[PATCH] fix /boot/loader for extended slices

[walt]

On Mon, 20 Dec 2004, Joerg Sonnenberger wrote:

> On Sun, Dec 19, 2004 at 01:50:14PM -0800, Matthew Dillon wrote:
> >     So what we need is some serious testing, e.g. like creating multiple
> >     extended dos partitions two recursions deep and seeing if the partition
> >     data is properly passed between the loader and kernel.
>
> One important test case is an extended partition starting at sector 0,
> read: the MBR itself. That used to be a fairly easy way to kill DOS, NT
> and Linux a while back :)

Forgive me for not trying that test case.  But I did go so far as to
re-install OpenBSD to try Jonas's idea and I did find that the OpenBSD
fdisk is like Alice's Restaurant:  you can get anything you want.

So I created a second primary slice of type 0x0F and created a logical
partition in it -- and found that DFly's revised loader correctly
detects it and numbers it starting where the first group of logical
partitions left off.  I didn't actually install anything in the
partition to boot, but if the kernel numbers partitions in the same
way then it should all 'just work'.  (Seems to me this kind of disk
setup should be extremely rare.)  (And OpenBSD users are not very
much interested in multi-booting anyway ;o)

JFYI, the WinXP Disk Manager now ignores the first extended DOS slice
and recognizes the newly created second one as one huge FAT fs and
ignores the new logical disk that it contains.  i.e. it is totally
befuddled.  Linux fdisk ignores the new logical partition completely
and continues to honor the original ones.