It's correct.
Each disk, when added to a new vdev, is partitioned into two devices.
The first (2G) is used for Linux "mdadm", as a member device of a raid1 "mirror". This mirror is then encrypted with dm-crypt and used as a swap device. (Nothing to do with ZFS.)
The second partition (remainder) is used as a vdev member, for mirror, RAIDZ1, RAIDZ2, or RAIDZ3. (This is ZFS.)