> [...] we've got to prevent processes taking one more fault per lifetime.
it's just now that i realize that we dont take this extra fault in
'normal' processes, as signal return state is set up with HUGE_CS? Cool,
and ignore my comments ...
-- mingo