Subject: Re: Sun 3/260 and 3/110 blues on last snapshot
To: NetBSD Bob <nbsdbob@weedcon1.cropsci.ncsu.edu>
From: Todd Whitesel <toddpw@best.com>
List: port-sun3
Date: 01/09/2000 01:00:52
> > dagobah:11# ls /mnt/dev
> 
> Say hello to the force and the spirit of Yoda, for us...... when you next
> seek out the dagobah system.....(:+}}...

yoda's the next68k, currently out of action waiting for more free time.
The mac68k is soda, and if I ever get the other two IIci's going again
they will be cola and lola. (Bonus trivia points to anyone who can figure
out where I got the names.)

> > So I am stumped, I don't know what I could be doing wrong here.
> 
> Does it need a set of block (sdxxx) as opposed to raw (rsdxxx) mounts?

I have no idea. Perhaps the sun3 ex-portmaster can be contacted and give
us some insight here.

In general I'm not going to have much time available for general bug fixing.
I'm hoping I can rely on the users and the portmasters for this. Right now
I can grind out buggy snapshots for alpha/i386/arm32/mac68k/sparc/sun3, and
with no crypto or X11 or packages. Eventually I want to be cranking out
weekly sets of everything on at least one port for every machine arch.
(Hardware pix at http://www.toddpw.org/arch-farm.html)

> > All my diskless setups (1.4, 1.4.1) required a MAKEDEV to be run by me
> > because the tar balls did not have most of the devices in them...
> 
> What exactly is a second MAKEDEV making that is not on the ramdisk?

In that bit, I wasn't talking about the ramdisk's MAKEDEV requirements.
The earlier question sounded like an NFS diskless setup where you unpack
everything on the server before you can netboot the new machine to mount
root over NFS. The etc.tgz file doesn't have most of the device nodes in it,
so you have to make them yourself somehow. That's all I was referring to.

There is a sort of 'second MAKEDEV' when you run sysinst; after it's
formatted the boot disk and unpacked everything, it runs 'sh MAKEDEV all'
in /mnt/dev. It doesn't copy /dev nodes from the ramdisk, nor does it expect
a .tgz file to include the complete set of device nodes.

Does that clear it up? I think I just confused the issue by switching
scenarios in mid-discussion, sorry...

> If I take the crude approach and dd miniroot into swap on sunos and boot
> from that, is that still known to work?

Got me. When dagobah finishes tonight's cross-compiler test I'll drop into
single-user and put a miniroot on sd0b to try out.

Todd Whitesel
toddpw @ best.com