1 files changed, 104 insertions, 61 deletions

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index c5d5e68940ea..76958570ae7f 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1061,6 +1061,12 @@ out:
 	}
 }
 
+static void __mptcp_clean_una_wakeup(struct sock *sk)
+{
+	__mptcp_clean_una(sk);
+	mptcp_write_space(sk);
+}
+
 static void mptcp_enter_memory_pressure(struct sock *sk)
 {
 	struct mptcp_subflow_context *subflow;
@@ -1189,6 +1195,7 @@ static bool mptcp_tx_cache_refill(struct sock *sk, int size,
 			 */
 			while (skbs->qlen > 1) {
 				skb = __skb_dequeue_tail(skbs);
+				*total_ts -= skb->truesize;
 				__kfree_skb(skb);
 			}
 			return skbs->qlen > 0;
@@ -1444,7 +1451,7 @@ static void mptcp_push_release(struct sock *sk, struct sock *ssk,
 	release_sock(ssk);
 }
 
-static void mptcp_push_pending(struct sock *sk, unsigned int flags)
+static void __mptcp_push_pending(struct sock *sk, unsigned int flags)
 {
 	struct sock *prev_ssk = NULL, *ssk = NULL;
 	struct mptcp_sock *msk = mptcp_sk(sk);
@@ -1696,14 +1703,14 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 
 wait_for_memory:
 		mptcp_set_nospace(sk);
-		mptcp_push_pending(sk, msg->msg_flags);
+		__mptcp_push_pending(sk, msg->msg_flags);
 		ret = sk_stream_wait_memory(sk, &timeo);
 		if (ret)
 			goto out;
 	}
 
 	if (copied)
-		mptcp_push_pending(sk, msg->msg_flags);
+		__mptcp_push_pending(sk, msg->msg_flags);
 
 out:
 	release_sock(sk);
@@ -2115,6 +2122,14 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk)
 	return backup;
 }
 
+static void mptcp_dispose_initial_subflow(struct mptcp_sock *msk)
+{
+	if (msk->subflow) {
+		iput(SOCK_INODE(msk->subflow));
+		msk->subflow = NULL;
+	}
+}
+
 /* subflow sockets can be either outgoing (connect) or incoming
  * (accept).
  *
@@ -2126,6 +2141,8 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk)
 static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 			      struct mptcp_subflow_context *subflow)
 {
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
 	list_del(&subflow->node);
 
 	lock_sock_nested(ssk, SINGLE_DEPTH_NESTING);
@@ -2154,6 +2171,18 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 	release_sock(ssk);
 
 	sock_put(ssk);
+
+	if (ssk == msk->last_snd)
+		msk->last_snd = NULL;
+
+	if (ssk == msk->ack_hint)
+		msk->ack_hint = NULL;
+
+	if (ssk == msk->first)
+		msk->first = NULL;
+
+	if (msk->subflow && ssk == msk->subflow->sk)
+		mptcp_dispose_initial_subflow(msk);
 }
 
 void mptcp_close_ssk(struct sock *sk, struct sock *ssk,
@@ -2238,14 +2267,58 @@ static void mptcp_check_fastclose(struct mptcp_sock *msk)
 	mptcp_close_wake_up(sk);
 }
 
-static void mptcp_worker(struct work_struct *work)
+static void __mptcp_retrans(struct sock *sk)
 {
-	struct mptcp_sock *msk = container_of(work, struct mptcp_sock, work);
-	struct sock *ssk, *sk = &msk->sk.icsk_inet.sk;
+	struct mptcp_sock *msk = mptcp_sk(sk);
 	struct mptcp_sendmsg_info info = {};
 	struct mptcp_data_frag *dfrag;
 	size_t copied = 0;
-	int state, ret;
+	struct sock *ssk;
+	int ret;
+
+	__mptcp_clean_una_wakeup(sk);
+	dfrag = mptcp_rtx_head(sk);
+	if (!dfrag)
+		return;
+
+	ssk = mptcp_subflow_get_retrans(msk);
+	if (!ssk)
+		goto reset_timer;
+
+	lock_sock(ssk);
+
+	/* limit retransmission to the bytes already sent on some subflows */
+	info.sent = 0;
+	info.limit = dfrag->already_sent;
+	while (info.sent < dfrag->already_sent) {
+		if (!mptcp_alloc_tx_skb(sk, ssk))
+			break;
+
+		ret = mptcp_sendmsg_frag(sk, ssk, dfrag, &info);
+		if (ret <= 0)
+			break;
+
+		MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_RETRANSSEGS);
+		copied += ret;
+		info.sent += ret;
+	}
+	if (copied)
+		tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
+			 info.size_goal);
+
+	mptcp_set_timeout(sk, ssk);
+	release_sock(ssk);
+
+reset_timer:
+	if (!mptcp_timer_pending(sk))
+		mptcp_reset_timer(sk);
+}
+
+static void mptcp_worker(struct work_struct *work)
+{
+	struct mptcp_sock *msk = container_of(work, struct mptcp_sock, work);
+	struct sock *sk = &msk->sk.icsk_inet.sk;
+	int state;
 
 	lock_sock(sk);
 	state = sk->sk_state;
@@ -2280,45 +2353,8 @@ static void mptcp_worker(struct work_struct *work)
 	if (test_and_clear_bit(MPTCP_WORK_CLOSE_SUBFLOW, &msk->flags))
 		__mptcp_close_subflow(msk);
 
-	if (!test_and_clear_bit(MPTCP_WORK_RTX, &msk->flags))
-		goto unlock;
-
-	__mptcp_clean_una(sk);
-	dfrag = mptcp_rtx_head(sk);
-	if (!dfrag)
-		goto unlock;
-
-	ssk = mptcp_subflow_get_retrans(msk);
-	if (!ssk)
-		goto reset_unlock;
-
-	lock_sock(ssk);
-
-	/* limit retransmission to the bytes already sent on some subflows */
-	info.sent = 0;
-	info.limit = dfrag->already_sent;
-	while (info.sent < dfrag->already_sent) {
-		if (!mptcp_alloc_tx_skb(sk, ssk))
-			break;
-
-		ret = mptcp_sendmsg_frag(sk, ssk, dfrag, &info);
-		if (ret <= 0)
-			break;
-
-		MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_RETRANSSEGS);
-		copied += ret;
-		info.sent += ret;
-	}
-	if (copied)
-		tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
-			 info.size_goal);
-
-	mptcp_set_timeout(sk, ssk);
-	release_sock(ssk);
-
-reset_unlock:
-	if (!mptcp_timer_pending(sk))
-		mptcp_reset_timer(sk);
+	if (test_and_clear_bit(MPTCP_WORK_RTX, &msk->flags))
+		__mptcp_retrans(sk);
 
 unlock:
 	release_sock(sk);
@@ -2523,12 +2559,6 @@ static void __mptcp_destroy_sock(struct sock *sk)
 
 	might_sleep();
 
-	/* dispose the ancillatory tcp socket, if any */
-	if (msk->subflow) {
-		iput(SOCK_INODE(msk->subflow));
-		msk->subflow = NULL;
-	}
-
 	/* be sure to always acquire the join list lock, to sync vs
 	 * mptcp_finish_join().
 	 */
@@ -2553,6 +2583,7 @@ static void __mptcp_destroy_sock(struct sock *sk)
 	sk_stream_kill_queues(sk);
 	xfrm_sk_free_policy(sk);
 	sk_refcnt_debug_release(sk);
+	mptcp_dispose_initial_subflow(msk);
 	sock_put(sk);
 }
 
@@ -2934,13 +2965,14 @@ static void mptcp_release_cb(struct sock *sk)
 {
 	unsigned long flags, nflags;
 
-	/* push_pending may touch wmem_reserved, do it before the later
-	 * cleanup
-	 */
-	if (test_and_clear_bit(MPTCP_CLEAN_UNA, &mptcp_sk(sk)->flags))
-		__mptcp_clean_una(sk);
-	if (test_and_clear_bit(MPTCP_PUSH_PENDING, &mptcp_sk(sk)->flags)) {
-		/* mptcp_push_pending() acquires the subflow socket lock
+	for (;;) {
+		flags = 0;
+		if (test_and_clear_bit(MPTCP_PUSH_PENDING, &mptcp_sk(sk)->flags))
+			flags |= MPTCP_PUSH_PENDING;
+		if (!flags)
+			break;
+
+		/* the following actions acquire the subflow socket lock
 		 *
 		 * 1) can't be invoked in atomic scope
 		 * 2) must avoid ABBA deadlock with msk socket spinlock: the RX
@@ -2949,13 +2981,21 @@ static void mptcp_release_cb(struct sock *sk)
 		 */
 
 		spin_unlock_bh(&sk->sk_lock.slock);
-		mptcp_push_pending(sk, 0);
+		if (flags & MPTCP_PUSH_PENDING)
+			__mptcp_push_pending(sk, 0);
+
+		cond_resched();
 		spin_lock_bh(&sk->sk_lock.slock);
 	}
+
+	if (test_and_clear_bit(MPTCP_CLEAN_UNA, &mptcp_sk(sk)->flags))
+		__mptcp_clean_una_wakeup(sk);
 	if (test_and_clear_bit(MPTCP_ERROR_REPORT, &mptcp_sk(sk)->flags))
 		__mptcp_error_report(sk);
 
-	/* clear any wmem reservation and errors */
+	/* push_pending may touch wmem_reserved, ensure we do the cleanup
+	 * later
+	 */
 	__mptcp_update_wmem(sk);
 	__mptcp_update_rmem(sk);
 
@@ -3285,6 +3325,9 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 		/* PM/worker can now acquire the first subflow socket
 		 * lock without racing with listener queue cleanup,
 		 * we can notify it, if needed.
+		 *
+		 * Even if remote has reset the initial subflow by now
+		 * the refcnt is still at least one.
 		 */
 		subflow = mptcp_subflow_ctx(msk->first);
 		list_add(&subflow->node, &msk->conn_list);