>...why not just call it cycletime()?  That should get us into the next
>century or so.


We have all these nice POSIX interfaces, with nanosecond resolution,
betwween userspace and the kernel.  Once these nanosecond-resolution
values get into the kernel, we throw away three orders of magnitude of
resolution by keeping everything internally in microseconds.

Cleaning up the in-kernel interfaces is a Good Thing, but we should
provide nanosecond resolution internally (in the kernel) from the
get-go.

In a separate but related issue, microsecond-resolution NTP code is
... getting pretty lame, by today's standards.  (Cycletime(), you say?
On which CPU, given an SMP system with multiple distinct CPU-clock
frequencies?)