On Tue, 14 Oct 1997, Wilko Bulte wrote:

> As Ross Harvey wrote...
> > 
> > Get the platform working with existing netboot code first. Only when
> > everything is working diskless should you screw around with a new driver.
> 
> Maybe I'm missing your point: this would mean that I need to get a DE422
> (that's what the Jensen has) driver working. 

Yes, eventually, although the first (console) and second (loaded
via TFTP, usually) stage bootstraps would normally just use the
console callbacks to load stuff, so you should actually be able to
load a kernel via the network without that device driver. Once
you've got the kernel loaded and running to the point where it will
get to single-user mode with a ramdisk root, then you can worry
about getting drivers working.

> I currently have D-Unix on the first disk, NetBSD 1.2 on the second.
> I can use D-Unix to drop test kernels, generated by the Multia, onto
> the Jensen's NetBSD disk. 

Ouch. This sounds like a lot more work to me, because you'll have
to be moving disks around and rebooting your development machine
all the time.

> Isn't that an easier approach than netbooting? It's probably easier
> for me to get a Adaptec 1742A driver working than a DE422 one (I've
> done SCSI drivers in the distant past, networking drivers are unchartered
> territory).

Networking drivers are way easier than SCSI drivers. However, the
1742 may `just work.' It's probably pretty close at any rate. It's
one of those things we've got support for, but it's never been
tested. Usually all you need to do for those drivers is fix up a
bunch of alignment stuff that the i386 didn't care about.

cjs

Curt Sampson    cjs@portal.ca	   Info at http://www.portal.ca/
Internet Portal Services, Inc.	   Through infinite myst, software reverberates
Vancouver, BC  (604) 257-9400	   In code possess'd of invisible folly.