From: Andrew Morton <akpm@osdl.org>
I converted the workqueues to per-device while I was there.  It seems
strange to create a new kernel thread (on each CPU!) and to then only
have a single global work to ever be queued upon it.
Plus without this, I'd have to use the _NAR stuff, gawd help me.
Does that workqueue really need to be per-cpu?
Does that workqueue really need to exist?  Why not use keventd?
Cc: Takashi Iwai <tiwai@suse.de>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jaroslav Kysela <perex@suse.cz>