> From: David Jones <dej@eecg.toronto.edu>
> Date: 	Wed, 8 Mar 1995 00:09:05 -0500

> I am trying to get netbsd to boot on my 3/260.
> 
> I have prepared root and usr filesystems on sd2 (SunOS nomenclature,
> it's SCSI ID 1) by copying my Amiga binaries over, running MAKEDEV
> from dev.tar.gz, and installing zupdate on top of all that.  I also
> did the installboot.
> 
> So I have a Quantum 105 on ID 1.  If I try
> 
> b sd(0,1,0)
[ (0,8,0), (0,16,0) ]

> the machine sits there for a few seconds, then returns to the boot
> monitor.  I get the same effect if I try to boot from a disk whose
> ID is not equal to any ID in the system.

The disk has a sunos disk label, right?

> Can I boot from my Quantum?  What numbers do I use?  I have SunOS
> installed on sd0 so I don't relish fiddling with the prom.

Try copying a SunOS kernel onto the quantum.  If both the disklabel
and /boot are correct, it should load SunOS and then panic.
You could use:  b sd(0,8,0)vmunix -a
so it will ask you where to get root, etc.

> Oh, well, screw that.  Let's try netbooting.

> I've tried booting a kernel I've compiled myself, and the netbsd-gen.gz,
> fetched from ftp.netbsd.org on March 7, fresh.  On my kernel, I get a
> 
> Watchdog reset!
> 
> just after it prints "Total mem: %d".  At this point, the keyboard is
> dead - I can't even L1-A back to the boot rom.

The PROM does that when the kernel touches VA=0 on the Sun3/260 and
I have not been able to figure out why.  On the 3/60 that takes you
into ddb instead. (MUCH more useful...)

> If I try the generic kernel, I get:
> 
> panic: pa_to_pvp: bad pa: %p

Was no physical address printed?

> I thought this problem was corrected!

I thought so too!  That was the problem with "junk" being left in
the MMU that our pmap code does not expect to find there...

> Try again: it works, sort of:
> 
> ie0: TDR detected an open 0 clocks away.
> ie0: TDR detected a short 0 clocks away.

I see that too on the ie - must be a bug.  In fact, it does it LOTS
when I run tcpdump on the 3/260 (output to the serial console).

> (My Ethernet works fine under SunOS)
Ditto.

> It then tried to mount NFS root and swap.  Is there a way I can have it
> mount root, etc. from the local SCSI disk?

Yes.  Use "b ... -a"  Useful flags are:
-a	Ask me about root, swap, etc.
-d	Debug (stop in ddb so you can set some breakpoints)
-s	Single-user mode

> OK, try again, netbooting.  Same result, but this time NO SCSI devices
> were found!!!

The SCSI driver is flaky about getting things reset if the SunOS
kernel has been running.  Seems OK after a power up...
I am working on a new scsi driver... slowly.  (Busy at my real job.)

> OK, try again.  Now the SCSI devices are found, but the Quantum appears
> on ALL luns!  (It appeared only at LUN 0 the first time, and all the
> time under SunOS)

That means your Quantum needs to be added to the "rogue" list in:
src/sys/scsi/scsiconf.c (look for the other rogue drives).

Gordon