Question

Hello,

I didn't find in the user manual information on the watchdog capability of the RUT955.

How to automatically restart a hung RUT955 ?

Answer 1

Hi,

This watchdog is implemented in SoC (it seems RUT955 is using atheros AR9334). This watchdog, when it is set up, has to be pinged in certain period of time.

You can check its parameters in ubus: ubus call system watchdog '{"stop": false}'

ubus call system watchdog '{"stop": true}' WILL STOP watchdog pinging, and system will restart in about 30 seconds.

If watchdog has not been pinged in "timeout" time (Userspace process does this every "frequency" seconds), that means system is in bad state, and it restarts. It has to be very major system crash, to watchdog to restart the system. If some of the processeses  are hung, but watchdog process is still working, system will not restart.

Comments

Is it possible to ping this watchdog from outside and/or generate an alarm ?

---

Without changes in sdk - no, you have to make changes in sdk sources, and build executable

procd opens watchdog device file in /dev/watchdog, and no other process can open it, even if it is stopped by ubus command. Procd is in charge of pinging it. 

Here is a useful link to custom watchdog project on similar SoC: https://github.com/sudomesh/watchpuppy

However, modified version of procd is needed, which you have to build from sdk