> commit! commit! ok!

OK!  Done!

> I would say allow the devices to attach in whatever order they normally do. 
> When the auxio2 device attaches, I would turn the power on for all the various
> devices.  I would also create external functions to flip the power on and off
> for the various devices, so that the devices can elect to turn themselves on.
> 
> If you wanted to tackle it.. you might make it so that opening the serial port
> device, flips the power switch on for the serial ports on the tadpole.  Perhaps
> a final close could turn them off.

Yes, I was thinking about this today.  It seems sensible to have an open
on the serial ports power them on and a (last) close power them off.  Because
the various devices are driven off both aux1 and aux2, does it make sense
for a single device to control this, and preferably all the power related
functions too?  I think I'd prefer that to having to open/ioctl/close many
devices (and the aux registers don't have /dev entries at the moment).  Add
this to the tctrl device?

> Great work!

Thanks!  Now, if someone else can get them working perfectly ...

A little more poking - I was wrong when I said that <backspace> and <escape>
generate odd characters first time.  Any key pressed generates an odd
character first time (at a guess it's that character with the top bit set).
After that, typing without pausing (for approx. 1 second) works as expected.
As soon as I pause, the stream of odd characters appears and back to first
press generates an odd character.  Also, I noticed that disconnecting DTR
makes the stream of odd characters much longer - from 5-20 characters to
maybe 300.  Finally `echo Some_text > /dev/ttya` generates garbage.  Is there
some timing issue with the control lines, I wonder?

J

-- 
                    My other computer also runs NetBSD
                          http://www.netbsd.org/