> Great! Is there (or will there be) also an "swapoff" command, for
> flushing and then removing swap-devices?

Not yet. The (syscall) interface is in place, but it turns out that
the implementation has a few hard parts to crack, given the current
interaction between the swap pager (which hasn't changed much) and
vm_swap.c (the swap device manager).

two possible approaches are:

	- make vm_swap.c and swap_pager.c co-operate to identify all
	  blocks allocated on a swap device that is to be removed.

	- have vm_swap.c set up a placeholder map that is able to
	  redirect swap I/O requests to the remaining swap devices.

-pk