Re: initial (non working) port to sb-2000 commited

[Eduardo Horvath <eeh%NetBSD.org@localhost>]

On Mon, 20 Jul 2009, Jochen Kunz wrote:

> On Wed, 10 Dec 2008 18:32:23 +1100
> matthew green <mrg%eterna.com.au@localhost> wrote:
> 
> > i've commited my work-in-progress of the openbsd schizo PCI driver.
> > it doesn't work yet.  BUS_SPACE_MAP_LINEAR mappings generate faults
> > when accessed and interrupts aren't done.
> > 
> > anyone who can help, please do :)
> Yesterday I got a Sun Fire V240. It seems to be quite similar to the
> Blade 2000. (US III and schizo(4).) I just tried to boot -current, see
> dmesg below. So the obvious questions are:

I did a little bit of work in this area several years ago, but never 
managet to get anything clean enough to consider checking in.

> What is the state of US III support?

There are signinficant changes in both the MMU and caches between USI/USII 
and USIII processors.  This means quite a bit of nasty assembly language 
work in locore.s as well as pmap.  What makes this a bit more tricky is 
that the code for USIII will be different from that required for USI/USIII 
processors, so it will need to determine the processor type at run-time 
and select which sets of routines to use, either by patching the trap 
table, setting and checking some variable, or patching code.  (This 
dynamic requirement along with very limited access to a USIII machine at 
the time is why I never checked in the USIII code I was working on.)

Until this is done any schizo support will not do much good since you 
won't be able to get far enough to execute it.

Eduardo