* truncated or holepunched since swap was confirmed.
* shmem_undo_range() will have done some of the
* unaccounting, now delete_from_swap_cache() will do
- * the rest (including mem_cgroup_uncharge_swapcache).
+ * the rest.
* Reset swap.val? No, leave it so "failed" goes back to
* "repeat": reading a hole and writing should succeed.
*/