Subject: running NetBSD/sun3 diskless and the "everything dumps core" problem....
To: None <port-sun3@NetBSD.ORG>
From: Greg A. Woods <woods@most.weird.com>
List: port-sun3
Date: 05/23/1998 17:54:24
[ On Fri, May 22, 1998 at 09:39:49 (-0500), Nathan Lane wrote: ]
> Subject: Re: Sun 3/60 TERMPWR/diskless
>
> I'm having excellent luck with a 3/110 w/8MB as my network time server
> and network management workstation.  A friend has a 3/60 that has been
> working pretty well - he did have some "everything core dumps all the
> time" problems but after he replaced the upper 4MB of SIMMs (with 3-chip
> no less) it seemed to clear up - only happens every few days now instead
> of once every few minutes.  He's running 1.3 and I'm running 1.3.1.

That's most incredible!

I've had no luck at all with 1.3, or various incarnations of -current
that were available before 1.3.1 was cut.

I've not yet tried 1.3.1, nor -current, since, but hope to soon.

I wonder how NetBSD/sun3 could possibly utilize memory differently than
SunOS-4 does.  I know that SunOS-4's parity detection code worked.  I
had a bad SIMM on my 3/60 some time ago (it was further aggravated by
heat too).  I've not yet investigated the code in NetBSD to see if it
should detect and report parity errors on the 3/60 (it seems as though
it should for VME memory, though not well), nor placed that bad SIMM
back in a 3/60 to see if NetBSD actually detects problems with it.

The 3-chip SIMMs probably are (must be?) the kind that generate parity
on the fly, thus the only possible data corruption parity from them
could detect is from noise on the connector or motherboard.  The major
problem with them is you can never know when random failures are due to
bad hardware -- they go so far as to generate good parity for corrupted
bytes, an insightful look into the mind of a PC hardware hacker and the
lengths they go to just to save a few pennies at the expense of the
consumer.  Capitalism isn't all that great when unchecked by safety
regulations and such and fed by ignorant consumers!  ;-)

I resorted to running the SunOS-4 based Xkernel on my 3/60 (instead of
booting SunOS-4 off my NetBSD server!).  It's working A-OK.  The only
thing I miss is access to my serial ports (one has my Caller-ID device
still connected to it, and the other is still my emergency dial-out
modem).

I'll power up a couple of my 3/260's and my other 3/60 again soon and
try 1.3.2 and -current again to see if the everything dumps core has
gone away for other reasons....

-- 
							Greg A. Woods

+1 416 443-1734      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>