From: Oleg Nesterov <oleg@xxxxxxxxxx>
[ Upstream commit 4121b4337146b64560d1e46ebec77196d9287802 ]
David Howells says:
(2) afs_lookup_volume_rcu().
There can be a lot of volumes known by a system. A thousand would
require a 10-step walk and this is drivable by remote operation, so I
think this should probably take a lock on the second pass too.
Make the "seq" counter odd on the 2nd pass, otherwise read_seqbegin_or_lock()
never takes the lock.
Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx>
Signed-off-by: David Howells <dhowells@xxxxxxxxxx>
cc: Marc Dionne <marc.dionne@xxxxxxxxxxxx>
cc: linux-afs@xxxxxxxxxxxxxxxxxxx
Link: https://lore.kernel.org/r/20231130115606.GA21571@xxxxxxxxxx/
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
fs/afs/callback.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/fs/afs/callback.c b/fs/afs/callback.c
index 7d9b23d981bf..229308c7f744 100644
--- a/fs/afs/callback.c
+++ b/fs/afs/callback.c
@@ -70,13 +70,14 @@ static struct afs_volume *afs_lookup_volume_rcu(struct afs_cell *cell,
{
struct afs_volume *volume = NULL;
struct rb_node *p;
- int seq = 0;
+ int seq = 1;
do {
/* Unfortunately, rbtree walking doesn't give reliable results
* under just the RCU read lock, so we have to check for
* changes.
*/
+ seq++; /* 2 on the 1st/lockless path, otherwise odd */
read_seqbegin_or_lock(&cell->volume_lock, &seq);
p = rcu_dereference_raw(cell->volumes.rb_node);
--
2.43.0
