espintcp: remove encap socket caching to avoid reference leak

The current scheme for caching the encap socket can lead to reference
leaks when we try to delete the netns.

The reference chain is: xfrm_state -> enacp_sk -> netns

Since the encap socket is a userspace socket, it holds a reference on
the netns. If we delete the espintcp state (through flush or
individual delete) before removing the netns, the reference on the
socket is dropped and the netns is correctly deleted. Otherwise, the
netns may not be reachable anymore (if all processes within the ns
have terminated), so we cannot delete the xfrm state to drop its
reference on the socket.

This patch results in a small (~2% in my tests) performance
regression.

A GC-type mechanism could be added for the socket cache, to clear
references if the state hasn't been used "recently", but it's a lot
more complex than just not caching the socket.

Fixes: e27cca9 ("xfrm: add espintcp (RFC 8229)")
Signed-off-by: Sabrina Dubroca <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
Signed-off-by: Steffen Klassert <[email protected]>

Author: qsn

include/net/xfrm.h

@@ -236,7 +236,6 @@ struct xfrm_state {
 
 	/* Data for encapsulator */
 	struct xfrm_encap_tmpl	*encap;
-	struct sock __rcu	*encap_sk;
 
 	/* NAT keepalive */
 	u32			nat_keepalive_interval; /* seconds */

net/ipv4/esp4.c

@@ -120,47 +120,16 @@ static void esp_ssg_unref(struct xfrm_state *x, void *tmp, struct sk_buff *skb)
 }
 
 #ifdef CONFIG_INET_ESPINTCP
-struct esp_tcp_sk {
-	struct sock *sk;
-	struct rcu_head rcu;
-};
-
-static void esp_free_tcp_sk(struct rcu_head *head)
-{
-	struct esp_tcp_sk *esk = container_of(head, struct esp_tcp_sk, rcu);
-
-	sock_put(esk->sk);
-	kfree(esk);
-}
-
 static struct sock *esp_find_tcp_sk(struct xfrm_state *x)
 {
 	struct xfrm_encap_tmpl *encap = x->encap;
 	struct net *net = xs_net(x);
-	struct esp_tcp_sk *esk;
 	__be16 sport, dport;
-	struct sock *nsk;
 	struct sock *sk;
 
-	sk = rcu_dereference(x->encap_sk);
-	if (sk && sk->sk_state == TCP_ESTABLISHED)
-		return sk;
-
 	spin_lock_bh(&x->lock);
 	sport = encap->encap_sport;
 	dport = encap->encap_dport;
-	nsk = rcu_dereference_protected(x->encap_sk,
-					lockdep_is_held(&x->lock));
-	if (sk && sk == nsk) {
-		esk = kmalloc(sizeof(*esk), GFP_ATOMIC);
-		if (!esk) {
-			spin_unlock_bh(&x->lock);
-			return ERR_PTR(-ENOMEM);
-		}
-		RCU_INIT_POINTER(x->encap_sk, NULL);
-		esk->sk = sk;
-		call_rcu(&esk->rcu, esp_free_tcp_sk);
-	}
 	spin_unlock_bh(&x->lock);
 
 	sk = inet_lookup_established(net, net->ipv4.tcp_death_row.hashinfo, x->id.daddr.a4,
@@ -173,20 +142,6 @@ static struct sock *esp_find_tcp_sk(struct xfrm_state *x)
 		return ERR_PTR(-EINVAL);
 	}
 
-	spin_lock_bh(&x->lock);
-	nsk = rcu_dereference_protected(x->encap_sk,
-					lockdep_is_held(&x->lock));
-	if (encap->encap_sport != sport ||
-	    encap->encap_dport != dport) {
-		sock_put(sk);
-		sk = nsk ?: ERR_PTR(-EREMCHG);
-	} else if (sk == nsk) {
-		sock_put(sk);
-	} else {
-		rcu_assign_pointer(x->encap_sk, sk);
-	}
-	spin_unlock_bh(&x->lock);
-
 	return sk;
 }
 
@@ -211,6 +166,8 @@ static int esp_output_tcp_finish(struct xfrm_state *x, struct sk_buff *skb)
 		err = espintcp_push_skb(sk, skb);
 	bh_unlock_sock(sk);
 
+	sock_put(sk);
+
 out:
 	rcu_read_unlock();
 	return err;
@@ -394,6 +351,8 @@ static struct ip_esp_hdr *esp_output_tcp_encap(struct xfrm_state *x,
 	if (IS_ERR(sk))
 		return ERR_CAST(sk);
 
+	sock_put(sk);
+
 	*lenp = htons(len);
 	esph = (struct ip_esp_hdr *)(lenp + 1);
 

net/ipv6/esp6.c

@@ -137,47 +137,16 @@ static void esp_ssg_unref(struct xfrm_state *x, void *tmp, struct sk_buff *skb)
 }
 
 #ifdef CONFIG_INET6_ESPINTCP
-struct esp_tcp_sk {
-	struct sock *sk;
-	struct rcu_head rcu;
-};
-
-static void esp_free_tcp_sk(struct rcu_head *head)
-{
-	struct esp_tcp_sk *esk = container_of(head, struct esp_tcp_sk, rcu);
-
-	sock_put(esk->sk);
-	kfree(esk);
-}
-
 static struct sock *esp6_find_tcp_sk(struct xfrm_state *x)
 {
 	struct xfrm_encap_tmpl *encap = x->encap;
 	struct net *net = xs_net(x);
-	struct esp_tcp_sk *esk;
 	__be16 sport, dport;
-	struct sock *nsk;
 	struct sock *sk;
 
-	sk = rcu_dereference(x->encap_sk);
-	if (sk && sk->sk_state == TCP_ESTABLISHED)
-		return sk;
-
 	spin_lock_bh(&x->lock);
 	sport = encap->encap_sport;
 	dport = encap->encap_dport;
-	nsk = rcu_dereference_protected(x->encap_sk,
-					lockdep_is_held(&x->lock));
-	if (sk && sk == nsk) {
-		esk = kmalloc(sizeof(*esk), GFP_ATOMIC);
-		if (!esk) {
-			spin_unlock_bh(&x->lock);
-			return ERR_PTR(-ENOMEM);
-		}
-		RCU_INIT_POINTER(x->encap_sk, NULL);
-		esk->sk = sk;
-		call_rcu(&esk->rcu, esp_free_tcp_sk);
-	}
 	spin_unlock_bh(&x->lock);
 
 	sk = __inet6_lookup_established(net, net->ipv4.tcp_death_row.hashinfo, &x->id.daddr.in6,
@@ -190,20 +159,6 @@ static struct sock *esp6_find_tcp_sk(struct xfrm_state *x)
 		return ERR_PTR(-EINVAL);
 	}
 
-	spin_lock_bh(&x->lock);
-	nsk = rcu_dereference_protected(x->encap_sk,
-					lockdep_is_held(&x->lock));
-	if (encap->encap_sport != sport ||
-	    encap->encap_dport != dport) {
-		sock_put(sk);
-		sk = nsk ?: ERR_PTR(-EREMCHG);
-	} else if (sk == nsk) {
-		sock_put(sk);
-	} else {
-		rcu_assign_pointer(x->encap_sk, sk);
-	}
-	spin_unlock_bh(&x->lock);
-
 	return sk;
 }
 
@@ -228,6 +183,8 @@ static int esp_output_tcp_finish(struct xfrm_state *x, struct sk_buff *skb)
 		err = espintcp_push_skb(sk, skb);
 	bh_unlock_sock(sk);
 
+	sock_put(sk);
+
 out:
 	rcu_read_unlock();
 	return err;
@@ -424,6 +381,8 @@ static struct ip_esp_hdr *esp6_output_tcp_encap(struct xfrm_state *x,
 	if (IS_ERR(sk))
 		return ERR_CAST(sk);
 
+	sock_put(sk);
+
 	*lenp = htons(len);
 	esph = (struct ip_esp_hdr *)(lenp + 1);
 

net/xfrm/xfrm_state.c

@@ -838,9 +838,6 @@ int __xfrm_state_delete(struct xfrm_state *x)
 		xfrm_nat_keepalive_state_updated(x);
 		spin_unlock(&net->xfrm.xfrm_state_lock);
 
-		if (x->encap_sk)
-			sock_put(rcu_dereference_raw(x->encap_sk));
-
 		xfrm_dev_state_delete(x);
 
 		/* All xfrm_state objects are created by xfrm_state_alloc.