[PATCH RFC 09/15] mm, swap: support flexible batch freeing of slots in different memcg

[Kairui Song via B4 Relay <devnull+kasong.tencent.com@kernel.org>]

From: Kairui Song <kasong@tencent.com>

Instead of let the caller ensures all slots are in the same memcg, the
make it be able to handle different memcg at once.

Signed-off-by: Kairui Song <kasong@tencent.com>
---
 mm/swapfile.c | 31 +++++++++++++++++++++++++------
 1 file changed, 25 insertions(+), 6 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 2cd3e260f1bf..cd2d3b2ca6f0 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -1883,10 +1883,13 @@ void __swap_cluster_free_entries(struct swap_info_struct *si,
 				 struct swap_cluster_info *ci,
 				 unsigned int ci_start, unsigned int nr_pages)
 {
+	void *shadow;
 	unsigned long old_tb;
-	unsigned short id;
+	unsigned int type = si->type;
+	unsigned int id = 0, id_iter, id_check;
 	unsigned int ci_off = ci_start, ci_end = ci_start + nr_pages;
-	unsigned long offset = cluster_offset(si, ci) + ci_start;
+	unsigned long offset = cluster_offset(si, ci);
+	unsigned int ci_batch = ci_off;
 
 	VM_WARN_ON(ci->count < nr_pages);
 
@@ -1896,13 +1899,29 @@ void __swap_cluster_free_entries(struct swap_info_struct *si,
 		/* Release the last ref, or after swap cache is dropped */
 		VM_WARN_ON(!swp_tb_is_shadow(old_tb) || __swp_tb_get_count(old_tb) > 1);
 		__swap_table_set(ci, ci_off, null_to_swp_tb());
+
+		shadow = swp_tb_to_shadow(old_tb);
+		id_iter = shadow_to_memcgid(shadow);
+		if (id != id_iter) {
+			if (id) {
+				id_check = swap_cgroup_clear(swp_entry(type, offset + ci_batch),
+							     ci_off - ci_batch);
+				WARN_ON(id != id_check);
+				mem_cgroup_uncharge_swap(id, ci_off - ci_batch);
+			}
+			id = id_iter;
+			ci_batch = ci_off;
+		}
 	} while (++ci_off < ci_end);
 
-	id = swap_cgroup_clear(swp_entry(si->type, offset), nr_pages);
-	if (id)
-		mem_cgroup_uncharge_swap(id, nr_pages);
+	if (id) {
+		id_check = swap_cgroup_clear(swp_entry(type, offset + ci_batch),
+					     ci_off - ci_batch);
+		WARN_ON(id != id_check);
+		mem_cgroup_uncharge_swap(id, ci_off - ci_batch);
+	}
 
-	swap_range_free(si, offset, nr_pages);
+	swap_range_free(si, offset + ci_start, nr_pages);
 	swap_cluster_assert_empty(ci, ci_start, nr_pages, false);
 
 	if (!ci->count)

-- 
2.53.0