xref: /llvm-project/llvm/lib/BinaryFormat/Magic.cpp (revision 8cc9fd31e64981d2c1aaf71d7b05ea928ef499f6)
1 //===- llvm/BinaryFormat/Magic.cpp - File magic identification --*- C++ -*-===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 
10 #include "llvm/BinaryFormat/Magic.h"
11 
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 
18 #if !defined(_MSC_VER) && !defined(__MINGW32__)
19 #include <unistd.h>
20 #else
21 #include <io.h>
22 #endif
23 
24 using namespace llvm;
25 using namespace llvm::support::endian;
26 using namespace llvm::sys::fs;
27 
28 template <size_t N>
29 static bool startswith(StringRef Magic, const char (&S)[N]) {
30   return Magic.startswith(StringRef(S, N - 1));
31 }
32 
33 /// @brief Identify the magic in magic.
34 file_magic llvm::identify_magic(StringRef Magic) {
35   if (Magic.size() < 4)
36     return file_magic::unknown;
37   switch ((unsigned char)Magic[0]) {
38   case 0x00: {
39     // COFF bigobj, CL.exe's LTO object file, or short import library file
40     if (startswith(Magic, "\0\0\xFF\xFF")) {
41       size_t MinSize =
42           offsetof(COFF::BigObjHeader, UUID) + sizeof(COFF::BigObjMagic);
43       if (Magic.size() < MinSize)
44         return file_magic::coff_import_library;
45 
46       const char *Start = Magic.data() + offsetof(COFF::BigObjHeader, UUID);
47       if (memcmp(Start, COFF::BigObjMagic, sizeof(COFF::BigObjMagic)) == 0)
48         return file_magic::coff_object;
49       if (memcmp(Start, COFF::ClGlObjMagic, sizeof(COFF::BigObjMagic)) == 0)
50         return file_magic::coff_cl_gl_object;
51       return file_magic::coff_import_library;
52     }
53     // Windows resource file
54     if (startswith(Magic, "\0\0\0\0\x20\0\0\0\xFF"))
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   case 0xDE: // 0x0B17C0DE = BC wraper
64     if (startswith(Magic, "\xDE\xC0\x17\x0B"))
65       return file_magic::bitcode;
66     break;
67   case 'B':
68     if (startswith(Magic, "BC\xC0\xDE"))
69       return file_magic::bitcode;
70     break;
71   case '!':
72     if (startswith(Magic, "!<arch>\n") || startswith(Magic, "!<thin>\n"))
73       return file_magic::archive;
74     break;
75 
76   case '\177':
77     if (startswith(Magic, "\177ELF") && Magic.size() >= 18) {
78       bool Data2MSB = Magic[5] == 2;
79       unsigned high = Data2MSB ? 16 : 17;
80       unsigned low = Data2MSB ? 17 : 16;
81       if (Magic[high] == 0) {
82         switch (Magic[low]) {
83         default:
84           return file_magic::elf;
85         case 1:
86           return file_magic::elf_relocatable;
87         case 2:
88           return file_magic::elf_executable;
89         case 3:
90           return file_magic::elf_shared_object;
91         case 4:
92           return file_magic::elf_core;
93         }
94       }
95       // It's still some type of ELF file.
96       return file_magic::elf;
97     }
98     break;
99 
100   case 0xCA:
101     if (startswith(Magic, "\xCA\xFE\xBA\xBE") ||
102         startswith(Magic, "\xCA\xFE\xBA\xBF")) {
103       // This is complicated by an overlap with Java class files.
104       // See the Mach-O section in /usr/share/file/magic for details.
105       if (Magic.size() >= 8 && Magic[7] < 43)
106         return file_magic::macho_universal_binary;
107     }
108     break;
109 
110   // The two magic numbers for mach-o are:
111   // 0xfeedface - 32-bit mach-o
112   // 0xfeedfacf - 64-bit mach-o
113   case 0xFE:
114   case 0xCE:
115   case 0xCF: {
116     uint16_t type = 0;
117     if (startswith(Magic, "\xFE\xED\xFA\xCE") ||
118         startswith(Magic, "\xFE\xED\xFA\xCF")) {
119       /* Native endian */
120       size_t MinSize;
121       if (Magic[3] == char(0xCE))
122         MinSize = sizeof(MachO::mach_header);
123       else
124         MinSize = sizeof(MachO::mach_header_64);
125       if (Magic.size() >= MinSize)
126         type = Magic[12] << 24 | Magic[13] << 12 | Magic[14] << 8 | Magic[15];
127     } else if (startswith(Magic, "\xCE\xFA\xED\xFE") ||
128                startswith(Magic, "\xCF\xFA\xED\xFE")) {
129       /* Reverse endian */
130       size_t MinSize;
131       if (Magic[0] == char(0xCE))
132         MinSize = sizeof(MachO::mach_header);
133       else
134         MinSize = sizeof(MachO::mach_header_64);
135       if (Magic.size() >= MinSize)
136         type = Magic[15] << 24 | Magic[14] << 12 | Magic[13] << 8 | Magic[12];
137     }
138     switch (type) {
139     default:
140       break;
141     case 1:
142       return file_magic::macho_object;
143     case 2:
144       return file_magic::macho_executable;
145     case 3:
146       return file_magic::macho_fixed_virtual_memory_shared_lib;
147     case 4:
148       return file_magic::macho_core;
149     case 5:
150       return file_magic::macho_preload_executable;
151     case 6:
152       return file_magic::macho_dynamically_linked_shared_lib;
153     case 7:
154       return file_magic::macho_dynamic_linker;
155     case 8:
156       return file_magic::macho_bundle;
157     case 9:
158       return file_magic::macho_dynamically_linked_shared_lib_stub;
159     case 10:
160       return file_magic::macho_dsym_companion;
161     case 11:
162       return file_magic::macho_kext_bundle;
163     }
164     break;
165   }
166   case 0xF0: // PowerPC Windows
167   case 0x83: // Alpha 32-bit
168   case 0x84: // Alpha 64-bit
169   case 0x66: // MPS R4000 Windows
170   case 0x50: // mc68K
171   case 0x4c: // 80386 Windows
172   case 0xc4: // ARMNT Windows
173     if (Magic[1] == 0x01)
174       return file_magic::coff_object;
175     LLVM_FALLTHROUGH;
176 
177   case 0x90: // PA-RISC Windows
178   case 0x68: // mc68K Windows
179     if (Magic[1] == 0x02)
180       return file_magic::coff_object;
181     break;
182 
183   case 'M': // Possible MS-DOS stub on Windows PE file
184     if (startswith(Magic, "MZ")) {
185       uint32_t off = read32le(Magic.data() + 0x3c);
186       // PE/COFF file, either EXE or DLL.
187       if (off < Magic.size() &&
188           memcmp(Magic.data() + off, COFF::PEMagic, sizeof(COFF::PEMagic)) == 0)
189         return file_magic::pecoff_executable;
190     }
191     break;
192 
193   case 0x64: // x86-64 or ARM64 Windows.
194     if (Magic[1] == char(0x86) || Magic[1] == char(0xaa))
195       return file_magic::coff_object;
196     break;
197 
198   default:
199     break;
200   }
201   return file_magic::unknown;
202 }
203 
204 std::error_code llvm::identify_magic(const Twine &Path, file_magic &Result) {
205   int FD;
206   if (std::error_code EC = openFileForRead(Path, FD))
207     return EC;
208 
209   char Buffer[32];
210   int Length = read(FD, Buffer, sizeof(Buffer));
211   if (close(FD) != 0 || Length < 0)
212     return std::error_code(errno, std::generic_category());
213 
214   Result = identify_magic(StringRef(Buffer, Length));
215   return std::error_code();
216 }
217