> These cahnges seem fairly simple and sane. They should drop into the zs
> driver fairly well, too.
> [ ... ]
> Oh, how about 3. :-)  Maybe throw the variables into the tty structure
> while leaving the ioctl code in the com driver.  That way there's an M.I.
> way to get at the timestamps and little duplicated code, but a driver
> which doesn't support it will reject the ioctl, never setting
> dcd_timestamping.

This sounds like a Good Idea, to me, if something like this is going
to be done.

It's also worth noting that I think the existing interface is totally
broken, and should be fixed.

Enabling/disabling timestamp collection should be done separately
(i.e. different ioctl) from getting the timestamp. (how do you
disable, now, anyway?)

Is this an interface that lots of other systems have (i.e. is it a
common, accepted interface), or just FreeBSD?

If the former, well, we should probably stick with it.  If the latter,
it probably makes sense to fix it, if only so that time stamping can
be disabled.


cgd