net: add a refcount tracker for kernel sockets

Commit ffa84b5ffb37 ("net: add netns refcount tracker to struct sock")
added a tracker to sockets, but did not track kernel sockets.

We still have syzbot reports hinting about netns being destroyed
while some kernel TCP sockets had not been dismantled.

This patch tracks kernel sockets, and adds a ref_tracker_dir_print()
call to net_free() right before the netns is freed.

Normally, each layer is responsible for properly releasing its
kernel sockets before last call to net_free().

This debugging facility is enabled with CONFIG_NET_NS_REFCNT_TRACKER=y

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Tested-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 14 insertions, 0 deletions

diff --git a/net/core/sock.c b/net/core/sock.c
index a3ba0358c77c..aa608dc0930b 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -2094,6 +2094,9 @@ struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
 		if (likely(sk->sk_net_refcnt)) {
 			get_net_track(net, &sk->ns_tracker, priority);
 			sock_inuse_add(net, 1);
+		} else {
+			__netns_tracker_alloc(net, &sk->ns_tracker,
+					      false, priority);
 		}
 
 		sock_net_set(sk, net);
@@ -2149,6 +2152,9 @@ static void __sk_destruct(struct rcu_head *head)
 
 	if (likely(sk->sk_net_refcnt))
 		put_net_track(sock_net(sk), &sk->ns_tracker);
+	else
+		__netns_tracker_free(sock_net(sk), &sk->ns_tracker, false);
+
 	sk_prot_free(sk->sk_prot_creator, sk);
 }
 
@@ -2237,6 +2243,14 @@ struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority)
 	if (likely(newsk->sk_net_refcnt)) {
 		get_net_track(sock_net(newsk), &newsk->ns_tracker, priority);
 		sock_inuse_add(sock_net(newsk), 1);
+	} else {
+		/* Kernel sockets are not elevating the struct net refcount.
+		 * Instead, use a tracker to more easily detect if a layer
+		 * is not properly dismantling its kernel sockets at netns
+		 * destroy time.
+		 */
+		__netns_tracker_alloc(sock_net(newsk), &newsk->ns_tracker,
+				      false, priority);
 	}
 	sk_node_init(&newsk->sk_node);
 	sock_lock_init(newsk);