|
| 1 | +import ctypes |
| 2 | +import ctypes.util |
| 3 | +import re |
| 4 | +from logging import getLogger |
| 5 | +from typing import Generator, NamedTuple, List, Optional, Union |
| 6 | +from ptrlib.binary.encoding import str2bytes |
| 7 | + |
| 8 | +logger = getLogger(__name__) |
| 9 | + |
| 10 | + |
| 11 | +class LinuxMemoryRegion(NamedTuple): |
| 12 | + start: int |
| 13 | + end: int |
| 14 | + perm: str |
| 15 | + offset: int |
| 16 | + path: Optional[str] |
| 17 | + |
| 18 | + def __repr__(self): |
| 19 | + return f"LinuxMemoryRegion('{str(self)}')" |
| 20 | + |
| 21 | + def __str__(self): |
| 22 | + return f"0x{self.start:016x}-0x{self.end:016x} {self.perm} {self.path}" |
| 23 | + |
| 24 | +class LinuxProcessMemory(object): |
| 25 | + """Memory inspector for Linux |
| 26 | + """ |
| 27 | + def __init__(self, pid: int, sudo: bool=True): |
| 28 | + """ |
| 29 | + Args: |
| 30 | + pid (int): Process ID to attach |
| 31 | + sudo (bool): If this parameter is set to true and a permission error occurs, ask pkexec prompt to get root privilege (Default is true) |
| 32 | + """ |
| 33 | + self._libc = ctypes.CDLL(ctypes.util.find_library("c"), use_errno=True) |
| 34 | + self._pid = pid |
| 35 | + |
| 36 | + self._libc.strerror.argtypes = [ctypes.c_int] |
| 37 | + self._libc.strerror.restype = ctypes.c_char_p |
| 38 | + |
| 39 | + self._libc.process_vm_readv.argtypes = [ |
| 40 | + ctypes.c_int, |
| 41 | + ctypes.POINTER(ctypes.c_void_p), |
| 42 | + ctypes.c_ulong, |
| 43 | + ctypes.POINTER(ctypes.c_void_p), |
| 44 | + ctypes.c_ulong, |
| 45 | + ctypes.c_ulong, |
| 46 | + ] |
| 47 | + self._libc.process_vm_readv.restype = ctypes.c_ssize_t |
| 48 | + |
| 49 | + self._libc.process_vm_writev.argtypes = self._libc.process_vm_readv.argtypes |
| 50 | + self._libc.process_vm_writev.resype = ctypes.c_ssize_t |
| 51 | + |
| 52 | + @property |
| 53 | + def vmmap(self) -> List[LinuxMemoryRegion]: |
| 54 | + maps = [] |
| 55 | + with open(f"/proc/{self._pid}/maps", "r") as f: |
| 56 | + for line in f: |
| 57 | + m = re.match(r"^([0-9a-f]+)-([0-9a-f]+) ([-rwxsp]+) ([0-9a-f]+) \S+ \d+\s+(.*)", line) |
| 58 | + start = int(m.groups()[0], 16) |
| 59 | + end = int(m.groups()[1], 16) |
| 60 | + perm = m.groups()[2] |
| 61 | + offset = int(m.groups()[3], 16) |
| 62 | + path = m.groups()[4] |
| 63 | + maps.append(LinuxMemoryRegion(start, end, perm, offset, path)) |
| 64 | + return maps |
| 65 | + |
| 66 | + def read(self, addr: int, size: int) -> bytes: |
| 67 | + """Attempt to read memory |
| 68 | +
|
| 69 | + Args: |
| 70 | + addr (int): Remote address to read data from |
| 71 | + size (int): Size to read |
| 72 | +
|
| 73 | + Returns: |
| 74 | + bytes: Data read from the memory |
| 75 | + """ |
| 76 | + e1 = e2 = None |
| 77 | + |
| 78 | + # 1. /proc/pid/mem is the most reliable |
| 79 | + try: |
| 80 | + return self.proc_mem_read(addr, size) |
| 81 | + except OSError as e: |
| 82 | + e2 = e |
| 83 | + |
| 84 | + # 2. process_vm_readv can bypass anti-debug |
| 85 | + try: |
| 86 | + return self.process_vm_read(addr, size) |
| 87 | + except OSError as e: |
| 88 | + e1 = e |
| 89 | + |
| 90 | + # 3. TODO: ptrace |
| 91 | + raise e1 or e2 |
| 92 | + |
| 93 | + def write(self, addr: int, data: bytes) -> int: |
| 94 | + """Attempt to write memory |
| 95 | +
|
| 96 | + Args: |
| 97 | + addr (int): Remote address to write data to |
| 98 | + data (bytes): Data to write |
| 99 | +
|
| 100 | + Returns: |
| 101 | + int: Number of bytes written to the memory |
| 102 | + """ |
| 103 | + e1 = e2 = None |
| 104 | + |
| 105 | + # 1. /proc/pid/mem is the most reliable |
| 106 | + try: |
| 107 | + return self.proc_mem_write(addr, data) |
| 108 | + except OSError as e: |
| 109 | + e2 = e |
| 110 | + |
| 111 | + # 2. process_vm_writev can bypass anti-debug |
| 112 | + try: |
| 113 | + return self.process_vm_write(addr, data) |
| 114 | + except OSError as e: |
| 115 | + e1 = e |
| 116 | + |
| 117 | + # 3. TODO: ptrace |
| 118 | + raise e1 or e2 |
| 119 | + |
| 120 | + def search(self, |
| 121 | + data: Union[str, bytes], |
| 122 | + start: Optional[int]=None, |
| 123 | + end: Optional[int]=None, |
| 124 | + length: Optional[int]=None) -> Generator[int, None, None]: |
| 125 | + """Search for memory |
| 126 | +
|
| 127 | + Args: |
| 128 | + data (bytes): Data to search |
| 129 | + start (int): Lower bound for search |
| 130 | + end (int): Upper bound for search |
| 131 | + len (int): Length of region to search (Requires either `start` or `end`) |
| 132 | +
|
| 133 | + Returns: |
| 134 | + generator: A generator to yield matching addresses |
| 135 | + """ |
| 136 | + if isinstance(data, str): |
| 137 | + data = str2bytes(data) |
| 138 | + |
| 139 | + if length is not None: |
| 140 | + if (start or end) is None: |
| 141 | + raise ValueError("`len` is specified but neither `start` nor `end` is set") |
| 142 | + elif start is None: |
| 143 | + start = end - length |
| 144 | + else: |
| 145 | + end = start + length |
| 146 | + |
| 147 | + if start is None: |
| 148 | + start = 0 |
| 149 | + if end is None: |
| 150 | + end = 1 << 64 # Works fine both for 32-bit and 64-bit |
| 151 | + |
| 152 | + prev_end = -1 |
| 153 | + region_start = 0 |
| 154 | + offset = 0 |
| 155 | + haystack = b'' |
| 156 | + for mem in self.vmmap: |
| 157 | + if mem.end <= start or mem.start > end: |
| 158 | + continue |
| 159 | + elif mem.start >= 0x8000_0000_0000: # Skip non-canonical and kernel memory |
| 160 | + continue |
| 161 | + elif mem.path == '[vvar]': # NOTE: Heuristic skip |
| 162 | + continue |
| 163 | + |
| 164 | + if mem.start != prev_end: |
| 165 | + region_start = mem.start |
| 166 | + offset = 0 |
| 167 | + haystack = b'' |
| 168 | + prev_end = mem.end |
| 169 | + |
| 170 | + # Search page by page |
| 171 | + for addr in range(mem.start, mem.end, 0x1000): |
| 172 | + try: |
| 173 | + haystack += self.read(addr, 0x1000) |
| 174 | + except OSError as e: |
| 175 | + logger.warning(f"Could not read memory ({addr}, {addr+0x1000}): {e}") |
| 176 | + continue |
| 177 | + |
| 178 | + # TODO: Implement stream KMP |
| 179 | + while True: |
| 180 | + found = haystack.find(data, offset) |
| 181 | + if found == -1: break |
| 182 | + yield region_start + found |
| 183 | + offset = found + 1 |
| 184 | + |
| 185 | + if offset <= len(haystack) - len(data): |
| 186 | + offset = len(haystack) - len(data) + 1 |
| 187 | + |
| 188 | + def proc_mem_read(self, addr: int, size: int): |
| 189 | + """Read memory with using /proc/pid/mem |
| 190 | +
|
| 191 | + Args: |
| 192 | + addr (int): Remote address to read data from |
| 193 | + size (int): Size to read |
| 194 | +
|
| 195 | + Returns: |
| 196 | + bytes: Data read from the memory |
| 197 | + """ |
| 198 | + with open(f"/proc/{self._pid}/mem", "rb") as f: |
| 199 | + f.seek(addr, 0) |
| 200 | + return f.read(size) |
| 201 | + |
| 202 | + def proc_mem_write(self, addr: int, data: Union[str, bytes]) -> int: |
| 203 | + """Write memory with using process_vm_writev |
| 204 | +
|
| 205 | + Args: |
| 206 | + addr (int): Remote address to write data to |
| 207 | + data (bytes): Data to write |
| 208 | +
|
| 209 | + Returns: |
| 210 | + int: Number of bytes written to the memory |
| 211 | + """ |
| 212 | + if isinstance(data, str): |
| 213 | + data = str2bytes(data) |
| 214 | + |
| 215 | + with open(f"/proc/{self._pid}/mem", "wb") as f: |
| 216 | + f.seek(addr, 0) |
| 217 | + return f.write(data) |
| 218 | + |
| 219 | + def process_vm_read(self, addr: int, size: int) -> bytes: |
| 220 | + """Read memory with using process_vm_readv |
| 221 | +
|
| 222 | + Args: |
| 223 | + addr (int): Remote address to read data from |
| 224 | + size (int): Size to read |
| 225 | +
|
| 226 | + Returns: |
| 227 | + bytes: Data read from the memory |
| 228 | + """ |
| 229 | + buf = ctypes.create_string_buffer(size) |
| 230 | + local_iov = (ctypes.c_void_p * 2)(ctypes.addressof(buf), size) |
| 231 | + remote_iov = (ctypes.c_void_p * 2)(addr, size) |
| 232 | + |
| 233 | + n_read = self._libc.process_vm_readv( |
| 234 | + self._pid, local_iov, 1, remote_iov, 1, 0 |
| 235 | + ) |
| 236 | + if n_read == -1: |
| 237 | + e = ctypes.get_errno() |
| 238 | + s = self._libc.strerror(e).decode() |
| 239 | + raise OSError(e, f"process_vm_readv failed: {s}") |
| 240 | + |
| 241 | + return buf.raw |
| 242 | + |
| 243 | + def process_vm_write(self, addr: int, data: bytes) -> int: |
| 244 | + """Write memory with using process_vm_writev |
| 245 | +
|
| 246 | + Args: |
| 247 | + addr (int): Remote address to write data to |
| 248 | + data (bytes): Data to write |
| 249 | +
|
| 250 | + Returns: |
| 251 | + int: Number of bytes written to the memory |
| 252 | + """ |
| 253 | + buf = ctypes.create_string_buffer(data) |
| 254 | + local_iov = (ctypes.c_void_p * 2)(ctypes.addressof(buf), len(data)) |
| 255 | + remote_iov = (ctypes.c_void_p * 2)(addr, len(data)) |
| 256 | + |
| 257 | + n_written = self._libc.process_vm_writev( |
| 258 | + self._pid, local_iov, 1, remote_iov, 1, 0 |
| 259 | + ) |
| 260 | + if n_written == -1: |
| 261 | + e = ctypes.get_errno() |
| 262 | + s = self._libc.strerror(e).decode() |
| 263 | + raise OSError(e, f"process_vm_writev failed: {s}") |
| 264 | + |
| 265 | + return n_written |
| 266 | + |
0 commit comments