> From: Geoff Adams <gadams@avernus.com>
> Date: Wed, 8 Feb 1995 02:20:52 -0500

> I've been trying for a while to get NetBSD -current to run on a sun 3.
> So far, no luck.  I thought I'd keep my mouth shut and wait for progress,
> though, since no doubt I'm not the only one with troubles.
> 
> I get the idea, though, that I should be able to run programs such as
> cc on a 3/50 or a 3/60 under -current.  Yet, I cannot.  I am running the
> system diskless (and in fact, I'm mounting my file systems from a NetBSD/
> sparc running -current).  I assume that my problems are ethernet driver
> troubles causing corrupted NFS data.  I thought I'd verify that I should
> expect things to work better than they are, and if so, I'll work on
> putting NetBSD onto a SCSI disk.
> 
> On my 3/60, the most recent kernel (#137, Jan 25) consistently panics
> while configuring le0:
> 
> le0 at obio0 addr 0x120000 level 3panic: pmap:pa_to_pvp: bas pa=0x1f800000

I thought that was fixed, but maybe not.  That is netbsd-gen.gz right?
You then land in ddb, I assume.  Please do a "trace" and send the result.

I'll update that kernel again too, just to be sure it's OK.

> This kernel, running on a 3/50, keeps giving ether framing errors:
> 
> le0: framing error
> le0: crc mismatch

I see these occasionally too.  They are probably real hardware errors.

> It also hangs when doing any database work (like updating the password
> files or building the kernel databases during multi-user startup).

Ah, that is your server dropping packets that NetBSD/sun3 sends.
Add the following line to your config file:

	# XXX - Work-around for slow SunOS/Sun3 servers (sigh...)
	options 	NFS_BOOT_RWSIZE=1024

> Using an older kernel (#427, Oct 25), I can get the 3/60 to come up
> multi-user.  I can compile some smaller .c files, but linking will
> always cause a hang or cause init to crash.  (This causes the machine
> to come up multi-user *again* without an intervening shutdown!  Core
> files everywhere.)
> 
> This kernel, on a 3/50, has deeper problems involving 0xdeadbeef and a
> subsequent Watchdog reset.  This may have been before the 3/50's memory
> was properly mapped to allow use of the bw2 as the console.

The old buggy kernel has VM problems that cause core dumps and worse.
I should delete that old kernel.  Don't use netbsd-old.gz anymore.

> Are these problems all caused by NFS troubles?  It certainly seems
> possible.  If I should be getting a clean boot in any of these cases,
> and especially if the system is actually working well for other people,
> I'd like to know.  I'd been assuming for a while that the port just
> wasn't that close to running, yet!

I have it running on 3/50, 3/60, 3/260, both diskless and with SCSI.

> Also, running NetBSD seems to have wiped the default boot device
> from my eeproms (on three machines!).  I don't remember the address
> to open to change it, though.  No biggie for now.

Hmmm... Never seen that.  Maybe a result of the old kernel (netbsd-old.gz)
letting things write using crossed-up page mappings...

> I'm very excited about the possibility of running NetBSD on these
> machines!  I have so many plans for what might otherwise be wasted
> hardware...  :-)

Great!  Let us know if you create anything worth sharing.

> Thanks for any help!

Always welcome.
Gordon