On Tue, Aug 29, 2006 at 10:46:45AM +1000, Daniel Carosone wrote:
> It also seems possible that the kernel could close the device once the last
> process owning pages in the swap space exits, long after a userland swapoff
> might have a chance to make a difference.  There was a discussion a while
> back about shutdown hook ordering that is relevant here, where swap, raid,
> cgd, etc devices might be stacked in different orders.

Maybe we should not unconfigure swap but just mark the parity as clean for
that device when shutting down, or somehow ignore dirty parity on swap
partitions.  After all, you'll never have to read those bits from disk again
after a reboot (or when you unconfigured the swapspace).

	Geert