naywatch: fix procd handling

Sometimes, naywatch can not handover the watchdog to procd again using
ubus. We need to call the same ubus command multiple times until procd
takes over control again.

Signed-off-by: Nick Hainke <vincent@systemli.org>

naywatch/files/naywatch.init

@@ -1,5 +1,7 @@
 #!/bin/sh /etc/rc.common
 
+. /usr/share/libubox/jshn.sh
+
 USE_PROCD=1
 START=95
 STOP=01
@@ -58,8 +60,31 @@ stop_service() {
     sync && wait
 }
 
+watchdog_procd_runnig() {
+	watch=$(ubus call system watchdog)
+	json_load "$watch"
+	json_get_var watchdogstatus status
+	if [[ "$watchdogstatus" == "running" ]] ; then
+		echo "1"
+        return
+	fi
+    echo "0"
+}
+
 service_stopped() {
     log "Naywatch Stopped!"
-    log "Handover Watchdog to procd again:"
-    ubus call system watchdog '{"magicclose":true,"stop":false}' > /dev/null
+    log "Try to handover watchdog to procd again."
+    for i in 1 2 3 4 5 6 7 8 9 10 ; do
+        sleep 5
+        ubus call system watchdog '{"stop":false}' > /dev/null
+        if [[ $(watchdog_procd_runnig) == "1" ]] ; then
+            log "procd running"
+            break
+        fi
+    done
+    if [[ $(watchdog_procd_runnig) == "1" ]] ; then
+        log "Handover sucessfully!"
+    else
+        log "Handover to procd failed! Device can reboot!"
+    fi
 }