diff mbox series

rbd: use kref_put_lock() to fix potential UAF

Message ID 20210722040216.4090-1-thunder.leizhen@huawei.com
State New
Headers show
Series rbd: use kref_put_lock() to fix potential UAF | expand

Commit Message

Leizhen (ThunderTown) July 22, 2021, 4:02 a.m. UTC 
When the refcount is decreased to 0, the rbd_client_release() is called.
Before CPU0 reaches the race point (1), CPU1 may obtain the spinlock
and traverse the linked list to find 'rbdc'. Although CPU1 will call
kref_get() to increase the refcount, it is obviously too late. CPU0 will
release 'rbdc' directly, CPU1 then accesses 'rbdc' and triggers UAF.

Use kref_put_lock() to ensure that both the operations of decrease
refcount to 0 and link deletion are lock protected eliminates this risk.

     CPU0                      CPU1
rbd_client_release():
			    <-------- (1)
spin_lock(&rbd_client_list_lock);
list_del(&rbdc->node);
spin_unlock(&rbd_client_list_lock);

kfree(rbdc);
			    <-------- use-after-free

Fixes: 602adf400201 ("rbd: introduce rados block device (rbd), based on libceph")
Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>

---
 drivers/block/rbd.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

-- 
2.25.1
diff mbox series

Patch

diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 531d390902dd..3a8989bf8e9c 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -875,11 +875,11 @@  static void rbd_client_release(struct kref *kref)
 {
 	struct rbd_client *rbdc = container_of(kref, struct rbd_client, kref);
 
-	dout("%s: rbdc %p\n", __func__, rbdc);
-	spin_lock(&rbd_client_list_lock);
 	list_del(&rbdc->node);
 	spin_unlock(&rbd_client_list_lock);
 
+	dout("%s: rbdc %p\n", __func__, rbdc);
+
 	ceph_destroy_client(rbdc->client);
 	kfree(rbdc);
 }
@@ -891,7 +891,8 @@  static void rbd_client_release(struct kref *kref)
 static void rbd_put_client(struct rbd_client *rbdc)
 {
 	if (rbdc)
-		kref_put(&rbdc->kref, rbd_client_release);
+		kref_put_lock(&rbdc->kref,
+			      rbd_client_release, &rbd_client_list_lock);
 }
 
 /*