I think, you should use another word.

What you mean is a timout handler.

A watchdog was used to recover, if the software or hardware goes crazy.

A timeout handler was used to solve the state, that a task was not finished during a certain time.

E.g. on establish an Ethernet connection you need such a timeout handler to handle the state, that a packet was lost and a retransmission must be done.


But typically another approach was used:

A sheduler was started if you give him the time value which must expire and the address of the timeout handler routine.

And if the task was finished before the time was gone, the timeout handler was removed from the sheduler.

The advantage of this solution is, you can add several different timeout handlers to the sheduler. And you need no crazy (and dangerous) stack manipulations.


Peter