It looks like Scott Reynolds changed the timing in adb.c so there
is a longer delay for adb_op_sync().  With the longer delay,
everything just works for me:

adb0 (direct, II series): 3 targets
aed0 at adb0 addr 0: ADB Event device
akbd0 at adb0 addr 2: extended keyboard
wskbd0 at akbd0: console keyboard, using wsdisplay0
ams0 at adb0 addr 3: Mouse Systems A3 mouse, 3-button, 300 dpi
wsmouse0 at ams0
akbd1 at adb0 addr 2: extended keyboard
wskbd1 at akbd1: console keyboard, using wsdisplay0

Thanks, Scott.

Bob,
I'm not sure what's up with the Performa 550.  It might be like you
said and there just be something weird about that hardware that
makes it not work right with -current.  If you want to try a new
kernel, I've got one at:

ftp://ftp9.ba.best.com/pub.s/sbrown/NetBSD/kernels/netbsd.marchhare.gz

I think that one is compiled without netatalk and a few other things,
but should have the basics necessary to see if it will boot or panic
on your system.

Steve


Bob Nestor wrote:

> Stephen C. Brown wrote:
>
> >I don't know about it being pulled into the 1.5 release.  Perhaps Scott
> >Reynolds can comment on that.   But, the "current" release does not
> >panic with an A3 mouse attached.  It does, however, misidentify it
> >as a single button 100-dpi mouse.  I think that's a timing issue with
> >the adb_op_sync().  I haven't had time to track this further as I've
> >been busy with real life things(getting engaged, etc. ;-) ).   If
> >someone has suggestions as to changes to make, etc.  I might
> >have a chance to test this again this weekend.
>
> Hi Steve,
>
> The current releases still panic on my Performa-550 with the A3 mouse.  I
> tried the timing patches that were in a test kernel Scott built, but
> still have the same problem.  I also tried building my own kernels with
> -current code and moving the timing value all the way out to 20,000 with
> no success.  I supect there's another timing problem lurking in there as
> I've also seen some strange stuff with the keyboard at times.  It does
> however always seem to get the mouse identifed.
>
> John Wittkowski had a terrible time trying to get this system running
> with his original HW Direct code.  As I recall he ended up putting some
> sort of timing hack in the code that didn't get removed until it was
> recently reworked for wscons support.  I ended up sending him the
> complete system one summer so he could debug with it and finish up the HW
> Direct code back then.  Looks like this system will have to make another
> cross-country trip. ;-)
>
> -bob
>
> p.s.  I'm still running with the patched kernel you sent me when you
> first found the A3 work-around.