diff --git a/include/net/sock.h b/include/net/sock.h
index 78c292f15ffc..efb7c0ff30d8 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -378,7 +378,7 @@ struct sock {
 #ifdef CONFIG_XFRM
 	struct xfrm_policy __rcu *sk_policy[2];
 #endif
-	struct dst_entry	*sk_rx_dst;
+	struct dst_entry __rcu	*sk_rx_dst;
 	struct dst_entry __rcu	*sk_dst_cache;
 	/* Note: 32bit hole on 64bit arches */
 	atomic_t		sk_wmem_alloc;
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 970a498c1166..c897d3fd457e 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -156,7 +156,7 @@ void inet_sock_destruct(struct sock *sk)
 
 	kfree(rcu_dereference_protected(inet->inet_opt, 1));
 	dst_release(rcu_dereference_check(sk->sk_dst_cache, 1));
-	dst_release(sk->sk_rx_dst);
+	dst_release(rcu_dereference_protected(sk->sk_rx_dst, 1));
 	sk_refcnt_debug_dec(sk);
 }
 EXPORT_SYMBOL(inet_sock_destruct);
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 4252aa0b5143..29b45910fafe 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2318,8 +2318,7 @@ int tcp_disconnect(struct sock *sk, int flags)
 	tcp_init_send_head(sk);
 	memset(&tp->rx_opt, 0, sizeof(tp->rx_opt));
 	__sk_dst_reset(sk);
-	dst_release(sk->sk_rx_dst);
-	sk->sk_rx_dst = NULL;
+	dst_release(xchg((__force struct dst_entry **)&sk->sk_rx_dst, NULL));
 	tcp_saved_syn_free(tp);
 	tp->segs_in = 0;
 	tp->segs_out = 0;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 2029e7a36cbb..9c7f716aab44 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -5481,7 +5481,7 @@ void tcp_rcv_established(struct sock *sk, struct sk_buff *skb,
 {
 	struct tcp_sock *tp = tcp_sk(sk);
 
-	if (unlikely(!sk->sk_rx_dst))
+	if (unlikely(!rcu_access_pointer(sk->sk_rx_dst)))
 		inet_csk(sk)->icsk_af_ops->sk_rx_dst_set(sk, skb);
 	/*
 	 *	Header prediction.
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 6f895694cca1..1fd1d590fa2b 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1404,15 +1404,18 @@ int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 	struct sock *rsk;
 
 	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
-		struct dst_entry *dst = sk->sk_rx_dst;
+		struct dst_entry *dst;
+
+		dst = rcu_dereference_protected(sk->sk_rx_dst,
+						lockdep_sock_is_held(sk));
 
 		sock_rps_save_rxhash(sk, skb);
 		sk_mark_napi_id(sk, skb);
 		if (dst) {
 			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
 			    !dst->ops->check(dst, 0)) {
+				RCU_INIT_POINTER(sk->sk_rx_dst, NULL);
 				dst_release(dst);
-				sk->sk_rx_dst = NULL;
 			}
 		}
 		tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len);
@@ -1489,7 +1492,7 @@ void tcp_v4_early_demux(struct sk_buff *skb)
 		skb->sk = sk;
 		skb->destructor = sock_edemux;
 		if (sk_fullsock(sk)) {
-			struct dst_entry *dst = READ_ONCE(sk->sk_rx_dst);
+			struct dst_entry *dst = rcu_dereference(sk->sk_rx_dst);
 
 			if (dst)
 				dst = dst_check(dst, 0);
@@ -1524,7 +1527,7 @@ bool tcp_prequeue(struct sock *sk, struct sk_buff *skb)
 	 * Instead of doing full sk_rx_dst validity here, let's perform
 	 * an optimistic check.
 	 */
-	if (likely(sk->sk_rx_dst))
+	if (likely(rcu_access_pointer(sk->sk_rx_dst)))
 		skb_dst_drop(skb);
 	else
 		skb_dst_force_safe(skb);
@@ -1818,7 +1821,7 @@ void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
 	struct dst_entry *dst = skb_dst(skb);
 
 	if (dst && dst_hold_safe(dst)) {
-		sk->sk_rx_dst = dst;
+		rcu_assign_pointer(sk->sk_rx_dst, dst);
 		inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
 	}
 }
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 8770966a564b..888c6eb8f0a0 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -1630,7 +1630,7 @@ static void udp_sk_rx_dst_set(struct sock *sk, struct dst_entry *dst)
 	struct dst_entry *old;
 
 	dst_hold(dst);
-	old = xchg(&sk->sk_rx_dst, dst);
+	old = xchg((__force struct dst_entry **)&sk->sk_rx_dst, dst);
 	dst_release(old);
 }
 
@@ -1815,7 +1815,7 @@ int __udp4_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
 		struct dst_entry *dst = skb_dst(skb);
 		int ret;
 
-		if (unlikely(sk->sk_rx_dst != dst))
+		if (unlikely(rcu_dereference(sk->sk_rx_dst) != dst))
 			udp_sk_rx_dst_set(sk, dst);
 
 		ret = udp_unicast_rcv_skb(sk, skb, uh);
@@ -1974,7 +1974,7 @@ void udp_v4_early_demux(struct sk_buff *skb)
 
 	skb->sk = sk;
 	skb->destructor = sock_efree;
-	dst = READ_ONCE(sk->sk_rx_dst);
+	dst = rcu_dereference(sk->sk_rx_dst);
 
 	if (dst)
 		dst = dst_check(dst, 0);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 6b8da8d4d1cc..d8b907910cc9 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -95,7 +95,7 @@ static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
 	if (dst && dst_hold_safe(dst)) {
 		const struct rt6_info *rt = (const struct rt6_info *)dst;
 
-		sk->sk_rx_dst = dst;
+		rcu_assign_pointer(sk->sk_rx_dst, dst);
 		inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
 		inet6_sk(sk)->rx_dst_cookie = rt6_get_cookie(rt);
 	}
@@ -1285,15 +1285,18 @@ static int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
 		opt_skb = skb_clone(skb, sk_gfp_mask(sk, GFP_ATOMIC));
 
 	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
-		struct dst_entry *dst = sk->sk_rx_dst;
+		struct dst_entry *dst;
+
+		dst = rcu_dereference_protected(sk->sk_rx_dst,
+						lockdep_sock_is_held(sk));
 
 		sock_rps_save_rxhash(sk, skb);
 		sk_mark_napi_id(sk, skb);
 		if (dst) {
 			if (inet_sk(sk)->rx_dst_ifindex != skb->skb_iif ||
 			    dst->ops->check(dst, np->rx_dst_cookie) == NULL) {
+				RCU_INIT_POINTER(sk->sk_rx_dst, NULL);
 				dst_release(dst);
-				sk->sk_rx_dst = NULL;
 			}
 		}
 
@@ -1621,7 +1624,7 @@ static void tcp_v6_early_demux(struct sk_buff *skb)
 		skb->sk = sk;
 		skb->destructor = sock_edemux;
 		if (sk_fullsock(sk)) {
-			struct dst_entry *dst = READ_ONCE(sk->sk_rx_dst);
+			struct dst_entry *dst = rcu_dereference(sk->sk_rx_dst);
 
 			if (dst)
 				dst = dst_check(dst, inet6_sk(sk)->rx_dst_cookie);