mptcp: invoke MP_FAIL response when needed

mptcp_mp_fail_no_response shouldn't be invoked on each worker run, it
should be invoked only when MP_FAIL response timeout occurs.

This patch refactors the MP_FAIL response logic.

Add fail_tout in mptcp_sock to record the MP_FAIL timestamp. Check it
in mptcp_worker() before invoking mptcp_mp_fail_no_response(). Drop
the code to reuse sk_timer for MP_FAIL response.

Add fail_ssk struct member in struct mptcp_sock to record the MP_FAIL
subsocket. It can replace the mp_fail_response_expect flag in struct
mptcp_subflow_context. Drop mp_fail_response_expect_subflow() helper,
just use this fail_ssk in mptcp_mp_fail_no_response() to reset the
subflow.

Closes: #281
Fixes: d9fb797 ("mptcp: Do not traverse the subflow connection list without lock")
Acked-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Geliang Tang <geliang.tang@suse.com>

net/mptcp/pm.c

@@ -299,23 +299,20 @@ void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
 	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
-	struct sock *s = (struct sock *)msk;
 
 	pr_debug("fail_seq=%llu", fail_seq);
 
 	if (!READ_ONCE(msk->allow_infinite_fallback))
 		return;
 
-	if (!READ_ONCE(subflow->mp_fail_response_expect)) {
+	if (!msk->fail_ssk) {
 		pr_debug("send MP_FAIL response and infinite map");
 
 		subflow->send_mp_fail = 1;
 		MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_MPFAILTX);
 		subflow->send_infinite_map = 1;
-	} else if (!sock_flag(sk, SOCK_DEAD)) {
+	} else {
 		pr_debug("MP_FAIL response received");
-
-		sk_stop_timer(s, &s->sk_timer);
 	}
 }
 

net/mptcp/protocol.c

@@ -2175,21 +2175,6 @@ static void mptcp_retransmit_timer(struct timer_list *t)
 	sock_put(sk);
 }
 
-static struct mptcp_subflow_context *
-mp_fail_response_expect_subflow(struct mptcp_sock *msk)
-{
-	struct mptcp_subflow_context *subflow, *ret = NULL;
-
-	mptcp_for_each_subflow(msk, subflow) {
-		if (READ_ONCE(subflow->mp_fail_response_expect)) {
-			ret = subflow;
-			break;
-		}
-	}
-
-	return ret;
-}
-
 static void mptcp_timeout_timer(struct timer_list *t)
 {
 	struct sock *sk = from_timer(sk, t, sk_timer);
@@ -2520,19 +2505,16 @@ static void __mptcp_retrans(struct sock *sk)
 
 static void mptcp_mp_fail_no_response(struct mptcp_sock *msk)
 {
-	struct mptcp_subflow_context *subflow;
-	struct sock *ssk;
+	struct sock *ssk = msk->fail_ssk;
 	bool slow;
 
-	subflow = mp_fail_response_expect_subflow(msk);
-	if (subflow) {
-		pr_debug("MP_FAIL doesn't respond, reset the subflow");
+	pr_debug("MP_FAIL doesn't respond, reset the subflow");
 
-		ssk = mptcp_subflow_tcp_sock(subflow);
-		slow = lock_sock_fast(ssk);
-		mptcp_subflow_reset(ssk);
-		unlock_sock_fast(ssk, slow);
-	}
+	slow = lock_sock_fast(ssk);
+	mptcp_subflow_reset(ssk);
+	unlock_sock_fast(ssk, slow);
+
+	msk->fail_ssk = NULL;
 }
 
 static void mptcp_worker(struct work_struct *work)
@@ -2575,7 +2557,8 @@ static void mptcp_worker(struct work_struct *work)
 	if (test_and_clear_bit(MPTCP_WORK_RTX, &msk->flags))
 		__mptcp_retrans(sk);
 
-	mptcp_mp_fail_no_response(msk);
+	if (msk->fail_ssk && time_after(jiffies, msk->fail_tout))
+		mptcp_mp_fail_no_response(msk);
 
 unlock:
 	release_sock(sk);
@@ -2602,6 +2585,8 @@ static int __mptcp_init_sock(struct sock *sk)
 	WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
 	WRITE_ONCE(msk->allow_infinite_fallback, true);
 	msk->recovery = false;
+	msk->fail_ssk = NULL;
+	msk->fail_tout = 0;
 
 	mptcp_pm_data_init(msk);
 

net/mptcp/protocol.h

@@ -306,6 +306,8 @@ struct mptcp_sock {
 
 	u32 setsockopt_seq;
 	char		ca_name[TCP_CA_NAME_MAX];
+	struct sock	*fail_ssk;
+	unsigned long	fail_tout;
 };
 
 #define mptcp_data_lock(sk) spin_lock_bh(&(sk)->sk_lock.slock)
@@ -468,7 +470,6 @@ struct mptcp_subflow_context {
 		local_id_valid : 1, /* local_id is correctly initialized */
 		valid_csum_seen : 1;        /* at least one csum validated */
 	enum mptcp_data_avail data_avail;
-	bool	mp_fail_response_expect;
 	u32	remote_nonce;
 	u64	thmac;
 	u32	local_nonce;

net/mptcp/subflow.c

@@ -974,7 +974,6 @@ static enum mapping_status get_mapping_status(struct sock *ssk,
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk);
 	bool csum_reqd = READ_ONCE(msk->csum_enabled);
-	struct sock *sk = (struct sock *)msk;
 	struct mptcp_ext *mpext;
 	struct sk_buff *skb;
 	u16 data_len;
@@ -1016,9 +1015,6 @@ static enum mapping_status get_mapping_status(struct sock *ssk,
 		pr_debug("infinite mapping received");
 		MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_INFINITEMAPRX);
 		subflow->map_data_len = 0;
-		if (!sock_flag(ssk, SOCK_DEAD))
-			sk_stop_timer(sk, &sk->sk_timer);
-
 		return MAPPING_INVALID;
 	}
 
@@ -1238,11 +1234,9 @@ static bool subflow_check_data_avail(struct sock *ssk)
 				tcp_send_active_reset(ssk, GFP_ATOMIC);
 				while ((skb = skb_peek(&ssk->sk_receive_queue)))
 					sk_eat_skb(ssk, skb);
-			} else if (!sock_flag(ssk, SOCK_DEAD)) {
-				WRITE_ONCE(subflow->mp_fail_response_expect, true);
-				sk_reset_timer((struct sock *)msk,
-					       &((struct sock *)msk)->sk_timer,
-					       jiffies + TCP_RTO_MAX);
+			} else {
+				msk->fail_ssk = ssk;
+				msk->fail_tout = jiffies + TCP_RTO_MAX;
 			}
 			WRITE_ONCE(subflow->data_avail, MPTCP_SUBFLOW_NODATA);
 			return true;