diff options
| author | Dimitry Andric <dim@FreeBSD.org> | 2011-02-20 12:57:14 +0000 | 
|---|---|---|
| committer | Dimitry Andric <dim@FreeBSD.org> | 2011-02-20 12:57:14 +0000 | 
| commit | cf099d11218cb6f6c5cce947d6738e347f07fb12 (patch) | |
| tree | d2b61ce94e654cb01a254d2195259db5f9cc3f3c /lib/Object | |
| parent | 49011b52fcba02a6051957b84705159f52fae4e4 (diff) | |
Notes
Diffstat (limited to 'lib/Object')
| -rw-r--r-- | lib/Object/CMakeLists.txt | 6 | ||||
| -rw-r--r-- | lib/Object/COFFObjectFile.cpp | 375 | ||||
| -rw-r--r-- | lib/Object/ELFObjectFile.cpp | 686 | ||||
| -rw-r--r-- | lib/Object/MachOObject.cpp | 342 | ||||
| -rw-r--r-- | lib/Object/Makefile | 14 | ||||
| -rw-r--r-- | lib/Object/ObjectFile.cpp | 71 | 
6 files changed, 1494 insertions, 0 deletions
diff --git a/lib/Object/CMakeLists.txt b/lib/Object/CMakeLists.txt new file mode 100644 index 000000000000..6a6814fd37d9 --- /dev/null +++ b/lib/Object/CMakeLists.txt @@ -0,0 +1,6 @@ +add_llvm_library(LLVMObject +  MachOObject.cpp +  ObjectFile.cpp +  COFFObjectFile.cpp +  ELFObjectFile.cpp +  ) diff --git a/lib/Object/COFFObjectFile.cpp b/lib/Object/COFFObjectFile.cpp new file mode 100644 index 000000000000..cfee82a0b217 --- /dev/null +++ b/lib/Object/COFFObjectFile.cpp @@ -0,0 +1,375 @@ +//===- COFFObjectFile.cpp - COFF object file implementation -----*- C++ -*-===// +// +//                     The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file declares the COFFObjectFile class. +// +//===----------------------------------------------------------------------===// + +#include "llvm/ADT/StringSwitch.h" +#include "llvm/ADT/Triple.h" +#include "llvm/Object/ObjectFile.h" +#include "llvm/Support/COFF.h" +#include "llvm/Support/Endian.h" + +using namespace llvm; +using namespace object; + +namespace { +using support::ulittle8_t; +using support::ulittle16_t; +using support::ulittle32_t; +using support::little16_t; +} + +namespace { +struct coff_file_header { +  ulittle16_t Machine; +  ulittle16_t NumberOfSections; +  ulittle32_t TimeDateStamp; +  ulittle32_t PointerToSymbolTable; +  ulittle32_t NumberOfSymbols; +  ulittle16_t SizeOfOptionalHeader; +  ulittle16_t Characteristics; +}; +} + +extern char coff_file_header_layout_static_assert +            [sizeof(coff_file_header) == 20 ? 1 : -1]; + +namespace { +struct coff_symbol { +  struct StringTableOffset { +    ulittle32_t Zeroes; +    ulittle32_t Offset; +  }; + +  union { +    char ShortName[8]; +    StringTableOffset Offset; +  } Name; + +  ulittle32_t Value; +  little16_t SectionNumber; + +  struct { +    ulittle8_t BaseType; +    ulittle8_t ComplexType; +  } Type; + +  ulittle8_t  StorageClass; +  ulittle8_t  NumberOfAuxSymbols; +}; +} + +extern char coff_coff_symbol_layout_static_assert +            [sizeof(coff_symbol) == 18 ? 1 : -1]; + +namespace { +struct coff_section { +  char Name[8]; +  ulittle32_t VirtualSize; +  ulittle32_t VirtualAddress; +  ulittle32_t SizeOfRawData; +  ulittle32_t PointerToRawData; +  ulittle32_t PointerToRelocations; +  ulittle32_t PointerToLinenumbers; +  ulittle16_t NumberOfRelocations; +  ulittle16_t NumberOfLinenumbers; +  ulittle32_t Characteristics; +}; +} + +extern char coff_coff_section_layout_static_assert +            [sizeof(coff_section) == 40 ? 1 : -1]; + +namespace { +class COFFObjectFile : public ObjectFile { +private: +  const coff_file_header *Header; +  const coff_section     *SectionTable; +  const coff_symbol      *SymbolTable; +  const char             *StringTable; + +  const coff_section     *getSection(std::size_t index) const; +  const char             *getString(std::size_t offset) const; + +protected: +  virtual SymbolRef getSymbolNext(DataRefImpl Symb) const; +  virtual StringRef getSymbolName(DataRefImpl Symb) const; +  virtual uint64_t  getSymbolAddress(DataRefImpl Symb) const; +  virtual uint64_t  getSymbolSize(DataRefImpl Symb) const; +  virtual char      getSymbolNMTypeChar(DataRefImpl Symb) const; +  virtual bool      isSymbolInternal(DataRefImpl Symb) const; + +  virtual SectionRef getSectionNext(DataRefImpl Sec) const; +  virtual StringRef  getSectionName(DataRefImpl Sec) const; +  virtual uint64_t   getSectionAddress(DataRefImpl Sec) const; +  virtual uint64_t   getSectionSize(DataRefImpl Sec) const; +  virtual StringRef  getSectionContents(DataRefImpl Sec) const; +  virtual bool       isSectionText(DataRefImpl Sec) const; + +public: +  COFFObjectFile(MemoryBuffer *Object); +  virtual symbol_iterator begin_symbols() const; +  virtual symbol_iterator end_symbols() const; +  virtual section_iterator begin_sections() const; +  virtual section_iterator end_sections() const; + +  virtual uint8_t getBytesInAddress() const; +  virtual StringRef getFileFormatName() const; +  virtual unsigned getArch() const; +}; +} // end namespace + +SymbolRef COFFObjectFile::getSymbolNext(DataRefImpl Symb) const { +  const coff_symbol *symb = reinterpret_cast<const coff_symbol*>(Symb.p); +  symb += 1 + symb->NumberOfAuxSymbols; +  Symb.p = reinterpret_cast<intptr_t>(symb); +  return SymbolRef(Symb, this); +} + +StringRef COFFObjectFile::getSymbolName(DataRefImpl Symb) const { +  const coff_symbol *symb = reinterpret_cast<const coff_symbol*>(Symb.p); +  // Check for string table entry. First 4 bytes are 0. +  if (symb->Name.Offset.Zeroes == 0) { +    uint32_t Offset = symb->Name.Offset.Offset; +    return StringRef(getString(Offset)); +  } + +  if (symb->Name.ShortName[7] == 0) +    // Null terminated, let ::strlen figure out the length. +    return StringRef(symb->Name.ShortName); +  // Not null terminated, use all 8 bytes. +  return StringRef(symb->Name.ShortName, 8); +} + +uint64_t COFFObjectFile::getSymbolAddress(DataRefImpl Symb) const { +  const coff_symbol *symb = reinterpret_cast<const coff_symbol*>(Symb.p); +  const coff_section *Section = getSection(symb->SectionNumber); +  char Type = getSymbolNMTypeChar(Symb); +  if (Type == 'U' || Type == 'w') +    return UnknownAddressOrSize; +  if (Section) +    return Section->VirtualAddress + symb->Value; +  return symb->Value; +} + +uint64_t COFFObjectFile::getSymbolSize(DataRefImpl Symb) const { +  // FIXME: Return the correct size. This requires looking at all the symbols +  //        in the same section as this symbol, and looking for either the next +  //        symbol, or the end of the section. +  const coff_symbol *symb = reinterpret_cast<const coff_symbol*>(Symb.p); +  const coff_section *Section = getSection(symb->SectionNumber); +  char Type = getSymbolNMTypeChar(Symb); +  if (Type == 'U' || Type == 'w') +    return UnknownAddressOrSize; +  if (Section) +    return Section->SizeOfRawData - symb->Value; +  return 0; +} + +char COFFObjectFile::getSymbolNMTypeChar(DataRefImpl Symb) const { +  const coff_symbol *symb = reinterpret_cast<const coff_symbol*>(Symb.p); +  char ret = StringSwitch<char>(getSymbolName(Symb)) +    .StartsWith(".debug", 'N') +    .StartsWith(".sxdata", 'N') +    .Default('?'); + +  if (ret != '?') +    return ret; + +  uint32_t Characteristics = 0; +  uint32_t PointerToRawData = 0; +  const coff_section *Section = getSection(symb->SectionNumber); +  if (Section) { +    Characteristics = Section->Characteristics; +    PointerToRawData = Section->PointerToRawData; +  } + +  switch (symb->SectionNumber) { +  case COFF::IMAGE_SYM_UNDEFINED: +    // Check storage classes. +    if (symb->StorageClass == COFF::IMAGE_SYM_CLASS_WEAK_EXTERNAL) +      return 'w'; // Don't do ::toupper. +    else +      ret = 'u'; +    break; +  case COFF::IMAGE_SYM_ABSOLUTE: +    ret = 'a'; +    break; +  case COFF::IMAGE_SYM_DEBUG: +    ret = 'n'; +    break; +  default: +    // Check section type. +    if (Characteristics & COFF::IMAGE_SCN_CNT_CODE) +      ret = 't'; +    else if (  Characteristics & COFF::IMAGE_SCN_MEM_READ +            && ~Characteristics & COFF::IMAGE_SCN_MEM_WRITE) // Read only. +      ret = 'r'; +    else if (Characteristics & COFF::IMAGE_SCN_CNT_INITIALIZED_DATA) +      ret = 'd'; +    else if (Characteristics & COFF::IMAGE_SCN_CNT_UNINITIALIZED_DATA) +      ret = 'b'; +    else if (Characteristics & COFF::IMAGE_SCN_LNK_INFO) +      ret = 'i'; + +    // Check for section symbol. +    else if (  symb->StorageClass == COFF::IMAGE_SYM_CLASS_STATIC +            && symb->Value == 0) +       ret = 's'; +  } + +  if (symb->StorageClass == COFF::IMAGE_SYM_CLASS_EXTERNAL) +    ret = ::toupper(ret); + +  return ret; +} + +bool COFFObjectFile::isSymbolInternal(DataRefImpl Symb) const { +  return false; +} + +SectionRef COFFObjectFile::getSectionNext(DataRefImpl Sec) const { +  const coff_section *sec = reinterpret_cast<const coff_section*>(Sec.p); +  sec += 1; +  Sec.p = reinterpret_cast<intptr_t>(sec); +  return SectionRef(Sec, this); +} + +StringRef COFFObjectFile::getSectionName(DataRefImpl Sec) const { +  const coff_section *sec = reinterpret_cast<const coff_section*>(Sec.p); +  StringRef name; +  if (sec->Name[7] == 0) +    // Null terminated, let ::strlen figure out the length. +    name = sec->Name; +  else +    // Not null terminated, use all 8 bytes. +    name = StringRef(sec->Name, 8); + +  // Check for string table entry. First byte is '/'. +  if (name[0] == '/') { +    uint32_t Offset; +    name.getAsInteger(10, Offset); +    return StringRef(getString(Offset)); +  } + +  // It's just a normal name. +  return name; +} + +uint64_t COFFObjectFile::getSectionAddress(DataRefImpl Sec) const { +  const coff_section *sec = reinterpret_cast<const coff_section*>(Sec.p); +  return sec->VirtualAddress; +} + +uint64_t COFFObjectFile::getSectionSize(DataRefImpl Sec) const { +  const coff_section *sec = reinterpret_cast<const coff_section*>(Sec.p); +  return sec->SizeOfRawData; +} + +StringRef COFFObjectFile::getSectionContents(DataRefImpl Sec) const { +  const coff_section *sec = reinterpret_cast<const coff_section*>(Sec.p); +  return StringRef(reinterpret_cast<const char *>(base + sec->PointerToRawData), +                   sec->SizeOfRawData); +} + +bool COFFObjectFile::isSectionText(DataRefImpl Sec) const { +  const coff_section *sec = reinterpret_cast<const coff_section*>(Sec.p); +  return sec->Characteristics & COFF::IMAGE_SCN_CNT_CODE; +} + +COFFObjectFile::COFFObjectFile(MemoryBuffer *Object) +  : ObjectFile(Object) { +  Header = reinterpret_cast<const coff_file_header *>(base); +  SectionTable = +    reinterpret_cast<const coff_section *>( base +                                          + sizeof(coff_file_header) +                                          + Header->SizeOfOptionalHeader); +  SymbolTable = +    reinterpret_cast<const coff_symbol *>(base + Header->PointerToSymbolTable); + +  // Find string table. +  StringTable = reinterpret_cast<const char *>(base) +              + Header->PointerToSymbolTable +              + Header->NumberOfSymbols * 18; +} + +ObjectFile::symbol_iterator COFFObjectFile::begin_symbols() const { +  DataRefImpl ret; +  ret.p = reinterpret_cast<intptr_t>(SymbolTable); +  return symbol_iterator(SymbolRef(ret, this)); +} + +ObjectFile::symbol_iterator COFFObjectFile::end_symbols() const { +  // The symbol table ends where the string table begins. +  DataRefImpl ret; +  ret.p = reinterpret_cast<intptr_t>(StringTable); +  return symbol_iterator(SymbolRef(ret, this)); +} + +ObjectFile::section_iterator COFFObjectFile::begin_sections() const { +  DataRefImpl ret; +  ret.p = reinterpret_cast<intptr_t>(SectionTable); +  return section_iterator(SectionRef(ret, this)); +} + +ObjectFile::section_iterator COFFObjectFile::end_sections() const { +  DataRefImpl ret; +  ret.p = reinterpret_cast<intptr_t>(SectionTable + Header->NumberOfSections); +  return section_iterator(SectionRef(ret, this)); +} + +uint8_t COFFObjectFile::getBytesInAddress() const { +  return getArch() == Triple::x86_64 ? 8 : 4; +} + +StringRef COFFObjectFile::getFileFormatName() const { +  switch(Header->Machine) { +  case COFF::IMAGE_FILE_MACHINE_I386: +    return "COFF-i386"; +  case COFF::IMAGE_FILE_MACHINE_AMD64: +    return "COFF-x86-64"; +  default: +    return "COFF-<unknown arch>"; +  } +} + +unsigned COFFObjectFile::getArch() const { +  switch(Header->Machine) { +  case COFF::IMAGE_FILE_MACHINE_I386: +    return Triple::x86; +  case COFF::IMAGE_FILE_MACHINE_AMD64: +    return Triple::x86_64; +  default: +    return Triple::UnknownArch; +  } +} + +const coff_section *COFFObjectFile::getSection(std::size_t index) const { +  if (index > 0 && index <= Header->NumberOfSections) +    return SectionTable + (index - 1); +  return 0; +} + +const char *COFFObjectFile::getString(std::size_t offset) const { +  const ulittle32_t *StringTableSize = +    reinterpret_cast<const ulittle32_t *>(StringTable); +  if (offset < *StringTableSize) +    return StringTable + offset; +  return 0; +} + +namespace llvm { + +  ObjectFile *ObjectFile::createCOFFObjectFile(MemoryBuffer *Object) { +    return new COFFObjectFile(Object); +  } + +} // end namespace llvm diff --git a/lib/Object/ELFObjectFile.cpp b/lib/Object/ELFObjectFile.cpp new file mode 100644 index 000000000000..682be770f48f --- /dev/null +++ b/lib/Object/ELFObjectFile.cpp @@ -0,0 +1,686 @@ +//===- ELFObjectFile.cpp - ELF object file implementation -------*- C++ -*-===// +// +//                     The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines the ELFObjectFile class. +// +//===----------------------------------------------------------------------===// + +#include "llvm/ADT/SmallVector.h" +#include "llvm/ADT/StringSwitch.h" +#include "llvm/ADT/Triple.h" +#include "llvm/Object/ObjectFile.h" +#include "llvm/Support/ELF.h" +#include "llvm/Support/Endian.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/MemoryBuffer.h" +#include <limits> +#include <utility> + +using namespace llvm; +using namespace object; + +// Templates to choose Elf_Addr and Elf_Off depending on is64Bits. +namespace { +template<support::endianness target_endianness> +struct ELFDataTypeTypedefHelperCommon { +  typedef support::detail::packed_endian_specific_integral +    <uint16_t, target_endianness, support::aligned> Elf_Half; +  typedef support::detail::packed_endian_specific_integral +    <uint32_t, target_endianness, support::aligned> Elf_Word; +  typedef support::detail::packed_endian_specific_integral +    <int32_t, target_endianness, support::aligned> Elf_Sword; +  typedef support::detail::packed_endian_specific_integral +    <uint64_t, target_endianness, support::aligned> Elf_Xword; +  typedef support::detail::packed_endian_specific_integral +    <int64_t, target_endianness, support::aligned> Elf_Sxword; +}; +} + +namespace { +template<support::endianness target_endianness, bool is64Bits> +struct ELFDataTypeTypedefHelper; + +/// ELF 32bit types. +template<support::endianness target_endianness> +struct ELFDataTypeTypedefHelper<target_endianness, false> +  : ELFDataTypeTypedefHelperCommon<target_endianness> { +  typedef support::detail::packed_endian_specific_integral +    <uint32_t, target_endianness, support::aligned> Elf_Addr; +  typedef support::detail::packed_endian_specific_integral +    <uint32_t, target_endianness, support::aligned> Elf_Off; +}; + +/// ELF 64bit types. +template<support::endianness target_endianness> +struct ELFDataTypeTypedefHelper<target_endianness, true> +  : ELFDataTypeTypedefHelperCommon<target_endianness>{ +  typedef support::detail::packed_endian_specific_integral +    <uint64_t, target_endianness, support::aligned> Elf_Addr; +  typedef support::detail::packed_endian_specific_integral +    <uint64_t, target_endianness, support::aligned> Elf_Off; +}; +} + +// I really don't like doing this, but the alternative is copypasta. +#define LLVM_ELF_IMPORT_TYPES(target_endianness, is64Bits) \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Addr Elf_Addr; \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Off Elf_Off; \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Half Elf_Half; \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Word Elf_Word; \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Sword Elf_Sword; \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Xword Elf_Xword; \ +typedef typename \ +  ELFDataTypeTypedefHelper<target_endianness, is64Bits>::Elf_Sxword Elf_Sxword; + +  // Section header. +namespace { +template<support::endianness target_endianness, bool is64Bits> +struct Elf_Shdr_Base; + +template<support::endianness target_endianness> +struct Elf_Shdr_Base<target_endianness, false> { +  LLVM_ELF_IMPORT_TYPES(target_endianness, false) +  Elf_Word sh_name;     // Section name (index into string table) +  Elf_Word sh_type;     // Section type (SHT_*) +  Elf_Word sh_flags;    // Section flags (SHF_*) +  Elf_Addr sh_addr;     // Address where section is to be loaded +  Elf_Off  sh_offset;   // File offset of section data, in bytes +  Elf_Word sh_size;     // Size of section, in bytes +  Elf_Word sh_link;     // Section type-specific header table index link +  Elf_Word sh_info;     // Section type-specific extra information +  Elf_Word sh_addralign;// Section address alignment +  Elf_Word sh_entsize;  // Size of records contained within the section +}; + +template<support::endianness target_endianness> +struct Elf_Shdr_Base<target_endianness, true> { +  LLVM_ELF_IMPORT_TYPES(target_endianness, true) +  Elf_Word  sh_name;     // Section name (index into string table) +  Elf_Word  sh_type;     // Section type (SHT_*) +  Elf_Xword sh_flags;    // Section flags (SHF_*) +  Elf_Addr  sh_addr;     // Address where section is to be loaded +  Elf_Off   sh_offset;   // File offset of section data, in bytes +  Elf_Xword sh_size;     // Size of section, in bytes +  Elf_Word  sh_link;     // Section type-specific header table index link +  Elf_Word  sh_info;     // Section type-specific extra information +  Elf_Xword sh_addralign;// Section address alignment +  Elf_Xword sh_entsize;  // Size of records contained within the section +}; + +template<support::endianness target_endianness, bool is64Bits> +struct Elf_Shdr_Impl : Elf_Shdr_Base<target_endianness, is64Bits> { +  using Elf_Shdr_Base<target_endianness, is64Bits>::sh_entsize; +  using Elf_Shdr_Base<target_endianness, is64Bits>::sh_size; + +  /// @brief Get the number of entities this section contains if it has any. +  unsigned getEntityCount() const { +    if (sh_entsize == 0) +      return 0; +    return sh_size / sh_entsize; +  } +}; +} + +namespace { +template<support::endianness target_endianness, bool is64Bits> +struct Elf_Sym_Base; + +template<support::endianness target_endianness> +struct Elf_Sym_Base<target_endianness, false> { +  LLVM_ELF_IMPORT_TYPES(target_endianness, false) +  Elf_Word      st_name;  // Symbol name (index into string table) +  Elf_Addr      st_value; // Value or address associated with the symbol +  Elf_Word      st_size;  // Size of the symbol +  unsigned char st_info;  // Symbol's type and binding attributes +  unsigned char st_other; // Must be zero; reserved +  Elf_Half      st_shndx; // Which section (header table index) it's defined in +}; + +template<support::endianness target_endianness> +struct Elf_Sym_Base<target_endianness, true> { +  LLVM_ELF_IMPORT_TYPES(target_endianness, true) +  Elf_Word      st_name;  // Symbol name (index into string table) +  unsigned char st_info;  // Symbol's type and binding attributes +  unsigned char st_other; // Must be zero; reserved +  Elf_Half      st_shndx; // Which section (header table index) it's defined in +  Elf_Addr      st_value; // Value or address associated with the symbol +  Elf_Xword     st_size;  // Size of the symbol +}; + +template<support::endianness target_endianness, bool is64Bits> +struct Elf_Sym_Impl : Elf_Sym_Base<target_endianness, is64Bits> { +  using Elf_Sym_Base<target_endianness, is64Bits>::st_info; + +  // These accessors and mutators correspond to the ELF32_ST_BIND, +  // ELF32_ST_TYPE, and ELF32_ST_INFO macros defined in the ELF specification: +  unsigned char getBinding() const { return st_info >> 4; } +  unsigned char getType() const { return st_info & 0x0f; } +  void setBinding(unsigned char b) { setBindingAndType(b, getType()); } +  void setType(unsigned char t) { setBindingAndType(getBinding(), t); } +  void setBindingAndType(unsigned char b, unsigned char t) { +    st_info = (b << 4) + (t & 0x0f); +  } +}; +} + +namespace { +template<support::endianness target_endianness, bool is64Bits> +class ELFObjectFile : public ObjectFile { +  LLVM_ELF_IMPORT_TYPES(target_endianness, is64Bits) + +  typedef Elf_Shdr_Impl<target_endianness, is64Bits> Elf_Shdr; +  typedef Elf_Sym_Impl<target_endianness, is64Bits> Elf_Sym; + +  struct Elf_Ehdr { +    unsigned char e_ident[ELF::EI_NIDENT]; // ELF Identification bytes +    Elf_Half e_type;     // Type of file (see ET_*) +    Elf_Half e_machine;  // Required architecture for this file (see EM_*) +    Elf_Word e_version;  // Must be equal to 1 +    Elf_Addr e_entry;    // Address to jump to in order to start program +    Elf_Off  e_phoff;    // Program header table's file offset, in bytes +    Elf_Off  e_shoff;    // Section header table's file offset, in bytes +    Elf_Word e_flags;    // Processor-specific flags +    Elf_Half e_ehsize;   // Size of ELF header, in bytes +    Elf_Half e_phentsize;// Size of an entry in the program header table +    Elf_Half e_phnum;    // Number of entries in the program header table +    Elf_Half e_shentsize;// Size of an entry in the section header table +    Elf_Half e_shnum;    // Number of entries in the section header table +    Elf_Half e_shstrndx; // Section header table index of section name +                                  // string table +    bool checkMagic() const { +      return (memcmp(e_ident, ELF::ElfMagic, strlen(ELF::ElfMagic))) == 0; +    } +    unsigned char getFileClass() const { return e_ident[ELF::EI_CLASS]; } +    unsigned char getDataEncoding() const { return e_ident[ELF::EI_DATA]; } +  }; + +  typedef SmallVector<const Elf_Shdr*, 1> SymbolTableSections_t; + +  const Elf_Ehdr *Header; +  const Elf_Shdr *SectionHeaderTable; +  const Elf_Shdr *dot_shstrtab_sec; // Section header string table. +  const Elf_Shdr *dot_strtab_sec;   // Symbol header string table. +  SymbolTableSections_t SymbolTableSections; + +  void            validateSymbol(DataRefImpl Symb) const; +  const Elf_Sym  *getSymbol(DataRefImpl Symb) const; +  const Elf_Shdr *getSection(DataRefImpl index) const; +  const Elf_Shdr *getSection(uint16_t index) const; +  const char     *getString(uint16_t section, uint32_t offset) const; +  const char     *getString(const Elf_Shdr *section, uint32_t offset) const; + +protected: +  virtual SymbolRef getSymbolNext(DataRefImpl Symb) const; +  virtual StringRef getSymbolName(DataRefImpl Symb) const; +  virtual uint64_t  getSymbolAddress(DataRefImpl Symb) const; +  virtual uint64_t  getSymbolSize(DataRefImpl Symb) const; +  virtual char      getSymbolNMTypeChar(DataRefImpl Symb) const; +  virtual bool      isSymbolInternal(DataRefImpl Symb) const; + +  virtual SectionRef getSectionNext(DataRefImpl Sec) const; +  virtual StringRef  getSectionName(DataRefImpl Sec) const; +  virtual uint64_t   getSectionAddress(DataRefImpl Sec) const; +  virtual uint64_t   getSectionSize(DataRefImpl Sec) const; +  virtual StringRef  getSectionContents(DataRefImpl Sec) const; +  virtual bool       isSectionText(DataRefImpl Sec) const; + +public: +  ELFObjectFile(MemoryBuffer *Object); +  virtual symbol_iterator begin_symbols() const; +  virtual symbol_iterator end_symbols() const; +  virtual section_iterator begin_sections() const; +  virtual section_iterator end_sections() const; + +  virtual uint8_t getBytesInAddress() const; +  virtual StringRef getFileFormatName() const; +  virtual unsigned getArch() const; +}; +} // end namespace + +template<support::endianness target_endianness, bool is64Bits> +void ELFObjectFile<target_endianness, is64Bits> +                  ::validateSymbol(DataRefImpl Symb) const { +  const Elf_Sym  *symb = getSymbol(Symb); +  const Elf_Shdr *SymbolTableSection = SymbolTableSections[Symb.d.b]; +  // FIXME: We really need to do proper error handling in the case of an invalid +  //        input file. Because we don't use exceptions, I think we'll just pass +  //        an error object around. +  if (!(  symb +        && SymbolTableSection +        && symb >= (const Elf_Sym*)(base +                   + SymbolTableSection->sh_offset) +        && symb <  (const Elf_Sym*)(base +                   + SymbolTableSection->sh_offset +                   + SymbolTableSection->sh_size))) +    // FIXME: Proper error handling. +    report_fatal_error("Symb must point to a valid symbol!"); +} + +template<support::endianness target_endianness, bool is64Bits> +SymbolRef ELFObjectFile<target_endianness, is64Bits> +                       ::getSymbolNext(DataRefImpl Symb) const { +  validateSymbol(Symb); +  const Elf_Shdr *SymbolTableSection = SymbolTableSections[Symb.d.b]; + +  ++Symb.d.a; +  // Check to see if we are at the end of this symbol table. +  if (Symb.d.a >= SymbolTableSection->getEntityCount()) { +    // We are at the end. If there are other symbol tables, jump to them. +    ++Symb.d.b; +    Symb.d.a = 1; // The 0th symbol in ELF is fake. +    // Otherwise return the terminator. +    if (Symb.d.b >= SymbolTableSections.size()) { +      Symb.d.a = std::numeric_limits<uint32_t>::max(); +      Symb.d.b = std::numeric_limits<uint32_t>::max(); +    } +  } + +  return SymbolRef(Symb, this); +} + +template<support::endianness target_endianness, bool is64Bits> +StringRef ELFObjectFile<target_endianness, is64Bits> +                       ::getSymbolName(DataRefImpl Symb) const { +  validateSymbol(Symb); +  const Elf_Sym  *symb = getSymbol(Symb); +  if (symb->st_name == 0) { +    const Elf_Shdr *section = getSection(symb->st_shndx); +    if (!section) +      return ""; +    return getString(dot_shstrtab_sec, section->sh_name); +  } + +  // Use the default symbol table name section. +  return getString(dot_strtab_sec, symb->st_name); +} + +template<support::endianness target_endianness, bool is64Bits> +uint64_t ELFObjectFile<target_endianness, is64Bits> +                      ::getSymbolAddress(DataRefImpl Symb) const { +  validateSymbol(Symb); +  const Elf_Sym  *symb = getSymbol(Symb); +  const Elf_Shdr *Section; +  switch (symb->st_shndx) { +  case ELF::SHN_COMMON: +   // Undefined symbols have no address yet. +  case ELF::SHN_UNDEF: return UnknownAddressOrSize; +  case ELF::SHN_ABS: return symb->st_value; +  default: Section = getSection(symb->st_shndx); +  } + +  switch (symb->getType()) { +  case ELF::STT_SECTION: return Section ? Section->sh_addr +                                        : UnknownAddressOrSize; +  case ELF::STT_FUNC: +  case ELF::STT_OBJECT: +  case ELF::STT_NOTYPE: +    return symb->st_value; +  default: return UnknownAddressOrSize; +  } +} + +template<support::endianness target_endianness, bool is64Bits> +uint64_t ELFObjectFile<target_endianness, is64Bits> +                      ::getSymbolSize(DataRefImpl Symb) const { +  validateSymbol(Symb); +  const Elf_Sym  *symb = getSymbol(Symb); +  if (symb->st_size == 0) +    return UnknownAddressOrSize; +  return symb->st_size; +} + +template<support::endianness target_endianness, bool is64Bits> +char ELFObjectFile<target_endianness, is64Bits> +                  ::getSymbolNMTypeChar(DataRefImpl Symb) const { +  validateSymbol(Symb); +  const Elf_Sym  *symb = getSymbol(Symb); +  const Elf_Shdr *Section = getSection(symb->st_shndx); + +  char ret = '?'; + +  if (Section) { +    switch (Section->sh_type) { +    case ELF::SHT_PROGBITS: +    case ELF::SHT_DYNAMIC: +      switch (Section->sh_flags) { +      case (ELF::SHF_ALLOC | ELF::SHF_EXECINSTR): +        ret = 't'; break; +      case (ELF::SHF_ALLOC | ELF::SHF_WRITE): +        ret = 'd'; break; +      case ELF::SHF_ALLOC: +      case (ELF::SHF_ALLOC | ELF::SHF_MERGE): +      case (ELF::SHF_ALLOC | ELF::SHF_MERGE | ELF::SHF_STRINGS): +        ret = 'r'; break; +      } +      break; +    case ELF::SHT_NOBITS: ret = 'b'; +    } +  } + +  switch (symb->st_shndx) { +  case ELF::SHN_UNDEF: +    if (ret == '?') +      ret = 'U'; +    break; +  case ELF::SHN_ABS: ret = 'a'; break; +  case ELF::SHN_COMMON: ret = 'c'; break; +  } + +  switch (symb->getBinding()) { +  case ELF::STB_GLOBAL: ret = ::toupper(ret); break; +  case ELF::STB_WEAK: +    if (symb->st_shndx == ELF::SHN_UNDEF) +      ret = 'w'; +    else +      if (symb->getType() == ELF::STT_OBJECT) +        ret = 'V'; +      else +        ret = 'W'; +  } + +  if (ret == '?' && symb->getType() == ELF::STT_SECTION) +    return StringSwitch<char>(getSymbolName(Symb)) +      .StartsWith(".debug", 'N') +      .StartsWith(".note", 'n'); + +  return ret; +} + +template<support::endianness target_endianness, bool is64Bits> +bool ELFObjectFile<target_endianness, is64Bits> +                  ::isSymbolInternal(DataRefImpl Symb) const { +  validateSymbol(Symb); +  const Elf_Sym  *symb = getSymbol(Symb); + +  if (  symb->getType() == ELF::STT_FILE +     || symb->getType() == ELF::STT_SECTION) +    return true; +  return false; +} + +template<support::endianness target_endianness, bool is64Bits> +SectionRef ELFObjectFile<target_endianness, is64Bits> +                        ::getSectionNext(DataRefImpl Sec) const { +  const uint8_t *sec = reinterpret_cast<const uint8_t *>(Sec.p); +  sec += Header->e_shentsize; +  Sec.p = reinterpret_cast<intptr_t>(sec); +  return SectionRef(Sec, this); +} + +template<support::endianness target_endianness, bool is64Bits> +StringRef ELFObjectFile<target_endianness, is64Bits> +                       ::getSectionName(DataRefImpl Sec) const { +  const Elf_Shdr *sec = reinterpret_cast<const Elf_Shdr *>(Sec.p); +  return StringRef(getString(dot_shstrtab_sec, sec->sh_name)); +} + +template<support::endianness target_endianness, bool is64Bits> +uint64_t ELFObjectFile<target_endianness, is64Bits> +                      ::getSectionAddress(DataRefImpl Sec) const { +  const Elf_Shdr *sec = reinterpret_cast<const Elf_Shdr *>(Sec.p); +  return sec->sh_addr; +} + +template<support::endianness target_endianness, bool is64Bits> +uint64_t ELFObjectFile<target_endianness, is64Bits> +                      ::getSectionSize(DataRefImpl Sec) const { +  const Elf_Shdr *sec = reinterpret_cast<const Elf_Shdr *>(Sec.p); +  return sec->sh_size; +} + +template<support::endianness target_endianness, bool is64Bits> +StringRef ELFObjectFile<target_endianness, is64Bits> +                       ::getSectionContents(DataRefImpl Sec) const { +  const Elf_Shdr *sec = reinterpret_cast<const Elf_Shdr *>(Sec.p); +  const char *start = (char*)base + sec->sh_offset; +  return StringRef(start, sec->sh_size); +} + +template<support::endianness target_endianness, bool is64Bits> +bool ELFObjectFile<target_endianness, is64Bits> +                  ::isSectionText(DataRefImpl Sec) const { +  const Elf_Shdr *sec = reinterpret_cast<const Elf_Shdr *>(Sec.p); +  if (sec->sh_flags & ELF::SHF_EXECINSTR) +    return true; +  return false; +} + +template<support::endianness target_endianness, bool is64Bits> +ELFObjectFile<target_endianness, is64Bits>::ELFObjectFile(MemoryBuffer *Object) +  : ObjectFile(Object) +  , SectionHeaderTable(0) +  , dot_shstrtab_sec(0) +  , dot_strtab_sec(0) { +  Header = reinterpret_cast<const Elf_Ehdr *>(base); + +  if (Header->e_shoff == 0) +    return; + +  SectionHeaderTable = +    reinterpret_cast<const Elf_Shdr *>(base + Header->e_shoff); +  uint32_t SectionTableSize = Header->e_shnum * Header->e_shentsize; +  if (!(  (const uint8_t *)SectionHeaderTable + SectionTableSize +         <= base + MapFile->getBufferSize())) +    // FIXME: Proper error handling. +    report_fatal_error("Section table goes past end of file!"); + + +  // To find the symbol tables we walk the section table to find SHT_STMTAB. +  for (const char *i = reinterpret_cast<const char *>(SectionHeaderTable), +                  *e = i + Header->e_shnum * Header->e_shentsize; +                   i != e; i += Header->e_shentsize) { +    const Elf_Shdr *sh = reinterpret_cast<const Elf_Shdr*>(i); +    if (sh->sh_type == ELF::SHT_SYMTAB) { +      SymbolTableSections.push_back(sh); +    } +  } + +  // Get string table sections. +  dot_shstrtab_sec = getSection(Header->e_shstrndx); +  if (dot_shstrtab_sec) { +    // Verify that the last byte in the string table in a null. +    if (((const char*)base + dot_shstrtab_sec->sh_offset) +        [dot_shstrtab_sec->sh_size - 1] != 0) +      // FIXME: Proper error handling. +      report_fatal_error("String table must end with a null terminator!"); +  } + +  // Merge this into the above loop. +  for (const char *i = reinterpret_cast<const char *>(SectionHeaderTable), +                  *e = i + Header->e_shnum * Header->e_shentsize; +                   i != e; i += Header->e_shentsize) { +    const Elf_Shdr *sh = reinterpret_cast<const Elf_Shdr*>(i); +    if (sh->sh_type == ELF::SHT_STRTAB) { +      StringRef SectionName(getString(dot_shstrtab_sec, sh->sh_name)); +      if (SectionName == ".strtab") { +        if (dot_strtab_sec != 0) +          // FIXME: Proper error handling. +          report_fatal_error("Already found section named .strtab!"); +        dot_strtab_sec = sh; +        const char *dot_strtab = (const char*)base + sh->sh_offset; +          if (dot_strtab[sh->sh_size - 1] != 0) +            // FIXME: Proper error handling. +            report_fatal_error("String table must end with a null terminator!"); +      } +    } +  } +} + +template<support::endianness target_endianness, bool is64Bits> +ObjectFile::symbol_iterator ELFObjectFile<target_endianness, is64Bits> +                                         ::begin_symbols() const { +  DataRefImpl SymbolData; +  memset(&SymbolData, 0, sizeof(SymbolData)); +  if (SymbolTableSections.size() == 0) { +    SymbolData.d.a = std::numeric_limits<uint32_t>::max(); +    SymbolData.d.b = std::numeric_limits<uint32_t>::max(); +  } else { +    SymbolData.d.a = 1; // The 0th symbol in ELF is fake. +    SymbolData.d.b = 0; +  } +  return symbol_iterator(SymbolRef(SymbolData, this)); +} + +template<support::endianness target_endianness, bool is64Bits> +ObjectFile::symbol_iterator ELFObjectFile<target_endianness, is64Bits> +                                         ::end_symbols() const { +  DataRefImpl SymbolData; +  memset(&SymbolData, 0, sizeof(SymbolData)); +  SymbolData.d.a = std::numeric_limits<uint32_t>::max(); +  SymbolData.d.b = std::numeric_limits<uint32_t>::max(); +  return symbol_iterator(SymbolRef(SymbolData, this)); +} + +template<support::endianness target_endianness, bool is64Bits> +ObjectFile::section_iterator ELFObjectFile<target_endianness, is64Bits> +                                          ::begin_sections() const { +  DataRefImpl ret; +  ret.p = reinterpret_cast<intptr_t>(base + Header->e_shoff); +  return section_iterator(SectionRef(ret, this)); +} + +template<support::endianness target_endianness, bool is64Bits> +ObjectFile::section_iterator ELFObjectFile<target_endianness, is64Bits> +                                          ::end_sections() const { +  DataRefImpl ret; +  ret.p = reinterpret_cast<intptr_t>(base +                                     + Header->e_shoff +                                     + (Header->e_shentsize * Header->e_shnum)); +  return section_iterator(SectionRef(ret, this)); +} + +template<support::endianness target_endianness, bool is64Bits> +uint8_t ELFObjectFile<target_endianness, is64Bits>::getBytesInAddress() const { +  return is64Bits ? 8 : 4; +} + +template<support::endianness target_endianness, bool is64Bits> +StringRef ELFObjectFile<target_endianness, is64Bits> +                       ::getFileFormatName() const { +  switch(Header->e_ident[ELF::EI_CLASS]) { +  case ELF::ELFCLASS32: +    switch(Header->e_machine) { +    case ELF::EM_386: +      return "ELF32-i386"; +    case ELF::EM_X86_64: +      return "ELF32-x86-64"; +    default: +      return "ELF32-unknown"; +    } +  case ELF::ELFCLASS64: +    switch(Header->e_machine) { +    case ELF::EM_386: +      return "ELF64-i386"; +    case ELF::EM_X86_64: +      return "ELF64-x86-64"; +    default: +      return "ELF64-unknown"; +    } +  default: +    // FIXME: Proper error handling. +    report_fatal_error("Invalid ELFCLASS!"); +  } +} + +template<support::endianness target_endianness, bool is64Bits> +unsigned ELFObjectFile<target_endianness, is64Bits>::getArch() const { +  switch(Header->e_machine) { +  case ELF::EM_386: +    return Triple::x86; +  case ELF::EM_X86_64: +    return Triple::x86_64; +  default: +    return Triple::UnknownArch; +  } +} + +template<support::endianness target_endianness, bool is64Bits> +const typename ELFObjectFile<target_endianness, is64Bits>::Elf_Sym * +ELFObjectFile<target_endianness, is64Bits>::getSymbol(DataRefImpl Symb) const { +  const Elf_Shdr *sec = SymbolTableSections[Symb.d.b]; +  return reinterpret_cast<const Elf_Sym *>( +           base +           + sec->sh_offset +           + (Symb.d.a * sec->sh_entsize)); +} + +template<support::endianness target_endianness, bool is64Bits> +const typename ELFObjectFile<target_endianness, is64Bits>::Elf_Shdr * +ELFObjectFile<target_endianness, is64Bits>::getSection(DataRefImpl Symb) const { +  const Elf_Shdr *sec = getSection(Symb.d.b); +  if (sec->sh_type != ELF::SHT_SYMTAB) +    // FIXME: Proper error handling. +    report_fatal_error("Invalid symbol table section!"); +  return sec; +} + +template<support::endianness target_endianness, bool is64Bits> +const typename ELFObjectFile<target_endianness, is64Bits>::Elf_Shdr * +ELFObjectFile<target_endianness, is64Bits>::getSection(uint16_t index) const { +  if (index == 0 || index >= ELF::SHN_LORESERVE) +    return 0; +  if (!SectionHeaderTable || index >= Header->e_shnum) +    // FIXME: Proper error handling. +    report_fatal_error("Invalid section index!"); + +  return reinterpret_cast<const Elf_Shdr *>( +         reinterpret_cast<const char *>(SectionHeaderTable) +         + (index * Header->e_shentsize)); +} + +template<support::endianness target_endianness, bool is64Bits> +const char *ELFObjectFile<target_endianness, is64Bits> +                         ::getString(uint16_t section, +                                     ELF::Elf32_Word offset) const { +  return getString(getSection(section), offset); +} + +template<support::endianness target_endianness, bool is64Bits> +const char *ELFObjectFile<target_endianness, is64Bits> +                         ::getString(const Elf_Shdr *section, +                                     ELF::Elf32_Word offset) const { +  assert(section && section->sh_type == ELF::SHT_STRTAB && "Invalid section!"); +  if (offset >= section->sh_size) +    // FIXME: Proper error handling. +    report_fatal_error("Sybol name offset outside of string table!"); +  return (const char *)base + section->sh_offset + offset; +} + +// EI_CLASS, EI_DATA. +static std::pair<unsigned char, unsigned char> +getElfArchType(MemoryBuffer *Object) { +  if (Object->getBufferSize() < ELF::EI_NIDENT) +    return std::make_pair((uint8_t)ELF::ELFCLASSNONE,(uint8_t)ELF::ELFDATANONE); +  return std::make_pair( (uint8_t)Object->getBufferStart()[ELF::EI_CLASS] +                       , (uint8_t)Object->getBufferStart()[ELF::EI_DATA]); +} + +namespace llvm { + +  ObjectFile *ObjectFile::createELFObjectFile(MemoryBuffer *Object) { +    std::pair<unsigned char, unsigned char> Ident = getElfArchType(Object); +    if (Ident.first == ELF::ELFCLASS32 && Ident.second == ELF::ELFDATA2LSB) +      return new ELFObjectFile<support::little, false>(Object); +    else if (Ident.first == ELF::ELFCLASS32 && Ident.second == ELF::ELFDATA2MSB) +      return new ELFObjectFile<support::big, false>(Object); +    else if (Ident.first == ELF::ELFCLASS64 && Ident.second == ELF::ELFDATA2LSB) +      return new ELFObjectFile<support::little, true>(Object); +    else if (Ident.first == ELF::ELFCLASS64 && Ident.second == ELF::ELFDATA2MSB) +      return new ELFObjectFile<support::big, true>(Object); +    // FIXME: Proper error handling. +    report_fatal_error("Not an ELF object file!"); +  } + +} // end namespace llvm diff --git a/lib/Object/MachOObject.cpp b/lib/Object/MachOObject.cpp new file mode 100644 index 000000000000..5e64d6323288 --- /dev/null +++ b/lib/Object/MachOObject.cpp @@ -0,0 +1,342 @@ +//===- MachOObject.cpp - Mach-O Object File Wrapper -----------------------===// +// +//                     The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "llvm/Object/MachOObject.h" +#include "llvm/ADT/StringRef.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Host.h" +#include "llvm/Support/SwapByteOrder.h" + +using namespace llvm; +using namespace llvm::object; + +/* Translation Utilities */ + +template<typename T> +static void SwapValue(T &Value) { +  Value = sys::SwapByteOrder(Value); +} + +template<typename T> +static void SwapStruct(T &Value); + +template<typename T> +static void ReadInMemoryStruct(const MachOObject &MOO, +                               StringRef Buffer, uint64_t Base, +                               InMemoryStruct<T> &Res) { +  typedef T struct_type; +  uint64_t Size = sizeof(struct_type); + +  // Check that the buffer contains the expected data. +  if (Base + Size >  Buffer.size()) { +    Res = 0; +    return; +  } + +  // Check whether we can return a direct pointer. +  struct_type *Ptr = (struct_type *) (Buffer.data() + Base); +  if (!MOO.isSwappedEndian()) { +    Res = Ptr; +    return; +  } + +  // Otherwise, copy the struct and translate the values. +  Res = *Ptr; +  SwapStruct(*Res); +} + +/* *** */ + +MachOObject::MachOObject(MemoryBuffer *Buffer_, bool IsLittleEndian_, +                         bool Is64Bit_) +  : Buffer(Buffer_), IsLittleEndian(IsLittleEndian_), Is64Bit(Is64Bit_), +    IsSwappedEndian(IsLittleEndian != sys::isLittleEndianHost()), +    HasStringTable(false), LoadCommands(0), NumLoadedCommands(0) { +  // Load the common header. +  memcpy(&Header, Buffer->getBuffer().data(), sizeof(Header)); +  if (IsSwappedEndian) { +    SwapValue(Header.Magic); +    SwapValue(Header.CPUType); +    SwapValue(Header.CPUSubtype); +    SwapValue(Header.FileType); +    SwapValue(Header.NumLoadCommands); +    SwapValue(Header.SizeOfLoadCommands); +    SwapValue(Header.Flags); +  } + +  if (is64Bit()) { +    memcpy(&Header64Ext, Buffer->getBuffer().data() + sizeof(Header), +           sizeof(Header64Ext)); +    if (IsSwappedEndian) { +      SwapValue(Header64Ext.Reserved); +    } +  } + +  // Create the load command array if sane. +  if (getHeader().NumLoadCommands < (1 << 20)) +    LoadCommands = new LoadCommandInfo[getHeader().NumLoadCommands]; +} + +MachOObject::~MachOObject() { +  delete [] LoadCommands; +} + +MachOObject *MachOObject::LoadFromBuffer(MemoryBuffer *Buffer, +                                         std::string *ErrorStr) { +  // First, check the magic value and initialize the basic object info. +  bool IsLittleEndian = false, Is64Bit = false; +  StringRef Magic = Buffer->getBuffer().slice(0, 4); +  if (Magic == "\xFE\xED\xFA\xCE") { +  }  else if (Magic == "\xCE\xFA\xED\xFE") { +    IsLittleEndian = true; +  } else if (Magic == "\xFE\xED\xFA\xCF") { +    Is64Bit = true; +  } else if (Magic == "\xCF\xFA\xED\xFE") { +    IsLittleEndian = true; +    Is64Bit = true; +  } else { +    if (ErrorStr) *ErrorStr = "not a Mach object file (invalid magic)"; +    return 0; +  } + +  // Ensure that the at least the full header is present. +  unsigned HeaderSize = Is64Bit ? macho::Header64Size : macho::Header32Size; +  if (Buffer->getBufferSize() < HeaderSize) { +    if (ErrorStr) *ErrorStr = "not a Mach object file (invalid header)"; +    return 0; +  } + +  OwningPtr<MachOObject> Object(new MachOObject(Buffer, IsLittleEndian, +                                                Is64Bit)); + +  // Check for bogus number of load commands. +  if (Object->getHeader().NumLoadCommands >= (1 << 20)) { +    if (ErrorStr) *ErrorStr = "not a Mach object file (unreasonable header)"; +    return 0; +  } + +  if (ErrorStr) *ErrorStr = ""; +  return Object.take(); +} + +StringRef MachOObject::getData(size_t Offset, size_t Size) const { +  return Buffer->getBuffer().substr(Offset,Size); +} + +void MachOObject::RegisterStringTable(macho::SymtabLoadCommand &SLC) { +  HasStringTable = true; +  StringTable = Buffer->getBuffer().substr(SLC.StringTableOffset, +                                           SLC.StringTableSize); +} + +const MachOObject::LoadCommandInfo & +MachOObject::getLoadCommandInfo(unsigned Index) const { +  assert(Index < getHeader().NumLoadCommands && "Invalid index!"); + +  // Load the command, if necessary. +  if (Index >= NumLoadedCommands) { +    uint64_t Offset; +    if (Index == 0) { +      Offset = getHeaderSize(); +    } else { +      const LoadCommandInfo &Prev = getLoadCommandInfo(Index - 1); +      Offset = Prev.Offset + Prev.Command.Size; +    } + +    LoadCommandInfo &Info = LoadCommands[Index]; +    memcpy(&Info.Command, Buffer->getBuffer().data() + Offset, +           sizeof(macho::LoadCommand)); +    if (IsSwappedEndian) { +      SwapValue(Info.Command.Type); +      SwapValue(Info.Command.Size); +    } +    Info.Offset = Offset; +    NumLoadedCommands = Index + 1; +  } + +  return LoadCommands[Index]; +} + +template<> +void SwapStruct(macho::SegmentLoadCommand &Value) { +  SwapValue(Value.Type); +  SwapValue(Value.Size); +  SwapValue(Value.VMAddress); +  SwapValue(Value.VMSize); +  SwapValue(Value.FileOffset); +  SwapValue(Value.FileSize); +  SwapValue(Value.MaxVMProtection); +  SwapValue(Value.InitialVMProtection); +  SwapValue(Value.NumSections); +  SwapValue(Value.Flags); +} +void MachOObject::ReadSegmentLoadCommand(const LoadCommandInfo &LCI, +                         InMemoryStruct<macho::SegmentLoadCommand> &Res) const { +  ReadInMemoryStruct(*this, Buffer->getBuffer(), LCI.Offset, Res); +} + +template<> +void SwapStruct(macho::Segment64LoadCommand &Value) { +  SwapValue(Value.Type); +  SwapValue(Value.Size); +  SwapValue(Value.VMAddress); +  SwapValue(Value.VMSize); +  SwapValue(Value.FileOffset); +  SwapValue(Value.FileSize); +  SwapValue(Value.MaxVMProtection); +  SwapValue(Value.InitialVMProtection); +  SwapValue(Value.NumSections); +  SwapValue(Value.Flags); +} +void MachOObject::ReadSegment64LoadCommand(const LoadCommandInfo &LCI, +                       InMemoryStruct<macho::Segment64LoadCommand> &Res) const { +  ReadInMemoryStruct(*this, Buffer->getBuffer(), LCI.Offset, Res); +} + +template<> +void SwapStruct(macho::SymtabLoadCommand &Value) { +  SwapValue(Value.Type); +  SwapValue(Value.Size); +  SwapValue(Value.SymbolTableOffset); +  SwapValue(Value.NumSymbolTableEntries); +  SwapValue(Value.StringTableOffset); +  SwapValue(Value.StringTableSize); +} +void MachOObject::ReadSymtabLoadCommand(const LoadCommandInfo &LCI, +                          InMemoryStruct<macho::SymtabLoadCommand> &Res) const { +  ReadInMemoryStruct(*this, Buffer->getBuffer(), LCI.Offset, Res); +} + +template<> +void SwapStruct(macho::DysymtabLoadCommand &Value) { +  SwapValue(Value.Type); +  SwapValue(Value.Size); +  SwapValue(Value.LocalSymbolsIndex); +  SwapValue(Value.NumLocalSymbols); +  SwapValue(Value.ExternalSymbolsIndex); +  SwapValue(Value.NumExternalSymbols); +  SwapValue(Value.UndefinedSymbolsIndex); +  SwapValue(Value.NumUndefinedSymbols); +  SwapValue(Value.TOCOffset); +  SwapValue(Value.NumTOCEntries); +  SwapValue(Value.ModuleTableOffset); +  SwapValue(Value.NumModuleTableEntries); +  SwapValue(Value.ReferenceSymbolTableOffset); +  SwapValue(Value.NumReferencedSymbolTableEntries); +  SwapValue(Value.IndirectSymbolTableOffset); +  SwapValue(Value.NumIndirectSymbolTableEntries); +  SwapValue(Value.ExternalRelocationTableOffset); +  SwapValue(Value.NumExternalRelocationTableEntries); +  SwapValue(Value.LocalRelocationTableOffset); +  SwapValue(Value.NumLocalRelocationTableEntries); +} +void MachOObject::ReadDysymtabLoadCommand(const LoadCommandInfo &LCI, +                        InMemoryStruct<macho::DysymtabLoadCommand> &Res) const { +  ReadInMemoryStruct(*this, Buffer->getBuffer(), LCI.Offset, Res); +} + +template<> +void SwapStruct(macho::IndirectSymbolTableEntry &Value) { +  SwapValue(Value.Index); +} +void +MachOObject::ReadIndirectSymbolTableEntry(const macho::DysymtabLoadCommand &DLC, +                                          unsigned Index, +                   InMemoryStruct<macho::IndirectSymbolTableEntry> &Res) const { +  uint64_t Offset = (DLC.IndirectSymbolTableOffset + +                     Index * sizeof(macho::IndirectSymbolTableEntry)); +  ReadInMemoryStruct(*this, Buffer->getBuffer(), Offset, Res); +} + + +template<> +void SwapStruct(macho::Section &Value) { +  SwapValue(Value.Address); +  SwapValue(Value.Size); +  SwapValue(Value.Offset); +  SwapValue(Value.Align); +  SwapValue(Value.RelocationTableOffset); +  SwapValue(Value.NumRelocationTableEntries); +  SwapValue(Value.Flags); +  SwapValue(Value.Reserved1); +  SwapValue(Value.Reserved2); +} +void MachOObject::ReadSection(const LoadCommandInfo &LCI, +                              unsigned Index, +                              InMemoryStruct<macho::Section> &Res) const { +  assert(LCI.Command.Type == macho::LCT_Segment && +         "Unexpected load command info!"); +  uint64_t Offset = (LCI.Offset + sizeof(macho::SegmentLoadCommand) + +                     Index * sizeof(macho::Section)); +  ReadInMemoryStruct(*this, Buffer->getBuffer(), Offset, Res); +} + +template<> +void SwapStruct(macho::Section64 &Value) { +  SwapValue(Value.Address); +  SwapValue(Value.Size); +  SwapValue(Value.Offset); +  SwapValue(Value.Align); +  SwapValue(Value.RelocationTableOffset); +  SwapValue(Value.NumRelocationTableEntries); +  SwapValue(Value.Flags); +  SwapValue(Value.Reserved1); +  SwapValue(Value.Reserved2); +  SwapValue(Value.Reserved3); +} +void MachOObject::ReadSection64(const LoadCommandInfo &LCI, +                                unsigned Index, +                                InMemoryStruct<macho::Section64> &Res) const { +  assert(LCI.Command.Type == macho::LCT_Segment64 && +         "Unexpected load command info!"); +  uint64_t Offset = (LCI.Offset + sizeof(macho::Segment64LoadCommand) + +                     Index * sizeof(macho::Section64)); +  ReadInMemoryStruct(*this, Buffer->getBuffer(), Offset, Res); +} + +template<> +void SwapStruct(macho::RelocationEntry &Value) { +  SwapValue(Value.Word0); +  SwapValue(Value.Word1); +} +void MachOObject::ReadRelocationEntry(uint64_t RelocationTableOffset, +                                      unsigned Index, +                            InMemoryStruct<macho::RelocationEntry> &Res) const { +  uint64_t Offset = (RelocationTableOffset + +                     Index * sizeof(macho::RelocationEntry)); +  ReadInMemoryStruct(*this, Buffer->getBuffer(), Offset, Res); +} + +template<> +void SwapStruct(macho::SymbolTableEntry &Value) { +  SwapValue(Value.StringIndex); +  SwapValue(Value.Flags); +  SwapValue(Value.Value); +} +void MachOObject::ReadSymbolTableEntry(uint64_t SymbolTableOffset, +                                       unsigned Index, +                           InMemoryStruct<macho::SymbolTableEntry> &Res) const { +  uint64_t Offset = (SymbolTableOffset + +                     Index * sizeof(macho::SymbolTableEntry)); +  ReadInMemoryStruct(*this, Buffer->getBuffer(), Offset, Res); +} + +template<> +void SwapStruct(macho::Symbol64TableEntry &Value) { +  SwapValue(Value.StringIndex); +  SwapValue(Value.Flags); +  SwapValue(Value.Value); +} +void MachOObject::ReadSymbol64TableEntry(uint64_t SymbolTableOffset, +                                       unsigned Index, +                         InMemoryStruct<macho::Symbol64TableEntry> &Res) const { +  uint64_t Offset = (SymbolTableOffset + +                     Index * sizeof(macho::Symbol64TableEntry)); +  ReadInMemoryStruct(*this, Buffer->getBuffer(), Offset, Res); +} diff --git a/lib/Object/Makefile b/lib/Object/Makefile new file mode 100644 index 000000000000..79388dc97f1a --- /dev/null +++ b/lib/Object/Makefile @@ -0,0 +1,14 @@ +##===- lib/Object/Makefile ---------------------------------*- Makefile -*-===## +# +#                     The LLVM Compiler Infrastructure +# +# This file is distributed under the University of Illinois Open Source +# License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## + +LEVEL = ../.. +LIBRARYNAME = LLVMObject +BUILD_ARCHIVE := 1 + +include $(LEVEL)/Makefile.common diff --git a/lib/Object/ObjectFile.cpp b/lib/Object/ObjectFile.cpp new file mode 100644 index 000000000000..161ae3a083f1 --- /dev/null +++ b/lib/Object/ObjectFile.cpp @@ -0,0 +1,71 @@ +//===- ObjectFile.cpp - File format independent object file -----*- C++ -*-===// +// +//                     The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines a file format independent ObjectFile class. +// +//===----------------------------------------------------------------------===// + +#include "llvm/Object/ObjectFile.h" +#include "llvm/ADT/OwningPtr.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Path.h" +#include "llvm/Support/system_error.h" + +using namespace llvm; +using namespace object; + +ObjectFile::ObjectFile(MemoryBuffer *Object) +  : MapFile(Object) { +  assert(MapFile && "Must be a valid MemoryBuffer!"); +  base = reinterpret_cast<const uint8_t *>(MapFile->getBufferStart()); +} + +ObjectFile::~ObjectFile() { +  delete MapFile; +} + +StringRef ObjectFile::getFilename() const { +  return MapFile->getBufferIdentifier(); +} + +ObjectFile *ObjectFile::createObjectFile(MemoryBuffer *Object) { +  if (!Object || Object->getBufferSize() < 64) +    return 0; +  sys::LLVMFileType type = sys::IdentifyFileType(Object->getBufferStart(), +                                static_cast<unsigned>(Object->getBufferSize())); +  switch (type) { +    case sys::ELF_Relocatable_FileType: +    case sys::ELF_Executable_FileType: +    case sys::ELF_SharedObject_FileType: +    case sys::ELF_Core_FileType: +      return createELFObjectFile(Object); +    case sys::Mach_O_Object_FileType: +    case sys::Mach_O_Executable_FileType: +    case sys::Mach_O_FixedVirtualMemorySharedLib_FileType: +    case sys::Mach_O_Core_FileType: +    case sys::Mach_O_PreloadExecutable_FileType: +    case sys::Mach_O_DynamicallyLinkedSharedLib_FileType: +    case sys::Mach_O_DynamicLinker_FileType: +    case sys::Mach_O_Bundle_FileType: +    case sys::Mach_O_DynamicallyLinkedSharedLibStub_FileType: +      return 0; +    case sys::COFF_FileType: +      return createCOFFObjectFile(Object); +    default: +      llvm_unreachable("Unknown Object File Type"); +  } +} + +ObjectFile *ObjectFile::createObjectFile(StringRef ObjectPath) { +  OwningPtr<MemoryBuffer> File; +  if (error_code ec = MemoryBuffer::getFile(ObjectPath, File)) +    return NULL; +  return createObjectFile(File.take()); +}  | 
