - } - } else { - if (Modifier == MCSymbolRefExpr::VK_GOT) { - Type = macho::RIT_X86_64_GOT; - } else if (Modifier == MCSymbolRefExpr::VK_GOTPCREL) { - // GOTPCREL is allowed as a modifier on non-PCrel instructions, in which - // case all we do is set the PCrel bit in the relocation entry; this is - // used with exception handling, for example. The source is required to - // include any necessary offset directly. - Type = macho::RIT_X86_64_GOT; - IsPCRel = 1; - } else if (Modifier == MCSymbolRefExpr::VK_TLVP) { - report_fatal_error("TLVP symbol modifier should have been rip-rel"); - } else if (Modifier != MCSymbolRefExpr::VK_None) - report_fatal_error("unsupported symbol modifier in relocation"); - else - Type = macho::RIT_X86_64_Unsigned; - } - } - - // x86_64 always writes custom values into the fixups. - FixedValue = Value; - - // struct relocation_info (8 bytes) - macho::RelocationEntry MRE; - MRE.Word0 = FixupOffset; - MRE.Word1 = ((Index << 0) | - (IsPCRel << 24) | - (Log2Size << 25) | - (IsExtern << 27) | - (Type << 28)); - Relocations[Fragment->getParent()].push_back(MRE); -} - -void MachObjectWriter::RecordScatteredRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, - MCValue Target, - unsigned Log2Size, - uint64_t &FixedValue) { - uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset(); - unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind()); - unsigned Type = macho::RIT_Vanilla; - - // See . - const MCSymbol *A = &Target.getSymA()->getSymbol(); - MCSymbolData *A_SD = &Asm.getSymbolData(*A); - - if (!A_SD->getFragment()) - report_fatal_error("symbol '" + A->getName() + - "' can not be undefined in a subtraction expression"); - - uint32_t Value = getSymbolAddress(A_SD, Layout); - uint64_t SecAddr = getSectionAddress(A_SD->getFragment()->getParent()); - FixedValue += SecAddr; - uint32_t Value2 = 0; - - if (const MCSymbolRefExpr *B = Target.getSymB()) { - MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol()); - - if (!B_SD->getFragment()) - report_fatal_error("symbol '" + B->getSymbol().getName() + - "' can not be undefined in a subtraction expression"); - - // Select the appropriate difference relocation type. - // - // Note that there is no longer any semantic difference between these two - // relocation types from the linkers point of view, this is done solely for - // pedantic compatibility with 'as'. - Type = A_SD->isExternal() ? (unsigned)macho::RIT_Difference : - (unsigned)macho::RIT_Generic_LocalDifference; - Value2 = getSymbolAddress(B_SD, Layout); - FixedValue -= getSectionAddress(B_SD->getFragment()->getParent()); - } - - // Relocations are written out in reverse order, so the PAIR comes first. - if (Type == macho::RIT_Difference || - Type == macho::RIT_Generic_LocalDifference) { - macho::RelocationEntry MRE; - MRE.Word0 = ((0 << 0) | - (macho::RIT_Pair << 24) | - (Log2Size << 28) | - (IsPCRel << 30) | - macho::RF_Scattered); - MRE.Word1 = Value2; - Relocations[Fragment->getParent()].push_back(MRE); - } - - macho::RelocationEntry MRE; - MRE.Word0 = ((FixupOffset << 0) | - (Type << 24) | - (Log2Size << 28) | - (IsPCRel << 30) | - macho::RF_Scattered); - MRE.Word1 = Value; - Relocations[Fragment->getParent()].push_back(MRE); -} - -void MachObjectWriter::RecordARMScatteredRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, - MCValue Target, - unsigned Log2Size, - uint64_t &FixedValue) { - uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset(); - unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind()); - unsigned Type = macho::RIT_Vanilla; - - // See . - const MCSymbol *A = &Target.getSymA()->getSymbol(); - MCSymbolData *A_SD = &Asm.getSymbolData(*A); - - if (!A_SD->getFragment()) - report_fatal_error("symbol '" + A->getName() + - "' can not be undefined in a subtraction expression"); - - uint32_t Value = getSymbolAddress(A_SD, Layout); - uint64_t SecAddr = getSectionAddress(A_SD->getFragment()->getParent()); - FixedValue += SecAddr; - uint32_t Value2 = 0; - - if (const MCSymbolRefExpr *B = Target.getSymB()) { - MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol()); - - if (!B_SD->getFragment()) - report_fatal_error("symbol '" + B->getSymbol().getName() + - "' can not be undefined in a subtraction expression"); - - // Select the appropriate difference relocation type. - Type = macho::RIT_Difference; - Value2 = getSymbolAddress(B_SD, Layout); - FixedValue -= getSectionAddress(B_SD->getFragment()->getParent()); - } - - // Relocations are written out in reverse order, so the PAIR comes first. - if (Type == macho::RIT_Difference || - Type == macho::RIT_Generic_LocalDifference) { - macho::RelocationEntry MRE; - MRE.Word0 = ((0 << 0) | - (macho::RIT_Pair << 24) | - (Log2Size << 28) | - (IsPCRel << 30) | - macho::RF_Scattered); - MRE.Word1 = Value2; - Relocations[Fragment->getParent()].push_back(MRE); - } - - macho::RelocationEntry MRE; - MRE.Word0 = ((FixupOffset << 0) | - (Type << 24) | - (Log2Size << 28) | - (IsPCRel << 30) | - macho::RF_Scattered); - MRE.Word1 = Value; - Relocations[Fragment->getParent()].push_back(MRE); -} - -void MachObjectWriter::RecordARMMovwMovtRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, - MCValue Target, - uint64_t &FixedValue) { - uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset(); - unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind()); - unsigned Type = macho::RIT_ARM_Half; - - // See . - const MCSymbol *A = &Target.getSymA()->getSymbol(); - MCSymbolData *A_SD = &Asm.getSymbolData(*A); - - if (!A_SD->getFragment()) - report_fatal_error("symbol '" + A->getName() + - "' can not be undefined in a subtraction expression"); - - uint32_t Value = getSymbolAddress(A_SD, Layout); - uint32_t Value2 = 0; - uint64_t SecAddr = getSectionAddress(A_SD->getFragment()->getParent()); - FixedValue += SecAddr; - - if (const MCSymbolRefExpr *B = Target.getSymB()) { - MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol()); - - if (!B_SD->getFragment()) - report_fatal_error("symbol '" + B->getSymbol().getName() + - "' can not be undefined in a subtraction expression"); - - // Select the appropriate difference relocation type. - Type = macho::RIT_ARM_HalfDifference; - Value2 = getSymbolAddress(B_SD, Layout); - FixedValue -= getSectionAddress(B_SD->getFragment()->getParent()); - } - - // Relocations are written out in reverse order, so the PAIR comes first. - // ARM_RELOC_HALF and ARM_RELOC_HALF_SECTDIFF abuse the r_length field: - // - // For these two r_type relocations they always have a pair following them and - // the r_length bits are used differently. The encoding of the r_length is as - // follows: - // low bit of r_length: - // 0 - :lower16: for movw instructions - // 1 - :upper16: for movt instructions - // high bit of r_length: - // 0 - arm instructions - // 1 - thumb instructions - // the other half of the relocated expression is in the following pair - // relocation entry in the the low 16 bits of r_address field. - unsigned ThumbBit = 0; - unsigned MovtBit = 0; - switch ((unsigned)Fixup.getKind()) { - default: break; - case ARM::fixup_arm_movt_hi16: - case ARM::fixup_arm_movt_hi16_pcrel: - MovtBit = 1; - break; - case ARM::fixup_t2_movt_hi16: - case ARM::fixup_t2_movt_hi16_pcrel: - MovtBit = 1; - // Fallthrough - case ARM::fixup_t2_movw_lo16: - case ARM::fixup_t2_movw_lo16_pcrel: - ThumbBit = 1; - break; - } - - - if (Type == macho::RIT_ARM_HalfDifference) { - uint32_t OtherHalf = MovtBit - ? (FixedValue & 0xffff) : ((FixedValue & 0xffff0000) >> 16); - - macho::RelocationEntry MRE; - MRE.Word0 = ((OtherHalf << 0) | - (macho::RIT_Pair << 24) | - (MovtBit << 28) | - (ThumbBit << 29) | - (IsPCRel << 30) | - macho::RF_Scattered); - MRE.Word1 = Value2; - Relocations[Fragment->getParent()].push_back(MRE); - } - - macho::RelocationEntry MRE; - MRE.Word0 = ((FixupOffset << 0) | - (Type << 24) | - (MovtBit << 28) | - (ThumbBit << 29) | - (IsPCRel << 30) | - macho::RF_Scattered); - MRE.Word1 = Value; - Relocations[Fragment->getParent()].push_back(MRE); -} - -void MachObjectWriter::RecordTLVPRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, - MCValue Target, - uint64_t &FixedValue) { - assert(Target.getSymA()->getKind() == MCSymbolRefExpr::VK_TLVP && - !is64Bit() && - "Should only be called with a 32-bit TLVP relocation!"); - - unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind()); - uint32_t Value = Layout.getFragmentOffset(Fragment)+Fixup.getOffset(); - unsigned IsPCRel = 0; - - // Get the symbol data. - MCSymbolData *SD_A = &Asm.getSymbolData(Target.getSymA()->getSymbol()); - unsigned Index = SD_A->getIndex(); - - // We're only going to have a second symbol in pic mode and it'll be a - // subtraction from the picbase. For 32-bit pic the addend is the difference - // between the picbase and the next address. For 32-bit static the addend is - // zero. - if (Target.getSymB()) { - // If this is a subtraction then we're pcrel. - uint32_t FixupAddress = - getFragmentAddress(Fragment, Layout) + Fixup.getOffset(); - MCSymbolData *SD_B = &Asm.getSymbolData(Target.getSymB()->getSymbol()); - IsPCRel = 1; - FixedValue = (FixupAddress - getSymbolAddress(SD_B, Layout) + - Target.getConstant()); - FixedValue += 1ULL << Log2Size; - } else { - FixedValue = 0; - } - - // struct relocation_info (8 bytes) - macho::RelocationEntry MRE; - MRE.Word0 = Value; - MRE.Word1 = ((Index << 0) | - (IsPCRel << 24) | - (Log2Size << 25) | - (1 << 27) | // Extern - (macho::RIT_Generic_TLV << 28)); // Type - Relocations[Fragment->getParent()].push_back(MRE); -} - -bool MachObjectWriter::getARMFixupKindMachOInfo(unsigned Kind, - unsigned &RelocType, - unsigned &Log2Size) { - RelocType = unsigned(macho::RIT_Vanilla); - Log2Size = ~0U; - - switch (Kind) { - default: - return false; - - case FK_Data_1: - Log2Size = llvm::Log2_32(1); - return true; - case FK_Data_2: - Log2Size = llvm::Log2_32(2); - return true; - case FK_Data_4: - Log2Size = llvm::Log2_32(4); - return true; - case FK_Data_8: - Log2Size = llvm::Log2_32(8); - return true; - - // Handle 24-bit branch kinds. - case ARM::fixup_arm_ldst_pcrel_12: - case ARM::fixup_arm_pcrel_10: - case ARM::fixup_arm_adr_pcrel_12: - case ARM::fixup_arm_condbranch: - case ARM::fixup_arm_uncondbranch: - RelocType = unsigned(macho::RIT_ARM_Branch24Bit); - // Report as 'long', even though that is not quite accurate. - Log2Size = llvm::Log2_32(4); - return true; - - // Handle Thumb branches. - case ARM::fixup_arm_thumb_br: - RelocType = unsigned(macho::RIT_ARM_ThumbBranch22Bit); - Log2Size = llvm::Log2_32(2); - return true; +void MachObjectWriter::WriteLinkeditLoadCommand(uint32_t Type, + uint32_t DataOffset, + uint32_t DataSize) { + uint64_t Start = OS.tell(); + (void) Start; - case ARM::fixup_t2_uncondbranch: - case ARM::fixup_arm_thumb_bl: - case ARM::fixup_arm_thumb_blx: - RelocType = unsigned(macho::RIT_ARM_ThumbBranch22Bit); - Log2Size = llvm::Log2_32(4); - return true; + Write32(Type); + Write32(macho::LinkeditLoadCommandSize); + Write32(DataOffset); + Write32(DataSize); - case ARM::fixup_arm_movt_hi16: - case ARM::fixup_arm_movt_hi16_pcrel: - case ARM::fixup_t2_movt_hi16: - case ARM::fixup_t2_movt_hi16_pcrel: - RelocType = unsigned(macho::RIT_ARM_HalfDifference); - // Report as 'long', even though that is not quite accurate. - Log2Size = llvm::Log2_32(4); - return true; - - case ARM::fixup_arm_movw_lo16: - case ARM::fixup_arm_movw_lo16_pcrel: - case ARM::fixup_t2_movw_lo16: - case ARM::fixup_t2_movw_lo16_pcrel: - RelocType = unsigned(macho::RIT_ARM_Half); - // Report as 'long', even though that is not quite accurate. - Log2Size = llvm::Log2_32(4); - return true; - } + assert(OS.tell() - Start == macho::LinkeditLoadCommandSize); } -void MachObjectWriter::RecordARMRelocation(const MCAssembler &Asm, - const MCAsmLayout &Layout, - const MCFragment *Fragment, - const MCFixup &Fixup, - MCValue Target, - uint64_t &FixedValue) { - unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind()); - unsigned Log2Size; - unsigned RelocType = macho::RIT_Vanilla; - if (!getARMFixupKindMachOInfo(Fixup.getKind(), RelocType, Log2Size)) { - report_fatal_error("unknown ARM fixup kind!"); - return; - } - - // If this is a difference or a defined symbol plus an offset, then we need a - // scattered relocation entry. Differences always require scattered - // relocations. - if (Target.getSymB()) { - if (RelocType == macho::RIT_ARM_Half || - RelocType == macho::RIT_ARM_HalfDifference) - return RecordARMMovwMovtRelocation(Asm, Layout, Fragment, Fixup, - Target, FixedValue); - return RecordARMScatteredRelocation(Asm, Layout, Fragment, Fixup, - Target, Log2Size, FixedValue); - } - - // Get the symbol data, if any. - MCSymbolData *SD = 0; - if (Target.getSymA()) - SD = &Asm.getSymbolData(Target.getSymA()->getSymbol()); - - // FIXME: For other platforms, we need to use scattered relocations for - // internal relocations with offsets. If this is an internal relocation with - // an offset, it also needs a scattered relocation entry. - // - // Is this right for ARM? - uint32_t Offset = Target.getConstant(); - if (IsPCRel && RelocType == macho::RIT_Vanilla) - Offset += 1 << Log2Size; - if (Offset && SD && !doesSymbolRequireExternRelocation(SD)) - return RecordARMScatteredRelocation(Asm, Layout, Fragment, Fixup, Target, - Log2Size, FixedValue); - - // See . - uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset(); - unsigned Index = 0; - unsigned IsExtern = 0; - unsigned Type = 0; - - if (Target.isAbsolute()) { // constant - // FIXME! - report_fatal_error("FIXME: relocations to absolute targets " - "not yet implemented"); - } else { - // Resolve constant variables. - if (SD->getSymbol().isVariable()) { - int64_t Res; - if (SD->getSymbol().getVariableValue()->EvaluateAsAbsolute( - Res, Layout, SectionAddress)) { - FixedValue = Res; - return; - } - } - - // Check whether we need an external or internal relocation. - if (doesSymbolRequireExternRelocation(SD)) { - IsExtern = 1; - Index = SD->getIndex(); - - // For external relocations, make sure to offset the fixup value to - // compensate for the addend of the symbol address, if it was - // undefined. This occurs with weak definitions, for example. - if (!SD->Symbol->isUndefined()) - FixedValue -= Layout.getSymbolOffset(SD); - } else { - // The index is the section ordinal (1-based). - const MCSectionData &SymSD = Asm.getSectionData( - SD->getSymbol().getSection()); - Index = SymSD.getOrdinal() + 1; - FixedValue += getSectionAddress(&SymSD); - } - if (IsPCRel) - FixedValue -= getSectionAddress(Fragment->getParent()); - // The type is determined by the fixup kind. - Type = RelocType; - } - - // struct relocation_info (8 bytes) - macho::RelocationEntry MRE; - MRE.Word0 = FixupOffset; - MRE.Word1 = ((Index << 0) | - (IsPCRel << 24) | - (Log2Size << 25) | - (IsExtern << 27) | - (Type << 28)); - Relocations[Fragment->getParent()].push_back(MRE); -} void MachObjectWriter::RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout, @@ -1230,101 +373,8 @@ void MachObjectWriter::RecordRelocation(const MCAssembler &Asm, const MCFixup &Fixup, MCValue Target, uint64_t &FixedValue) { - // FIXME: These needs to be factored into the target Mach-O writer. - if (isARM()) { - RecordARMRelocation(Asm, Layout, Fragment, Fixup, Target, FixedValue); - return; - } - if (is64Bit()) { - RecordX86_64Relocation(Asm, Layout, Fragment, Fixup, Target, FixedValue); - return; - } - - unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind()); - unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind()); - - // If this is a 32-bit TLVP reloc it's handled a bit differently. - if (Target.getSymA() && - Target.getSymA()->getKind() == MCSymbolRefExpr::VK_TLVP) { - RecordTLVPRelocation(Asm, Layout, Fragment, Fixup, Target, FixedValue); - return; - } - - // If this is a difference or a defined symbol plus an offset, then we need a - // scattered relocation entry. Differences always require scattered - // relocations. - if (Target.getSymB()) - return RecordScatteredRelocation(Asm, Layout, Fragment, Fixup, - Target, Log2Size, FixedValue); - - // Get the symbol data, if any. - MCSymbolData *SD = 0; - if (Target.getSymA()) - SD = &Asm.getSymbolData(Target.getSymA()->getSymbol()); - - // If this is an internal relocation with an offset, it also needs a scattered - // relocation entry. - uint32_t Offset = Target.getConstant(); - if (IsPCRel) - Offset += 1 << Log2Size; - if (Offset && SD && !doesSymbolRequireExternRelocation(SD)) - return RecordScatteredRelocation(Asm, Layout, Fragment, Fixup, - Target, Log2Size, FixedValue); - - // See . - uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset(); - unsigned Index = 0; - unsigned IsExtern = 0; - unsigned Type = 0; - - if (Target.isAbsolute()) { // constant - // SymbolNum of 0 indicates the absolute section. - // - // FIXME: Currently, these are never generated (see code below). I cannot - // find a case where they are actually emitted. - Type = macho::RIT_Vanilla; - } else { - // Resolve constant variables. - if (SD->getSymbol().isVariable()) { - int64_t Res; - if (SD->getSymbol().getVariableValue()->EvaluateAsAbsolute( - Res, Layout, SectionAddress)) { - FixedValue = Res; - return; - } - } - - // Check whether we need an external or internal relocation. - if (doesSymbolRequireExternRelocation(SD)) { - IsExtern = 1; - Index = SD->getIndex(); - // For external relocations, make sure to offset the fixup value to - // compensate for the addend of the symbol address, if it was - // undefined. This occurs with weak definitions, for example. - if (!SD->Symbol->isUndefined()) - FixedValue -= Layout.getSymbolOffset(SD); - } else { - // The index is the section ordinal (1-based). - const MCSectionData &SymSD = Asm.getSectionData( - SD->getSymbol().getSection()); - Index = SymSD.getOrdinal() + 1; - FixedValue += getSectionAddress(&SymSD); - } - if (IsPCRel) - FixedValue -= getSectionAddress(Fragment->getParent()); - - Type = macho::RIT_Vanilla; - } - - // struct relocation_info (8 bytes) - macho::RelocationEntry MRE; - MRE.Word0 = FixupOffset; - MRE.Word1 = ((Index << 0) | - (IsPCRel << 24) | - (Log2Size << 25) | - (IsExtern << 27) | - (Type << 28)); - Relocations[Fragment->getParent()].push_back(MRE); + TargetObjectWriter->RecordRelocation(this, Asm, Layout, Fragment, Fixup, + Target, FixedValue); } void MachObjectWriter::BindIndirectSymbols(MCAssembler &Asm) { @@ -1550,20 +600,41 @@ IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm, // requires the compiler to use .set to absolutize the differences between // symbols which the compiler knows to be assembly time constants, so we // don't need to worry about considering symbol differences fully resolved. + // + // If the file isn't using sub-sections-via-symbols, we can make the + // same assumptions about any symbol that we normally make about + // assembler locals. if (!Asm.getBackend().hasReliableSymbolDifference()) { - if (!SA.isTemporary() || !SA.isInSection() || &SecA != &SecB) + if (!SA.isInSection() || &SecA != &SecB || + (!SA.isTemporary() && + FB.getAtom() != Asm.getSymbolData(SA).getFragment()->getAtom() && + Asm.getSubsectionsViaSymbols())) return false; return true; } + // For Darwin x86_64, there is one special case when the reference IsPCRel. + // If the fragment with the reference does not have a base symbol but meets + // the simple way of dealing with this, in that it is a temporary symbol in + // the same atom then it is assumed to be fully resolved. This is needed so + // a relocation entry is not created and so the static linker does not + // mess up the reference later. + else if(!FB.getAtom() && + SA.isTemporary() && SA.isInSection() && &SecA == &SecB){ + return true; + } } else { if (!TargetObjectWriter->useAggressiveSymbolFolding()) return false; } - const MCFragment &FA = *Asm.getSymbolData(SA).getFragment(); + const MCFragment *FA = Asm.getSymbolData(SA).getFragment(); + + // Bail if the symbol has no fragment. + if (!FA) + return false; - A_Base = FA.getAtom(); + A_Base = FA->getAtom(); if (!A_Base) return false; @@ -1579,7 +650,8 @@ IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm, return false; } -void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) { +void MachObjectWriter::WriteObject(MCAssembler &Asm, + const MCAsmLayout &Layout) { unsigned NumSections = Asm.size(); // The section data starts after the header, the segment load command (and @@ -1598,6 +670,13 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) macho::DysymtabLoadCommandSize); } + // Add the data-in-code load command size, if used. + unsigned NumDataRegions = Asm.getDataRegions().size(); + if (NumDataRegions) { + ++NumLoadCommands; + LoadCommandsSize += macho::LinkeditLoadCommandSize; + } + // Compute the total size of the section data, as well as its file size and vm // size. uint64_t SectionDataStart = (is64Bit() ? macho::Header64Size : @@ -1645,6 +724,15 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) RelocTableEnd += NumRelocs * macho::RelocationInfoSize; } + // Write the data-in-code load command, if used. + uint64_t DataInCodeTableEnd = RelocTableEnd + NumDataRegions * 8; + if (NumDataRegions) { + uint64_t DataRegionsOffset = RelocTableEnd; + uint64_t DataRegionsSize = NumDataRegions * 8; + WriteLinkeditLoadCommand(macho::LCT_DataInCode, DataRegionsOffset, + DataRegionsSize); + } + // Write the symbol table load command, if used. if (NumSymbols) { unsigned FirstLocalSymbol = 0; @@ -1661,10 +749,10 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) // If used, the indirect symbols are written after the section data. if (NumIndirectSymbols) - IndirectSymbolOffset = RelocTableEnd; + IndirectSymbolOffset = DataInCodeTableEnd; // The symbol table is written after the indirect symbol data. - uint64_t SymbolTableOffset = RelocTableEnd + IndirectSymbolSize; + uint64_t SymbolTableOffset = DataInCodeTableEnd + IndirectSymbolSize; // The string table is written after symbol table. uint64_t StringTableOffset = @@ -1682,7 +770,7 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) // Write the actual section data. for (MCAssembler::const_iterator it = Asm.begin(), ie = Asm.end(); it != ie; ++it) { - Asm.WriteSectionData(it, Layout); + Asm.writeSectionData(it, Layout); uint64_t Pad = getPaddingSize(it, Layout); for (unsigned int i = 0; i < Pad; ++i) @@ -1704,6 +792,23 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) } } + // Write out the data-in-code region payload, if there is one. + for (MCAssembler::const_data_region_iterator + it = Asm.data_region_begin(), ie = Asm.data_region_end(); + it != ie; ++it) { + const DataRegionData *Data = &(*it); + uint64_t Start = getSymbolAddress(&Layout.getAssembler().getSymbolData(*Data->Start), Layout); + uint64_t End = getSymbolAddress(&Layout.getAssembler().getSymbolData(*Data->End), Layout); + DEBUG(dbgs() << "data in code region-- kind: " << Data->Kind + << " start: " << Start << "(" << Data->Start->getName() << ")" + << " end: " << End << "(" << Data->End->getName() << ")" + << " size: " << End - Start + << "\n"); + Write32(Start); + Write16(End - Start); + Write16(Data->Kind); + } + // Write the symbol table data, if used. if (NumSymbols) { // Write the indirect symbol entries.