// Copyright (c) 2006, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // macho_id.cc: Functions to gather identifying information from a macho file // // See macho_id.h for documentation // // Author: Dan Waylonis extern "C" { // necessary for Leopard #include <fcntl.h> #include <mach-o/loader.h> #include <mach-o/swap.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/time.h> #include <sys/types.h> #include <unistd.h> } #include "common/mac/macho_id.h" #include "common/mac/macho_walker.h" #include "common/mac/macho_utilities.h" namespace MacFileUtilities { using google_breakpad::MD5Init; using google_breakpad::MD5Update; using google_breakpad::MD5Final; MachoID::MachoID(const char *path) : memory_(0), memory_size_(0), crc_(0), md5_context_(), update_function_(NULL) { strlcpy(path_, path, sizeof(path_)); } MachoID::MachoID(const char *path, void *memory, size_t size) : memory_(memory), memory_size_(size), crc_(0), md5_context_(), update_function_(NULL) { strlcpy(path_, path, sizeof(path_)); } MachoID::~MachoID() { } // The CRC info is from http://en.wikipedia.org/wiki/Adler-32 // With optimizations from http://www.zlib.net/ // The largest prime smaller than 65536 #define MOD_ADLER 65521 // MAX_BLOCK is the largest n such that 255n(n+1)/2 + (n+1)(MAX_BLOCK-1) <= 2^32-1 #define MAX_BLOCK 5552 void MachoID::UpdateCRC(unsigned char *bytes, size_t size) { // Unrolled loops for summing #define DO1(buf,i) {sum1 += (buf)[i]; sum2 += sum1;} #define DO2(buf,i) DO1(buf,i); DO1(buf,i+1); #define DO4(buf,i) DO2(buf,i); DO2(buf,i+2); #define DO8(buf,i) DO4(buf,i); DO4(buf,i+4); #define DO16(buf) DO8(buf,0); DO8(buf,8); // Split up the crc uint32_t sum1 = crc_ & 0xFFFF; uint32_t sum2 = (crc_ >> 16) & 0xFFFF; // Do large blocks while (size >= MAX_BLOCK) { size -= MAX_BLOCK; int block_count = MAX_BLOCK / 16; do { DO16(bytes); bytes += 16; } while (--block_count); sum1 %= MOD_ADLER; sum2 %= MOD_ADLER; } // Do remaining bytes if (size) { while (size >= 16) { size -= 16; DO16(bytes); bytes += 16; } while (size--) { sum1 += *bytes++; sum2 += sum1; } sum1 %= MOD_ADLER; sum2 %= MOD_ADLER; crc_ = (sum2 << 16) | sum1; } } void MachoID::UpdateMD5(unsigned char *bytes, size_t size) { MD5Update(&md5_context_, bytes, static_cast<unsigned>(size)); } void MachoID::Update(MachoWalker *walker, off_t offset, size_t size) { if (!update_function_ || !size) return; // Read up to 4k bytes at a time unsigned char buffer[4096]; size_t buffer_size; off_t file_offset = offset; while (size > 0) { if (size > sizeof(buffer)) { buffer_size = sizeof(buffer); size -= buffer_size; } else { buffer_size = size; size = 0; } if (!walker->ReadBytes(buffer, buffer_size, file_offset)) return; (this->*update_function_)(buffer, buffer_size); file_offset += buffer_size; } } bool MachoID::UUIDCommand(cpu_type_t cpu_type, cpu_subtype_t cpu_subtype, unsigned char bytes[16]) { struct breakpad_uuid_command uuid_cmd; uuid_cmd.cmd = 0; if (!WalkHeader(cpu_type, cpu_subtype, UUIDWalkerCB, &uuid_cmd)) return false; // If we found the command, we'll have initialized the uuid_command // structure if (uuid_cmd.cmd == LC_UUID) { memcpy(bytes, uuid_cmd.uuid, sizeof(uuid_cmd.uuid)); return true; } return false; } bool MachoID::IDCommand(cpu_type_t cpu_type, cpu_subtype_t cpu_subtype, unsigned char identifier[16]) { struct dylib_command dylib_cmd; dylib_cmd.cmd = 0; if (!WalkHeader(cpu_type, cpu_subtype, IDWalkerCB, &dylib_cmd)) return false; // If we found the command, we'll have initialized the dylib_command // structure if (dylib_cmd.cmd == LC_ID_DYLIB) { // Take the hashed filename, version, and compatability version bytes // to form the first 12 bytes, pad the rest with zeros // create a crude hash of the filename to generate the first 4 bytes identifier[0] = 0; identifier[1] = 0; identifier[2] = 0; identifier[3] = 0; for (int j = 0, i = (int)strlen(path_)-1; i>=0 && path_[i]!='/'; ++j, --i) { identifier[j%4] += path_[i]; } identifier[4] = (dylib_cmd.dylib.current_version >> 24) & 0xFF; identifier[5] = (dylib_cmd.dylib.current_version >> 16) & 0xFF; identifier[6] = (dylib_cmd.dylib.current_version >> 8) & 0xFF; identifier[7] = dylib_cmd.dylib.current_version & 0xFF; identifier[8] = (dylib_cmd.dylib.compatibility_version >> 24) & 0xFF; identifier[9] = (dylib_cmd.dylib.compatibility_version >> 16) & 0xFF; identifier[10] = (dylib_cmd.dylib.compatibility_version >> 8) & 0xFF; identifier[11] = dylib_cmd.dylib.compatibility_version & 0xFF; identifier[12] = (cpu_type >> 24) & 0xFF; identifier[13] = (cpu_type >> 16) & 0xFF; identifier[14] = (cpu_type >> 8) & 0xFF; identifier[15] = cpu_type & 0xFF; return true; } return false; } uint32_t MachoID::Adler32(cpu_type_t cpu_type, cpu_subtype_t cpu_subtype) { update_function_ = &MachoID::UpdateCRC; crc_ = 0; if (!WalkHeader(cpu_type, cpu_subtype, WalkerCB, this)) return 0; return crc_; } bool MachoID::MD5(cpu_type_t cpu_type, cpu_subtype_t cpu_subtype, unsigned char identifier[16]) { update_function_ = &MachoID::UpdateMD5; MD5Init(&md5_context_); if (!WalkHeader(cpu_type, cpu_subtype, WalkerCB, this)) return false; MD5Final(identifier, &md5_context_); return true; } bool MachoID::WalkHeader(cpu_type_t cpu_type, cpu_subtype_t cpu_subtype, MachoWalker::LoadCommandCallback callback, void *context) { if (memory_) { MachoWalker walker(memory_, memory_size_, callback, context); return walker.WalkHeader(cpu_type, cpu_subtype); } else { MachoWalker walker(path_, callback, context); return walker.WalkHeader(cpu_type, cpu_subtype); } } // static bool MachoID::WalkerCB(MachoWalker *walker, load_command *cmd, off_t offset, bool swap, void *context) { MachoID *macho_id = (MachoID *)context; if (cmd->cmd == LC_SEGMENT) { struct segment_command seg; if (!walker->ReadBytes(&seg, sizeof(seg), offset)) return false; if (swap) swap_segment_command(&seg, NXHostByteOrder()); struct mach_header_64 header; off_t header_offset; if (!walker->CurrentHeader(&header, &header_offset)) return false; // Process segments that have sections: // (e.g., __TEXT, __DATA, __IMPORT, __OBJC) offset += sizeof(struct segment_command); struct section sec; for (unsigned long i = 0; i < seg.nsects; ++i) { if (!walker->ReadBytes(&sec, sizeof(sec), offset)) return false; if (swap) swap_section(&sec, 1, NXHostByteOrder()); // sections of type S_ZEROFILL are "virtual" and contain no data // in the file itself if ((sec.flags & SECTION_TYPE) != S_ZEROFILL && sec.offset != 0) macho_id->Update(walker, header_offset + sec.offset, sec.size); offset += sizeof(struct section); } } else if (cmd->cmd == LC_SEGMENT_64) { struct segment_command_64 seg64; if (!walker->ReadBytes(&seg64, sizeof(seg64), offset)) return false; if (swap) breakpad_swap_segment_command_64(&seg64, NXHostByteOrder()); struct mach_header_64 header; off_t header_offset; if (!walker->CurrentHeader(&header, &header_offset)) return false; // Process segments that have sections: // (e.g., __TEXT, __DATA, __IMPORT, __OBJC) offset += sizeof(struct segment_command_64); struct section_64 sec64; for (unsigned long i = 0; i < seg64.nsects; ++i) { if (!walker->ReadBytes(&sec64, sizeof(sec64), offset)) return false; if (swap) breakpad_swap_section_64(&sec64, 1, NXHostByteOrder()); // sections of type S_ZEROFILL are "virtual" and contain no data // in the file itself if ((sec64.flags & SECTION_TYPE) != S_ZEROFILL && sec64.offset != 0) macho_id->Update(walker, header_offset + sec64.offset, (size_t)sec64.size); offset += sizeof(struct section_64); } } // Continue processing return true; } // static bool MachoID::UUIDWalkerCB(MachoWalker *walker, load_command *cmd, off_t offset, bool swap, void *context) { if (cmd->cmd == LC_UUID) { struct breakpad_uuid_command *uuid_cmd = (struct breakpad_uuid_command *)context; if (!walker->ReadBytes(uuid_cmd, sizeof(struct breakpad_uuid_command), offset)) return false; if (swap) breakpad_swap_uuid_command(uuid_cmd, NXHostByteOrder()); return false; } // Continue processing return true; } // static bool MachoID::IDWalkerCB(MachoWalker *walker, load_command *cmd, off_t offset, bool swap, void *context) { if (cmd->cmd == LC_ID_DYLIB) { struct dylib_command *dylib_cmd = (struct dylib_command *)context; if (!walker->ReadBytes(dylib_cmd, sizeof(struct dylib_command), offset)) return false; if (swap) swap_dylib_command(dylib_cmd, NXHostByteOrder()); return false; } // Continue processing return true; } } // namespace MacFileUtilities