Subject: Memory detection (was Re: CPU memory read ECC error)
To: Aaron J. Grier <agrier@poofy.goof.com>
From: Simon Burge <simonb@netbsd.org>
List: port-pmax
Date: 08/15/1999 15:10:03
Simon Burge wrote:

> If you can live with it a bit longer, I'd like to make a patch that
> would show the module number.  Can you test this?  If so, what version
> of NetBSD are you running?  I can build 1.4.1 or -current kernels, or
> give you a source patch to try.

Hmm, I can't find a reference anywhere for determining if we have 8MB
or 32MB simms in a 5000/2{4,6}0 - only for a 5000/200.  The Ultrix
headers don't give any useful info either.

After poking around at a disassembly of the Ultrix /200 and /240 error
handling routines, I'm gonna guess that the CSR (Control and Status
Register) info we have in kn03.h is wrong (nothing uses it anyway).  Can
you boot the kernel in

	ftp://ftp.NetBSD.ORG/pub/NetBSD/misc/simonb/netbsd-csrtest.gz

?  Just into single user mode should be fine.  Before the copyright
messages should be a line like:

	csr = 0x00002400

That's on my /240 with 32MB modules.  Hopefully yours will be 0x00002000
with 8MB modules.


Does anyone have any doco on 5000/240 internals?  Looking at the Ultrix
includes, it sounds like we should be able to use the REX getbitmap
call to workout what memory is installed.  Can we handle non-contiguous
memory on the pmax?  I've seen reference to other ports that can, so
maybe this is possible...

While on a testing sort-of-topic, has anyone played with LKMs on the
pmax?  I've never used them, but it sould be possible to put the code to
(for example) query the CSR without building a new kernel, shouldn't it?

Simon.