> > The trouble with checking if the sense register is zero is that when MacOS
> > is collecting its sense info, the last part of the extended sense code
> > could have been zero. That value sticks around past NetBSD boot and suddenly
> > we think there's no monitor. Not good :-)
> 
> So what you're saying is that we won't handle extended codes right.
> Wouldn't a better fix for things be to indicate what models support
> extended sense codes (in the machdep stuff), and change this test for
> those models?
>
> Can we support reading the extended sense codes before we get the intvid
> stuff? If so, can't we just hack on the sense code reading code so we read
> the right thing?

I've got code now that reads the extended sense codes properly on the Quadras
and putting it on the other machines that can do it should be a piece of cake.

However, I'm expecting to make my first intvidtest2 release at the end of
this month into the first week of next month (my next break). So I figure
you'll get sense codes along with the rest of the stuff everybody's been
waiting for. Until then, though, its a good idea to comment out the faulty
code in the kernel.

I did a send-pr on the patch. It should be in the source tree proper
eventually.