Subject: Re: New bootblocks available for testing
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Simon Burge <simonb@telstra.com.au>
List: port-pmax
Date: 03/27/1999 09:45:23
Jason Thorpe wrote:

> On Fri, 26 Mar 1999 15:30:53 +1100 
>  Simon Burge <simonb@netbsd.org> wrote:
> 
>  > I'm looking for guinea pigs to test other models.  These bootblocks can
>  > boot a.out, ECOFF and ELF format kernels, with ELF being the preferred
>  > format.  Symbol table loading for DDB is supported on all a.out kernels
>  > and new (as of last night's sup) ELF kernels, but it's possible old
> 
> Awesome :-)

Glad you're happy - you've been lightly badgering me to do two stage
bootblocks for long enough :-)

>  > The tar file contains /boot, /usr/mdec/bootrz and /usr/mdec/rzboot.
>  > Just grab the tar file, cd to / and extract it.  To build from source,
>  > you'll need to uncomment the SRCS= line in .../stand/Makefile - there's
>  > currently a problem with making symlinks during a "make depend" I
>  > haven't got nutted out yet.  The just "make install".
>  > 
>  > Once you have done either of the above steps, install the new first
>  > stage boot into the disk you want to test with (it must a valid root
>  > filesystem!)
>  > 
>  > 	disklabel -B rzX
> 
> BTW, it's probably an immensely good idea to write an installboot program
> so that you don't have to use disklabel to install the boot block.
> 
> I plan on doing this for NetBSD/hp300, as well...

Ahh, so I can just copy what you do.  Thanks :-)

>  > where X is the SCSI id of the disk, and copy the /boot file to the root
>  > partition on that disk.  You DO NOT need to use an installboot-style
>  > program with those bootblocks - indeed, there isn't one.  Then boot of
>  > the test disk.  You should see something like:
> 
> I like this approach... but I think that "installboot" should install the
> first stage (not necessarily patch block #'s into it tho :-)

Michael Hitch has toyed with the idea of doing a fixed block stage one,
more for proof of concept than anything else.  On the pmax, it should be
_extremely_ simple - the first sector contains a "boot map", that lists
61 groups of continuous blocks, a load address and an exec address...

Simon.