[ On Thu, February 19, 1998 at 17:16:45 (-0500), Gordon W. Ross wrote: ]
> Subject: Re: UVM for sun3?
>
> Sorry for taking so long to respond.  Busy at work again...

No problem!  I've been very busy too!

> > Date: Sun, 15 Feb 98 22:18:24 -0500 (EST)
> > From: woods@kuma.web.net (Greg A. Woods)
> 
> [ implement MACHINE_NEW_NONCONTIG ... ]
> > 
> > Yeah, I thought so.  I think there are already the beginnings of hints
> > as to how to do this in sun3/pmap.c too, at least in what needs
> > replacing....
> 
> That part is done.

Bonus!

> > However first off I need to get the current kernel running as-is.  It's
> > dying because of a test that results in a panic() that's only there when
> > DIAGNOSTIC is set.  At the moment because of my unfamiliarity with the
> > BSD kernel autoconf code I'm not even sure where this state that fails
> > the test is getting set.
> > 
> > The test is at the opening of bus_scan() in sun3/autoconf.c:
> [ if (cf->cf_fstate == FSTATE_STAR) ]
> 
> Oops.  I'm not sure that diagnostic is still needed, but
> I removed the offending "star" config lines from bw2.sun3.

Ah!  OK, well I'll hopefully get to try again later this week.  I'm
still busy copying over my entire world over from my old Sun-3 SunOS-4
system onto the newer SS-2 NetBSD system, and as part of that I'm
finally going to track NetBSD-current far more frequently and with a
local CVS repo.  I'll probably hold off on even sun3 kernel builds until
I can get working areas set up and the old sun3 converted to NetBSD.

One stumbling block I still have is in trying to get NetBSD/sun3 running
diskless so my workstations can also do away with SunOS-4.  I will be
putting some more memory into my main sun3 to try and further avoid
paging, but I'm not 100% sure that's the problem.  I am hoping that UVM
will at least shuffle the bugs around enough that it'll work more
reliably than it does now, or even be easier to debug than it is now.

-- 
							Greg A. Woods

+1 416 443-1734			VE3TCP			robohack!woods
Planix, Inc. <woods@planix.com>; Secrets Of The Weird <woods@weird.com>