Hm.   I actually like the idea of Control/ALT/DEL sending init a
signal better than I like the idea of it just doing an immediate
reboot, now that you mention it.

The only fly in the ointment, IMHO, is that this won't get you out of
a hang - if init can't run, it can't reboot.   Maybe if there's
already a SIGUSR1 posted to init, it should call cpu_reboot after all.

			       _MelloN_