stable-tracking test image

[Werner Almesberger]

Andy Green wrote:
> No I think this is bootloader business, as is some PMU init -- states on
> these signals impact power consumption at least.

Yeah, we have to take care of those that affect power consumption.

> 2442 docs anyway show Port Data levels are "undefined" at reset.

You had me scared for a moment :-) Most of them (B through J) are
input with pull-down active. Port A defaults to "function".

> We can explicitly set them all again in kernel but it's likely best as
> it is.

What I'm a bit concerned about is divergence between boot loader
and kernel. Right now, we can easily change Qi whenever necessary,
but as soon as GTA03 goes into production, we lose that privilege
(much like it's with NOR at the moment).

And then every setting that needs to change ends up in two places.
That's why I think it would be good if one didn't have to check
what whatever ancient version of Qi there is on the machine might
do as well.

- Werner