On Tue, 3 Dec 1996, Kevin P. Neal wrote:

> At 01:47 PM 12/2/96 +0100, Ignatios Souvatzis wrote:
> >Generally:
> >
> >the "yellow screen" is created by the early exception vectors, which
> >are installed before the graphics system is installed.
> >
> >After that, they are replaced by the Guru vector.
> 
> Put another way, it's an exception that strikes before the exception handling
> code is in place.

There are certain old games and other programs that deliberately set the
Amiga in a state whereby a warm start will cause a "yellow screen" while
rebooting.  Therefore, the cause can be a deliberate action.   (This was
done in order to keep some resident code snoopers from working to decipher
the game's protection schemes)

This implies that the condition Darren is experiencing could be caused by
some cleanup code that didn't clean up exactly the right thing (or doesn't
work properly on certain motherboard/chip revisions).  I suppose there is
some sort of cleanup code in the NetBSD kernel that is called just prior
to resetting the machine.  And it could be that the cleanup code is buggy.

If memory serves (from when I was a C= developer), a yellow screen means
that the MMU is enabled (or in a funky state) during ROM IPL.  Since
AmigaDOS doesn't use the MMU (making it free for apps to use it), it also
doesn't expect it to be enabled after a reset.  Normally, it wouldn't be
enabled after a reset, except for a warm reset.  Power cycling the
computer usually will resolve the problem in these cases.

The only fix for it is for NetBSD to try to set the machine's state as
close to the initial power up state as possible, and hope for the best.
It is interesting to recall that C= had reset problems in several various
Amiga models, and finally settled on publishing the "Proper Way" to do it,
one that they could have the IPL code depend on.  Perhaps I could dig up
some more info on this and get back to you all if there is a request for
it.

Hope this helps!

------------------
Bruce Drake
Synectics, Inc.
(919) 872-1275 USA