llvm-project

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/MachO.h"
14
#include "llvm/Support/Endian.h"
15
#include "llvm/Support/MemoryBuffer.h"
16

17
#if !defined(_MSC_VER) && !defined(__MINGW32__)
18
#include <unistd.h>
19
#else
20
#include <io.h>
21
#endif
22

23
using namespace llvm;
24
using namespace llvm::support::endian;
25
using namespace llvm::sys::fs;
26

27
template <size_t N>
28
static bool startswith(StringRef Magic, const char (&S)[N]) {
29
  return Magic.starts_with(StringRef(S, N - 1));
30
}
31

32
/// Identify the magic in magic.
33
file_magic llvm::identify_magic(StringRef Magic) {
34
  if (Magic.size() < 4)
35
    return file_magic::unknown;
36
  switch ((unsigned char)Magic[0]) {
37
  case 0x00: {
38
    // COFF bigobj, CL.exe's LTO object file, or short import library file
39
    if (startswith(Magic, "\0\0\xFF\xFF")) {
40
      size_t MinSize =
41
          offsetof(COFF::BigObjHeader, UUID) + sizeof(COFF::BigObjMagic);
42
      if (Magic.size() < MinSize)
43
        return file_magic::coff_import_library;
44

45
      const char *Start = Magic.data() + offsetof(COFF::BigObjHeader, UUID);
46
      if (memcmp(Start, COFF::BigObjMagic, sizeof(COFF::BigObjMagic)) == 0)
47
        return file_magic::coff_object;
48
      if (memcmp(Start, COFF::ClGlObjMagic, sizeof(COFF::BigObjMagic)) == 0)
49
        return file_magic::coff_cl_gl_object;
50
      return file_magic::coff_import_library;
51
    }
52
    // Windows resource file
53
    if (Magic.size() >= sizeof(COFF::WinResMagic) &&
54
        memcmp(Magic.data(), COFF::WinResMagic, sizeof(COFF::WinResMagic)) == 0)
55
      return file_magic::windows_resource;
56
    // 0x0000 = COFF unknown machine type
57
    if (Magic[1] == 0)
58
      return file_magic::coff_object;
59
    if (startswith(Magic, "\0asm"))
60
      return file_magic::wasm_object;
61
    break;
62
  }
63

64
  case 0x01:
65
    // XCOFF format
66
    if (startswith(Magic, "\x01\xDF"))
67
      return file_magic::xcoff_object_32;
68
    if (startswith(Magic, "\x01\xF7"))
69
      return file_magic::xcoff_object_64;
70
    break;
71

72
  case 0x03:
73
    if (startswith(Magic, "\x03\xF0\x00"))
74
      return file_magic::goff_object;
75
    // SPIR-V format in little-endian mode.
76
    if (startswith(Magic, "\x03\x02\x23\x07"))
77
      return file_magic::spirv_object;
78
    break;
79

80
  case 0x07: // SPIR-V format in big-endian mode.
81
    if (startswith(Magic, "\x07\x23\x02\x03"))
82
      return file_magic::spirv_object;
83
    break;
84

85
  case 0x10:
86
    if (startswith(Magic, "\x10\xFF\x10\xAD"))
87
      return file_magic::offload_binary;
88
    break;
89

90
  case 0xDE: // 0x0B17C0DE = BC wraper
91
    if (startswith(Magic, "\xDE\xC0\x17\x0B"))
92
      return file_magic::bitcode;
93
    break;
94
  case 'B':
95
    if (startswith(Magic, "BC\xC0\xDE"))
96
      return file_magic::bitcode;
97
    break;
98
  case 'C':
99
    if (startswith(Magic, "CCOB"))
100
      return file_magic::offload_bundle_compressed;
101
    if (startswith(Magic, "CPCH"))
102
      return file_magic::clang_ast;
103
    break;
104
  case '!':
105
    if (startswith(Magic, "!<arch>\n") || startswith(Magic, "!<thin>\n"))
106
      return file_magic::archive;
107
    break;
108
  case '<':
109
    if (startswith(Magic, "<bigaf>\n"))
110
      return file_magic::archive;
111
    break;
112
  case '\177':
113
    if (startswith(Magic, "\177ELF") && Magic.size() >= 18) {
114
      bool Data2MSB = Magic[5] == 2;
115
      unsigned high = Data2MSB ? 16 : 17;
116
      unsigned low = Data2MSB ? 17 : 16;
117
      if (Magic[high] == 0) {
118
        switch (Magic[low]) {
119
        default:
120
          return file_magic::elf;
121
        case 1:
122
          return file_magic::elf_relocatable;
123
        case 2:
124
          return file_magic::elf_executable;
125
        case 3:
126
          return file_magic::elf_shared_object;
127
        case 4:
128
          return file_magic::elf_core;
129
        }
130
      }
131
      // It's still some type of ELF file.
132
      return file_magic::elf;
133
    }
134
    break;
135

136
  case 0xCA:
137
    if (startswith(Magic, "\xCA\xFE\xBA\xBE") ||
138
        startswith(Magic, "\xCA\xFE\xBA\xBF")) {
139
      // This is complicated by an overlap with Java class files.
140
      // See the Mach-O section in /usr/share/file/magic for details.
141
      if (Magic.size() >= 8 && Magic[7] < 43)
142
        return file_magic::macho_universal_binary;
143
    }
144
    break;
145

146
  // The two magic numbers for mach-o are:
147
  // 0xfeedface - 32-bit mach-o
148
  // 0xfeedfacf - 64-bit mach-o
149
  case 0xFE:
150
  case 0xCE:
151
  case 0xCF: {
152
    uint16_t type = 0;
153
    if (startswith(Magic, "\xFE\xED\xFA\xCE") ||
154
        startswith(Magic, "\xFE\xED\xFA\xCF")) {
155
      /* Native endian */
156
      size_t MinSize;
157
      if (Magic[3] == char(0xCE))
158
        MinSize = sizeof(MachO::mach_header);
159
      else
160
        MinSize = sizeof(MachO::mach_header_64);
161
      if (Magic.size() >= MinSize)
162
        type = Magic[12] << 24 | Magic[13] << 12 | Magic[14] << 8 | Magic[15];
163
    } else if (startswith(Magic, "\xCE\xFA\xED\xFE") ||
164
               startswith(Magic, "\xCF\xFA\xED\xFE")) {
165
      /* Reverse endian */
166
      size_t MinSize;
167
      if (Magic[0] == char(0xCE))
168
        MinSize = sizeof(MachO::mach_header);
169
      else
170
        MinSize = sizeof(MachO::mach_header_64);
171
      if (Magic.size() >= MinSize)
172
        type = Magic[15] << 24 | Magic[14] << 12 | Magic[13] << 8 | Magic[12];
173
    }
174
    switch (type) {
175
    default:
176
      break;
177
    case 1:
178
      return file_magic::macho_object;
179
    case 2:
180
      return file_magic::macho_executable;
181
    case 3:
182
      return file_magic::macho_fixed_virtual_memory_shared_lib;
183
    case 4:
184
      return file_magic::macho_core;
185
    case 5:
186
      return file_magic::macho_preload_executable;
187
    case 6:
188
      return file_magic::macho_dynamically_linked_shared_lib;
189
    case 7:
190
      return file_magic::macho_dynamic_linker;
191
    case 8:
192
      return file_magic::macho_bundle;
193
    case 9:
194
      return file_magic::macho_dynamically_linked_shared_lib_stub;
195
    case 10:
196
      return file_magic::macho_dsym_companion;
197
    case 11:
198
      return file_magic::macho_kext_bundle;
199
    case 12:
200
      return file_magic::macho_file_set;
201
    }
202
    break;
203
  }
204
  case 0xF0: // PowerPC Windows
205
  case 0x83: // Alpha 32-bit
206
  case 0x84: // Alpha 64-bit
207
  case 0x66: // MPS R4000 Windows
208
  case 0x50: // mc68K
209
    if (startswith(Magic, "\x50\xed\x55\xba"))
210
      return file_magic::cuda_fatbinary;
211
    [[fallthrough]];
212

213
  case 0x4c: // 80386 Windows
214
  case 0xc4: // ARMNT Windows
215
    if (Magic[1] == 0x01)
216
      return file_magic::coff_object;
217
    [[fallthrough]];
218

219
  case 0x90: // PA-RISC Windows
220
  case 0x68: // mc68K Windows
221
    if (Magic[1] == 0x02)
222
      return file_magic::coff_object;
223
    break;
224

225
  case 'M': // Possible MS-DOS stub on Windows PE file, MSF/PDB file or a
226
            // Minidump file.
227
    if (startswith(Magic, "MZ") && Magic.size() >= 0x3c + 4) {
228
      uint32_t off = read32le(Magic.data() + 0x3c);
229
      // PE/COFF file, either EXE or DLL.
230
      if (Magic.substr(off).starts_with(
231
              StringRef(COFF::PEMagic, sizeof(COFF::PEMagic))))
232
        return file_magic::pecoff_executable;
233
    }
234
    if (Magic.starts_with("Microsoft C/C++ MSF 7.00\r\n"))
235
      return file_magic::pdb;
236
    if (startswith(Magic, "MDMP"))
237
      return file_magic::minidump;
238
    break;
239

240
  case 0x64: // x86-64 or ARM64 Windows.
241
    if (Magic[1] == char(0x86) || Magic[1] == char(0xaa))
242
      return file_magic::coff_object;
243
    break;
244

245
  case 0x2d: // YAML '-' MachO TBD.
246
    if (startswith(Magic, "--- !tapi") || startswith(Magic, "---\narchs:"))
247
      return file_magic::tapi_file;
248
    break;
249
  case 0x7b: // JSON '{' MachO TBD.
250
    return file_magic::tapi_file;
251
    break;
252

253
  case 'D': // DirectX container file - DXBC
254
    if (startswith(Magic, "DXBC"))
255
      return file_magic::dxcontainer_object;
256
    break;
257

258
  case 0x41: // ARM64EC windows
259
    if (Magic[1] == char(0xA6))
260
      return file_magic::coff_object;
261
    break;
262

263
  case 0x4e: // ARM64X windows
264
    if (Magic[1] == char(0xA6))
265
      return file_magic::coff_object;
266
    break;
267

268
  case '_': {
269
    const char OBMagic[] = "__CLANG_OFFLOAD_BUNDLE__";
270
    if (Magic.size() >= sizeof(OBMagic) && startswith(Magic, OBMagic))
271
      return file_magic::offload_bundle;
272
    break;
273
  }
274

275
  default:
276
    break;
277
  }
278
  return file_magic::unknown;
279
}
280

281
std::error_code llvm::identify_magic(const Twine &Path, file_magic &Result) {
282
  auto FileOrError = MemoryBuffer::getFile(Path, /*IsText=*/false,
283
                                           /*RequiresNullTerminator=*/false);
284
  if (!FileOrError)
285
    return FileOrError.getError();
286

287
  std::unique_ptr<MemoryBuffer> FileBuffer = std::move(*FileOrError);
288
  Result = identify_magic(FileBuffer->getBuffer());
289

290
  return std::error_code();
291
}
292