Re: Unsupported RS6k's and ofppc

On Fri, 10 Mar 2006 23:45:33 -0700 (MST)
Tim Rightnour <[email protected]> wrote:

> First.. from speaking with someone about the Powerstack II 3000, I was
> told that it has three different firmware images available for it. 
> Firmworks OF, PPC-BUG, and ARC.  The PPC-BUG manual states that it
> provides residual data. This means that if you can get PPC-BUG
> firmware on that machine, and the manual wasn't lying, then you can
> most likely boot NetBSD/prep.
Yes. The first thing I tried with the PowerStack II was port-prep as it
is a PReP machine. I had to notice that the OFW on my machine provides
no residual data so port-prep was out of question. Then I started to
hack on port-ofppc. PPC-BUG would be fine, but where to get and how to
install?

> Second.  The 7043-150 may or may not be a transitional machine like
> the 7025. If it is, I believe it also provides residual, and can boot
> and run prep.
The 7043-150 is CHRP only. No residual data and CHRP memory map.

> As for the B50.. I have no clue.. but stay tuned...
The B50 is mostly identical to the 7043-150. I have both machines. The
B50 is a 7043-150 mainboard in a 2U 19" box with a shorter PCI riser
card so that the two PCI slots fit in 2U. 

> Yeah.. this is the *exact* problem I had with the 7024.  What you need
> to do is find the interrupt map.  I've seen OFW node dumps of the
> 7043-150, and I know there is a node somewhere called "interrupt-map"
> or maybe its a property of a node.  This (once decoded) will tell you
> exactly how to wire the openpic.
The problem with the PowerStack II is not interrupt related. It seems I
can't access PCI memory space. PCI IO works fine. That is the reason why
ex(4) worked: ex(4) preferes PCI IO and does not provid CSRs in PCI MEM
region. tlp(4) and esiop(4) want PCI MEM.

> Also.. if the machine is prep bootable, then you have a second
> wonderful opportunity.  You can use prep to understand the
> architecture of the machine, and then make it work under ofppc.
I have no problem understanding the general machine architecture, I
suspect quirky hardware...

> > If there is someone with interrest I can port my old patches to
> > -current and put them on the net. I don't wane commit it in this
> > state as it will fsck up port-ofppc. (Hmmm. Maybe I chould commit
> > anyway. Maybe enough people get anoyed to help me fix the remaining
> > issues. ;-) )
> I'd be more than happy to help you with OpenPIC/MPIC issues.
You should not have said this. ;-)
OK, I'll bring my patches for port-ofppc up to -current and leave the
interrupt stuff to your amusement. There are other open issues like MD
disklabel stuff that I have to fix.
-- 


tschüß,
       Jochen

Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/