I've decided to try and be a little forward-looking in setting up a new
server, and wanted to use wedges (dk).  Reading through the wedge code
in sys/dev/dkwedge leads me to a couple questions though.

First, looking at dkwedge_mbr.c, it looks like there's no support for=20
the NetBSD partition type in there.  A quick glance at
dkwedge_bsdlabel.c shows that it looks for the BSD partition entry in
the MBR, and then looks for a disklabel to parse.

My question is: Given a non-EFI machine that's required to have an MBR,
what's the point of also having a disklabel?  Couldn't we just have
multiple MBR_PTYPE_NETBSD partitions, and let those get mapped directly
to wedges (similar to how GPT works)?

I guess there's an issue with picking which one to boot from, but aside
from that, why require a disklabel also?

Second, is it reasonable to use a GPT rather than a BSD disklabel on top
of Raidframe?  The wedge code looks like this should work, but I'm not
sure if I'm missing something or not.

Thirdly, if using GPT, is the expectation that NetBSD reuses the
*FREEBSD* types?  There don't seem to be any NetBSD types defined.

Thanks!

     ScottE