Subject: Re: Q: Compaq, *BSD and 'Linux-only' AlphaBIOS (fwd)
To: Andrew Reilly <reilly@zeta.org.au>
From: Mike Smith <msmith@freebsd.org>
List: port-alpha
Date: 12/04/1999 17:05:23
> On Sat, Dec 04, 1999 at 12:20:15AM -0800, Mike Smith wrote:
> > There are licensed components inside SRM that prevent this from
> > happening; we've been down this path with Dompaq already. However, there
> > does seem to be some fairly strong interest inside the organisation for
> > the production of an SRM replacement that would be open-sourced.
>
> Apart from boot-monitor stuff, there's PALcode, right? What
> does that actually do for you, that takes so much (proprietary)
> code? Isn't it just shortcuts to hide the TLB/memory management
> hardware under some sort of high-level API?
It depends on which footprint you're talking about. In the BSD (OSF)
case, it also covers things such as interrupt handling and
acknowledgement, and there are probably SMP primitives there as well.
> Put another way: is there anything about the various Alpha
> implementations that are insufficiently documented, or prevent
> us from doing these things ourselves, right on the chip itself?
Yes, there are. It took a lot of effort on the part of several quite
powerful people inside Digital just to liberate the documentation for the
old AlphaServer 2100 4/xxx systems. In some ways, things are getting
easier, but regardless the current situation is such that a) the
documentation is very hard to get, and b) the degree of skill and
understanding of the system architecture required is possessed by only a
very few people.
> The *BSD reliance on the SRM seems to basically limit the Alpha
> purchase choices to Compaq. I.e., not Samsung/AlphaProcessor.
This is currently correct. It is, however, erroneous to think that
either a) we're not aware of this, or b) we're somehow indifferent or
inactive on the subject. I know for a fact that the folks on the FreeBSD
side of the fence have been pursuing every avenue we've come across, and
I hardly expect the NetBSD people have been any less active. There are
simply some very substantial obstacles currently in the way of a
breakthrough.
--
\\ Give a man a fish, and you feed him for a day. \\ Mike Smith
\\ Tell him he should learn how to fish himself, \\ msmith@freebsd.org
\\ and he'll hate you for a lifetime. \\ msmith@cdrom.com