> Could someone test this and tell me whether or not it works?  I don't
> have a PS/2 mouse.
> 

It's not quite right yet.  As posted, the driver always locks up my keyboard
when a process reading from /dev/pms0 exits (on two different PC's with
different keyboards).

I noticed a difference between this new driver and the one from Shoji Yuen that
I've been using up til now is that in the pmsclose routine the order the
various pms_*_cmd routines were called is different.  By reverting to the order
used by Shoji, that problem went away.  Ie, the code in pmsclose now looks
like:

       /* Disable interrupts. */
       pms_dev_cmd(sc->sc_iobase, PMS_DEV_DISABLE);
       pms_pit_cmd(sc->sc_iobase, PMS_INT_DISABLE);
       pms_aux_cmd(sc->sc_iobase, PMS_AUX_DISABLE);

Unfortunately, I still can't make it work under X (v2.1.1), specifying both
"busmouse" and "logitech".  The mouse just doesn't move at all.

Interestingly, doing a "cat -u -v /dev/pms0" seems to show the correct byte
sequences are being generated (certainly different from the sequences generated
by Shoji's driver, and they seem to match with what the lms driver produces).

If no one else solves this before tomorrow (my time!) I'll dig a little deeper,
but it's late now, and I've got to build an X server binary with debugging
before I can see what's going on...

Duncan

------------------------------------------------------------------------------