[PATCH 1/3] fs: inode: Recycle volatile inode numbers from private slabs

[Chris Down]

One limitation to this approach is that slab recycling is currently only
per-memcg. This means workloads which heavily exercise get_next_ino with
the same memcg are most likely to benefit, rather than those with a wide
range of cgroups thrashing it. Depending on the workload, I've seen from
10%-50% recycle rate, which seems like a reasonable win with no
significant increase in code complexity, although it of course doesn't
fix the problem entirely.

Signed-off-by: Chris Down <chris@xxxxxxxxxxxxxx>
Reported-by: Phyllipe Medeiros <phyllipe@xxxxxx>
Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
Cc: Matthew Wilcox <willy@xxxxxxxxxxxxx>
Cc: Amir Goldstein <amir73il@xxxxxxxxx>
Cc: Jeff Layton <jlayton@xxxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: linux-fsdevel@xxxxxxxxxxxxxxx
Cc: linux-kernel@xxxxxxxxxxxxxxx
Cc: kernel-team@xxxxxx
---
 fs/hugetlbfs/inode.c | 4 +++-
 fs/inode.c           | 5 +++++
 mm/shmem.c           | 4 +++-
 3 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index d5c2a3158610..7b8fc84299c8 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -732,7 +732,9 @@ static struct inode *hugetlbfs_get_root(struct super_block *sb,
 
 	inode = new_inode(sb);
 	if (inode) {
-		inode->i_ino = get_next_ino();
+		/* Recycle to avoid 32-bit wraparound where possible */
+		if (!inode->i_ino)
+			inode->i_ino = get_next_ino();
 		inode->i_mode = S_IFDIR | ctx->mode;
 		inode->i_uid = ctx->uid;
 		inode->i_gid = ctx->gid;
diff --git a/fs/inode.c b/fs/inode.c
index aff2b5831168..255a4ae81b65 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -880,6 +880,11 @@ static struct inode *find_inode_fast(struct super_block *sb,
 #define LAST_INO_BATCH 1024
 static DEFINE_PER_CPU(unsigned int, last_ino);
 
+/*
+ * As get_next_ino returns a type with a small width (typically 32 bits),
+ * consider reusing inode numbers in your filesystem if you have a private inode
+ * cache in order to reduce the risk of wraparound.
+ */
 unsigned int get_next_ino(void)
 {
 	unsigned int *p = &get_cpu_var(last_ino);
diff --git a/mm/shmem.c b/mm/shmem.c
index 165fa6332993..ff041cb15550 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -2247,7 +2247,9 @@ static struct inode *shmem_get_inode(struct super_block *sb, const struct inode
 
 	inode = new_inode(sb);
 	if (inode) {
-		inode->i_ino = get_next_ino();
+		/* Recycle to avoid 32-bit wraparound where possible */
+		if (!inode->i_ino)
+			inode->i_ino = get_next_ino();
 		inode_init_owner(inode, dir, mode);
 		inode->i_blocks = 0;
 		inode->i_atime = inode->i_mtime = inode->i_ctime = current_time(inode);
-- 
2.24.1