inet: move inet->transparent to inet->inet_flags

IP_TRANSPARENT socket option can now be set/read
without locking the socket.

v2: removed unused issk variable in mptcp_setsockopt_sol_ip_set_transparent()
v4: rebased after commit 3f326a82 ("mptcp: change the mpc check helper to return a sk")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/inet_sock.h

@@ -231,7 +231,6 @@ struct inet_sock {
 	__u8			mc_ttl;
 	__u8			pmtudisc;
 	__u8			is_icsk:1,
-				transparent:1,
 				nodefrag:1;
 	__u8			bind_address_no_port:1,
 				defer_connect:1; /* Indicates that fastopen_connect is set
@@ -271,6 +270,7 @@ enum {
 	INET_FLAGS_HDRINCL	= 12,
 	INET_FLAGS_MC_LOOP	= 13,
 	INET_FLAGS_MC_ALL	= 14,
+	INET_FLAGS_TRANSPARENT	= 15,
 };
 
 /* cmsg flags for inet */
@@ -397,7 +397,7 @@ static inline __u8 inet_sk_flowi_flags(const struct sock *sk)
 {
 	__u8 flags = 0;
 
-	if (inet_sk(sk)->transparent || inet_test_bit(HDRINCL, sk))
+	if (inet_test_bit(TRANSPARENT, sk) || inet_test_bit(HDRINCL, sk))
 		flags |= FLOWI_FLAG_ANYSRC;
 	return flags;
 }
@@ -424,7 +424,7 @@ static inline bool inet_can_nonlocal_bind(struct net *net,
 {
 	return READ_ONCE(net->ipv4.sysctl_ip_nonlocal_bind) ||
 		test_bit(INET_FLAGS_FREEBIND, &inet->inet_flags) ||
-		inet->transparent;
+		test_bit(INET_FLAGS_TRANSPARENT, &inet->inet_flags);
 }
 
 static inline bool inet_addr_valid_or_nonlocal(struct net *net,

include/net/ipv6.h

@@ -938,7 +938,7 @@ static inline bool ipv6_can_nonlocal_bind(struct net *net,
 {
 	return net->ipv6.sysctl.ip_nonlocal_bind ||
 		test_bit(INET_FLAGS_FREEBIND, &inet->inet_flags) ||
-		inet->transparent;
+		test_bit(INET_FLAGS_TRANSPARENT, &inet->inet_flags);
 }
 
 /* Sysctl settings for net ipv6.auto_flowlabels */

include/net/route.h

@@ -298,7 +298,7 @@ static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst,
 {
 	__u8 flow_flags = 0;
 
-	if (inet_sk(sk)->transparent)
+	if (inet_test_bit(TRANSPARENT, sk))
 		flow_flags |= FLOWI_FLAG_ANYSRC;
 
 	flowi4_init_output(fl4, oif, READ_ONCE(sk->sk_mark), ip_sock_rt_tos(sk),

include/net/tcp.h

@@ -2031,7 +2031,7 @@ static inline bool inet_sk_transparent(const struct sock *sk)
 	case TCP_NEW_SYN_RECV:
 		return inet_rsk(inet_reqsk(sk))->no_srccheck;
 	}
-	return inet_sk(sk)->transparent;
+	return inet_test_bit(TRANSPARENT, sk);
 }
 
 /* Determines whether this is a thin stream (which may suffer from

net/ipv4/inet_diag.c

@@ -187,7 +187,7 @@ int inet_diag_msg_attrs_fill(struct sock *sk, struct sk_buff *skb,
 	inet_sockopt.freebind	= inet_test_bit(FREEBIND, sk);
 	inet_sockopt.hdrincl	= inet_test_bit(HDRINCL, sk);
 	inet_sockopt.mc_loop	= inet_test_bit(MC_LOOP, sk);
-	inet_sockopt.transparent = inet->transparent;
+	inet_sockopt.transparent = inet_test_bit(TRANSPARENT, sk);
 	inet_sockopt.mc_all	= inet_test_bit(MC_ALL, sk);
 	inet_sockopt.nodefrag	= inet->nodefrag;
 	inet_sockopt.bind_address_no_port = inet->bind_address_no_port;

net/ipv4/inet_timewait_sock.c

@@ -203,7 +203,7 @@ struct inet_timewait_sock *inet_twsk_alloc(const struct sock *sk,
 		tw->tw_reuseport    = sk->sk_reuseport;
 		tw->tw_hash	    = sk->sk_hash;
 		tw->tw_ipv6only	    = 0;
-		tw->tw_transparent  = inet->transparent;
+		tw->tw_transparent  = inet_test_bit(TRANSPARENT, sk);
 		tw->tw_prot	    = sk->sk_prot_creator;
 		atomic64_set(&tw->tw_cookie, atomic64_read(&sk->sk_cookie));
 		twsk_net_set(tw, sock_net(sk));

net/ipv4/ip_sockglue.c

@@ -1005,7 +1005,16 @@ int do_ip_setsockopt(struct sock *sk, int level, int optname,
 			return -EINVAL;
 		inet_assign_bit(MC_ALL, sk, val);
 		return 0;
-
+	case IP_TRANSPARENT:
+		if (!!val && !sockopt_ns_capable(sock_net(sk)->user_ns, CAP_NET_RAW) &&
+		    !sockopt_ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)) {
+			err = -EPERM;
+			break;
+		}
+		if (optlen < 1)
+			goto e_inval;
+		inet_assign_bit(TRANSPARENT, sk, val);
+		return 0;
 	}
 
 	err = 0;
@@ -1319,17 +1328,6 @@ int do_ip_setsockopt(struct sock *sk, int level, int optname,
 		err = xfrm_user_policy(sk, optname, optval, optlen);
 		break;
 
-	case IP_TRANSPARENT:
-		if (!!val && !sockopt_ns_capable(sock_net(sk)->user_ns, CAP_NET_RAW) &&
-		    !sockopt_ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)) {
-			err = -EPERM;
-			break;
-		}
-		if (optlen < 1)
-			goto e_inval;
-		inet->transparent = !!val;
-		break;
-
 	case IP_MINTTL:
 		if (optlen < 1)
 			goto e_inval;
@@ -1585,6 +1583,9 @@ int do_ip_getsockopt(struct sock *sk, int level, int optname,
 	case IP_MULTICAST_ALL:
 		val = inet_test_bit(MC_ALL, sk);
 		goto copyval;
+	case IP_TRANSPARENT:
+		val = inet_test_bit(TRANSPARENT, sk);
+		goto copyval;
 	}
 
 	if (needs_rtnl)
@@ -1735,9 +1736,6 @@ int do_ip_getsockopt(struct sock *sk, int level, int optname,
 		len -= msg.msg_controllen;
 		return copy_to_sockptr(optlen, &len, sizeof(int));
 	}
-	case IP_TRANSPARENT:
-		val = inet->transparent;
-		break;
 	case IP_MINTTL:
 		val = inet->min_ttl;
 		break;

net/ipv4/tcp_input.c

@@ -7000,7 +7000,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 
 	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
 	tcp_openreq_init(req, &tmp_opt, skb, sk);
-	inet_rsk(req)->no_srccheck = inet_sk(sk)->transparent;
+	inet_rsk(req)->no_srccheck = inet_test_bit(TRANSPARENT, sk);
 
 	/* Note: tcp_v6_init_req() might override ir_iif for link locals */
 	inet_rsk(req)->ir_iif = inet_request_bound_dev_if(sk, skb);

net/ipv4/tcp_minisocks.c

@@ -289,9 +289,8 @@ void tcp_time_wait(struct sock *sk, int state, int timeo)
 	if (tw) {
 		struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
 		const int rto = (icsk->icsk_rto << 2) - (icsk->icsk_rto >> 1);
-		struct inet_sock *inet = inet_sk(sk);
 
-		tw->tw_transparent	= inet->transparent;
+		tw->tw_transparent	= inet_test_bit(TRANSPARENT, sk);
 		tw->tw_mark		= sk->sk_mark;
 		tw->tw_priority		= sk->sk_priority;
 		tw->tw_rcv_wscale	= tp->rx_opt.rcv_wscale;

net/ipv6/ipv6_sockglue.c

@@ -633,7 +633,7 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
 		if (optlen < sizeof(int))
 			goto e_inval;
 		/* we don't have a separate transparent bit for IPV6 we use the one in the IPv4 socket */
-		inet_sk(sk)->transparent = valbool;
+		inet_assign_bit(TRANSPARENT, sk, valbool);
 		retv = 0;
 		break;
 
@@ -1330,7 +1330,7 @@ int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 	}
 
 	case IPV6_TRANSPARENT:
-		val = inet_sk(sk)->transparent;
+		val = inet_test_bit(TRANSPARENT, sk);
 		break;
 
 	case IPV6_FREEBIND:

net/mptcp/sockopt.c

@@ -416,7 +416,8 @@ static int mptcp_setsockopt_v6(struct mptcp_sock *msk, int optname,
 			sk->sk_ipv6only = ssk->sk_ipv6only;
 			break;
 		case IPV6_TRANSPARENT:
-			inet_sk(sk)->transparent = inet_sk(ssk)->transparent;
+			inet_assign_bit(TRANSPARENT, sk,
+					inet_test_bit(TRANSPARENT, ssk));
 			break;
 		case IPV6_FREEBIND:
 			inet_assign_bit(FREEBIND, sk,
@@ -685,7 +686,6 @@ static int mptcp_setsockopt_sol_ip_set_transparent(struct mptcp_sock *msk, int o
 						   sockptr_t optval, unsigned int optlen)
 {
 	struct sock *sk = (struct sock *)msk;
-	struct inet_sock *issk;
 	struct sock *ssk;
 	int err;
 
@@ -701,14 +701,13 @@ static int mptcp_setsockopt_sol_ip_set_transparent(struct mptcp_sock *msk, int o
 		return PTR_ERR(ssk);
 	}
 
-	issk = inet_sk(ssk);
-
 	switch (optname) {
 	case IP_FREEBIND:
 		inet_assign_bit(FREEBIND, ssk, inet_test_bit(FREEBIND, sk));
 		break;
 	case IP_TRANSPARENT:
-		issk->transparent = inet_sk(sk)->transparent;
+		inet_assign_bit(TRANSPARENT, ssk,
+				inet_test_bit(TRANSPARENT, sk));
 		break;
 	default:
 		release_sock(sk);
@@ -1441,7 +1440,7 @@ static void sync_socket_options(struct mptcp_sock *msk, struct sock *ssk)
 	__tcp_sock_set_cork(ssk, !!msk->cork);
 	__tcp_sock_set_nodelay(ssk, !!msk->nodelay);
 
-	inet_sk(ssk)->transparent = inet_sk(sk)->transparent;
+	inet_assign_bit(TRANSPARENT, ssk, inet_test_bit(TRANSPARENT, sk));
 	inet_assign_bit(FREEBIND, ssk, inet_test_bit(FREEBIND, sk));
 }