Hi,

from the embOS manual:

pfResetFunc may be used to perform additional operations inside a callback function
prior to the reset of the microcontroller.

The reset function is optional and gives the chance to react on such a situation.
You could e.g. write a message in a log file or set a piece of hardware in safe state before resetting the μC.

The basic idea is that one hardware watchdog is not enough for an RTOS with multiple tasks.
If you would trigger the hardware watchdog in one task only all other tasks are not checked and you don't know if they are still running.
Therefore we have the unlimited software watchdogs which you can use with the OS_WD_*() API.
In the end a hardware watchdog is used, of course.

I hope that makes it more clear.

Best regards,
Til

Dear Til,

I don't understand your answer : "If you would
trigger the hardware watchdog in one task only all other tasks are not
checked and you don't know if they are still running."
Why Hardware watchdog reload in the idle loop is not possible ? Please explain .

In my mind, if any task is blocked, the idle loop is not reached anymore and the hardware watchdog is triggered.

What do you think about kicking the watchdog in the lowest priority task ?

Regards
Jluc

Dear JLuc,

In my mind, if any task is blocked, the idle loop is not reached anymore and the hardware watchdog is triggered.

There could be situations where one task is not running as expected but OS_Idle() still gets executed.
Also triggering the watchdog in OS_Idle() only does not help because there could be a task which runs for a long time and therefore you don't reach OS_Idle() on time.
I guess you think of situations only where the CPU "stops" in a task and no further code gets executed.
In that case the hardware watchdog would be sufficient.
But you also want to check that all tasks are running as expected.

What do you think about kicking the watchdog in the lowest priority task ?

Same here, with it you only know this task is still running but how do you want to know all other tasks are still running?

Best regards,
Til

Dear Til,

thanks a lot for these interesting answers. If I'm correct, my watchdog in Idle loop is efficient for:
-> trapped in OS_Error(),
-> infinite loops without OS_Delay() consuming all the resource,

but not the task timing.

Gone to upgrade to 4.38...
Jluc

If I'm correct, my watchdog in Idle loop is efficient for:
-> trapped in OS_Error(),
-> infinite loops without OS_Delay() consuming all the resource,

Yeah, that's it.

Usually you don't have the debug code and OS_Error() in your release firmware.
Although you can do so, of course!
But in that case you could also directly reset your microcontroller in OS_Error().