David Laight wrote:

> > When I first did lfs boot support for x86 you used to have to do
> > something like:
> > 
> > 	newfs_lfs -O 256 /dev/rwd0a
> > 	installboot -b 32 /dev/rwd0a /usr/mdec/bootxx_lfsv2
> > 
> > The -O to newfs_lfs says to start the filesystem 256 sectors after the
> > start of the disk, which leaves some room to put the primary bootstrap,
> > and then the -b 32 tells installboot to just blat the bootstrap starting
> > at block 32.
> 
> Erm -b doesn't mean that (I don't think it ever has!).

Hmmmm.  From installboot(8):

     -b s1bno    Install primary at block number s1bno instead of the default
                 location for the machine and file system type.  [alpha, pmax,
                 vax]

Note the [alpha, pmax, vax] at the bottom.  I did this too long ago,
obviously the memory is failing :-/

> Starting the fs anywhere other than 8k is a PITA - you need to be VERY
> careful to ensure that any old superblock (eg from an FFSv1 fs) is
> overwritten.  Otherwise mount and/or fsck WILL treat the disk as FFSv1
> with the expected consequences for your data.

Ouch.  Back before the bootblocks were split up there was no option to
boot lfs any other way (on x86).  Hopefully not a problem now that we
have split bootblocks.  Did you test LFS (v1 or v2) for booting?

Simon.
--
Simon Burge                            <simonb@wasabisystems.com>
NetBSD Support and Service:         http://www.wasabisystems.com/