mptcp: plug races between subflow fail and subflow creation

We have races similar to the one addressed by the previous patch between
subflow failing and additional subflow creation. They are just harder to
trigger.

The solution is similar. Use a separate flag to track the condition
'socket state prevent any additional subflow creation' protected by the
fallback lock.

The socket fallback makes such flag true, and also receiving or sending
an MP_FAIL option.

The field 'allow_infinite_fallback' is now always touched under the
relevant lock, we can drop the ONCE annotation on write.

Fixes: 478d770 ("mptcp: send out MP_FAIL when data checksum fails")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Link: https://patch.msgid.link/20250714-net-mptcp-fallback-races-v1-2-391aff963322@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Author: Paolo Abeni

net/mptcp/pm.c

@@ -765,8 +765,14 @@ void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
 
 	pr_debug("fail_seq=%llu\n", fail_seq);
 
-	if (!READ_ONCE(msk->allow_infinite_fallback))
+	/* After accepting the fail, we can't create any other subflows */
+	spin_lock_bh(&msk->fallback_lock);
+	if (!msk->allow_infinite_fallback) {
+		spin_unlock_bh(&msk->fallback_lock);
 		return;
+	}
+	msk->allow_subflows = false;
+	spin_unlock_bh(&msk->fallback_lock);
 
 	if (!subflow->fail_tout) {
 		pr_debug("send MP_FAIL response and infinite map\n");

net/mptcp/protocol.c

@@ -791,7 +791,7 @@ void mptcp_data_ready(struct sock *sk, struct sock *ssk)
 static void mptcp_subflow_joined(struct mptcp_sock *msk, struct sock *ssk)
 {
 	mptcp_subflow_ctx(ssk)->map_seq = READ_ONCE(msk->ack_seq);
-	WRITE_ONCE(msk->allow_infinite_fallback, false);
+	msk->allow_infinite_fallback = false;
 	mptcp_event(MPTCP_EVENT_SUB_ESTABLISHED, msk, ssk, GFP_ATOMIC);
 }
 
@@ -803,7 +803,7 @@ static bool __mptcp_finish_join(struct mptcp_sock *msk, struct sock *ssk)
 		return false;
 
 	spin_lock_bh(&msk->fallback_lock);
-	if (__mptcp_check_fallback(msk)) {
+	if (!msk->allow_subflows) {
 		spin_unlock_bh(&msk->fallback_lock);
 		return false;
 	}
@@ -2625,7 +2625,7 @@ static void __mptcp_retrans(struct sock *sk)
 				len = max(copied, len);
 				tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
 					 info.size_goal);
-				WRITE_ONCE(msk->allow_infinite_fallback, false);
+				msk->allow_infinite_fallback = false;
 			}
 			spin_unlock_bh(&msk->fallback_lock);
 
@@ -2753,7 +2753,8 @@ static void __mptcp_init_sock(struct sock *sk)
 	WRITE_ONCE(msk->first, NULL);
 	inet_csk(sk)->icsk_sync_mss = mptcp_sync_mss;
 	WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
-	WRITE_ONCE(msk->allow_infinite_fallback, true);
+	msk->allow_infinite_fallback = true;
+	msk->allow_subflows = true;
 	msk->recovery = false;
 	msk->subflow_id = 1;
 	msk->last_data_sent = tcp_jiffies32;
@@ -3549,7 +3550,7 @@ bool mptcp_finish_join(struct sock *ssk)
 	/* active subflow, already present inside the conn_list */
 	if (!list_empty(&subflow->node)) {
 		spin_lock_bh(&msk->fallback_lock);
-		if (__mptcp_check_fallback(msk)) {
+		if (!msk->allow_subflows) {
 			spin_unlock_bh(&msk->fallback_lock);
 			return false;
 		}

net/mptcp/protocol.h

@@ -346,13 +346,15 @@ struct mptcp_sock {
 		u64	rtt_us; /* last maximum rtt of subflows */
 	} rcvq_space;
 	u8		scaling_ratio;
+	bool		allow_subflows;
 
 	u32		subflow_id;
 	u32		setsockopt_seq;
 	char		ca_name[TCP_CA_NAME_MAX];
 
-	spinlock_t	fallback_lock;	/* protects fallback and
-					 * allow_infinite_fallback
+	spinlock_t	fallback_lock;	/* protects fallback,
+					 * allow_infinite_fallback and
+					 * allow_join
 					 */
 };
 
@@ -1232,6 +1234,7 @@ static inline bool __mptcp_try_fallback(struct mptcp_sock *msk)
 		return false;
 	}
 
+	msk->allow_subflows = false;
 	set_bit(MPTCP_FALLBACK_DONE, &msk->flags);
 	spin_unlock_bh(&msk->fallback_lock);
 	return true;

net/mptcp/subflow.c

@@ -1302,20 +1302,29 @@ static void subflow_sched_work_if_closed(struct mptcp_sock *msk, struct sock *ss
 		mptcp_schedule_work(sk);
 }
 
-static void mptcp_subflow_fail(struct mptcp_sock *msk, struct sock *ssk)
+static bool mptcp_subflow_fail(struct mptcp_sock *msk, struct sock *ssk)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk);
 	unsigned long fail_tout;
 
+	/* we are really failing, prevent any later subflow join */
+	spin_lock_bh(&msk->fallback_lock);
+	if (!msk->allow_infinite_fallback) {
+		spin_unlock_bh(&msk->fallback_lock);
+		return false;
+	}
+	msk->allow_subflows = false;
+	spin_unlock_bh(&msk->fallback_lock);
+
 	/* graceful failure can happen only on the MPC subflow */
 	if (WARN_ON_ONCE(ssk != READ_ONCE(msk->first)))
-		return;
+		return false;
 
 	/* since the close timeout take precedence on the fail one,
 	 * no need to start the latter when the first is already set
 	 */
 	if (sock_flag((struct sock *)msk, SOCK_DEAD))
-		return;
+		return true;
 
 	/* we don't need extreme accuracy here, use a zero fail_tout as special
 	 * value meaning no fail timeout at all;
@@ -1327,6 +1336,7 @@ static void mptcp_subflow_fail(struct mptcp_sock *msk, struct sock *ssk)
 	tcp_send_ack(ssk);
 
 	mptcp_reset_tout_timer(msk, subflow->fail_tout);
+	return true;
 }
 
 static bool subflow_check_data_avail(struct sock *ssk)
@@ -1387,12 +1397,11 @@ static bool subflow_check_data_avail(struct sock *ssk)
 		    (subflow->mp_join || subflow->valid_csum_seen)) {
 			subflow->send_mp_fail = 1;
 
-			if (!READ_ONCE(msk->allow_infinite_fallback)) {
+			if (!mptcp_subflow_fail(msk, ssk)) {
 				subflow->reset_transient = 0;
 				subflow->reset_reason = MPTCP_RST_EMIDDLEBOX;
 				goto reset;
 			}
-			mptcp_subflow_fail(msk, ssk);
 			WRITE_ONCE(subflow->data_avail, true);
 			return true;
 		}