Subject: Resolutions. Was Re: probs with newly acquired 380
To: None <>
From: NetBSD hp300 list <>
List: port-hp300
Date: 04/03/2000 01:26:31
Thank you all for the many replies and information.

I've now resolved my serial console issues and have installed
NetBSD on the 9000/380.

A few noteworthy comments on resolutions and observations:

> 2)      The keyboard doesn't work. 
>         When the machine powers up, the keyboard lights do indeed
>         flash.  I presume this is normal.  However, absolutely no
>         keypresses are recognised by either the boot ROM or with
>         SYS_INST when the console isn't directed to the serial port.

The keyboard is a model 46030A.  Indications from a number of people
indicate that it is indeed a much newer keyboard than the '380 appears
to be able to deal with.  At this point it is inconsequential since I
have resolved the serial console problem (see below).

>  Note that when using the serial port as the console for SYS_INST,
>  I still don't have access to the ROM startup via the serial port.
>  I would presume the ROM is still directing startup to a presumed
>  video board and motherboard keyboard port.  Until I can resolve
>  the keyboard issue, I presume I'm stuck with this situation.

As was clarified by Michael B. Wolfson, it is extremely easy to access
the console via the serial port on the CPU board: the PROM listens to
not only the keyboard port, but the serial port as well for input,
and will switch to the serial port when the keystrokes are recognised.
It was just too easy! :)  I don't know what PROM versions or machines
this is similarly the case on, but perhaps a note on the FAQ clarifying
this can help others as well--I myself had assumed all attempts to
switch the console via keystrokes mentioned there must come from the
CPU keyboard and not the serial port.  This switching to the serial
console is possible regardless of whether or not a keyboard or
video board is plugged in.

> 3)      I can (apparently) successfully disklabel drives with SYS_INST
>         however doing the miniroot step simply hangs after responding
>         to the "Server path?" prompt.  There is absolutely no attempt
>         at the server end to mount an NFS path, since there are
>         neither accept nor reject messages in the log.

This issue was never resolved.  I believe the SYS_INST code itself to
be inoperable on the model 380 for other than disklabelling and booting.
The miniroot function simply wouldn't even send packets out onto the
ethernet to attempt to access the server NFS partition.  I tried many
different Unix version NFS servers, so I don't believe this to be at
all a problem related to the NFS server used, nor are there any kind
of firewalls or other factors of concern.

This is how I managed to install on a drive finally (Thanks again to
Mr. Wolfson):

	o	rboot the SYS_INST image
	o	disklabel the disk

		Sample partition setup:
			8 partitions:
			     size   offset
			a:   2984063,    1024
			b:   1351680,    2985087
			c:   5688447,    0
			d:   1351680,    4336767
			e:   0,    0
			f:   0,    0
			g:   0,    0
			h:   0,    0

	o	move the drive to another Unix machine
		(In this case FreeBSD)

	o	dd(1) the miniroot to the "b" partition:

			dd if=miniroot.fs oseek=2985087 of=/dev/rda4

	o	move the drive back to the HP
	o	rboot SYS_INST
	o	boot the drive

The system is up and running now.

At this point I'm trying to build a CURRENT system from the 1.4.2 ALPHA
snapshot installed from.


There appear to be many issues preventing a CURRENT build from 1.4.2
due to egcs differences, as well as include file and other such changes.
A "make build" simply doesn't work.

I'm currently pursuing the following strategy:

	o	'make includes'
	o	make gcc, gas, etc. and install
	o	make libes (The current step I'm on) and install
	o	reattempt a 'make world', and resolve issues as they arise
	o	finally do 'make build' to ensure everything is in sync.

Any comments or direction on bringing this up to CURRENT?  I haven't been
able to locate a CURRENT snapshot dated after the 1.4.2 ALPHA release.  If
anyone has made one available, please let me know.

I plan to delve into the install issue once I've brought this machine up to
current sources.  If anyone else is already doing or considering this 
themselves, please let me know as well so I don't duplicate the effort.

Overall, I'm extremely pleased at the system performance and robustness.
Although I haven't performed any CPU comparisons for performance, I have
done rudimentary disk I/O speed using dd(1): about 1.8MB/s on a 15ms
SCSI drive.

Finally, a request:

Would anyone be able to part with two sets of the cables for installing
drives in the C2213A SCSI expansion cabinet?  This is a three-drive
full-height cabinet (mine are labelled "HP 6000 600S" on the front).
The cables desired are:

	(2) power connector cables HP no. C2212-60054
	    14-pin female (6 pins used) -> two 4-pin female drive-power

	(2) SCSI extensions HP no. C2212-60148
	    one male to two female (for stacked 1/2-height

Additionally, I'd like to track down a three-connector RGB -to- female
PC-style mini-15-pin video connector, in order to connect a PC non-BNC 
monitor to the box.

Once again, thanks to everyone for their quick and extremely informative
replies to my request for assistance.

Scott G. Akmentins-Taylor     InterNet:
MRY Systems	
    (Skots Gregorijs Akmentins-Teilors -- just call me "Skots")
		----- Labak miris neka sarkans -----