Comment 3 for bug 139514

Thanks! In and of itself, this patch looks fine, though it'll need a feature freeze exception.

Shouldn't there be some kind of control over spare devices here? http://cgi.cse.unsw.edu.au/~neilb/01093607424 says that Linux raid10 supports that, and since we're already doing that for RAID1 it seems to make sense to support it here too.