|
| 1 | +from binaryninja.binaryview import BinaryView, DataVariable |
| 2 | +from binaryninja.log import Logger |
| 3 | +from binaryninja.types import IntegerType, PointerType |
| 4 | + |
| 5 | +from typing import List |
| 6 | + |
| 7 | +logger = Logger(session_id=0, logger_name="rust_string_slicer") |
| 8 | + |
| 9 | + |
| 10 | +def recover_string_slices_from_readonly_data(bv: BinaryView): |
| 11 | + if bv.arch is None: |
| 12 | + logger.log_error("Could not get architecture of current binary view, exiting") |
| 13 | + return |
| 14 | + |
| 15 | + readonly_segments = list( |
| 16 | + filter( |
| 17 | + lambda segment: segment.readable |
| 18 | + and not segment.writable |
| 19 | + and not segment.executable, |
| 20 | + bv.segments, |
| 21 | + ) |
| 22 | + ) |
| 23 | + |
| 24 | + # Obtain all data vars which are pointers to data in readonly data segments |
| 25 | + data_vars_to_ro_segment_data: List[DataVariable] = [] |
| 26 | + for _data_var_addr, candidate_string_slice_data_ptr in bv.data_vars.items(): |
| 27 | + if isinstance(candidate_string_slice_data_ptr.type, PointerType): |
| 28 | + for readonly_segment in readonly_segments: |
| 29 | + if candidate_string_slice_data_ptr.value in readonly_segment: |
| 30 | + data_vars_to_ro_segment_data.append(candidate_string_slice_data_ptr) |
| 31 | + logger.log_debug( |
| 32 | + f"Found pointer var at {candidate_string_slice_data_ptr.address:#x} ({candidate_string_slice_data_ptr}) pointing to {candidate_string_slice_data_ptr.value:#x} " |
| 33 | + ) |
| 34 | + |
| 35 | + # Try to read an integer following the data var, |
| 36 | + # and treat it as a candidate for a string slice length. |
| 37 | + for candidate_string_slice_data_ptr in data_vars_to_ro_segment_data: |
| 38 | + candidate_string_slice_len_addr = ( |
| 39 | + candidate_string_slice_data_ptr.address |
| 40 | + + candidate_string_slice_data_ptr.type.width |
| 41 | + ) |
| 42 | + |
| 43 | + # Filter out anything at the candidate address |
| 44 | + # that's already defined as any data var type which is not an integer. |
| 45 | + existing_data_var_at_candidate_string_slice_len_addr = bv.get_data_var_at( |
| 46 | + candidate_string_slice_len_addr |
| 47 | + ) |
| 48 | + if existing_data_var_at_candidate_string_slice_len_addr is not None: |
| 49 | + if not isinstance( |
| 50 | + existing_data_var_at_candidate_string_slice_len_addr.type, IntegerType |
| 51 | + ): |
| 52 | + continue |
| 53 | + |
| 54 | + candidate_string_slice_len = bv.read_int( |
| 55 | + address=candidate_string_slice_len_addr, |
| 56 | + size=bv.arch.default_int_size, |
| 57 | + sign=False, |
| 58 | + endian=bv.arch.endianness, |
| 59 | + ) |
| 60 | + |
| 61 | + logger.log_debug( |
| 62 | + f"Pointer var at {candidate_string_slice_data_ptr.address:#x} is followed by integer with value {candidate_string_slice_len:#x}" |
| 63 | + ) |
| 64 | + |
| 65 | + # Filter out any potential string slice which has length 0 |
| 66 | + if candidate_string_slice_len == 0: |
| 67 | + continue |
| 68 | + |
| 69 | + # Attempt to read out the pointed to value as a string slice, with the length obtained above. |
| 70 | + candidate_string_slice = bv.read( |
| 71 | + addr=candidate_string_slice_data_ptr.value, |
| 72 | + length=candidate_string_slice_len, |
| 73 | + ) |
| 74 | + |
| 75 | + logger.log_debug( |
| 76 | + f"Obtained candidate string slice with addr {candidate_string_slice_data_ptr.value:#x}, len {candidate_string_slice_len:#x}: {candidate_string_slice}" |
| 77 | + ) |
| 78 | + |
| 79 | + # Sanity check whether the recovered string is valid UTF-8 |
| 80 | + try: |
| 81 | + candidate_utf8_string = candidate_string_slice.decode("utf-8") |
| 82 | + logger.log_info( |
| 83 | + f'Recovered string at addr {candidate_string_slice_data_ptr.value:#x}, len {candidate_string_slice_len:#x}: "{candidate_utf8_string}"' |
| 84 | + ) |
| 85 | + except UnicodeDecodeError as err: |
| 86 | + logger.log_warn( |
| 87 | + "Candidate string slice {candidate_string_slice} does not decode to a valid UTF-8 string; excluding from final results: {err}" |
| 88 | + ) |
| 89 | + continue |
| 90 | + |
| 91 | + |
| 92 | +recover_string_slices_from_readonly_data(bv) |
0 commit comments