Subject: hp300 ethernet problem.
To: None <port-hp300@NetBSD.ORG>
From: Richard Underwood <richard@duff.org>
List: port-hp300
Date: 12/02/1996 19:26:42
Hi,

	I have a hp300 - I'm not sure which model, it was upgraded to an
'040 with 16M (BOOT ROM Rev 2.0) - I understand it should work with
NetBSD.

	It's completely diskless so before I buy disks for it I was
trying to netboot it. I'm using a NetBSD machine to do rbootd, rarp &
bootparam and a linux machine (which has disk space!) to do the NFS.

	The problem I have is the machine seems to be losing
packets. Loading the SYS_NBOOT loader takes a long time & often fails.
If it does load then the rarp/bootparam/NFS mount all works but the
kernel loading fails. The first (32 byte?) block is usually read fine,
but then the 1024 byte reads often time out. The furthest I've seen
it get is 38k which isn't very good. Since then, I haven't managed to
get the SYS_NBOOT loaded again - Read error :LAN 21 or something -
I thought maybe it got worse as it warmed up. I'll try again some time
when it's cold.

	I don't think it's a Linux NFS problem as the SYS_NBOOT loading is
from the NetBSD machine and suffers similar problems. tcpdump doesn't
recognise RMP, so it's difficult to tell, but I see what look like
repeated blocks being sent. I'd guess 1 in 5 packets are being received.

	During NFS I get the following when it times out:

le0_poll: rmd status 0x6b00

	My guess is it's a hardware problem. I tried a tranceiver
instead of the 10base2 BNC connector, but that didn't help. After
switching the jumper back it seemed to work slightly better which makes
me think there may be a dodgy connection or something.

	I read on the hp300 list archive of people having similar problems
during the RMP boot requests, and shorter ethernet segments helping. I'm
not sure I have the same problem, but I'll try it.

	Can you think of anything I haven't thought of? Given that
the hardware seems to be working mostly I'm reluctant to junk the
whole motherboard, but I'm no hardware expert & would have difficulty
'debugging' the hardware.

	Many thanks for any help,

		Richard
--
Richard Underwood - richard@duff.org