Subject: Re: dd miniroot
To: None <thorpej@nas.nasa.gov>
From: Terry Lambert <terry@lambert.org>
List: port-hp300
Date: 12/09/1996 10:55:51
>  > 	Is it possible to just dd the miniroot onto a SCSI disk and be
>  > done with it?  Or do I need to dd a SYS_INST first (or netboot it
>  > anyways).  I haven't had any luck with trying to do a network NFS
>  > install...
> 
> Well, the problem is that you need to havea NetBSD-format disklabel
> on the disk, and the miniroot needs to be stashed in partition `b'.
> Also, if you use this method, you have to get a boot program onto the
> disk so that you can boot the miniroot.
> 
> SYS_INST takes care of the disklabel, and loads the miniroot kernel
> itself.
> 
> May I ask what sort of problems you were having?

Heh... send a message, and then this comes down the line.

He's probably having the same problems I am: the ROM-using SYS_INST
routines don't seem to work on SCSI disks.


As to the dd...

I'd like to see a 100M "system image" with intrinsic disklabel that
I could dd onto a raw disk (maybe by installing it in a x86 box?).
The disklabel and all would be written to the disk.

Yes, I know that 100M is small.  It's the largest "miniroot" I could
think of that would have a normal amount of swap... in a file, not
slice b.

Then I would put it in an HP box, boot, and disklabel -e to set the
real disk size, create a swap slice b, and recover the swapfile space
leaving a 100M /, an xxM swap, and an xxxM /usr, and so on.

It's really quite irrelevant at that point that the SCSI SYS_INST
seems to uniformly fail, as long as the kernel SCSI drivers work...
both of us (and anyone else with SCSI who is doing a direct install
of 1.2 instead of 1.0 and/or does not have a direct netboot capability
at their machine) could be up and hacking code... for things like
the SYS_INST (that area springs to mind for some reason... ;-)).


					Regards,
					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.