Terry R. Friedrichsen wrote:
> Sigh.  I should probably stay *way* outta this, and I'll bet poor Dean is
> feeling very much unloved at this point, but ...

Yes.  And the word 'scapegoat' comes to mind as well...

> Perhaps what needs to happen is an enumeration of the steps necessary to
> get from where we are to the twin goals cgd outlined:
> 
> >       (1) to become completely 'native',
> >       (2) to be clean.

I agree completely.  For the past while, I have been working on the following:

	1) getting my pmax to the state were I can use it as a maintainer.
	2) getting the kernel to the point where it will compile 'native'
	3) helping anyone who asks for help, or getting someone else to help.

> I, for one, would be satisfied, for the moment, with a roadmap of what steps
> one takes to get a NetBSD/pmax running native on a DECstation with an Ultrix
> assist.  I'm willing to worry about how to bootstrap from bare hardware later
> (your mileage may vary).

It is possible to get NetBSD/pmax running native, but the kernel won't
compile without a bunch of errors.  cc and as won't work perfectly and the
source for these in the tree won't compile for mips.  You'll be able to 
get Johnathans binary executables to run multi-user,  but you'll have
a hard time recompiling them.  Since you can't recompile the kernel you'll
have to stick with the miniroot one which of course does not swap.  Also,
every now and then the machine will hang or crash, most likely when you
are doing something important.  Possibly your partitions will be corrupted
to the point where you will have to start from scratch again.  Anyone
still interested in doing the installation?  BTW, this is the environment
that I have to work with.

As you may have guessed, I believe we should have a working native OS
before we start letting everyone loose on installations.  The time I spend
on helping people install a flakey OS could be better spent on making it
less flakey. (eg. I'd like to be able to have a working set of tools
(cc, as, ld, etc) and be able to rebuild them.  I would also like to be
able to build a kernel that works)

> Several folks seem to have figured this out.  Unless Dean or somebody else
> objects, I hereby volunteer to collect this information and create a docu-
> ment that describes the procedures and contains the code necessary to get
> NetBSD/pmax running native on DECstations with the help of Ultrix.  I'll
> verify that all this works by actually doing it on my 5000/25 while creating
> the documentation.
> 
> Only thing is, I'm sure I don't have all this information *now*, and I only
> have the barest idea how to do it.  This may make me unsuitable for the task;
> if you think so, please flame me.
> 
> OTOH, if you've done this already, and are willing to tolerate endless stupid
> questions from me, please send me mail.  I've got the necessary hardware all
> set up, and I'll make time to do the job with relative alacrity.
> 
> Dean, is this OK with you?  Or are you already working on this?

I'd like to see this happen, but right now I have to worry about all the 
support that will be needed because NetBSD/pmax isn't ready yet.  I'd gladly
help anyone who promised to contribute.

Ugh, I have to make one point that Johnathan probably won't like, and I hope
he doesn't get too upset with me, because he has been the most helpful person
yet.

I don't believe that it is necessary to have an ECOFF NetBSD kernel ever.
It is not needed for an NetBSD installation.  I had to install NetBSD on
the pmax I was given and I never used one.  Other people, like Herb,
installed it without having Ultrix!  I don't see how it would make things
easier then "cat bootblocks miniroot > /dev/rz#c".

Here is a possible installation sequence:
1) write bootblocks & miniroot onto start of the disk
2) boot it
3) relabel set up the partitions to be the size you want.
4) copy the miniroot to what will eventually be swap
5) reboot miniroot from swap.
6) newfs the partitions for /, /usr, /var, whatever.
7) from some external source, fill in /, /usr, /var...
8) reboot, setup /etc files, build a new kernel, and so on.

Some may note this is somewhat similar to a SunOS installation. (although
they have a bootable CDROM or tape which does most of these steps for you)

Note:
Unfortunately, the current miniroot kernel always assumes it is running
from the 'a' partition, so you have to use disklabel to swap 'a' and 'b'
partitions back and forth before rebooting.

Dean.