steam-backlog-enforcer/steam_backlog_enforcer/_screen_hold.py
Krzysztof kuhy Rudnicki b3e629e3f0 Bill gaming time only while the user is actually playing
A running game process was treated as proof of play. On 2026-08-28 the budget
billed ~2h23m while a gatelock screen locker covered the display and the
machine could not be touched at all -- the game was simply still resident.

A tick now bills only when all three hold, evaluated together rather than
short-circuited so the audit log can answer "was the screen ALSO locked?":

  * no gatelock guard owns the screen  (_screen_hold, via /proc/locks)
  * the focused window belongs to a qualifying process or its ancestry
  * some input source has been used within idle_grace_seconds (default 300)

Any probe that cannot answer BILLS the tick and is recorded in `degraded`.
A broken detector must never become an unlimited-gaming exploit.

Three findings from measuring this machine rather than assuming:

  * /dev/input/js0 here is an *ASRock LED Controller* -- the kernel exposes
    the motherboard RGB device as a joystick. A presence check would have
    disabled the idle criterion permanently, so _controller_input watches for
    real events instead (skipping joydev's synthetic open burst).
  * Aseprite's window carries no _NET_WM_PID at all, so focus attribution
    falls back to XRes, which asks the server who owns the window rather than
    trusting the client to have announced itself.
  * /proc/locks reports gatelock's hold without taking it; gatelock's own
    read_claim_if_held detects held-ness by acquiring, which would race a
    guard every 3 seconds.

Idle pauses also refund the grace period once per walk-away, on the
engaged->paused edge only, never across a day rollover, and never on the first
tick after a restart -- the only place the counter moves backwards.

Everything is recorded to /var/log/steam-backlog-enforcer/budget.jsonl, on
change plus a heartbeat rather than one line per 3s tick. Records name the
qualifying PIDs: reconstructing this incident from journald alone was only
possible by inferring accrual from the spacing of two warning notifications.

Also: refuse to bounce Steam while a game is running. ensure_steam_debug_port
relaunches Steam when the CDP port is closed, which kills the running game --
that is how restarting the daemon to deploy this change ended a live session.
It now defers exactly as it already did for in-flight game updates.

Talks to X through ctypes over one cached connection: at a 3s tick, forking
xprintidle and xprop would cost ~40 forks a minute.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019CEsG6BzHtYR9cHcdxPZqG
2026-08-28 19:24:14 +02:00

125 lines
3.6 KiB
Python

"""Detecting that a gatelock guard currently owns the screen.
The screen lockers (``screen_locker``, ``leetcode_guard``) coordinate through a
single advisory ``flock`` on ``$XDG_RUNTIME_DIR/gatelock/holder.lock``. The file's
*contents* cannot answer "is it held right now" — a stale record left by a dead
process is byte-identical to a live one — and gatelock's own
``read_claim_if_held`` answers the question by *taking* the lock.
The enforcer must not do that. Taking the lock every three seconds, even for
microseconds, would race a guard trying to claim the screen. ``/proc/locks``
reports the same fact by observation, so this reads that instead and never
touches the file.
Line format is ``idx: FLOCK ADVISORY WRITE <pid> <maj:min:inode> <start> <end>``,
with major and minor in hex. Lines containing ``->`` describe processes *waiting*
for a lock, not holding one, and are skipped.
"""
from __future__ import annotations
from dataclasses import dataclass
import os
from pathlib import Path
import re
from typing import Final
_PROC_LOCKS: Final = Path("/proc/locks")
_HOLDER_RELATIVE: Final = Path("gatelock/holder.lock")
# `maj:min:inode`, major and minor hex, inode decimal.
_DEVICE_FIELD: Final = re.compile(r"^[0-9a-f]+:[0-9a-f]+:\d+$")
_WAITER_MARKER: Final = "->"
_PID_FIELD_INDEX: Final = 4
class ScreenHoldError(RuntimeError):
"""Whether the screen is held could not be determined."""
@dataclass(frozen=True)
class ScreenHold:
"""Whether a guard owns the screen, and which process does."""
held: bool
holder_pid: int | None = None
def holder_lock_path(uid: int) -> Path:
"""Return the gatelock holder-lock path for *uid*.
Args:
uid: Owner of the desktop session.
Returns:
Path to ``holder.lock``.
"""
return Path(f"/run/user/{uid}") / _HOLDER_RELATIVE
def screen_hold(path: Path) -> ScreenHold:
"""Report whether *path*'s advisory lock is currently held.
Args:
path: The gatelock holder-lock file.
Returns:
The hold state, with the holding PID when there is one.
Raises:
ScreenHoldError: If ``/proc/locks`` cannot be read. A missing *path* is
not an error — it means no guard has ever run this boot.
"""
try:
stat = path.stat()
except OSError:
return ScreenHold(held=False)
target = f"{os.major(stat.st_dev):02x}:{os.minor(stat.st_dev):02x}:{stat.st_ino}"
try:
raw = _PROC_LOCKS.read_text(encoding="utf-8")
except OSError as exc:
msg = f"cannot read {_PROC_LOCKS}: {exc}"
raise ScreenHoldError(msg) from exc
return _find_holder(raw, target)
def _find_holder(raw: str, target: str) -> ScreenHold:
"""Scan ``/proc/locks`` text for a live lock on *target*.
Args:
raw: Contents of ``/proc/locks``.
target: ``maj:min:inode`` string to match.
Returns:
The hold state.
"""
for line in raw.splitlines():
fields = line.split()
if _WAITER_MARKER in fields:
continue
if not any(field == target for field in fields if _DEVICE_FIELD.match(field)):
continue
return ScreenHold(held=True, holder_pid=_pid_of(fields))
return ScreenHold(held=False)
def _pid_of(fields: list[str]) -> int | None:
"""Return the PID column of a ``/proc/locks`` row.
Args:
fields: Whitespace-split row.
Returns:
The PID, or ``None`` if the row is shaped unexpectedly.
"""
if len(fields) <= _PID_FIELD_INDEX:
return None
try:
return int(fields[_PID_FIELD_INDEX])
except ValueError:
return None