[PATCH 6/7] rhashtable: use cmpxchg() to protect ->future_tbl.

From: NeilBrown
Date: Sun Jun 17 2018 - 22:57:02 EST


Rather than borrowing one of the bucket locks to
protect ->future_tbl updates, use cmpxchg().
This gives more freedom to change how bucket locking
is implemented.

Acked-by: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: NeilBrown <neilb@xxxxxxxx>
---
lib/rhashtable.c | 15 ++++-----------
1 file changed, 4 insertions(+), 11 deletions(-)

diff --git a/lib/rhashtable.c b/lib/rhashtable.c
index 2aa41c15df17..52ec83212856 100644
--- a/lib/rhashtable.c
+++ b/lib/rhashtable.c
@@ -297,21 +297,14 @@ static int rhashtable_rehash_attach(struct rhashtable *ht,
struct bucket_table *old_tbl,
struct bucket_table *new_tbl)
{
- /* Protect future_tbl using the first bucket lock. */
- spin_lock_bh(old_tbl->locks);
-
- /* Did somebody beat us to it? */
- if (rcu_access_pointer(old_tbl->future_tbl)) {
- spin_unlock_bh(old_tbl->locks);
- return -EEXIST;
- }
-
/* Make insertions go into the new, empty table right away. Deletions
* and lookups will be attempted in both tables until we synchronize.
+ * As cmpxchg() provides strong barriers, we do not need
+ * rcu_assign_pointer().
*/
- rcu_assign_pointer(old_tbl->future_tbl, new_tbl);

- spin_unlock_bh(old_tbl->locks);
+ if (cmpxchg(&old_tbl->future_tbl, NULL, new_tbl) != NULL)
+ return -EEXIST;

return 0;
}