Subject: Re: MVME167A success/problems.
To: None <jdb8042@blkbox.com>
From: Steve Woodford <steve@mctavish.demon.co.uk>
List: port-mvme68k
Date: 05/30/1999 12:01:01
On Sun, 30 May 1999, John D. Baker wrote:

> Howdy!  Just joined the list the other day and would like to share my
> experiences (good and bad) with the mvme68k port of NetBSD.

Excellent! It's always nice to hear folks' experiences installing
NetBSD/mvme68k.

> I did have to replace the Timekeeper SRAM in the '147--got it from
> JDR, imagine that!

Yah, since most 2nd-hand '147s are pretty old, the Timekeeper RAM is
usually long gone.

> There didn't seem to be any documentation for how to use off-board
> (VMEbus) RAM--how to make the kernel see it, but I eventually
> figured out the magic words in the NVRAM.  Soon, I had 24MB of

Oops, I need to put the relevant info in the FAQ...

> While trying to solve a problem related to the fact that I then
> had more real memory than swap space (installed without extra
> memory), my '147 suddenly died :(.

Bummer. Any idea what went wrong?

> In spite of the excellent installation instructions, I had to figure
> out a few things on my own.  My '167 is so old (167Bug v1.3) that it
> did not support network I/O in the debugger.  Also, the ";T" option
> of "IOT" does not exist in this version of the debugger.

That's quite an old board you have there ;-)

> The '167's SCSI interface seems to be very unforgiving of marginal
> devices and excess bus length.  Eventually, I got NetBSD1.4 installed

Early '167s had real problems with the SCSI bus. Motorola re-worked
the board to be much more tolerant in that area.

Perhaps I should add an NVRAM option to disable synchronous SCSI
negotiation. That might make the early boards more usable, albeit a bit
slower.

> That's when I started running into problems.  The preliminary install
> (booting from tape, partitioning and labelling the disk, and installing
> the miniroot went as well as could be expected.  It was while I was
> installing the sets (over NFS) that I kept running into MMU Faults.

There's only one thing I think might be causing this right now.
Unfortunately, the installation instructions fail to mention the jumper
settings which NetBSD needs on a '167. (My fault)

There is a group of 8 jumpers, collectively named "J1" on the board. On
my two boards these are just behind the front panel LEDs. Your older board
might have them somewhere else. All jumpers should be in place *except*
the one between pins 1 and 2 (the pair nearest the top of the board).

If your board is in VMEbus slot 1, you should have J2 set. (located
beneath the group of 8 jumpers - J1) Otherwise, J2 *must* be removed.

[re-reads the install notes...]

I've also just noticed that the notes don't have any details on
configuring the '167 firmware using 'env' and 'cnfg'. I'll try to dig
out my notes and post them here...

> Soon thereafter, I was logged in and setting up some prefered items
> when the machine spontaneously panicked, again with an MMU Fault.

I'm at a loss to explain this, other than the possible problems caused by
the wrong jumper settings above. My home network's main server is a
16Mb mvme167 running stock 1.4. It has never once hiccupped. But I guess
that's of little help for you!

> I am going to re-install from scratch again soon and keep an eye out
> for the MMU Fault panic so I can provide some input on this problem.

I doubt re-installing will cure the fault. But, if you can capture the
whole panic output and send it to me (together with your kernel config
file if you're not using the GENERIC one) I might be able to figure out
what's going on.

Anyhow, thanks for sharing your experiences!

Cheers, Steve