Path: blob/main/contrib/llvm-project/llvm/lib/BinaryFormat/Magic.cpp
35232 views
//===- llvm/BinaryFormat/Magic.cpp - File magic identification --*- C++ -*-===//1//2// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.3// See https://llvm.org/LICENSE.txt for license information.4// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception5//6//===----------------------------------------------------------------------===//78#include "llvm/BinaryFormat/Magic.h"9#include "llvm/ADT/StringRef.h"10#include "llvm/ADT/Twine.h"11#include "llvm/BinaryFormat/COFF.h"12#include "llvm/BinaryFormat/MachO.h"13#include "llvm/Support/Endian.h"14#include "llvm/Support/MemoryBuffer.h"1516#if !defined(_MSC_VER) && !defined(__MINGW32__)17#include <unistd.h>18#else19#include <io.h>20#endif2122using namespace llvm;23using namespace llvm::support::endian;24using namespace llvm::sys::fs;2526template <size_t N>27static bool startswith(StringRef Magic, const char (&S)[N]) {28return Magic.starts_with(StringRef(S, N - 1));29}3031/// Identify the magic in magic.32file_magic llvm::identify_magic(StringRef Magic) {33if (Magic.size() < 4)34return file_magic::unknown;35switch ((unsigned char)Magic[0]) {36case 0x00: {37// COFF bigobj, CL.exe's LTO object file, or short import library file38if (startswith(Magic, "\0\0\xFF\xFF")) {39size_t MinSize =40offsetof(COFF::BigObjHeader, UUID) + sizeof(COFF::BigObjMagic);41if (Magic.size() < MinSize)42return file_magic::coff_import_library;4344const char *Start = Magic.data() + offsetof(COFF::BigObjHeader, UUID);45if (memcmp(Start, COFF::BigObjMagic, sizeof(COFF::BigObjMagic)) == 0)46return file_magic::coff_object;47if (memcmp(Start, COFF::ClGlObjMagic, sizeof(COFF::BigObjMagic)) == 0)48return file_magic::coff_cl_gl_object;49return file_magic::coff_import_library;50}51// Windows resource file52if (Magic.size() >= sizeof(COFF::WinResMagic) &&53memcmp(Magic.data(), COFF::WinResMagic, sizeof(COFF::WinResMagic)) == 0)54return file_magic::windows_resource;55// 0x0000 = COFF unknown machine type56if (Magic[1] == 0)57return file_magic::coff_object;58if (startswith(Magic, "\0asm"))59return file_magic::wasm_object;60break;61}6263case 0x01:64// XCOFF format65if (startswith(Magic, "\x01\xDF"))66return file_magic::xcoff_object_32;67if (startswith(Magic, "\x01\xF7"))68return file_magic::xcoff_object_64;69break;7071case 0x03:72if (startswith(Magic, "\x03\xF0\x00"))73return file_magic::goff_object;74// SPIR-V format in little-endian mode.75if (startswith(Magic, "\x03\x02\x23\x07"))76return file_magic::spirv_object;77break;7879case 0x07: // SPIR-V format in big-endian mode.80if (startswith(Magic, "\x07\x23\x02\x03"))81return file_magic::spirv_object;82break;8384case 0x10:85if (startswith(Magic, "\x10\xFF\x10\xAD"))86return file_magic::offload_binary;87break;8889case 0xDE: // 0x0B17C0DE = BC wraper90if (startswith(Magic, "\xDE\xC0\x17\x0B"))91return file_magic::bitcode;92break;93case 'B':94if (startswith(Magic, "BC\xC0\xDE"))95return file_magic::bitcode;96break;97case 'C':98if (startswith(Magic, "CCOB"))99return file_magic::offload_bundle_compressed;100if (startswith(Magic, "CPCH"))101return file_magic::clang_ast;102break;103case '!':104if (startswith(Magic, "!<arch>\n") || startswith(Magic, "!<thin>\n"))105return file_magic::archive;106break;107case '<':108if (startswith(Magic, "<bigaf>\n"))109return file_magic::archive;110break;111case '\177':112if (startswith(Magic, "\177ELF") && Magic.size() >= 18) {113bool Data2MSB = Magic[5] == 2;114unsigned high = Data2MSB ? 16 : 17;115unsigned low = Data2MSB ? 17 : 16;116if (Magic[high] == 0) {117switch (Magic[low]) {118default:119return file_magic::elf;120case 1:121return file_magic::elf_relocatable;122case 2:123return file_magic::elf_executable;124case 3:125return file_magic::elf_shared_object;126case 4:127return file_magic::elf_core;128}129}130// It's still some type of ELF file.131return file_magic::elf;132}133break;134135case 0xCA:136if (startswith(Magic, "\xCA\xFE\xBA\xBE") ||137startswith(Magic, "\xCA\xFE\xBA\xBF")) {138// This is complicated by an overlap with Java class files.139// See the Mach-O section in /usr/share/file/magic for details.140if (Magic.size() >= 8 && Magic[7] < 43)141return file_magic::macho_universal_binary;142}143break;144145// The two magic numbers for mach-o are:146// 0xfeedface - 32-bit mach-o147// 0xfeedfacf - 64-bit mach-o148case 0xFE:149case 0xCE:150case 0xCF: {151uint16_t type = 0;152if (startswith(Magic, "\xFE\xED\xFA\xCE") ||153startswith(Magic, "\xFE\xED\xFA\xCF")) {154/* Native endian */155size_t MinSize;156if (Magic[3] == char(0xCE))157MinSize = sizeof(MachO::mach_header);158else159MinSize = sizeof(MachO::mach_header_64);160if (Magic.size() >= MinSize)161type = Magic[12] << 24 | Magic[13] << 12 | Magic[14] << 8 | Magic[15];162} else if (startswith(Magic, "\xCE\xFA\xED\xFE") ||163startswith(Magic, "\xCF\xFA\xED\xFE")) {164/* Reverse endian */165size_t MinSize;166if (Magic[0] == char(0xCE))167MinSize = sizeof(MachO::mach_header);168else169MinSize = sizeof(MachO::mach_header_64);170if (Magic.size() >= MinSize)171type = Magic[15] << 24 | Magic[14] << 12 | Magic[13] << 8 | Magic[12];172}173switch (type) {174default:175break;176case 1:177return file_magic::macho_object;178case 2:179return file_magic::macho_executable;180case 3:181return file_magic::macho_fixed_virtual_memory_shared_lib;182case 4:183return file_magic::macho_core;184case 5:185return file_magic::macho_preload_executable;186case 6:187return file_magic::macho_dynamically_linked_shared_lib;188case 7:189return file_magic::macho_dynamic_linker;190case 8:191return file_magic::macho_bundle;192case 9:193return file_magic::macho_dynamically_linked_shared_lib_stub;194case 10:195return file_magic::macho_dsym_companion;196case 11:197return file_magic::macho_kext_bundle;198case 12:199return file_magic::macho_file_set;200}201break;202}203case 0xF0: // PowerPC Windows204case 0x83: // Alpha 32-bit205case 0x84: // Alpha 64-bit206case 0x66: // MPS R4000 Windows207case 0x50: // mc68K208if (startswith(Magic, "\x50\xed\x55\xba"))209return file_magic::cuda_fatbinary;210[[fallthrough]];211212case 0x4c: // 80386 Windows213case 0xc4: // ARMNT Windows214if (Magic[1] == 0x01)215return file_magic::coff_object;216[[fallthrough]];217218case 0x90: // PA-RISC Windows219case 0x68: // mc68K Windows220if (Magic[1] == 0x02)221return file_magic::coff_object;222break;223224case 'M': // Possible MS-DOS stub on Windows PE file, MSF/PDB file or a225// Minidump file.226if (startswith(Magic, "MZ") && Magic.size() >= 0x3c + 4) {227uint32_t off = read32le(Magic.data() + 0x3c);228// PE/COFF file, either EXE or DLL.229if (Magic.substr(off).starts_with(230StringRef(COFF::PEMagic, sizeof(COFF::PEMagic))))231return file_magic::pecoff_executable;232}233if (Magic.starts_with("Microsoft C/C++ MSF 7.00\r\n"))234return file_magic::pdb;235if (startswith(Magic, "MDMP"))236return file_magic::minidump;237break;238239case 0x64: // x86-64 or ARM64 Windows.240if (Magic[1] == char(0x86) || Magic[1] == char(0xaa))241return file_magic::coff_object;242break;243244case 0x2d: // YAML '-' MachO TBD.245if (startswith(Magic, "--- !tapi") || startswith(Magic, "---\narchs:"))246return file_magic::tapi_file;247break;248case 0x7b: // JSON '{' MachO TBD.249return file_magic::tapi_file;250break;251252case 'D': // DirectX container file - DXBC253if (startswith(Magic, "DXBC"))254return file_magic::dxcontainer_object;255break;256257case 0x41: // ARM64EC windows258if (Magic[1] == char(0xA6))259return file_magic::coff_object;260break;261262case 0x4e: // ARM64X windows263if (Magic[1] == char(0xA6))264return file_magic::coff_object;265break;266267case '_': {268const char OBMagic[] = "__CLANG_OFFLOAD_BUNDLE__";269if (Magic.size() >= sizeof(OBMagic) && startswith(Magic, OBMagic))270return file_magic::offload_bundle;271break;272}273274default:275break;276}277return file_magic::unknown;278}279280std::error_code llvm::identify_magic(const Twine &Path, file_magic &Result) {281auto FileOrError = MemoryBuffer::getFile(Path, /*IsText=*/false,282/*RequiresNullTerminator=*/false);283if (!FileOrError)284return FileOrError.getError();285286std::unique_ptr<MemoryBuffer> FileBuffer = std::move(*FileOrError);287Result = identify_magic(FileBuffer->getBuffer());288289return std::error_code();290}291292293