tools: wait for the freeze and capture the evidence at the moment it lands

Catching the in-mission freeze by hand costs a tool call every 25 s, and it
arrives anywhere between ten seconds and never - roughly half the runs that reach
flight freeze and the other half do not. freeze_watch.sh polls frozen.py, checks
the flight HUD is still on screen (the GAME OVER screen ANIMATES, so a run that
ended there is not a freeze and must not be scored as one), and on detection
snapshots what the stuck-wait probe has said.

Taking the probe snapshot at the moment of detection matters: the comparison
against the healthy-run baseline - one thread polling one Event at BE56BB5C - is
then made from the same instant rather than reconstructed afterwards from a log
that kept growing.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
This commit is contained in:
Sylpheed RE agent
2026-08-24 02:58:23 +00:00
parent 4b20b36919
commit 5542729802

View File

@@ -0,0 +1,48 @@
#!/usr/bin/env bash
# Wait for the in-mission freeze and capture the evidence at the moment it lands.
#
# Written because catching it by hand costs a tool call every 25 s and the freeze
# takes anywhere from ten seconds to never: roughly half the runs that reach
# flight freeze, and the other half do not
# (docs/re/mission-freeze-resume-spin.md).
#
# "Frozen" is two screenshots byte-identical (frozen.py); it is checked ONLY
# while the flight HUD is still on screen, because the GAME OVER screen animates
# and a run that ended there is not a freeze.
#
# On detection it snapshots what the stuck-wait probe has said, so the comparison
# against the healthy-run baseline (one thread polling one Event) is made from
# the same instant rather than reconstructed later.
set -u
export HOME=/sylph-home/re
DISP="${DISPLAY:-:98}"
SD="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
LOG="${1:-/sylph-home/re/shots/probe-canary.stdout}"
OUT="${2:-/sylph-home/re/logs/freeze-evidence.txt}"
DEADLINE=$(( SECONDS + ${3:-1800} ))
: > "$OUT"
while [ $SECONDS -lt $DEADLINE ]; do
if ! pgrep -x xenia_canary >/dev/null; then
echo "EMULATOR GONE at ${SECONDS}s" | tee -a "$OUT"; exit 4
fi
if python3 "$SD/frozen.py" 5 >/dev/null; then
if python3 -c "import sys; sys.path.insert(0,'$SD'); import frozen; sys.exit(0 if frozen.in_flight() else 1)"; then
{
echo "FROZEN IN FLIGHT at ${SECONDS}s"
echo
echo "=== stuck-wait probe: (thread, object) pairs ==="
grep -oE 'thread [0-9A-F]+ has timed out [0-9]+ times in a row on object [0-9A-F]+ \(type [0-9]+\)' "$LOG" \
| sed -E 's/has timed out [0-9]+ times in a row //' | sort | uniq -c | sort -rn
echo
echo "=== last 12 probe lines ==="
grep 'wait-probe' "$LOG" | tail -12
} | tee -a "$OUT"
exit 0
fi
echo "frozen but NOT in flight (mission over) at ${SECONDS}s" | tee -a "$OUT"
exit 5
fi
sleep 20
done
echo "NO FREEZE within ${3:-1800}s" | tee -a "$OUT"; exit 1