> [ ... RB_ASKNAME ... ]
> 2) Since libsa already has to deal with this, modify the boot blocks
> to allow specifying the root and dump partitions and pass in wedge
> information for both of them, and *remove* RB_ASKNAME from the kernel
> (except as part of the old boot block compatibility).  (While I'm sure
> this will inspire lots of flames, I note that *no* non-BSD systems I'm
> aware of have this functionality, and in practice I've never seen
> anyone else use it.  It didn't even completely work in 4.4BSD, and
> only gave you one shot as specifying the device names anyway.)

A couple of comments:

(1) There are systems on which it's "Hard" for boot blocks to read
anything other than the device that the firmware is told to boot from.
(I.e. if told to boot from the net, it's Hard to do anything but boot
from the network interface told to boot from.  if told to boot from
"SCSI disk X on bus Y on controller Z," it's Hard to read and/or write
to any other device.)  Doesn't that pretty much preclude specifying an
alternate root device via the boot block, because the boot blocks
can't look at the other devices to get their wedge information?

(2) "If the functionality wasn't desirable, why did we go to so much
trouble to fix it so that it works well?"  The ability to specify a
different root device (e.g. network/nfs, alternate root partition,
whatever), when booting is quite useful, especially for people who do
system development.  How many times, I wonder, have people been saved
a significatn amount of trouble by it?  I know that I have at least a
few...



cgd