On Fri, Aug 30, 2002 at 03:40:36PM +0100, David Laight wrote:

> Why? What are you timing?

Delays between two bytes send by the mouse while we expect more to come
for the same packet. The threshold originaly was ~5 ms, but we now moved
to 25ms, and rounding that up to 30 (with hz=100) would not do any harm,
I guess.

> > Since pmsinput is called from pckbcintr, which is a regular interrupt
> > handler, we will not have cli'd, so the sti in microtime does no harm.

> It certainly does some harm somewhere - since I had to fix that
> codepath in my faster microtime.

That may well be true, but if the call to microtime from pmsinput would cause
this, there would be something more elementary broken. We are not Linux,
so interrupt handlers do not run with all interrupts disabled.

> It won't help if you trully need better than 10ms accuracy.

I have a working callout based version now, will send patches for review
in a few minutes.

Martin