Merge branch 'udp-round-of-data-races-fixes'

Eric Dumazet says:

====================
udp: round of data-races fixes

This series is inspired by multiple syzbot reports.

Many udp fields reads or writes are racy.

Add a proper udp->udp_flags and move there all
flags needing atomic safety.

Also add missing READ_ONCE()/WRITE_ONCE() when
lockless readers need access to specific fields.
====================

Link: https://lore.kernel.org/r/20230912091730.1591459-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

5 files changed, 42 insertions, 43 deletions

diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index f39b9c844580..c3ff984b6354 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -714,7 +714,7 @@ int __udp4_lib_err(struct sk_buff *skb, u32 info, struct udp_table *udptable)
 			       iph->saddr, uh->source, skb->dev->ifindex,
 			       inet_sdif(skb), udptable, NULL);
 
-	if (!sk || udp_sk(sk)->encap_type) {
+	if (!sk || READ_ONCE(udp_sk(sk)->encap_type)) {
 		/* No socket for error: try tunnels before discarding */
 		if (static_branch_unlikely(&udp_encap_needed_key)) {
 			sk = __udp4_lib_err_encap(net, iph, uh, udptable, sk, skb,
@@ -1051,7 +1051,7 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	u8 tos, scope;
 	__be16 dport;
 	int err, is_udplite = IS_UDPLITE(sk);
-	int corkreq = READ_ONCE(up->corkflag) || msg->msg_flags&MSG_MORE;
+	int corkreq = udp_test_bit(CORK, sk) || msg->msg_flags & MSG_MORE;
 	int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
 	struct sk_buff *skb;
 	struct ip_options_data opt_copy;
@@ -1315,11 +1315,11 @@ void udp_splice_eof(struct socket *sock)
 	struct sock *sk = sock->sk;
 	struct udp_sock *up = udp_sk(sk);
 
-	if (!up->pending || READ_ONCE(up->corkflag))
+	if (!up->pending || udp_test_bit(CORK, sk))
 		return;
 
 	lock_sock(sk);
-	if (up->pending && !READ_ONCE(up->corkflag))
+	if (up->pending && !udp_test_bit(CORK, sk))
 		udp_push_pending_frames(sk);
 	release_sock(sk);
 }
@@ -1868,7 +1868,7 @@ try_again:
 						      (struct sockaddr *)sin);
 	}
 
-	if (udp_sk(sk)->gro_enabled)
+	if (udp_test_bit(GRO_ENABLED, sk))
 		udp_cmsg_recv(msg, sk, skb);
 
 	if (inet_cmsg_flags(inet))
@@ -2081,7 +2081,8 @@ static int udp_queue_rcv_one_skb(struct sock *sk, struct sk_buff *skb)
 	}
 	nf_reset_ct(skb);
 
-	if (static_branch_unlikely(&udp_encap_needed_key) && up->encap_type) {
+	if (static_branch_unlikely(&udp_encap_needed_key) &&
+	    READ_ONCE(up->encap_type)) {
 		int (*encap_rcv)(struct sock *sk, struct sk_buff *skb);
 
 		/*
@@ -2119,7 +2120,8 @@ static int udp_queue_rcv_one_skb(struct sock *sk, struct sk_buff *skb)
 	/*
 	 * 	UDP-Lite specific tests, ignored on UDP sockets
 	 */
-	if ((up->pcflag & UDPLITE_RECV_CC)  &&  UDP_SKB_CB(skb)->partial_cov) {
+	if (udp_test_bit(UDPLITE_RECV_CC, sk) && UDP_SKB_CB(skb)->partial_cov) {
+		u16 pcrlen = READ_ONCE(up->pcrlen);
 
 		/*
 		 * MIB statistics other than incrementing the error count are
@@ -2132,7 +2134,7 @@ static int udp_queue_rcv_one_skb(struct sock *sk, struct sk_buff *skb)
 		 * delivery of packets with coverage values less than a value
 		 * provided by the application."
 		 */
-		if (up->pcrlen == 0) {          /* full coverage was set  */
+		if (pcrlen == 0) {          /* full coverage was set  */
 			net_dbg_ratelimited("UDPLite: partial coverage %d while full coverage %d requested\n",
 					    UDP_SKB_CB(skb)->cscov, skb->len);
 			goto drop;
@@ -2143,9 +2145,9 @@ static int udp_queue_rcv_one_skb(struct sock *sk, struct sk_buff *skb)
 		 * that it wants x while sender emits packets of smaller size y.
 		 * Therefore the above ...()->partial_cov statement is essential.
 		 */
-		if (UDP_SKB_CB(skb)->cscov  <  up->pcrlen) {
+		if (UDP_SKB_CB(skb)->cscov < pcrlen) {
 			net_dbg_ratelimited("UDPLite: coverage %d too small, need min %d\n",
-					    UDP_SKB_CB(skb)->cscov, up->pcrlen);
+					    UDP_SKB_CB(skb)->cscov, pcrlen);
 			goto drop;
 		}
 	}
@@ -2618,7 +2620,7 @@ void udp_destroy_sock(struct sock *sk)
 			if (encap_destroy)
 				encap_destroy(sk);
 		}
-		if (up->encap_enabled)
+		if (udp_test_bit(ENCAP_ENABLED, sk))
 			static_branch_dec(&udp_encap_needed_key);
 	}
 }
@@ -2658,9 +2660,9 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
 	switch (optname) {
 	case UDP_CORK:
 		if (val != 0) {
-			WRITE_ONCE(up->corkflag, 1);
+			udp_set_bit(CORK, sk);
 		} else {
-			WRITE_ONCE(up->corkflag, 0);
+			udp_clear_bit(CORK, sk);
 			lock_sock(sk);
 			push_pending_frames(sk);
 			release_sock(sk);
@@ -2675,17 +2677,17 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
 		case UDP_ENCAP_ESPINUDP_NON_IKE:
 #if IS_ENABLED(CONFIG_IPV6)
 			if (sk->sk_family == AF_INET6)
-				up->encap_rcv = ipv6_stub->xfrm6_udp_encap_rcv;
+				WRITE_ONCE(up->encap_rcv,
+					   ipv6_stub->xfrm6_udp_encap_rcv);
 			else
 #endif
-				up->encap_rcv = xfrm4_udp_encap_rcv;
+				WRITE_ONCE(up->encap_rcv,
+					   xfrm4_udp_encap_rcv);
 #endif
 			fallthrough;
 		case UDP_ENCAP_L2TPINUDP:
-			up->encap_type = val;
-			lock_sock(sk);
-			udp_tunnel_encap_enable(sk->sk_socket);
-			release_sock(sk);
+			WRITE_ONCE(up->encap_type, val);
+			udp_tunnel_encap_enable(sk);
 			break;
 		default:
 			err = -ENOPROTOOPT;
@@ -2694,11 +2696,11 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
 		break;
 
 	case UDP_NO_CHECK6_TX:
-		up->no_check6_tx = valbool;
+		udp_set_no_check6_tx(sk, valbool);
 		break;
 
 	case UDP_NO_CHECK6_RX:
-		up->no_check6_rx = valbool;
+		udp_set_no_check6_rx(sk, valbool);
 		break;
 
 	case UDP_SEGMENT:
@@ -2708,14 +2710,12 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
 		break;
 
 	case UDP_GRO:
-		lock_sock(sk);
 
 		/* when enabling GRO, accept the related GSO packet type */
 		if (valbool)
-			udp_tunnel_encap_enable(sk->sk_socket);
-		up->gro_enabled = valbool;
-		up->accept_udp_l4 = valbool;
-		release_sock(sk);
+			udp_tunnel_encap_enable(sk);
+		udp_assign_bit(GRO_ENABLED, sk, valbool);
+		udp_assign_bit(ACCEPT_L4, sk, valbool);
 		break;
 
 	/*
@@ -2730,8 +2730,8 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
 			val = 8;
 		else if (val > USHRT_MAX)
 			val = USHRT_MAX;
-		up->pcslen = val;
-		up->pcflag |= UDPLITE_SEND_CC;
+		WRITE_ONCE(up->pcslen, val);
+		udp_set_bit(UDPLITE_SEND_CC, sk);
 		break;
 
 	/* The receiver specifies a minimum checksum coverage value. To make
@@ -2744,8 +2744,8 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
 			val = 8;
 		else if (val > USHRT_MAX)
 			val = USHRT_MAX;
-		up->pcrlen = val;
-		up->pcflag |= UDPLITE_RECV_CC;
+		WRITE_ONCE(up->pcrlen, val);
+		udp_set_bit(UDPLITE_RECV_CC, sk);
 		break;
 
 	default:
@@ -2783,19 +2783,19 @@ int udp_lib_getsockopt(struct sock *sk, int level, int optname,
 
 	switch (optname) {
 	case UDP_CORK:
-		val = READ_ONCE(up->corkflag);
+		val = udp_test_bit(CORK, sk);
 		break;
 
 	case UDP_ENCAP:
-		val = up->encap_type;
+		val = READ_ONCE(up->encap_type);
 		break;
 
 	case UDP_NO_CHECK6_TX:
-		val = up->no_check6_tx;
+		val = udp_get_no_check6_tx(sk);
 		break;
 
 	case UDP_NO_CHECK6_RX:
-		val = up->no_check6_rx;
+		val = udp_get_no_check6_rx(sk);
 		break;
 
 	case UDP_SEGMENT:
@@ -2803,17 +2803,17 @@ int udp_lib_getsockopt(struct sock *sk, int level, int optname,
 		break;
 
 	case UDP_GRO:
-		val = up->gro_enabled;
+		val = udp_test_bit(GRO_ENABLED, sk);
 		break;
 
 	/* The following two cannot be changed on UDP sockets, the return is
 	 * always 0 (which corresponds to the full checksum coverage of UDP). */
 	case UDPLITE_SEND_CSCOV:
-		val = up->pcslen;
+		val = READ_ONCE(up->pcslen);
 		break;
 
 	case UDPLITE_RECV_CSCOV:
-		val = up->pcrlen;
+		val = READ_ONCE(up->pcrlen);
 		break;
 
 	default:
diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
index 0f46b3c2e4ac..6c95d28d0c4a 100644
--- a/net/ipv4/udp_offload.c
+++ b/net/ipv4/udp_offload.c
@@ -557,10 +557,10 @@ struct sk_buff *udp_gro_receive(struct list_head *head, struct sk_buff *skb,
 	NAPI_GRO_CB(skb)->is_flist = 0;
 	if (!sk || !udp_sk(sk)->gro_receive) {
 		if (skb->dev->features & NETIF_F_GRO_FRAGLIST)
-			NAPI_GRO_CB(skb)->is_flist = sk ? !udp_sk(sk)->gro_enabled : 1;
+			NAPI_GRO_CB(skb)->is_flist = sk ? !udp_test_bit(GRO_ENABLED, sk) : 1;
 
 		if ((!sk && (skb->dev->features & NETIF_F_GRO_UDP_FWD)) ||
-		    (sk && udp_sk(sk)->gro_enabled) || NAPI_GRO_CB(skb)->is_flist)
+		    (sk && udp_test_bit(GRO_ENABLED, sk)) || NAPI_GRO_CB(skb)->is_flist)
 			return call_gro_receive(udp_gro_receive_segment, head, skb);
 
 		/* no GRO, be sure flush the current packet */
diff --git a/net/ipv4/udp_tunnel_core.c b/net/ipv4/udp_tunnel_core.c
index 9b18f371af0d..1e7e4aecdc48 100644
--- a/net/ipv4/udp_tunnel_core.c
+++ b/net/ipv4/udp_tunnel_core.c
@@ -78,7 +78,7 @@ void setup_udp_tunnel_sock(struct net *net, struct socket *sock,
 	udp_sk(sk)->gro_receive = cfg->gro_receive;
 	udp_sk(sk)->gro_complete = cfg->gro_complete;
 
-	udp_tunnel_encap_enable(sock);
+	udp_tunnel_encap_enable(sk);
 }
 EXPORT_SYMBOL_GPL(setup_udp_tunnel_sock);
 
diff --git a/net/ipv4/udplite.c b/net/ipv4/udplite.c
index 39ecdad1b50c..af37af3ab727 100644
--- a/net/ipv4/udplite.c
+++ b/net/ipv4/udplite.c
@@ -21,7 +21,6 @@ EXPORT_SYMBOL(udplite_table);
 static int udplite_sk_init(struct sock *sk)
 {
 	udp_init_sock(sk);
-	udp_sk(sk)->pcflag = UDPLITE_BIT;
 	pr_warn_once("UDP-Lite is deprecated and scheduled to be removed in 2025, "
 		     "please contact the netdev mailing list\n");
 	return 0;
diff --git a/net/ipv4/xfrm4_input.c b/net/ipv4/xfrm4_input.c
index eac206a290d0..183f6dc37242 100644
--- a/net/ipv4/xfrm4_input.c
+++ b/net/ipv4/xfrm4_input.c
@@ -85,11 +85,11 @@ int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb)
 	struct udphdr *uh;
 	struct iphdr *iph;
 	int iphlen, len;
-
 	__u8 *udpdata;
 	__be32 *udpdata32;
-	__u16 encap_type = up->encap_type;
+	u16 encap_type;
 
+	encap_type = READ_ONCE(up->encap_type);
 	/* if this is not encapsulated socket, then just return now */
 	if (!encap_type)
 		return 1;