From: Andrew Gallatin <gallatin@cs.duke.edu>
> Doug Rabson writes:
>  > On Sat, 23 Jan 1999, Andrew Gallatin wrote:
>  > > 
>  > > What's this $75 do-it-yourself SRM kit?  Does it actually include SRM
>  > > sources?!?  What, if any, licensing restrictions are there on
>  > > distributing binaries produced from it?
>  > 
>  > It sounds like the EBSDK (which is $75) but I didn't think that included
>  > SRM sources.
>
> Sigh.  I knew it sounded too good to be true :-(
>
> Speaking of too good to be true, it would be nice if Compaq would just 
> license the SRM console in source form & let people write the MD
> portions for 3rd party / low-end boards..
>
> Drew
> ------------------------------------------------------------------------------
> Andrew Gallatin, Sr Systems Programmer	http://www.cs.duke.edu/~gallatin

Since the point has frequently been missed in this thread, it might help
to point out the centerpiece of Compaq's alpha firmware strategy:  they
very intentionally and severely restrict SRM in order to prevent DU from
running on either low end Compaq systems or on any Samsung system at all.
Some older Samsung boards were so close as to be able to run the DEC SRM,
but that isn't likely to happen in the future. Samsung doesn't really care
because the NT and Linux markets are their targets anyway.

Now, I've been communicating with some Compaq staffers about the console
interests of freenix developers. We are making some progress and may have
some good news to report shortly. It will, however, only apply to the Compaq
boards, and probably not to anything from Samsung.

More sighs. I'm afraid I started some of the folklore that is circulating
about the MILO palcode.

Over a year ago I tried running NetBSD on top of the console I had written
(for the Avalon A12 parallel processor) and in which was incorporated the
free ebsdk palcode.

It occasionally made it to a `login:' prompt, but would quickly die due to
obscure TLB coherency bugs (or something that looked like that) in the free
palcode.

I was on the verge of fixing the free palcode when DEC just gave me (well,
Avalon) the real palcode, which worked great, running NetBSD as solid as
a rock. At that point I took what seemed like the obvious guess that versions
of MILO provided to alpha linux box retailers must have had a binary version
of the real palcode in them.  (Which would imply a GPL violation.)

After all, that was the deal Compaq made with me (except I wasn't bundling
GPL'ed code into my binary console :-) and so I guessed it was what they
did with others.

But in the course of the negotiations with Compaq over console products,
I talked to some insiders who swore up and down that MILO uses the public
ebsdk palcode and that all the sources to MILO are public. The claims and
catches are: 1. the linux kernel uses only the functions of the palcode that
work, and 2. partly because of the ebsdk palcode problems, MILO only works
on a few system types.

Part of the confusion stems from the fact that Compaq has (or had) several
MILO directories on gatekeeper...some of them truly ancient. A net surfer
stumbling on one of the old ones would readily conclude that the good stuff
is in a quietly arranged binary distribution.

So, `the truth is out there', and I have seen much more recent MILO's ...
perhaps this time there is no conspiracy. :-)

	Ross.Harvey@Computer.Org