Tool/software:
Hi there,
The steps that we used for watchdog are as follow:
1.Enable watchdog in the chip with a 30 Sec timeout period.
2.Ping the watchdog within 30 Sec when SW is functional.
3.When the "reboot" Linux system call is issued from user space the kernel driver will do the following tasks
3-1 A callback function for the 'reboot' command will be invoked which will write data to the watchdog HW module during its "windows closed period" and this should trigger system reboot immediately.
3-2 Stop pinging the watchdog after receiving the "reboot" command.
We proceeded step 1~2 for a few moment then performing step 3 to reboot the system.
We found that our system could be rebooted dozens of times (e.g. 15 times) and finally it did not be rebooted.
I think if step 3-1 didn't make it reboot then step 3-2 would also cause system to reboot finally.
Thanks,
Sean