mptcp: New MP_CAPABLE receiver side

Keys are now sent by the client after the 3-way handshake. Thus, we need
to delay some parts of the state initialization to that moment.

Signed-off-by: Christoph Paasch <cpaasch@apple.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>

include/net/mptcp.h

@@ -102,7 +102,8 @@ struct mptcp_request_sock {
 
 	u8				loc_id;
 	u8				rem_id; /* Address-id in the MP_JOIN */
-	u8				dss_csum:1,
+	u16				dss_csum:1,
+					rem_key_set:1,
 					is_sub:1, /* Is this a new subflow? */
 					low_prio:1, /* Interface set to low-prio? */
 					rcv_low_prio:1,
@@ -272,6 +273,7 @@ struct mptcp_cb {
 
 	u16	send_infinite_mapping:1,
 		send_mptcpv1_mpcapable:1,
+		rem_key_set:1,
 		in_time_wait:1,
 		list_rcvd:1, /* XXX TO REMOVE */
 		addr_signal:1, /* Path-manager wants us to call addr_signal */
@@ -811,10 +813,11 @@ void mptcp_options_write(__be32 *ptr, struct tcp_sock *tp,
 void mptcp_close(struct sock *meta_sk, long timeout);
 bool mptcp_doit(struct sock *sk);
 int mptcp_create_master_sk(struct sock *meta_sk, __u64 remote_key,
-			   __u8 mptcp_ver, u32 window);
+			   int rem_key_set, __u8 mptcp_ver, u32 window);
 int mptcp_check_req_fastopen(struct sock *child, struct request_sock *req);
 int mptcp_check_req_master(struct sock *sk, struct sock *child,
 			   struct request_sock *req, const struct sk_buff *skb,
+			   const struct mptcp_options_received *mopt,
 			   int drop, u32 tsoff);
 struct sock *mptcp_check_req_child(struct sock *meta_sk,
 				   struct sock *child,
@@ -838,6 +841,8 @@ void mptcp_sub_close_wq(struct work_struct *work);
 void mptcp_sub_close(struct sock *sk, unsigned long delay);
 struct sock *mptcp_select_ack_sock(const struct sock *meta_sk);
 void mptcp_prepare_for_backlog(struct sock *sk, struct sk_buff *skb);
+void mptcp_initialize_recv_vars(struct tcp_sock *meta_tp, struct mptcp_cb *mpcb,
+				__u64 remote_key);
 int mptcp_backlog_rcv(struct sock *meta_sk, struct sk_buff *skb);
 void mptcp_ack_handler(struct timer_list *t);
 bool mptcp_check_rtt(const struct tcp_sock *tp, int time);
@@ -1415,6 +1420,7 @@ static inline int mptcp_check_req_master(const struct sock *sk,
 					 const struct sock *child,
 					 const struct request_sock *req,
 					 const struct sk_buff *skb,
+					 const struct mptcp_options_received *mopt,
 					 int drop,
 					 u32 tsoff)
 {

include/net/tcp.h

@@ -582,6 +582,7 @@ void inet_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb);
 /* From syncookies.c */
 struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
 				 struct request_sock *req,
+				 const struct mptcp_options_received *mopt,
 				 struct dst_entry *dst, u32 tsoff);
 int __cookie_v4_check(const struct iphdr *iph, const struct tcphdr *th,
 		      u32 cookie);

net/ipv4/syncookies.c

@@ -203,6 +203,7 @@ EXPORT_SYMBOL_GPL(__cookie_v4_check);
 
 struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
 				 struct request_sock *req,
+				 const struct mptcp_options_received *mopt,
 				 struct dst_entry *dst, u32 tsoff)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
@@ -219,7 +220,7 @@ struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
 	if (!child)
 		goto listen_overflow;
 
-	ret = mptcp_check_req_master(sk, child, req, skb, 0, tsoff);
+	ret = mptcp_check_req_master(sk, child, req, skb, mopt, 0, tsoff);
 	if (ret < 0)
 		return NULL;
 
@@ -428,7 +429,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
 	ireq->rcv_wscale  = rcv_wscale;
 	ireq->ecn_ok = cookie_ecn_ok(&tcp_opt, sock_net(sk), &rt->dst);
 
-	ret = tcp_get_cookie_sock(sk, skb, req, &rt->dst, tsoff);
+	ret = tcp_get_cookie_sock(sk, skb, req, &mopt, &rt->dst, tsoff);
 	/* ip_queue_xmit() depends on our flow being setup
 	 * Normal sockets get it right from inet_csk_route_child_sock()
 	 */

net/ipv4/tcp_minisocks.c

@@ -828,7 +828,7 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 		goto listen_overflow;
 
 	if (own_req && !is_meta_sk(sk)) {
-		int ret = mptcp_check_req_master(sk, child, req, skb, 1, 0);
+		int ret = mptcp_check_req_master(sk, child, req, skb, &mopt, 1, 0);
 		if (ret < 0)
 			goto listen_overflow;
 

net/ipv6/syncookies.c

@@ -267,7 +267,7 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 	ireq->rcv_wscale = rcv_wscale;
 	ireq->ecn_ok = cookie_ecn_ok(&tcp_opt, sock_net(sk), dst);
 
-	ret = tcp_get_cookie_sock(sk, skb, req, dst, tsoff);
+	ret = tcp_get_cookie_sock(sk, skb, req, &mopt, dst, tsoff);
 out:
 	return ret;
 out_free:

net/mptcp/mptcp_ctrl.c

@@ -321,7 +321,11 @@ static void mptcp_reqsk_new_mptcp(struct request_sock *req,
 	spin_unlock(&mptcp_tk_hashlock);
 	local_bh_enable();
 	rcu_read_unlock();
-	mtreq->mptcp_rem_key = mopt->mptcp_sender_key;
+
+	if (mtreq->mptcp_ver == MPTCP_VERSION_0) {
+		mtreq->mptcp_rem_key = mopt->mptcp_sender_key;
+		mtreq->rem_key_set = 1;
+	}
 }
 
 static int mptcp_reqsk_new_cookie(struct request_sock *req,
@@ -357,7 +361,10 @@ static int mptcp_reqsk_new_cookie(struct request_sock *req,
 	local_bh_enable();
 	rcu_read_unlock();
 
-	mtreq->mptcp_rem_key = mopt->mptcp_sender_key;
+	if (mtreq->mptcp_ver == MPTCP_VERSION_0) {
+		mtreq->mptcp_rem_key = mopt->mptcp_sender_key;
+		mtreq->rem_key_set = 1;
+	}
 
 	return true;
 }
@@ -1236,14 +1243,33 @@ static const struct tcp_sock_ops mptcp_sub_specific = {
 	.set_cong_ctrl                  = __tcp_set_congestion_control,
 };
 
+void mptcp_initialize_recv_vars(struct tcp_sock *meta_tp, struct mptcp_cb *mpcb,
+				__u64 remote_key)
+{
+	u64 idsn;
+
+	mpcb->mptcp_rem_key = remote_key;
+	mpcb->rem_key_set = 1;
+	mptcp_key_sha1(mpcb->mptcp_rem_key, &mpcb->mptcp_rem_token, &idsn);
+
+	idsn++;
+	mpcb->rcv_high_order[0] = idsn >> 32;
+	mpcb->rcv_high_order[1] = mpcb->rcv_high_order[0] + 1;
+	meta_tp->copied_seq = (u32)idsn;
+	meta_tp->rcv_nxt = (u32)idsn;
+	meta_tp->rcv_wup = (u32)idsn;
+
+	meta_tp->snd_wl1 = meta_tp->rcv_nxt - 1;
+}
+
 static int mptcp_alloc_mpcb(struct sock *meta_sk, __u64 remote_key,
-			    __u8 mptcp_ver, u32 window)
+			    int rem_key_set, __u8 mptcp_ver, u32 window)
 {
 	struct mptcp_cb *mpcb;
 	struct sock *master_sk;
 	struct inet_connection_sock *meta_icsk = inet_csk(meta_sk);
 	struct tcp_sock *master_tp, *meta_tp = tcp_sk(meta_sk);
-	u64 snd_idsn, rcv_idsn;
+	u64 snd_idsn;
 
 	dst_release(meta_sk->sk_rx_dst);
 	meta_sk->sk_rx_dst = NULL;
@@ -1276,12 +1302,6 @@ static int mptcp_alloc_mpcb(struct sock *meta_sk, __u64 remote_key,
 	mpcb->snd_high_order[0] = snd_idsn >> 32;
 	mpcb->snd_high_order[1] = mpcb->snd_high_order[0] - 1;
 
-	mpcb->mptcp_rem_key = remote_key;
-	mptcp_key_sha1(mpcb->mptcp_rem_key, &mpcb->mptcp_rem_token, &rcv_idsn);
-	rcv_idsn++;
-	mpcb->rcv_high_order[0] = rcv_idsn >> 32;
-	mpcb->rcv_high_order[1] = mpcb->rcv_high_order[0] + 1;
-
 	mpcb->meta_sk = meta_sk;
 	mpcb->master_sk = master_sk;
 
@@ -1393,11 +1413,9 @@ static int mptcp_alloc_mpcb(struct sock *meta_sk, __u64 remote_key,
 	meta_tp->pushed_seq = meta_tp->write_seq;
 	meta_tp->snd_up = meta_tp->write_seq;
 
-	meta_tp->copied_seq = (u32)rcv_idsn;
-	meta_tp->rcv_nxt = (u32)rcv_idsn;
-	meta_tp->rcv_wup = (u32)rcv_idsn;
+	if (rem_key_set)
+		mptcp_initialize_recv_vars(meta_tp, mpcb, remote_key);
 
-	meta_tp->snd_wl1 = meta_tp->rcv_nxt - 1;
 	meta_tp->snd_wnd = window;
 	meta_tp->retrans_stamp = 0; /* Set in tcp_connect() */
 
@@ -2144,12 +2162,12 @@ bool mptcp_doit(struct sock *sk)
 }
 
 int mptcp_create_master_sk(struct sock *meta_sk, __u64 remote_key,
-			   __u8 mptcp_ver, u32 window)
+			   int rem_key_set, __u8 mptcp_ver, u32 window)
 {
 	struct tcp_sock *master_tp;
 	struct sock *master_sk;
 
-	if (mptcp_alloc_mpcb(meta_sk, remote_key, mptcp_ver, window))
+	if (mptcp_alloc_mpcb(meta_sk, remote_key, rem_key_set, mptcp_ver, window))
 		goto err_alloc_mpcb;
 
 	master_sk = tcp_sk(meta_sk)->mpcb->master_sk;
@@ -2177,6 +2195,7 @@ int mptcp_create_master_sk(struct sock *meta_sk, __u64 remote_key,
 }
 
 static int __mptcp_check_req_master(struct sock *child,
+				    const struct mptcp_options_received *mopt,
 				    struct request_sock *req)
 {
 	struct tcp_sock *child_tp = tcp_sk(child);
@@ -2188,6 +2207,8 @@ static int __mptcp_check_req_master(struct sock *child,
 	if (!inet_rsk(req)->mptcp_rqsk)
 		return 1;
 
+	mtreq = mptcp_rsk(req);
+
 	if (!inet_rsk(req)->saw_mpc) {
 		/* Fallback to regular TCP, because we saw one SYN without
 		 * MP_CAPABLE. In tcp_check_req we continue the regular path.
@@ -2199,15 +2220,21 @@ static int __mptcp_check_req_master(struct sock *child,
 		return 1;
 	}
 
+	/* mopt can be NULL when coming from FAST-OPEN */
+	if (mopt && mopt->saw_mpc && mtreq->mptcp_ver == MPTCP_VERSION_1) {
+		mtreq->mptcp_rem_key = mopt->mptcp_sender_key;
+		mtreq->rem_key_set = 1;
+	}
+
 	MPTCP_INC_STATS(sock_net(meta_sk), MPTCP_MIB_MPCAPABLEPASSIVEACK);
 
 	/* Just set this values to pass them to mptcp_alloc_mpcb */
-	mtreq = mptcp_rsk(req);
 	child_tp->mptcp_loc_key = mtreq->mptcp_loc_key;
 	child_tp->mptcp_loc_token = mtreq->mptcp_loc_token;
 
 	if (mptcp_create_master_sk(meta_sk, mtreq->mptcp_rem_key,
-				   mtreq->mptcp_ver, child_tp->snd_wnd)) {
+				   mtreq->rem_key_set, mtreq->mptcp_ver,
+				   child_tp->snd_wnd)) {
 		inet_csk_prepare_forced_close(meta_sk);
 		tcp_done(meta_sk);
 
@@ -2242,7 +2269,7 @@ int mptcp_check_req_fastopen(struct sock *child, struct request_sock *req)
 	u32 new_mapping;
 	int ret;
 
-	ret = __mptcp_check_req_master(child, req);
+	ret = __mptcp_check_req_master(child, NULL, req);
 	if (ret)
 		return ret;
 
@@ -2285,12 +2312,13 @@ int mptcp_check_req_fastopen(struct sock *child, struct request_sock *req)
 
 int mptcp_check_req_master(struct sock *sk, struct sock *child,
 			   struct request_sock *req, const struct sk_buff *skb,
+			   const struct mptcp_options_received *mopt,
 			   int drop, u32 tsoff)
 {
 	struct sock *meta_sk = child;
 	int ret;
 
-	ret = __mptcp_check_req_master(child, req);
+	ret = __mptcp_check_req_master(child, mopt, req);
 	if (ret)
 		return ret;
 	child = tcp_sk(child)->mpcb->master_sk;
@@ -2658,8 +2686,10 @@ void mptcp_cookies_reqsk_init(struct request_sock *req,
 	/* Absolutely need to always initialize this. */
 	mtreq->hash_entry.pprev = NULL;
 
+	mtreq->mptcp_ver = mopt->mptcp_ver;
 	mtreq->mptcp_rem_key = mopt->mptcp_sender_key;
 	mtreq->mptcp_loc_key = mopt->mptcp_receiver_key;
+	mtreq->rem_key_set = 1;
 
 	/* Generate the token */
 	mptcp_key_sha1(mtreq->mptcp_loc_key, &mtreq->mptcp_loc_token, NULL);