Subject: Re: Boot device confusion
To: matthew green <mrg@eterna.com.au>
From: Allen Briggs <briggs@netbsd.org>
List: tech-kern
Date: 06/20/2005 23:03:57
On Tue, Jun 21, 2005 at 09:44:48AM +1000, matthew green wrote:
> i'm not so sure.  i've used systems that have bios raid setup
> but the root was on one of these components and the bios raid
> was not used at all.

How did this work?  The BIOS RAID was actually using the disk
as a component, but root was on that component even though the
kernel configured the logical disk, too?  I'm genuinely curious
about this configuration.  Why would you do this?  Why not just
delete the logical disk and use the components?

> loader, or, is #2 (disk sizes) really that hard?  it could be
> used to avoid 99+% of these cases and seems reliable without
> breaking anything that currently is useful.

We currently don't have any geometry information in struct device
or struct disk, which are the two structures common to all disks.
Except the label, which we can't use because that's what we want
to compare to.  And I don't know of a way to query the disk for
the #sectors.  Hmmm...  Unless we call DIOCGDEFLABEL?  That might
work, too.  I'll try that.

-allen

-- 
                  Use NetBSD!  http://www.NetBSD.org/