ipv6: lockless IPV6_MULTICAST_LOOP implementation

Add inet6_{test|set|clear|assign}_bit() helpers.

Note that I am using bits from inet->inet_flags,
this might change in the future if we need more flags.

While solving data-races accessing np->mc_loop,
this patch also allows to implement lockless accesses
to np->mcast_hops in the following patch.

Also constify sk_mc_loop() argument.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 files changed, 30 insertions, 25 deletions

diff --git a/include/linux/ipv6.h b/include/linux/ipv6.h
index c2e087071384..68cf1ca94914 100644
--- a/include/linux/ipv6.h
+++ b/include/linux/ipv6.h
@@ -218,11 +218,9 @@ struct ipv6_pinfo {
 #if defined(__BIG_ENDIAN_BITFIELD)
 	/* Packed in 16bits. */
 	__s16			mcast_hops:9;
-	__u16			__unused_2:6,
-				mc_loop:1;
+	__u16			__unused_2:7,
 #else
-	__u16			mc_loop:1,
-				__unused_2:6;
+	__u16			__unused_2:7;
 	__s16			mcast_hops:9;
 #endif
 	int			ucast_oif;
@@ -283,6 +281,18 @@ struct ipv6_pinfo {
 	struct inet6_cork	cork;
 };
 
+/* We currently use available bits from inet_sk(sk)->inet_flags,
+ * this could change in the future.
+ */
+#define inet6_test_bit(nr, sk)			\
+	test_bit(INET_FLAGS_##nr, &inet_sk(sk)->inet_flags)
+#define inet6_set_bit(nr, sk)			\
+	set_bit(INET_FLAGS_##nr, &inet_sk(sk)->inet_flags)
+#define inet6_clear_bit(nr, sk)			\
+	clear_bit(INET_FLAGS_##nr, &inet_sk(sk)->inet_flags)
+#define inet6_assign_bit(nr, sk, val)		\
+	assign_bit(INET_FLAGS_##nr, &inet_sk(sk)->inet_flags, val)
+
 /* WARNING: don't change the layout of the members in {raw,udp,tcp}6_sock! */
 struct raw6_sock {
 	/* inet_sock has to be the first member of raw6_sock */
diff --git a/include/net/inet_sock.h b/include/net/inet_sock.h
index 2de0e4d4a027..b5a9dca92fb4 100644
--- a/include/net/inet_sock.h
+++ b/include/net/inet_sock.h
@@ -268,6 +268,7 @@ enum {
 	INET_FLAGS_NODEFRAG	= 17,
 	INET_FLAGS_BIND_ADDRESS_NO_PORT = 18,
 	INET_FLAGS_DEFER_CONNECT = 19,
+	INET_FLAGS_MC6_LOOP	= 20,
 };
 
 /* cmsg flags for inet */
diff --git a/include/net/sock.h b/include/net/sock.h
index 676146e9d181..56ac1abadea5 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -2238,7 +2238,7 @@ static inline void sock_confirm_neigh(struct sk_buff *skb, struct neighbour *n)
 	}
 }
 
-bool sk_mc_loop(struct sock *sk);
+bool sk_mc_loop(const struct sock *sk);
 
 static inline bool sk_can_gso(const struct sock *sk)
 {
diff --git a/net/core/sock.c b/net/core/sock.c
index bb89b88bc1e8..213a62ac13f2 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -759,7 +759,7 @@ out:
 	return ret;
 }
 
-bool sk_mc_loop(struct sock *sk)
+bool sk_mc_loop(const struct sock *sk)
 {
 	if (dev_recursion_level())
 		return false;
@@ -771,7 +771,7 @@ bool sk_mc_loop(struct sock *sk)
 		return inet_test_bit(MC_LOOP, sk);
 #if IS_ENABLED(CONFIG_IPV6)
 	case AF_INET6:
-		return inet6_sk(sk)->mc_loop;
+		return inet6_test_bit(MC6_LOOP, sk);
 #endif
 	}
 	WARN_ON_ONCE(1);
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 368824fe9719..bbd4aa1b96d0 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -217,7 +217,7 @@ lookup_protocol:
 	inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk);
 	np->hop_limit	= -1;
 	np->mcast_hops	= IPV6_DEFAULT_MCASTHOPS;
-	np->mc_loop	= 1;
+	inet6_set_bit(MC6_LOOP, sk);
 	np->mc_all	= 1;
 	np->pmtudisc	= IPV6_PMTUDISC_WANT;
 	np->repflow	= net->ipv6.sysctl.flowlabel_reflect & FLOWLABEL_REFLECT_ESTABLISHED;
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index f27993a1470d..755fac85a120 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -424,6 +424,13 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
 			return -EINVAL;
 		WRITE_ONCE(np->hop_limit, val);
 		return 0;
+	case IPV6_MULTICAST_LOOP:
+		if (optlen < sizeof(int))
+			return -EINVAL;
+		if (val != valbool)
+			return -EINVAL;
+		inet6_assign_bit(MC6_LOOP, sk, valbool);
+		return 0;
 	}
 	if (needs_rtnl)
 		rtnl_lock();
@@ -755,15 +762,6 @@ done:
 		retv = 0;
 		break;
 
-	case IPV6_MULTICAST_LOOP:
-		if (optlen < sizeof(int))
-			goto e_inval;
-		if (val != valbool)
-			goto e_inval;
-		np->mc_loop = valbool;
-		retv = 0;
-		break;
-
 	case IPV6_UNICAST_IF:
 	{
 		struct net_device *dev = NULL;
@@ -1367,7 +1365,7 @@ int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 	}
 
 	case IPV6_MULTICAST_LOOP:
-		val = np->mc_loop;
+		val = inet6_test_bit(MC6_LOOP, sk);
 		break;
 
 	case IPV6_MULTICAST_IF:
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index b554fd40bdc3..679443d7ecb5 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -1996,7 +1996,7 @@ static int __net_init ndisc_net_init(struct net *net)
 	np = inet6_sk(sk);
 	np->hop_limit = 255;
 	/* Do not loopback ndisc messages */
-	np->mc_loop = 0;
+	inet6_clear_bit(MC6_LOOP, sk);
 
 	return 0;
 }
diff --git a/net/netfilter/ipvs/ip_vs_sync.c b/net/netfilter/ipvs/ip_vs_sync.c
index da5af28ff57b..3c2251cabd04 100644
--- a/net/netfilter/ipvs/ip_vs_sync.c
+++ b/net/netfilter/ipvs/ip_vs_sync.c
@@ -1298,17 +1298,13 @@ static void set_sock_size(struct sock *sk, int mode, int val)
 static void set_mcast_loop(struct sock *sk, u_char loop)
 {
 	/* setsockopt(sock, SOL_IP, IP_MULTICAST_LOOP, &loop, sizeof(loop)); */
-	lock_sock(sk);
 	inet_assign_bit(MC_LOOP, sk, loop);
 #ifdef CONFIG_IP_VS_IPV6
-	if (sk->sk_family == AF_INET6) {
-		struct ipv6_pinfo *np = inet6_sk(sk);
-
+	if (READ_ONCE(sk->sk_family) == AF_INET6) {
 		/* IPV6_MULTICAST_LOOP */
-		np->mc_loop = loop ? 1 : 0;
+		inet6_assign_bit(MC6_LOOP, sk, loop);
 	}
 #endif
-	release_sock(sk);
 }
 
 /*