Skip to content

thingino-motors: fix S59motor stop path wedging motors-daemon #1511

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
catongates wants to merge 1 commit into
base: master
Choose a base branch
from
Open

thingino-motors: fix S59motor stop path wedging motors-daemon #1511

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
62 changes: 57 additions & 5 deletions package/thingino-motors/files/S59motor
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,15 +16,58 @@ start_daemon() {
start-stop-daemon -S -b -m -p "$PIDFILE" -x "$DAEMON" -- $@
}

# Returns 0 only if the daemon is genuinely gone afterwards, 1 otherwise.
# stop() depends on that contract -- see the rmmod guard there.
#
# The daemon self-daemonizes (double-fork + setsid under -d, see
# motor-daemon.c), so the PID start-stop-daemon records via -m is the
# intermediate child, which exits immediately: the pidfile is stale from
# the moment it is written and -K can never match the real process.
# Confirmed live -- this used to print "no /usr/bin/motors-daemon found;
# none killed" while ps plainly showed the daemon running, after which
# stop() went on to rmmod the module out from under it. Because motor_fops
# in ingenic-sdk's motor.c has no .owner = THIS_MODULE (see
# themactep/ingenic-sdk#45), opening /dev/motor never takes a module
# reference (/proc/modules shows use count 0 with a live opener), so that
# rmmod SUCCEEDS and frees the module's code, IRQ handler and completions
# under the running daemon -- leaving it in permanently uninterruptible D
# state, immune to SIGKILL, recoverable only by a power cycle.
#
# Match anchored on the binary path: an unanchored -f pattern also
# matches any shell/ssh command line that merely mentions the path
# (verified -- an unanchored pgrep matched this script's own caller).
stop_daemon() {
if [ -f "$PIDFILE" ]; then
start-stop-daemon -K -p "$PIDFILE" -x "$DAEMON"
status=$?
start-stop-daemon -K -q -p "$PIDFILE" -x "$DAEMON" 2>/dev/null
rm -f "$PIDFILE"
return $status
fi

start-stop-daemon -K -x "$DAEMON"
if ! pgrep -f "^$DAEMON" >/dev/null 2>&1; then
return 0
fi

pkill -TERM -f "^$DAEMON" 2>/dev/null

# 5s for a graceful exit, then escalate. Fractional sleep is
# supported by this firmware's busybox (verified).
i=0
while [ "$i" -lt 25 ]; do
pgrep -f "^$DAEMON" >/dev/null 2>&1 || return 0
i=$((i + 1))
sleep 0.2
done

echo "motors-daemon ignored SIGTERM, escalating to SIGKILL" >&2
pkill -KILL -f "^$DAEMON" 2>/dev/null

i=0
while [ "$i" -lt 25 ]; do
pgrep -f "^$DAEMON" >/dev/null 2>&1 || return 0
i=$((i + 1))
sleep 0.2
done

return 1
}

MOTORS_CONFIG="/etc/thingino.json"
Expand Down Expand Up @@ -173,7 +216,16 @@ start() {
stop() {
echo -c 15 "Stopping motors"

stop_daemon
# Refuse to unload the module while anything still holds /dev/motor.
# The kernel will NOT stop us -- motor_fops lacks .owner = THIS_MODULE
# (themactep/ingenic-sdk#45), so the module's use count stays 0 even
# with a live opener and rmmod succeeds, orphaning that opener in
# unkillable D state (power cycle only). Failing loudly here is
# strictly better than that.
if ! stop_daemon; then
echo "motors-daemon still running; refusing to rmmod (would wedge it)" >&2
exit 1
fi

if ! rmmod motor; then
echo "Failed to unload motor module." >&2
Expand Down