llvm/lib/BinaryFormat/Magic.cpp

   1 //===- llvm/BinaryFormat/Magic.cpp - File magic identification --*- C++ -*-===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #include "llvm/BinaryFormat/Magic.h"
  10 #include "llvm/ADT/StringRef.h"
  11 #include "llvm/ADT/Twine.h"
  12 #include "llvm/BinaryFormat/COFF.h"
  13 #include "llvm/BinaryFormat/ELF.h"
  14 #include "llvm/BinaryFormat/MachO.h"
  15 #include "llvm/Support/Endian.h"
  16 #include "llvm/Support/FileSystem.h"
  17 #include "llvm/Support/MemoryBuffer.h"
  18
  19 #if !defined(_MSC_VER) && !defined(__MINGW32__)
  20 #include <unistd.h>
  21 #else
  22 #include <io.h>
  23 #endif
  24
  25 using namespace llvm;
  26 using namespace llvm::support::endian;
  27 using namespace llvm::sys::fs;
  28
  29 template <size_t N>
  30 static bool startswith(StringRef Magic, const char (&S)[N]) {
  31   return Magic.startswith(StringRef(S, N - 1));
  32 }
  33
  34 /// Identify the magic in magic.
  35 file_magic llvm::identify_magic(StringRef Magic) {
  36   if (Magic.size() < 4)
  37     return file_magic::unknown;
  38   switch ((unsigned char)Magic[0]) {
  39   case 0x00: {
  40     // COFF bigobj, CL.exe's LTO object file, or short import library file
  41     if (startswith(Magic, "\0\0\xFF\xFF")) {
  42       size_t MinSize =
  43           offsetof(COFF::BigObjHeader, UUID) + sizeof(COFF::BigObjMagic);
  44       if (Magic.size() < MinSize)
  45         return file_magic::coff_import_library;
  46
  47       const char *Start = Magic.data() + offsetof(COFF::BigObjHeader, UUID);
  48       if (memcmp(Start, COFF::BigObjMagic, sizeof(COFF::BigObjMagic)) == 0)
  49         return file_magic::coff_object;
  50       if (memcmp(Start, COFF::ClGlObjMagic, sizeof(COFF::BigObjMagic)) == 0)
  51         return file_magic::coff_cl_gl_object;
  52       return file_magic::coff_import_library;
  53     }
  54     // Windows resource file
  55     if (Magic.size() >= sizeof(COFF::WinResMagic) &&
  56         memcmp(Magic.data(), COFF::WinResMagic, sizeof(COFF::WinResMagic)) == 0)
  57       return file_magic::windows_resource;
  58     // 0x0000 = COFF unknown machine type
  59     if (Magic[1] == 0)
  60       return file_magic::coff_object;
  61     if (startswith(Magic, "\0asm"))
  62       return file_magic::wasm_object;
  63     break;
  64   }
  65
  66   case 0x01:
  67     // XCOFF format
  68     if (startswith(Magic, "\x01\xDF"))
  69       return file_magic::xcoff_object_32;
  70     if (startswith(Magic, "\x01\xF7"))
  71       return file_magic::xcoff_object_64;
  72     break;
  73
  74   case 0x03:
  75     if (startswith(Magic, "\x03\xF0\x00"))
  76       return file_magic::goff_object;
  77     break;
  78
  79   case 0xDE: // 0x0B17C0DE = BC wraper
  80     if (startswith(Magic, "\xDE\xC0\x17\x0B"))
  81       return file_magic::bitcode;
  82     break;
  83   case 'B':
  84     if (startswith(Magic, "BC\xC0\xDE"))
  85       return file_magic::bitcode;
  86     break;
  87   case '!':
  88     if (startswith(Magic, "!<arch>\n") || startswith(Magic, "!<thin>\n"))
  89       return file_magic::archive;
  90     break;
  91
  92   case '\177':
  93     if (startswith(Magic, "\177ELF") && Magic.size() >= 18) {
  94       bool Data2MSB = Magic[5] == 2;
  95       unsigned high = Data2MSB ? 16 : 17;
  96       unsigned low = Data2MSB ? 17 : 16;
  97       if (Magic[high] == 0) {
  98         switch (Magic[low]) {
  99         default:
 100           return file_magic::elf;
 101         case 1:
 102           return file_magic::elf_relocatable;
 103         case 2:
 104           return file_magic::elf_executable;
 105         case 3:
 106           return file_magic::elf_shared_object;
 107         case 4:
 108           return file_magic::elf_core;
 109         }
 110       }
 111       // It's still some type of ELF file.
 112       return file_magic::elf;
 113     }
 114     break;
 115
 116   case 0xCA:
 117     if (startswith(Magic, "\xCA\xFE\xBA\xBE") ||
 118         startswith(Magic, "\xCA\xFE\xBA\xBF")) {
 119       // This is complicated by an overlap with Java class files.
 120       // See the Mach-O section in /usr/share/file/magic for details.
 121       if (Magic.size() >= 8 && Magic[7] < 43)
 122         return file_magic::macho_universal_binary;
 123     }
 124     break;
 125
 126   // The two magic numbers for mach-o are:
 127   // 0xfeedface - 32-bit mach-o
 128   // 0xfeedfacf - 64-bit mach-o
 129   case 0xFE:
 130   case 0xCE:
 131   case 0xCF: {
 132     uint16_t type = 0;
 133     if (startswith(Magic, "\xFE\xED\xFA\xCE") ||
 134         startswith(Magic, "\xFE\xED\xFA\xCF")) {
 135       /* Native endian */
 136       size_t MinSize;
 137       if (Magic[3] == char(0xCE))
 138         MinSize = sizeof(MachO::mach_header);
 139       else
 140         MinSize = sizeof(MachO::mach_header_64);
 141       if (Magic.size() >= MinSize)
 142         type = Magic[12] << 24 | Magic[13] << 12 | Magic[14] << 8 | Magic[15];
 143     } else if (startswith(Magic, "\xCE\xFA\xED\xFE") ||
 144                startswith(Magic, "\xCF\xFA\xED\xFE")) {
 145       /* Reverse endian */
 146       size_t MinSize;
 147       if (Magic[0] == char(0xCE))
 148         MinSize = sizeof(MachO::mach_header);
 149       else
 150         MinSize = sizeof(MachO::mach_header_64);
 151       if (Magic.size() >= MinSize)
 152         type = Magic[15] << 24 | Magic[14] << 12 | Magic[13] << 8 | Magic[12];
 153     }
 154     switch (type) {
 155     default:
 156       break;
 157     case 1:
 158       return file_magic::macho_object;
 159     case 2:
 160       return file_magic::macho_executable;
 161     case 3:
 162       return file_magic::macho_fixed_virtual_memory_shared_lib;
 163     case 4:
 164       return file_magic::macho_core;
 165     case 5:
 166       return file_magic::macho_preload_executable;
 167     case 6:
 168       return file_magic::macho_dynamically_linked_shared_lib;
 169     case 7:
 170       return file_magic::macho_dynamic_linker;
 171     case 8:
 172       return file_magic::macho_bundle;
 173     case 9:
 174       return file_magic::macho_dynamically_linked_shared_lib_stub;
 175     case 10:
 176       return file_magic::macho_dsym_companion;
 177     case 11:
 178       return file_magic::macho_kext_bundle;
 179     }
 180     break;
 181   }
 182   case 0xF0: // PowerPC Windows
 183   case 0x83: // Alpha 32-bit
 184   case 0x84: // Alpha 64-bit
 185   case 0x66: // MPS R4000 Windows
 186   case 0x50: // mc68K
 187   case 0x4c: // 80386 Windows
 188   case 0xc4: // ARMNT Windows
 189     if (Magic[1] == 0x01)
 190       return file_magic::coff_object;
 191     LLVM_FALLTHROUGH;
 192
 193   case 0x90: // PA-RISC Windows
 194   case 0x68: // mc68K Windows
 195     if (Magic[1] == 0x02)
 196       return file_magic::coff_object;
 197     break;
 198
 199   case 'M': // Possible MS-DOS stub on Windows PE file, MSF/PDB file or a
 200             // Minidump file.
 201     if (startswith(Magic, "MZ") && Magic.size() >= 0x3c + 4) {
 202       uint32_t off = read32le(Magic.data() + 0x3c);
 203       // PE/COFF file, either EXE or DLL.
 204       if (Magic.substr(off).startswith(
 205               StringRef(COFF::PEMagic, sizeof(COFF::PEMagic))))
 206         return file_magic::pecoff_executable;
 207     }
 208     if (Magic.startswith("Microsoft C/C++ MSF 7.00\r\n"))
 209       return file_magic::pdb;
 210     if (startswith(Magic, "MDMP"))
 211       return file_magic::minidump;
 212     break;
 213
 214   case 0x64: // x86-64 or ARM64 Windows.
 215     if (Magic[1] == char(0x86) || Magic[1] == char(0xaa))
 216       return file_magic::coff_object;
 217     break;
 218
 219   case 0x2d: // YAML '-'
 220     if (startswith(Magic, "--- !tapi") || startswith(Magic, "---\narchs:"))
 221       return file_magic::tapi_file;
 222     break;
 223
 224   default:
 225     break;
 226   }
 227   return file_magic::unknown;
 228 }
 229
 230 std::error_code llvm::identify_magic(const Twine &Path, file_magic &Result) {
 231   auto FileOrError = MemoryBuffer::getFile(Path, /*IsText=*/false,
 232                                            /*RequiresNullTerminator=*/false);
 233   if (!FileOrError)
 234     return FileOrError.getError();
 235
 236   std::unique_ptr<MemoryBuffer> FileBuffer = std::move(*FileOrError);
 237   Result = identify_magic(FileBuffer->getBuffer());
 238
 239   return std::error_code();
 240 }