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.startswith(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     break;
76 
77   case 0x10:
78     if (startswith(Magic, "\x10\xFF\x10\xAD"))
79       return file_magic::offload_binary;
80     break;
81 
82   case 0xDE: // 0x0B17C0DE = BC wraper
83     if (startswith(Magic, "\xDE\xC0\x17\x0B"))
84       return file_magic::bitcode;
85     break;
86   case 'B':
87     if (startswith(Magic, "BC\xC0\xDE"))
88       return file_magic::bitcode;
89     break;
90   case '!':
91     if (startswith(Magic, "!<arch>\n") || startswith(Magic, "!<thin>\n"))
92       return file_magic::archive;
93     break;
94   case '<':
95     if (startswith(Magic, "<bigaf>\n"))
96       return file_magic::archive;
97     break;
98   case '\177':
99     if (startswith(Magic, "\177ELF") && Magic.size() >= 18) {
100       bool Data2MSB = Magic[5] == 2;
101       unsigned high = Data2MSB ? 16 : 17;
102       unsigned low = Data2MSB ? 17 : 16;
103       if (Magic[high] == 0) {
104         switch (Magic[low]) {
105         default:
106           return file_magic::elf;
107         case 1:
108           return file_magic::elf_relocatable;
109         case 2:
110           return file_magic::elf_executable;
111         case 3:
112           return file_magic::elf_shared_object;
113         case 4:
114           return file_magic::elf_core;
115         }
116       }
117       // It's still some type of ELF file.
118       return file_magic::elf;
119     }
120     break;
121 
122   case 0xCA:
123     if (startswith(Magic, "\xCA\xFE\xBA\xBE") ||
124         startswith(Magic, "\xCA\xFE\xBA\xBF")) {
125       // This is complicated by an overlap with Java class files.
126       // See the Mach-O section in /usr/share/file/magic for details.
127       if (Magic.size() >= 8 && Magic[7] < 43)
128         return file_magic::macho_universal_binary;
129     }
130     break;
131 
132   // The two magic numbers for mach-o are:
133   // 0xfeedface - 32-bit mach-o
134   // 0xfeedfacf - 64-bit mach-o
135   case 0xFE:
136   case 0xCE:
137   case 0xCF: {
138     uint16_t type = 0;
139     if (startswith(Magic, "\xFE\xED\xFA\xCE") ||
140         startswith(Magic, "\xFE\xED\xFA\xCF")) {
141       /* Native endian */
142       size_t MinSize;
143       if (Magic[3] == char(0xCE))
144         MinSize = sizeof(MachO::mach_header);
145       else
146         MinSize = sizeof(MachO::mach_header_64);
147       if (Magic.size() >= MinSize)
148         type = Magic[12] << 24 | Magic[13] << 12 | Magic[14] << 8 | Magic[15];
149     } else if (startswith(Magic, "\xCE\xFA\xED\xFE") ||
150                startswith(Magic, "\xCF\xFA\xED\xFE")) {
151       /* Reverse endian */
152       size_t MinSize;
153       if (Magic[0] == char(0xCE))
154         MinSize = sizeof(MachO::mach_header);
155       else
156         MinSize = sizeof(MachO::mach_header_64);
157       if (Magic.size() >= MinSize)
158         type = Magic[15] << 24 | Magic[14] << 12 | Magic[13] << 8 | Magic[12];
159     }
160     switch (type) {
161     default:
162       break;
163     case 1:
164       return file_magic::macho_object;
165     case 2:
166       return file_magic::macho_executable;
167     case 3:
168       return file_magic::macho_fixed_virtual_memory_shared_lib;
169     case 4:
170       return file_magic::macho_core;
171     case 5:
172       return file_magic::macho_preload_executable;
173     case 6:
174       return file_magic::macho_dynamically_linked_shared_lib;
175     case 7:
176       return file_magic::macho_dynamic_linker;
177     case 8:
178       return file_magic::macho_bundle;
179     case 9:
180       return file_magic::macho_dynamically_linked_shared_lib_stub;
181     case 10:
182       return file_magic::macho_dsym_companion;
183     case 11:
184       return file_magic::macho_kext_bundle;
185     case 12:
186       return file_magic::macho_file_set;
187     }
188     break;
189   }
190   case 0xF0: // PowerPC Windows
191   case 0x83: // Alpha 32-bit
192   case 0x84: // Alpha 64-bit
193   case 0x66: // MPS R4000 Windows
194   case 0x50: // mc68K
195     if (startswith(Magic, "\x50\xed\x55\xba"))
196       return file_magic::cuda_fatbinary;
197     [[fallthrough]];
198 
199   case 0x4c: // 80386 Windows
200   case 0xc4: // ARMNT Windows
201     if (Magic[1] == 0x01)
202       return file_magic::coff_object;
203     [[fallthrough]];
204 
205   case 0x90: // PA-RISC Windows
206   case 0x68: // mc68K Windows
207     if (Magic[1] == 0x02)
208       return file_magic::coff_object;
209     break;
210 
211   case 'M': // Possible MS-DOS stub on Windows PE file, MSF/PDB file or a
212             // Minidump file.
213     if (startswith(Magic, "MZ") && Magic.size() >= 0x3c + 4) {
214       uint32_t off = read32le(Magic.data() + 0x3c);
215       // PE/COFF file, either EXE or DLL.
216       if (Magic.substr(off).startswith(
217               StringRef(COFF::PEMagic, sizeof(COFF::PEMagic))))
218         return file_magic::pecoff_executable;
219     }
220     if (Magic.startswith("Microsoft C/C++ MSF 7.00\r\n"))
221       return file_magic::pdb;
222     if (startswith(Magic, "MDMP"))
223       return file_magic::minidump;
224     break;
225 
226   case 0x64: // x86-64 or ARM64 Windows.
227     if (Magic[1] == char(0x86) || Magic[1] == char(0xaa))
228       return file_magic::coff_object;
229     break;
230 
231   case 0x2d: // YAML '-'
232     if (startswith(Magic, "--- !tapi") || startswith(Magic, "---\narchs:"))
233       return file_magic::tapi_file;
234     break;
235 
236   case 'D': // DirectX container file - DXBC
237     if (startswith(Magic, "DXBC"))
238       return file_magic::dxcontainer_object;
239     break;
240 
241   case 0x41: // ARM64EC windows
242     if (Magic[1] == char(0xA6))
243       return file_magic::coff_object;
244     break;
245 
246   default:
247     break;
248   }
249   return file_magic::unknown;
250 }
251 
252 std::error_code llvm::identify_magic(const Twine &Path, file_magic &Result) {
253   auto FileOrError = MemoryBuffer::getFile(Path, /*IsText=*/false,
254                                            /*RequiresNullTerminator=*/false);
255   if (!FileOrError)
256     return FileOrError.getError();
257 
258   std::unique_ptr<MemoryBuffer> FileBuffer = std::move(*FileOrError);
259   Result = identify_magic(FileBuffer->getBuffer());
260 
261   return std::error_code();
262 }
263