net: better track kernel sockets lifetime

While kernel sockets are dismantled during pernet_operations->exit(),
their freeing can be delayed by any tx packets still held in qdisc
or device queues, due to skb_set_owner_w() prior calls.

This then trigger the following warning from ref_tracker_dir_exit() [1]

To fix this, make sure that kernel sockets own a reference on net->passive.

Add sk_net_refcnt_upgrade() helper, used whenever a kernel socket
is converted to a refcounted one.

[1]

[  136.263918][   T35] ref_tracker: net notrefcnt@ffff8880638f01e0 has 1/2 users at
[  136.263918][   T35]      sk_alloc+0x2b3/0x370
[  136.263918][   T35]      inet6_create+0x6ce/0x10f0
[  136.263918][   T35]      __sock_create+0x4c0/0xa30
[  136.263918][   T35]      inet_ctl_sock_create+0xc2/0x250
[  136.263918][   T35]      igmp6_net_init+0x39/0x390
[  136.263918][   T35]      ops_init+0x31e/0x590
[  136.263918][   T35]      setup_net+0x287/0x9e0
[  136.263918][   T35]      copy_net_ns+0x33f/0x570
[  136.263918][   T35]      create_new_namespaces+0x425/0x7b0
[  136.263918][   T35]      unshare_nsproxy_namespaces+0x124/0x180
[  136.263918][   T35]      ksys_unshare+0x57d/0xa70
[  136.263918][   T35]      __x64_sys_unshare+0x38/0x40
[  136.263918][   T35]      do_syscall_64+0xf3/0x230
[  136.263918][   T35]      entry_SYSCALL_64_after_hwframe+0x77/0x7f
[  136.263918][   T35]
[  136.343488][   T35] ref_tracker: net notrefcnt@ffff8880638f01e0 has 1/2 users at
[  136.343488][   T35]      sk_alloc+0x2b3/0x370
[  136.343488][   T35]      inet6_create+0x6ce/0x10f0
[  136.343488][   T35]      __sock_create+0x4c0/0xa30
[  136.343488][   T35]      inet_ctl_sock_create+0xc2/0x250
[  136.343488][   T35]      ndisc_net_init+0xa7/0x2b0
[  136.343488][   T35]      ops_init+0x31e/0x590
[  136.343488][   T35]      setup_net+0x287/0x9e0
[  136.343488][   T35]      copy_net_ns+0x33f/0x570
[  136.343488][   T35]      create_new_namespaces+0x425/0x7b0
[  136.343488][   T35]      unshare_nsproxy_namespaces+0x124/0x180
[  136.343488][   T35]      ksys_unshare+0x57d/0xa70
[  136.343488][   T35]      __x64_sys_unshare+0x38/0x40
[  136.343488][   T35]      do_syscall_64+0xf3/0x230
[  136.343488][   T35]      entry_SYSCALL_64_after_hwframe+0x77/0x7f

Fixes: 0cafd77 ("net: add a refcount tracker for kernel sockets")
Reported-by: [email protected]
Closes: https://lore.kernel.org/netdev/[email protected]/T/#u
Signed-off-by: Eric Dumazet <[email protected]>
Reviewed-by: Kuniyuki Iwashima <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Author: edumazet

include/net/sock.h

@@ -1751,6 +1751,7 @@ static inline bool sock_allow_reclassification(const struct sock *csk)
 struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
 		      struct proto *prot, int kern);
 void sk_free(struct sock *sk);
+void sk_net_refcnt_upgrade(struct sock *sk);
 void sk_destruct(struct sock *sk);
 struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority);
 void sk_free_unlock_clone(struct sock *sk);

net/core/sock.c

@@ -2246,6 +2246,7 @@ struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
 			get_net_track(net, &sk->ns_tracker, priority);
 			sock_inuse_add(net, 1);
 		} else {
+			net_passive_inc(net);
 			__netns_tracker_alloc(net, &sk->ns_tracker,
 					      false, priority);
 		}
@@ -2270,6 +2271,7 @@ EXPORT_SYMBOL(sk_alloc);
 static void __sk_destruct(struct rcu_head *head)
 {
 	struct sock *sk = container_of(head, struct sock, sk_rcu);
+	struct net *net = sock_net(sk);
 	struct sk_filter *filter;
 
 	if (sk->sk_destruct)
@@ -2301,14 +2303,28 @@ static void __sk_destruct(struct rcu_head *head)
 	put_cred(sk->sk_peer_cred);
 	put_pid(sk->sk_peer_pid);
 
-	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);
-
+	if (likely(sk->sk_net_refcnt)) {
+		put_net_track(net, &sk->ns_tracker);
+	} else {
+		__netns_tracker_free(net, &sk->ns_tracker, false);
+		net_passive_dec(net);
+	}
 	sk_prot_free(sk->sk_prot_creator, sk);
 }
 
+void sk_net_refcnt_upgrade(struct sock *sk)
+{
+	struct net *net = sock_net(sk);
+
+	WARN_ON_ONCE(sk->sk_net_refcnt);
+	__netns_tracker_free(net, &sk->ns_tracker, false);
+	net_passive_dec(net);
+	sk->sk_net_refcnt = 1;
+	get_net_track(net, &sk->ns_tracker, GFP_KERNEL);
+	sock_inuse_add(net, 1);
+}
+EXPORT_SYMBOL_GPL(sk_net_refcnt_upgrade);
+
 void sk_destruct(struct sock *sk)
 {
 	bool use_call_rcu = sock_flag(sk, SOCK_RCU_FREE);
@@ -2405,6 +2421,7 @@ struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority)
 		 * is not properly dismantling its kernel sockets at netns
 		 * destroy time.
 		 */
+		net_passive_inc(sock_net(newsk));
 		__netns_tracker_alloc(sock_net(newsk), &newsk->ns_tracker,
 				      false, priority);
 	}

net/mptcp/subflow.c

@@ -1772,10 +1772,7 @@ int mptcp_subflow_create_socket(struct sock *sk, unsigned short family,
 	 * needs it.
 	 * Update ns_tracker to current stack trace and refcounted tracker.
 	 */
-	__netns_tracker_free(net, &sf->sk->ns_tracker, false);
-	sf->sk->sk_net_refcnt = 1;
-	get_net_track(net, &sf->sk->ns_tracker, GFP_KERNEL);
-	sock_inuse_add(net, 1);
+	sk_net_refcnt_upgrade(sf->sk);
 	err = tcp_set_ulp(sf->sk, "mptcp");
 	if (err)
 		goto err_free;

net/netlink/af_netlink.c

@@ -795,16 +795,6 @@ static int netlink_release(struct socket *sock)
 
 	sock_prot_inuse_add(sock_net(sk), &netlink_proto, -1);
 
-	/* Because struct net might disappear soon, do not keep a pointer. */
-	if (!sk->sk_net_refcnt && sock_net(sk) != &init_net) {
-		__netns_tracker_free(sock_net(sk), &sk->ns_tracker, false);
-		/* Because of deferred_put_nlk_sk and use of work queue,
-		 * it is possible  netns will be freed before this socket.
-		 */
-		sock_net_set(sk, &init_net);
-		__netns_tracker_alloc(&init_net, &sk->ns_tracker,
-				      false, GFP_KERNEL);
-	}
 	call_rcu(&nlk->rcu, deferred_put_nlk_sk);
 	return 0;
 }

net/rds/tcp.c

@@ -504,12 +504,8 @@ bool rds_tcp_tune(struct socket *sock)
 			release_sock(sk);
 			return false;
 		}
-		/* Update ns_tracker to current stack trace and refcounted tracker */
-		__netns_tracker_free(net, &sk->ns_tracker, false);
-
-		sk->sk_net_refcnt = 1;
-		netns_tracker_alloc(net, &sk->ns_tracker, GFP_KERNEL);
-		sock_inuse_add(net, 1);
+		sk_net_refcnt_upgrade(sk);
+		put_net(net);
 	}
 	rtn = net_generic(net, rds_tcp_netid);
 	if (rtn->sndbuf_size > 0) {

net/smc/af_smc.c

@@ -3337,10 +3337,7 @@ int smc_create_clcsk(struct net *net, struct sock *sk, int family)
 	 * which need net ref.
 	 */
 	sk = smc->clcsock->sk;
-	__netns_tracker_free(net, &sk->ns_tracker, false);
-	sk->sk_net_refcnt = 1;
-	get_net_track(net, &sk->ns_tracker, GFP_KERNEL);
-	sock_inuse_add(net, 1);
+	sk_net_refcnt_upgrade(sk);
 	return 0;
 }
 

net/sunrpc/svcsock.c

@@ -1541,10 +1541,7 @@ static struct svc_xprt *svc_create_socket(struct svc_serv *serv,
 	newlen = error;
 
 	if (protocol == IPPROTO_TCP) {
-		__netns_tracker_free(net, &sock->sk->ns_tracker, false);
-		sock->sk->sk_net_refcnt = 1;
-		get_net_track(net, &sock->sk->ns_tracker, GFP_KERNEL);
-		sock_inuse_add(net, 1);
+		sk_net_refcnt_upgrade(sock->sk);
 		if ((error = kernel_listen(sock, 64)) < 0)
 			goto bummer;
 	}

net/sunrpc/xprtsock.c

@@ -1941,12 +1941,8 @@ static struct socket *xs_create_sock(struct rpc_xprt *xprt,
 		goto out;
 	}
 
-	if (protocol == IPPROTO_TCP) {
-		__netns_tracker_free(xprt->xprt_net, &sock->sk->ns_tracker, false);
-		sock->sk->sk_net_refcnt = 1;
-		get_net_track(xprt->xprt_net, &sock->sk->ns_tracker, GFP_KERNEL);
-		sock_inuse_add(xprt->xprt_net, 1);
-	}
+	if (protocol == IPPROTO_TCP)
+		sk_net_refcnt_upgrade(sock->sk);
 
 	filp = sock_alloc_file(sock, O_NONBLOCK, NULL);
 	if (IS_ERR(filp))