On Tue, 10 Aug 1999, Roger Ivie wrote:

> I'm having a bit of fun with a 3100/30 and thought I'd share the wealth.
> I've installed the 1.4.1 preview snapshot (see below), which is the first
> time I've attempted NetBSD on a 3100. The kernel hangs when it attempts to
> probe the SCSI bus containing a disk (I know this because I've moved the 
> disk between the busses and it follows the disk). I eventually see the 
> message "ncr0: no REQ for next phase, abort". The disk involved is an
> RZ56 from which I am attempting to boot using the "boot rom()netbsd -d"
> trick. I do have other disks I can attempt if you folks think the RZ56
> is the problem. Perhaps I should try a different kernel?


	Hmm - you might want to try a different disk, or even netbooting
	from the PC?

> [...]

> I thought I might be able to take advantage of this situation to bring NetBSD
> up on my 3100 without requiring a pass through my MicroVAX II (the last time
> I installed NetBSD on a VAX, it was 1.3 by tape on a MicroVAX II; not fun,
> so I decided to try a different approach).
> 
	1.4.1 should be a _lot_ less painful to install :)

> I have a spare RRD42 lying about and one of the Macintoshes in the office
> has a CD writer, so I wrote the miniroot to a CD and attempted to boot it
> on the 3100. Yes, this means that I have a CD to which I have written a
> whole 1.2 MB. I was able to read the bootstrap from the CD and coax it into
> reading the kernel using "boot rom()netbsd.gz -d". However, the system
> halted after displaying the numbers like so:
> 
>    number+number+number+[number
> 
> which I attributed to a lack of RAM. The system has only the stock 4MB on
> the motherboard. While this trick worked in general, I clearly was not
> going to be able to install the system using the miniroot.
> 
	Hmm - we've booted a 4MB 3100 with the 1.4.1 miniroot here. It fell
	over with a kernel panic which I believe Ragge may have fixed.
	We have since pruned the miniroot down to a fraction under 1MB
	which may help, though 4MB may still be too tight. Certainly 8MB
	should work.

> up the bootloader from the hard disk. Once again, I used the "boot
> rom()netbsd -d" trick to get the bootloader to read the kernel from the
> disk, which appears to have been successful.
> 
	The full 1.4.1 bootblocks should do the rom() trick automatically :)

> Except that the system seems to have hung while probing the SCSI bus, of
> course.

	I've not seen that myself, though I'm sure if anyone else has they
	will chime in :)

		David/absolute
	(Apologies for possible terse nature of above - have one arm in sling)

                  -=-  and team B will be... Kenny.  -=-