Files
mvt/src/mvt/android/artifacts/settings.py
T
Donncha Ó Cearbhaill 858496e60b End a settings record at a blank line
dumpsys prints a blank line after every namespace block and after a
change history, and the generation registry and any vendor dumps follow
the last block before the section trailer. A record ran until the next
`_id:` line, heading or trailer, so the last row of the section
absorbed those dumps into whichever field came last.

A blank line now closes the record being read; lines that follow it
without an `_id:` are skipped until the next row or heading. The fixture
carries a generation registry after its last block, modelled on the
AOSP dump, and the last row is pinned to exactly its own fields.
2026-09-05 16:17:43 +02:00

305 lines
10 KiB
Python

# Mobile Verification Toolkit (MVT)
# Copyright (c) 2021-2023 The MVT Authors.
# Use of this software is governed by the MVT License 1.1 that can be found at
# https://license.mvt.re/1.1/
import re
from datetime import datetime
from typing import Optional, Sequence
from mvt.common.module_types import ModuleAtomicResult, ModuleSerializedResult
from mvt.common.utils import convert_datetime_to_iso
from .artifact import AndroidArtifact
ANDROID_DANGEROUS_SETTINGS = [
{
"description": "disabled Google Play Services apps verification",
"key": "verifier_verify_adb_installs",
"safe_value": "1",
},
{
"description": "disabled Google Play Protect",
"key": "package_verifier_enable",
"safe_value": "1",
},
{
"description": "disabled APK package verification",
"key": "package_verifier_state",
"safe_value": "1",
},
{
"description": "disabled Google Play Protect",
"key": "package_verifier_user_consent",
"safe_value": "1",
},
{
"description": "disabled Google Play Protect",
"key": "upload_apk_enable",
"safe_value": "1",
},
{
"description": "disabled confirmation of adb apps installation",
"key": "adb_install_need_confirm",
"safe_value": "1",
},
{
"description": "disabled sharing of security reports",
"key": "send_security_reports",
"safe_value": "1",
},
{
"description": "disabled sharing of crash logs with manufacturer",
"key": "samsung_errorlog_agree",
"safe_value": "1",
},
{
"description": "disabled applications errors reports",
"key": "send_action_app_error",
"safe_value": "1",
},
{
"description": "enabled accessibility services",
"key": "accessibility_enabled",
"safe_value": "0",
},
]
# dumpsys prints the fields of a setting record, and of a change history entry,
# always in this order and separated by a single space. After the value come
# `default:` and `defaultSystemSet:` when a default is recorded, then `tag:`;
# some vendor builds add whether the value survives a restore, either as
# `isValuePreservedInRestore:` or as a bare `notPreservedInRestore` token.
SETTING_FIELDS = (
"_id",
"name",
"pkg",
"value",
"default",
"defaultSystemSet",
"tag",
"isValuePreservedInRestore",
)
HISTORY_FIELDS = ("time", "mode", "oldValue", "newValue", "package")
NAMESPACE_PATTERN = re.compile(
r"^(CONFIG|GLOBAL|SECURE|SYSTEM) SETTINGS \(user (\d+)\)$"
)
SECTION_END_PATTERN = re.compile(r"ending at: (\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2})")
class Settings(AndroidArtifact):
"""Parser for the `dumpsys settings` output.
Every row of the settings provider becomes one result, keeping the fields
dumpsys prints alongside the value: the row id, the package which recorded
the setting, the default, the tag, and the change history. A setting name
can appear more than once within a namespace, so results are a list rather
than a mapping.
"""
def serialize(self, result: ModuleAtomicResult) -> ModuleSerializedResult:
records = []
for entry in result.get("history", []):
if not entry.get("timestamp"):
continue
records.append(
{
"timestamp": entry["timestamp"],
"module": self.__class__.__name__,
"event": "settings_change",
"data": (
f"{result.get('namespace')} setting "
f'"{result.get("name")}" changed from '
f'"{entry.get("oldValue")}" to "{entry.get("newValue")}" '
f"by {entry.get('pkg')}"
),
}
)
return records
def check_indicators(self) -> None:
for result in self.results:
name = result.get("name")
value = result.get("value")
for danger in ANDROID_DANGEROUS_SETTINGS:
# Check if one of the dangerous settings is using an unsafe
# value (different than the one specified).
if danger["key"] != name or danger["safe_value"] == value:
continue
history = result.get("history") or []
self.alertstore.medium(
f'Found suspicious "{result.get("namespace")}" setting '
f'"{name} = {value}" ({danger["description"]})',
history[-1]["timestamp"] if history else "",
result,
)
break
def parse(self, content: str) -> None:
self.results: list[ModuleAtomicResult] = []
section_end = self._parse_section_end(content)
namespace: Optional[str] = None
user: Optional[str] = None
record_lines: list[str] = []
history_lines: list[str] = []
in_history = False
def flush() -> None:
nonlocal record_lines, history_lines, in_history
if record_lines:
self.results.append(
self._build_record(
namespace, user, record_lines, history_lines, section_end
)
)
record_lines = []
history_lines = []
in_history = False
for line in content.splitlines():
heading = NAMESPACE_PATTERN.match(line.strip())
if heading:
flush()
namespace = heading.group(1).lower()
user = heading.group(2)
continue
if line.startswith("--------- "):
# dumpsys closes every section with a duration trailer.
flush()
namespace = None
continue
if namespace is None:
continue
if not line.strip():
# dumpsys prints a blank line after every namespace block and
# after a change history, and other dumps such as the
# generation registry follow the last block, so a blank line
# closes the record being read.
flush()
continue
if line.startswith("_id:"):
flush()
record_lines = [line]
continue
if not record_lines:
continue
stripped = line.strip()
if stripped.startswith("History ("):
in_history = True
continue
if in_history:
if stripped.startswith("time:"):
history_lines.append(stripped)
elif stripped and history_lines:
# A history entry can be wrapped over several lines.
history_lines[-1] += " " + stripped
continue
# Anything else continues the value of the record being read.
record_lines.append(line)
flush()
@staticmethod
def _split_fields(text: str, keys: Sequence[str]) -> dict[str, str]:
"""Split the `key:value` fields of one record.
Values are free-form and may contain spaces and newlines, so a field
runs up to the start of the next key which is actually present. Keys
dumpsys did not print are skipped.
"""
fields: dict[str, str] = {}
key = keys[0]
if not text.startswith(f"{key}:"):
return fields
remainder = text[len(key) + 1 :]
for next_key in keys[1:]:
value, separator, rest = remainder.partition(f" {next_key}:")
if separator:
fields[key] = value
key, remainder = next_key, rest
fields[key] = remainder
return fields
@staticmethod
def _parse_section_end(content: str) -> Optional[datetime]:
"""Return the time the settings section was dumped, if reported."""
match = SECTION_END_PATTERN.search(content)
if not match:
return None
try:
return datetime.strptime(match.group(1), "%Y-%m-%d %H:%M:%S")
except ValueError:
return None
@staticmethod
def _resolve_timestamp(
value: str, section_end: Optional[datetime]
) -> Optional[str]:
"""Add the missing year to a `MM-DD HH:MM:SS.mmm` history timestamp.
dumpsys prints the change history without a year, so it is resolved
against the time the section was dumped: the most recent matching date
at or before that time.
"""
if section_end is None:
return None
try:
partial = datetime.strptime(value, "%m-%d %H:%M:%S.%f")
timestamp = partial.replace(year=section_end.year)
if timestamp > section_end:
timestamp = partial.replace(year=section_end.year - 1)
except ValueError:
return None
return convert_datetime_to_iso(timestamp)
def _parse_history(
self, line: str, section_end: Optional[datetime]
) -> ModuleAtomicResult:
fields = self._split_fields(line, HISTORY_FIELDS)
return {
"timestamp": self._resolve_timestamp(fields.get("time", ""), section_end),
"oldValue": fields.get("oldValue"),
"newValue": fields.get("newValue"),
"pkg": fields.get("package"),
}
def _build_record(
self,
namespace: Optional[str],
user: Optional[str],
record_lines: list[str],
history_lines: list[str],
section_end: Optional[datetime],
) -> ModuleAtomicResult:
text = "\n".join(record_lines).rstrip()
# The bare `notPreservedInRestore` token has no `key:` shape and is
# printed last, so peel it off before splitting the fields.
head = text.removesuffix(" notPreservedInRestore")
record: ModuleAtomicResult = {"namespace": namespace, "user": user}
record.update(self._split_fields(head, SETTING_FIELDS))
if head != text:
record["isValuePreservedInRestore"] = "false"
record["history"] = [
self._parse_history(entry, section_end) for entry in history_lines
]
return record