So we can skip the locking. Also renames sw_int to
ring_int, cause that better matches its purpose.
Signed-off-by: NChristian Koenig <christian.koenig@amd.com>