IMHO the answer is no, because dmpartx (or whatever it's called) can do
this today. It can use dm tables to slice up any existing block device,
whether it's a raw disk or a dm-constructed RAID set.
Open questions:
---------------
o do we need to prioritize on device-mapper targets for higher RAID levels
(in particular we'ld need RAID5 to support some ATARAID formats) ?
Personally I would be thrilled to see this work move ahead; in fact, I'd
be happy to see this dmraid tool support the Linux MD metadata so I can
use dm targets for my MD RAID-1 systems (thereby reducing the amount of
code I build into my kernel).