l2tp: close all race conditions in l2tp_tunnel_register()

The code in l2tp_tunnel_register() is racy in several ways:

1. It modifies the tunnel socket _after_ publishing it.

2. It calls setup_udp_tunnel_sock() on an existing socket without
   locking.

3. It changes sock lock class on fly, which triggers many syzbot
   reports.

This patch amends all of them by moving socket initialization code
before publishing and under sock lock. As suggested by Jakub, the
l2tp lockdep class is not necessary as we can just switch to
bh_lock_sock_nested().

Fixes: 37159ef2 ("l2tp: fix a lockdep splat")
Fixes: 6b9f3423 ("l2tp: fix races in tunnel creation")
Reported-by: syzbot+52866e24647f9a23403f@syzkaller.appspotmail.com
Reported-by: syzbot+94cc2a66fc228b23f360@syzkaller.appspotmail.com
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Guillaume Nault <gnault@redhat.com>
Cc: Jakub Sitnicki <jakub@cloudflare.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Tom Parkin <tparkin@katalix.com>
Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Reviewed-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/l2tp/l2tp_core.c

@@ -1041,7 +1041,7 @@ static int l2tp_xmit_core(struct l2tp_session *session, struct sk_buff *skb, uns
 	IPCB(skb)->flags &= ~(IPSKB_XFRM_TUNNEL_SIZE | IPSKB_XFRM_TRANSFORMED | IPSKB_REROUTED);
 	nf_reset_ct(skb);
 
-	bh_lock_sock(sk);
+	bh_lock_sock_nested(sk);
 	if (sock_owned_by_user(sk)) {
 		kfree_skb(skb);
 		ret = NET_XMIT_DROP;
@@ -1385,8 +1385,6 @@ static int l2tp_tunnel_sock_create(struct net *net,
 	return err;
 }
 
-static struct lock_class_key l2tp_socket_class;
-
 int l2tp_tunnel_create(int fd, int version, u32 tunnel_id, u32 peer_tunnel_id,
 		       struct l2tp_tunnel_cfg *cfg, struct l2tp_tunnel **tunnelp)
 {
@@ -1482,21 +1480,16 @@ int l2tp_tunnel_register(struct l2tp_tunnel *tunnel, struct net *net,
 	}
 
 	sk = sock->sk;
+	lock_sock(sk);
 	write_lock_bh(&sk->sk_callback_lock);
 	ret = l2tp_validate_socket(sk, net, tunnel->encap);
-	if (ret < 0)
+	if (ret < 0) {
+		release_sock(sk);
 		goto err_inval_sock;
+	}
 	rcu_assign_sk_user_data(sk, tunnel);
 	write_unlock_bh(&sk->sk_callback_lock);
 
-	sock_hold(sk);
-	tunnel->sock = sk;
-	tunnel->l2tp_net = net;
-
-	spin_lock_bh(&pn->l2tp_tunnel_idr_lock);
-	idr_replace(&pn->l2tp_tunnel_idr, tunnel, tunnel->tunnel_id);
-	spin_unlock_bh(&pn->l2tp_tunnel_idr_lock);
-
 	if (tunnel->encap == L2TP_ENCAPTYPE_UDP) {
 		struct udp_tunnel_sock_cfg udp_cfg = {
 			.sk_user_data = tunnel,
@@ -1510,9 +1503,16 @@ int l2tp_tunnel_register(struct l2tp_tunnel *tunnel, struct net *net,
 
 	tunnel->old_sk_destruct = sk->sk_destruct;
 	sk->sk_destruct = &l2tp_tunnel_destruct;
-	lockdep_set_class_and_name(&sk->sk_lock.slock, &l2tp_socket_class,
-				   "l2tp_sock");
 	sk->sk_allocation = GFP_ATOMIC;
+	release_sock(sk);
+
+	sock_hold(sk);
+	tunnel->sock = sk;
+	tunnel->l2tp_net = net;
+
+	spin_lock_bh(&pn->l2tp_tunnel_idr_lock);
+	idr_replace(&pn->l2tp_tunnel_idr, tunnel, tunnel->tunnel_id);
+	spin_unlock_bh(&pn->l2tp_tunnel_idr_lock);
 
 	trace_register_tunnel(tunnel);