Kernel Summit 2006: Realtime
Posted Jul 31, 2006 16:44 UTC (Mon) by Blaisorblade
In reply to: Kernel Summit 2006: Realtime
Parent article: Kernel Summit 2006: Realtime
There is already a soft watchdog in the kernel almost exactly like that - the only difference from your proposal is that instead of one high-priority process it is the timer interrupt to do the check; I also think that since the time interrupt should remain a raw interrupt without a task checking for it (it's used for the scheduler) it is the same thing.
And both things aren't safe when a process disables interrupts but that's not interesting (you really need a hardware watchdog for that).
to post comments)