On Wed, Jul 23, 2003 at 10:45:40AM +0100, Patrick Welche wrote:
> ======
> /var/backups/work/disklabel.wd0 diffs (OLD < > NEW)
> ======
> --- /var/backups/work/disklabel.wd0.current     2002-09-07 02:16:10.000000000
> ++0000
> +++ /var/backups/work/disklabel.wd0     2003-07-23 02:16:28.000000000 +0000
> @@ -21,7 +21,7 @@
>  #        size    offset     fstype  [fsize bsize cpg/sgs]
>   a:   1048320         0     4.2BSD   1024  8192    86   # (Cyl.    0 - 1039)
>   b:    524160   1048320       swap                      # (Cyl. 1040 - 1559)
> - c:  39102336         0     unused      0     0         # (Cyl.    0 - 38791)  
> + c:        16         0     unused      0     0         # (Cyl.    0 - 0*)
>   d:  39102336         0     unused      0     0         # (Cyl.    0 - 38791)  
>   e:   4194288   1572480     4.2BSD   1024  8192    86   # (Cyl. 1560 - 5720)
>   f:  10485216   5766768     4.2BSD   1024  8192    86   # (Cyl. 5721 - 16122)
> 
> ======
> 
> Where does that wd0c 16 come from? I thought it was right before.. This is
> after booting a kernel yesterday made from yesteray's -current/i386. It was
> running a 30th June before then. (Still 30th June userland)

What does fdisk say about that disk?
The 'c' partition is defined to be the area of the disk that matches
the NetBSD MBR partition.
The new kernel might be enforcing that.

It's bounds are actually rather irrelevant as they are only used in the
check that (attempts to) stop you overwriting the NetBSD disklabel.

FWIW having partitions start in sector 0 is slightly dangerous (it won't
work with some BIOS.  I believe the PC spec reserves the first track for
the BIOS...)

	David

-- 
David Laight: david@l8s.co.uk