Jason Thorpe writes:

>you're wrong, but that's because you've been looking at the OLD SCC DRIVER,
>which has incorrect console attach semantics for the alpha!  (This is why
>the DEC 3000s still need to use the PROM console!)


Jason,

You are confusing your guesses of what I'm doing with what i'm
actually doing.  Your guesses are just wrong.  I went back and looked
at old mail from cgd, and from ken hornstein, and I looked at the new
alpha driver.

I may be technically wrong about what *all* the Alpha models do, but
if so, were those other people, and not all of them were looking at
the old scc driver.  Some of them had even looked at the PDF docs cgd
put on the ALpha web page.  

And it's still technically true that there are a finite number of
behaviours and required console-attach semantics; and that the
console-attach semantics could be controlled via flags specified at
attach time from model-dependent front-end attach wrappers which call
an ``MI'' (i.e., alpha or pmax) `zs at ioasic' backend, which can
handle truly shared aspects -- like the DMA engines we've talked about
on the phone -- which in turn talks to the MI zs backend.

That *does* result in code sharing.  Undeniably so.

Jason, please stop misrepresenting me like this on the port-pmax
mailing list.  That is unprofessional and simply not acceptable.