Merge branch 'mptcp-avoid-workqueue-usage-for-data'

Paolo Abeni says:

====================
mptcp: avoid workqueue usage for data

The current locking schema used to protect the MPTCP data-path
requires the usage of the MPTCP workqueue to process the incoming
data, depending on trylock result.

The above poses scalability limits and introduces random delays
in MPTCP-level acks.

With this series we use a single spinlock to protect the MPTCP
data-path, removing the need for workqueue and delayed ack usage.

This additionally reduces the number of atomic operations required
per packet and cleans-up considerably the poll/wake-up code.
====================

Link: https://lore.kernel.org/r/cover.1606413118.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

7 files changed, 598 insertions, 235 deletions

diff --git a/include/net/sock.h b/include/net/sock.h
index 80469c2c448d..f59764614e30 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -1590,6 +1590,7 @@ static inline void lock_sock(struct sock *sk)
 	lock_sock_nested(sk, 0);
 }
 
+void __lock_sock(struct sock *sk);
 void __release_sock(struct sock *sk);
 void release_sock(struct sock *sk);
 
diff --git a/net/core/sock.c b/net/core/sock.c
index 9badbe7bb4e4..f0f096852876 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -2486,7 +2486,7 @@ bool sk_page_frag_refill(struct sock *sk, struct page_frag *pfrag)
 }
 EXPORT_SYMBOL(sk_page_frag_refill);
 
-static void __lock_sock(struct sock *sk)
+void __lock_sock(struct sock *sk)
 	__releases(&sk->sk_lock.slock)
 	__acquires(&sk->sk_lock.slock)
 {
diff --git a/net/mptcp/mptcp_diag.c b/net/mptcp/mptcp_diag.c
index 5f390a97f556..b70ae4ba3000 100644
--- a/net/mptcp/mptcp_diag.c
+++ b/net/mptcp/mptcp_diag.c
@@ -140,7 +140,7 @@ static void mptcp_diag_get_info(struct sock *sk, struct inet_diag_msg *r,
 	info->mptcpi_flags = flags;
 	info->mptcpi_token = READ_ONCE(msk->token);
 	info->mptcpi_write_seq = READ_ONCE(msk->write_seq);
-	info->mptcpi_snd_una = atomic64_read(&msk->snd_una);
+	info->mptcpi_snd_una = READ_ONCE(msk->snd_una);
 	info->mptcpi_rcv_nxt = READ_ONCE(msk->ack_seq);
 	unlock_sock_fast(sk, slow);
 }
diff --git a/net/mptcp/options.c b/net/mptcp/options.c
index 8a59b3e44599..6b7b4b67f18c 100644
--- a/net/mptcp/options.c
+++ b/net/mptcp/options.c
@@ -830,18 +830,20 @@ static u64 expand_ack(u64 old_ack, u64 cur_ack, bool use_64bit)
 }
 
 static void ack_update_msk(struct mptcp_sock *msk,
-			   const struct sock *ssk,
+			   struct sock *ssk,
 			   struct mptcp_options_received *mp_opt)
 {
-	u64 new_snd_una, snd_una, old_snd_una = atomic64_read(&msk->snd_una);
-	u64 new_wnd_end, wnd_end, old_wnd_end = atomic64_read(&msk->wnd_end);
-	u64 snd_nxt = READ_ONCE(msk->snd_nxt);
+	u64 new_wnd_end, new_snd_una, snd_nxt = READ_ONCE(msk->snd_nxt);
 	struct sock *sk = (struct sock *)msk;
+	u64 old_snd_una;
+
+	mptcp_data_lock(sk);
 
 	/* avoid ack expansion on update conflict, to reduce the risk of
 	 * wrongly expanding to a future ack sequence number, which is way
 	 * more dangerous than missing an ack
 	 */
+	old_snd_una = msk->snd_una;
 	new_snd_una = expand_ack(old_snd_una, mp_opt->data_ack, mp_opt->ack64);
 
 	/* ACK for data not even sent yet? Ignore. */
@@ -850,26 +852,16 @@ static void ack_update_msk(struct mptcp_sock *msk,
 
 	new_wnd_end = new_snd_una + tcp_sk(ssk)->snd_wnd;
 
-	while (after64(new_wnd_end, old_wnd_end)) {
-		wnd_end = old_wnd_end;
-		old_wnd_end = atomic64_cmpxchg(&msk->wnd_end, wnd_end,
-					       new_wnd_end);
-		if (old_wnd_end == wnd_end) {
-			if (mptcp_send_head(sk))
-				mptcp_schedule_work(sk);
-			break;
-		}
+	if (after64(new_wnd_end, msk->wnd_end)) {
+		msk->wnd_end = new_wnd_end;
+		__mptcp_wnd_updated(sk, ssk);
 	}
 
-	while (after64(new_snd_una, old_snd_una)) {
-		snd_una = old_snd_una;
-		old_snd_una = atomic64_cmpxchg(&msk->snd_una, snd_una,
-					       new_snd_una);
-		if (old_snd_una == snd_una) {
-			mptcp_data_acked(sk);
-			break;
-		}
+	if (after64(new_snd_una, old_snd_una)) {
+		msk->snd_una = new_snd_una;
+		__mptcp_data_acked(sk);
 	}
+	mptcp_data_unlock(sk);
 }
 
 bool mptcp_update_rcv_data_fin(struct mptcp_sock *msk, u64 data_fin_seq, bool use_64bit)
@@ -922,8 +914,19 @@ void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb)
 	struct mptcp_options_received mp_opt;
 	struct mptcp_ext *mpext;
 
-	if (__mptcp_check_fallback(msk))
+	if (__mptcp_check_fallback(msk)) {
+		/* Keep it simple and unconditionally trigger send data cleanup and
+		 * pending queue spooling. We will need to acquire the data lock
+		 * for more accurate checks, and once the lock is acquired, such
+		 * helpers are cheap.
+		 */
+		mptcp_data_lock(subflow->conn);
+		if (mptcp_send_head(subflow->conn))
+			__mptcp_wnd_updated(subflow->conn, sk);
+		__mptcp_data_acked(subflow->conn);
+		mptcp_data_unlock(subflow->conn);
 		return;
+	}
 
 	mptcp_get_options(skb, &mp_opt);
 	if (!check_fully_established(msk, sk, subflow, skb, &mp_opt))
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 16e9cb1c79cc..221f7cdd416b 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -60,7 +60,7 @@ static struct socket *__mptcp_nmpc_socket(const struct mptcp_sock *msk)
 /* Returns end sequence number of the receiver's advertised window */
 static u64 mptcp_wnd_end(const struct mptcp_sock *msk)
 {
-	return atomic64_read(&msk->wnd_end);
+	return READ_ONCE(msk->wnd_end);
 }
 
 static bool mptcp_is_tcpsk(struct sock *sk)
@@ -348,17 +348,22 @@ static void mptcp_close_wake_up(struct sock *sk)
 		sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
 }
 
-static void mptcp_check_data_fin_ack(struct sock *sk)
+static bool mptcp_pending_data_fin_ack(struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 
-	if (__mptcp_check_fallback(msk))
-		return;
+	return !__mptcp_check_fallback(msk) &&
+	       ((1 << sk->sk_state) &
+		(TCPF_FIN_WAIT1 | TCPF_CLOSING | TCPF_LAST_ACK)) &&
+	       msk->write_seq == READ_ONCE(msk->snd_una);
+}
+
+static void mptcp_check_data_fin_ack(struct sock *sk)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
 
 	/* Look for an acknowledged DATA_FIN */
-	if (((1 << sk->sk_state) &
-	     (TCPF_FIN_WAIT1 | TCPF_CLOSING | TCPF_LAST_ACK)) &&
-	    msk->write_seq == atomic64_read(&msk->snd_una)) {
+	if (mptcp_pending_data_fin_ack(sk)) {
 		mptcp_stop_timer(sk);
 
 		WRITE_ONCE(msk->snd_data_fin_enable, 0);
@@ -453,15 +458,15 @@ static bool mptcp_subflow_cleanup_rbuf(struct sock *ssk)
 
 static void mptcp_cleanup_rbuf(struct mptcp_sock *msk)
 {
+	struct sock *ack_hint = READ_ONCE(msk->ack_hint);
 	struct mptcp_subflow_context *subflow;
 
 	/* if the hinted ssk is still active, try to use it */
-	if (likely(msk->ack_hint)) {
+	if (likely(ack_hint)) {
 		mptcp_for_each_subflow(msk, subflow) {
 			struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 
-			if (msk->ack_hint == ssk &&
-			    mptcp_subflow_cleanup_rbuf(ssk))
+			if (ack_hint == ssk && mptcp_subflow_cleanup_rbuf(ssk))
 				return;
 		}
 	}
@@ -614,13 +619,13 @@ static bool __mptcp_move_skbs_from_subflow(struct mptcp_sock *msk,
 			break;
 		}
 	} while (more_data_avail);
-	msk->ack_hint = ssk;
+	WRITE_ONCE(msk->ack_hint, ssk);
 
 	*bytes += moved;
 	return done;
 }
 
-static bool mptcp_ofo_queue(struct mptcp_sock *msk)
+static bool __mptcp_ofo_queue(struct mptcp_sock *msk)
 {
 	struct sock *sk = (struct sock *)msk;
 	struct sk_buff *skb, *tail;
@@ -666,34 +671,27 @@ static bool mptcp_ofo_queue(struct mptcp_sock *msk)
 /* In most cases we will be able to lock the mptcp socket.  If its already
  * owned, we need to defer to the work queue to avoid ABBA deadlock.
  */
-static bool move_skbs_to_msk(struct mptcp_sock *msk, struct sock *ssk)
+static void move_skbs_to_msk(struct mptcp_sock *msk, struct sock *ssk)
 {
 	struct sock *sk = (struct sock *)msk;
 	unsigned int moved = 0;
 
-	if (READ_ONCE(sk->sk_lock.owned))
-		return false;
-
-	if (unlikely(!spin_trylock_bh(&sk->sk_lock.slock)))
-		return false;
-
-	/* must re-check after taking the lock */
-	if (!READ_ONCE(sk->sk_lock.owned)) {
-		__mptcp_move_skbs_from_subflow(msk, ssk, &moved);
-		mptcp_ofo_queue(msk);
+	if (inet_sk_state_load(sk) == TCP_CLOSE)
+		return;
 
-		/* If the moves have caught up with the DATA_FIN sequence number
-		 * it's time to ack the DATA_FIN and change socket state, but
-		 * this is not a good place to change state. Let the workqueue
-		 * do it.
-		 */
-		if (mptcp_pending_data_fin(sk, NULL))
-			mptcp_schedule_work(sk);
-	}
+	mptcp_data_lock(sk);
 
-	spin_unlock_bh(&sk->sk_lock.slock);
+	__mptcp_move_skbs_from_subflow(msk, ssk, &moved);
+	__mptcp_ofo_queue(msk);
 
-	return moved > 0;
+	/* If the moves have caught up with the DATA_FIN sequence number
+	 * it's time to ack the DATA_FIN and change socket state, but
+	 * this is not a good place to change state. Let the workqueue
+	 * do it.
+	 */
+	if (mptcp_pending_data_fin(sk, NULL))
+		mptcp_schedule_work(sk);
+	mptcp_data_unlock(sk);
 }
 
 void mptcp_data_ready(struct sock *sk, struct sock *ssk)
@@ -771,16 +769,6 @@ bool mptcp_schedule_work(struct sock *sk)
 	return false;
 }
 
-void mptcp_data_acked(struct sock *sk)
-{
-	mptcp_reset_timer(sk);
-
-	if ((test_bit(MPTCP_NOSPACE, &mptcp_sk(sk)->flags) ||
-	     mptcp_send_head(sk) ||
-	     (inet_sk_state_load(sk) != TCP_ESTABLISHED)))
-		mptcp_schedule_work(sk);
-}
-
 void mptcp_subflow_eof(struct sock *sk)
 {
 	if (!test_and_set_bit(MPTCP_WORK_EOF, &mptcp_sk(sk)->flags))
@@ -825,16 +813,6 @@ static void mptcp_check_for_eof(struct mptcp_sock *msk)
 	mptcp_close_wake_up(sk);
 }
 
-static bool mptcp_ext_cache_refill(struct mptcp_sock *msk)
-{
-	const struct sock *sk = (const struct sock *)msk;
-
-	if (!msk->cached_ext)
-		msk->cached_ext = __skb_ext_alloc(sk->sk_allocation);
-
-	return !!msk->cached_ext;
-}
-
 static struct sock *mptcp_subflow_recv_lookup(const struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow;
@@ -873,6 +851,121 @@ static bool mptcp_frag_can_collapse_to(const struct mptcp_sock *msk,
 		df->data_seq + df->data_len == msk->write_seq;
 }
 
+static int mptcp_wmem_with_overhead(struct sock *sk, int size)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+	int ret, skbs;
+
+	ret = size + ((sizeof(struct mptcp_data_frag) * size) >> PAGE_SHIFT);
+	skbs = (msk->tx_pending_data + size) / msk->size_goal_cache;
+	if (skbs < msk->skb_tx_cache.qlen)
+		return ret;
+
+	return ret + (skbs - msk->skb_tx_cache.qlen) * SKB_TRUESIZE(MAX_TCP_HEADER);
+}
+
+static void __mptcp_wmem_reserve(struct sock *sk, int size)
+{
+	int amount = mptcp_wmem_with_overhead(sk, size);
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	WARN_ON_ONCE(msk->wmem_reserved);
+	if (amount <= sk->sk_forward_alloc)
+		goto reserve;
+
+	/* under memory pressure try to reserve at most a single page
+	 * otherwise try to reserve the full estimate and fallback
+	 * to a single page before entering the error path
+	 */
+	if ((tcp_under_memory_pressure(sk) && amount > PAGE_SIZE) ||
+	    !sk_wmem_schedule(sk, amount)) {
+		if (amount <= PAGE_SIZE)
+			goto nomem;
+
+		amount = PAGE_SIZE;
+		if (!sk_wmem_schedule(sk, amount))
+			goto nomem;
+	}
+
+reserve:
+	msk->wmem_reserved = amount;
+	sk->sk_forward_alloc -= amount;
+	return;
+
+nomem:
+	/* we will wait for memory on next allocation */
+	msk->wmem_reserved = -1;
+}
+
+static void __mptcp_update_wmem(struct sock *sk)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	if (!msk->wmem_reserved)
+		return;
+
+	if (msk->wmem_reserved < 0)
+		msk->wmem_reserved = 0;
+	if (msk->wmem_reserved > 0) {
+		sk->sk_forward_alloc += msk->wmem_reserved;
+		msk->wmem_reserved = 0;
+	}
+}
+
+static bool mptcp_wmem_alloc(struct sock *sk, int size)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	/* check for pre-existing error condition */
+	if (msk->wmem_reserved < 0)
+		return false;
+
+	if (msk->wmem_reserved >= size)
+		goto account;
+
+	mptcp_data_lock(sk);
+	if (!sk_wmem_schedule(sk, size)) {
+		mptcp_data_unlock(sk);
+		return false;
+	}
+
+	sk->sk_forward_alloc -= size;
+	msk->wmem_reserved += size;
+	mptcp_data_unlock(sk);
+
+account:
+	msk->wmem_reserved -= size;
+	return true;
+}
+
+static void mptcp_wmem_uncharge(struct sock *sk, int size)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	if (msk->wmem_reserved < 0)
+		msk->wmem_reserved = 0;
+	msk->wmem_reserved += size;
+}
+
+static void mptcp_mem_reclaim_partial(struct sock *sk)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	/* if we are experiencing a transint allocation error,
+	 * the forward allocation memory has been already
+	 * released
+	 */
+	if (msk->wmem_reserved < 0)
+		return;
+
+	mptcp_data_lock(sk);
+	sk->sk_forward_alloc += msk->wmem_reserved;
+	sk_mem_reclaim_partial(sk);
+	msk->wmem_reserved = sk->sk_forward_alloc;
+	sk->sk_forward_alloc = 0;
+	mptcp_data_unlock(sk);
+}
+
 static void dfrag_uncharge(struct sock *sk, int len)
 {
 	sk_mem_uncharge(sk, len);
@@ -888,7 +981,7 @@ static void dfrag_clear(struct sock *sk, struct mptcp_data_frag *dfrag)
 	put_page(dfrag->page);
 }
 
-static void mptcp_clean_una(struct sock *sk)
+static void __mptcp_clean_una(struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	struct mptcp_data_frag *dtmp, *dfrag;
@@ -899,10 +992,9 @@ static void mptcp_clean_una(struct sock *sk)
 	 * plain TCP
 	 */
 	if (__mptcp_check_fallback(msk))
-		atomic64_set(&msk->snd_una, msk->snd_nxt);
-
-	snd_una = atomic64_read(&msk->snd_una);
+		msk->snd_una = READ_ONCE(msk->snd_nxt);
 
+	snd_una = msk->snd_una;
 	list_for_each_entry_safe(dfrag, dtmp, &msk->rtx_queue, list) {
 		if (after64(dfrag->data_seq + dfrag->data_len, snd_una))
 			break;
@@ -930,36 +1022,34 @@ static void mptcp_clean_una(struct sock *sk)
 	}
 
 out:
-	if (cleaned)
-		sk_mem_reclaim_partial(sk);
-}
-
-static void mptcp_clean_una_wakeup(struct sock *sk)
-{
-	struct mptcp_sock *msk = mptcp_sk(sk);
+	if (cleaned) {
+		if (tcp_under_memory_pressure(sk)) {
+			__mptcp_update_wmem(sk);
+			sk_mem_reclaim_partial(sk);
+		}
 
-	mptcp_clean_una(sk);
+		if (sk_stream_is_writeable(sk)) {
+			/* pairs with memory barrier in mptcp_poll */
+			smp_mb();
+			if (test_and_clear_bit(MPTCP_NOSPACE, &msk->flags))
+				sk_stream_write_space(sk);
+		}
+	}
 
-	/* Only wake up writers if a subflow is ready */
-	if (sk_stream_is_writeable(sk)) {
-		clear_bit(MPTCP_NOSPACE, &msk->flags);
-		sk_stream_write_space(sk);
+	if (snd_una == READ_ONCE(msk->snd_nxt)) {
+		if (msk->timer_ival)
+			mptcp_stop_timer(sk);
+	} else {
+		mptcp_reset_timer(sk);
 	}
 }
 
-/* ensure we get enough memory for the frag hdr, beyond some minimal amount of
- * data
- */
-static bool mptcp_page_frag_refill(struct sock *sk, struct page_frag *pfrag)
+static void mptcp_enter_memory_pressure(struct sock *sk)
 {
 	struct mptcp_subflow_context *subflow;
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	bool first = true;
 
-	if (likely(skb_page_frag_refill(32U + sizeof(struct mptcp_data_frag),
-					pfrag, sk->sk_allocation)))
-		return true;
-
 	sk_stream_moderate_sndbuf(sk);
 	mptcp_for_each_subflow(msk, subflow) {
 		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
@@ -969,6 +1059,18 @@ static bool mptcp_page_frag_refill(struct sock *sk, struct page_frag *pfrag)
 		sk_stream_moderate_sndbuf(ssk);
 		first = false;
 	}
+}
+
+/* ensure we get enough memory for the frag hdr, beyond some minimal amount of
+ * data
+ */
+static bool mptcp_page_frag_refill(struct sock *sk, struct page_frag *pfrag)
+{
+	if (likely(skb_page_frag_refill(32U + sizeof(struct mptcp_data_frag),
+					pfrag, sk->sk_allocation)))
+		return true;
+
+	mptcp_enter_memory_pressure(sk);
 	return false;
 }
 
@@ -1015,6 +1117,128 @@ static int mptcp_check_allowed_size(struct mptcp_sock *msk, u64 data_seq,
 	return avail_size;
 }
 
+static bool __mptcp_add_ext(struct sk_buff *skb, gfp_t gfp)
+{
+	struct skb_ext *mpext = __skb_ext_alloc(gfp);
+
+	if (!mpext)
+		return false;
+	__skb_ext_set(skb, SKB_EXT_MPTCP, mpext);
+	return true;
+}
+
+static struct sk_buff *__mptcp_do_alloc_tx_skb(struct sock *sk, gfp_t gfp)
+{
+	struct sk_buff *skb;
+
+	skb = alloc_skb_fclone(MAX_TCP_HEADER, gfp);
+	if (likely(skb)) {
+		if (likely(__mptcp_add_ext(skb, gfp))) {
+			skb_reserve(skb, MAX_TCP_HEADER);
+			skb->reserved_tailroom = skb->end - skb->tail;
+			return skb;
+		}
+		__kfree_skb(skb);
+	} else {
+		mptcp_enter_memory_pressure(sk);
+	}
+	return NULL;
+}
+
+static bool mptcp_tx_cache_refill(struct sock *sk, int size,
+				  struct sk_buff_head *skbs, int *total_ts)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+	struct sk_buff *skb;
+	int space_needed;
+
+	if (unlikely(tcp_under_memory_pressure(sk))) {
+		mptcp_mem_reclaim_partial(sk);
+
+		/* under pressure pre-allocate at most a single skb */
+		if (msk->skb_tx_cache.qlen)
+			return true;
+		space_needed = msk->size_goal_cache;
+	} else {
+		space_needed = msk->tx_pending_data + size -
+			       msk->skb_tx_cache.qlen * msk->size_goal_cache;
+	}
+
+	while (space_needed > 0) {
+		skb = __mptcp_do_alloc_tx_skb(sk, sk->sk_allocation);
+		if (unlikely(!skb)) {
+			/* under memory pressure, try to pass the caller a
+			 * single skb to allow forward progress
+			 */
+			while (skbs->qlen > 1) {
+				skb = __skb_dequeue_tail(skbs);
+				__kfree_skb(skb);
+			}
+			return skbs->qlen > 0;
+		}
+
+		*total_ts += skb->truesize;
+		__skb_queue_tail(skbs, skb);
+		space_needed -= msk->size_goal_cache;
+	}
+	return true;
+}
+
+static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+	struct sk_buff *skb;
+
+	if (ssk->sk_tx_skb_cache) {
+		skb = ssk->sk_tx_skb_cache;
+		if (unlikely(!skb_ext_find(skb, SKB_EXT_MPTCP) &&
+			     !__mptcp_add_ext(skb, gfp)))
+			return false;
+		return true;
+	}
+
+	skb = skb_peek(&msk->skb_tx_cache);
+	if (skb) {
+		if (likely(sk_wmem_schedule(ssk, skb->truesize))) {
+			skb = __skb_dequeue(&msk->skb_tx_cache);
+			if (WARN_ON_ONCE(!skb))
+				return false;
+
+			mptcp_wmem_uncharge(sk, skb->truesize);
+			ssk->sk_tx_skb_cache = skb;
+			return true;
+		}
+
+		/* over memory limit, no point to try to allocate a new skb */
+		return false;
+	}
+
+	skb = __mptcp_do_alloc_tx_skb(sk, gfp);
+	if (!skb)
+		return false;
+
+	if (likely(sk_wmem_schedule(ssk, skb->truesize))) {
+		ssk->sk_tx_skb_cache = skb;
+		return true;
+	}
+	kfree_skb(skb);
+	return false;
+}
+
+static bool mptcp_must_reclaim_memory(struct sock *sk, struct sock *ssk)
+{
+	return !ssk->sk_tx_skb_cache &&
+	       !skb_peek(&mptcp_sk(sk)->skb_tx_cache) &&
+	       tcp_under_memory_pressure(sk);
+}
+
+static bool mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk)
+{
+	if (unlikely(mptcp_must_reclaim_memory(sk, ssk)))
+		mptcp_mem_reclaim_partial(sk);
+	return __mptcp_alloc_tx_skb(sk, ssk, sk->sk_allocation);
+}
+
 static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 			      struct mptcp_data_frag *dfrag,
 			      struct mptcp_sendmsg_info *info)
@@ -1026,7 +1250,7 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 	struct sk_buff *skb, *tail;
 	bool can_collapse = false;
 	int avail_size;
-	size_t ret;
+	size_t ret = 0;
 
 	pr_debug("msk=%p ssk=%p sending dfrag at seq=%lld len=%d already sent=%d",
 		 msk, ssk, dfrag->data_seq, dfrag->data_len, info->sent);
@@ -1034,6 +1258,7 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 	/* compute send limit */
 	info->mss_now = tcp_send_mss(ssk, &info->size_goal, info->flags);
 	avail_size = info->size_goal;
+	msk->size_goal_cache = info->size_goal;
 	skb = tcp_write_queue_tail(ssk);
 	if (skb) {
 		/* Limit the write to the size available in the
@@ -1054,10 +1279,12 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 	/* Zero window and all data acked? Probe. */
 	avail_size = mptcp_check_allowed_size(msk, data_seq, avail_size);
 	if (avail_size == 0) {
-		if (skb || atomic64_read(&msk->snd_una) != msk->snd_nxt)
+		u64 snd_una = READ_ONCE(msk->snd_una);
+
+		if (skb || snd_una != msk->snd_nxt)
 			return 0;
 		zero_window_probe = true;
-		data_seq = atomic64_read(&msk->snd_una) - 1;
+		data_seq = snd_una - 1;
 		avail_size = 1;
 	}
 
@@ -1082,8 +1309,11 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 		goto out;
 	}
 
-	mpext = __skb_ext_set(tail, SKB_EXT_MPTCP, msk->cached_ext);
-	msk->cached_ext = NULL;
+	mpext = skb_ext_find(tail, SKB_EXT_MPTCP);
+	if (WARN_ON_ONCE(!mpext)) {
+		/* should never reach here, stream corrupted */
+		return -EINVAL;
+	}
 
 	memset(mpext, 0, sizeof(*mpext));
 	mpext->data_seq = data_seq;
@@ -1107,31 +1337,6 @@ out:
 	return ret;
 }
 
-static void mptcp_nospace(struct mptcp_sock *msk)
-{
-	struct mptcp_subflow_context *subflow;
-
-	set_bit(MPTCP_NOSPACE, &msk->flags);
-	smp_mb__after_atomic(); /* msk->flags is changed by write_space cb */
-
-	mptcp_for_each_subflow(msk, subflow) {
-		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
-		bool ssk_writeable = sk_stream_is_writeable(ssk);
-		struct socket *sock = READ_ONCE(ssk->sk_socket);
-
-		if (ssk_writeable || !sock)
-			continue;
-
-		/* enables ssk->write_space() callbacks */
-		set_bit(SOCK_NOSPACE, &sock->flags);
-	}
-
-	/* mptcp_data_acked() could run just before we set the NOSPACE bit,
-	 * so explicitly check for snd_una value
-	 */
-	mptcp_clean_una((struct sock *)msk);
-}
-
 #define MPTCP_SEND_BURST_SIZE		((1 << 16) - \
 					 sizeof(struct tcphdr) - \
 					 MAX_TCP_OPTION_SPACE - \
@@ -1156,9 +1361,6 @@ static struct sock *mptcp_subflow_get_send(struct mptcp_sock *msk,
 	sock_owned_by_me((struct sock *)msk);
 
 	*sndbuf = 0;
-	if (!mptcp_ext_cache_refill(msk))
-		return NULL;
-
 	if (__mptcp_check_fallback(msk)) {
 		if (!msk->first)
 			return NULL;
@@ -1267,6 +1469,15 @@ static void mptcp_push_pending(struct sock *sk, unsigned int flags)
 			if (ssk != prev_ssk || !prev_ssk)
 				lock_sock(ssk);
 
+			/* keep it simple and always provide a new skb for the
+			 * subflow, even if we will not use it when collapsing
+			 * on the pending one
+			 */
+			if (!mptcp_alloc_tx_skb(sk, ssk)) {
+				mptcp_push_release(sk, ssk, &info);
+				goto out;
+			}
+
 			ret = mptcp_sendmsg_frag(sk, ssk, dfrag, &info);
 			if (ret <= 0) {
 				mptcp_push_release(sk, ssk, &info);
@@ -1277,6 +1488,7 @@ static void mptcp_push_pending(struct sock *sk, unsigned int flags)
 			dfrag->already_sent += ret;
 			msk->snd_nxt += ret;
 			msk->snd_burst -= ret;
+			msk->tx_pending_data -= ret;
 			copied += ret;
 			len -= ret;
 		}
@@ -1296,6 +1508,63 @@ out:
 	}
 }
 
+static void __mptcp_subflow_push_pending(struct sock *sk, struct sock *ssk)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+	struct mptcp_sendmsg_info info;
+	struct mptcp_data_frag *dfrag;
+	int len, copied = 0;
+
+	info.flags = 0;
+	while ((dfrag = mptcp_send_head(sk))) {
+		info.sent = dfrag->already_sent;
+		info.limit = dfrag->data_len;
+		len = dfrag->data_len - dfrag->already_sent;
+		while (len > 0) {
+			int ret = 0;
+
+			/* do auto tuning */
+			if (!(sk->sk_userlocks & SOCK_SNDBUF_LOCK) &&
+			    ssk->sk_sndbuf > READ_ONCE(sk->sk_sndbuf))
+				WRITE_ONCE(sk->sk_sndbuf, ssk->sk_sndbuf);
+
+			if (unlikely(mptcp_must_reclaim_memory(sk, ssk))) {
+				__mptcp_update_wmem(sk);
+				sk_mem_reclaim_partial(sk);
+			}
+			if (!__mptcp_alloc_tx_skb(sk, ssk, GFP_ATOMIC))
+				goto out;
+
+			ret = mptcp_sendmsg_frag(sk, ssk, dfrag, &info);
+			if (ret <= 0)
+				goto out;
+
+			info.sent += ret;
+			dfrag->already_sent += ret;
+			msk->snd_nxt += ret;
+			msk->snd_burst -= ret;
+			msk->tx_pending_data -= ret;
+			copied += ret;
+			len -= ret;
+		}
+		WRITE_ONCE(msk->first_pending, mptcp_send_next(sk));
+	}
+
+out:
+	/* __mptcp_alloc_tx_skb could have released some wmem and we are
+	 * not going to flush it via release_sock()
+	 */
+	__mptcp_update_wmem(sk);
+	if (copied) {
+		mptcp_set_timeout(sk, ssk);
+		tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
+			 info.size_goal);
+		if (msk->snd_data_fin_enable &&
+		    msk->snd_nxt + 1 == msk->write_seq)
+			mptcp_schedule_work(sk);
+	}
+}
+
 static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
@@ -1307,7 +1576,7 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	if (msg->msg_flags & ~(MSG_MORE | MSG_DONTWAIT | MSG_NOSIGNAL))
 		return -EOPNOTSUPP;
 
-	lock_sock(sk);
+	mptcp_lock_sock(sk, __mptcp_wmem_reserve(sk, len));
 
 	timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
 
@@ -1318,11 +1587,11 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	}
 
 	pfrag = sk_page_frag(sk);
-	mptcp_clean_una(sk);
 
 	while (msg_data_left(msg)) {
+		int total_ts, frag_truesize = 0;
 		struct mptcp_data_frag *dfrag;
-		int frag_truesize = 0;
+		struct sk_buff_head skbs;
 		bool dfrag_collapsed;
 		size_t psize, offset;
 
@@ -1337,11 +1606,9 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		dfrag = mptcp_pending_tail(sk);
 		dfrag_collapsed = mptcp_frag_can_collapse_to(msk, pfrag, dfrag);
 		if (!dfrag_collapsed) {
-			if (!sk_stream_memory_free(sk)) {
-				mptcp_push_pending(sk, msg->msg_flags);
-				if (!sk_stream_memory_free(sk))
-					goto wait_for_memory;
-			}
+			if (!sk_stream_memory_free(sk))
+				goto wait_for_memory;
+
 			if (!mptcp_page_frag_refill(sk, pfrag))
 				goto wait_for_memory;
 
@@ -1356,11 +1623,20 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		offset = dfrag->offset + dfrag->data_len;
 		psize = pfrag->size - offset;
 		psize = min_t(size_t, psize, msg_data_left(msg));
-		if (!sk_wmem_schedule(sk, psize + frag_truesize))
+		total_ts = psize + frag_truesize;
+		__skb_queue_head_init(&skbs);
+		if (!mptcp_tx_cache_refill(sk, psize, &skbs, &total_ts))
 			goto wait_for_memory;
 
+		if (!mptcp_wmem_alloc(sk, total_ts)) {
+			__skb_queue_purge(&skbs);
+			goto wait_for_memory;
+		}
+
+		skb_queue_splice_tail(&skbs, &msk->skb_tx_cache);
 		if (copy_page_from_iter(dfrag->page, offset, psize,
 					&msg->msg_iter) != psize) {
+			mptcp_wmem_uncharge(sk, psize + frag_truesize);
 			ret = -EFAULT;
 			goto out;
 		}
@@ -1376,7 +1652,6 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		 * Note: we charge such data both to sk and ssk
 		 */
 		sk_wmem_queued_add(sk, frag_truesize);
-		sk->sk_forward_alloc -= frag_truesize;
 		if (!dfrag_collapsed) {
 			get_page(dfrag->page);
 			list_add_tail(&dfrag->list, &msk->rtx_queue);
@@ -1387,21 +1662,20 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 			 dfrag->data_seq, dfrag->data_len, dfrag->already_sent,
 			 !dfrag_collapsed);
 
-		if (!mptcp_ext_cache_refill(msk))
-			goto wait_for_memory;
 		continue;
 
 wait_for_memory:
-		mptcp_nospace(msk);
-		if (mptcp_timer_pending(sk))
-			mptcp_reset_timer(sk);
+		set_bit(MPTCP_NOSPACE, &msk->flags);
+		mptcp_push_pending(sk, msg->msg_flags);
 		ret = sk_stream_wait_memory(sk, &timeo);
 		if (ret)
 			goto out;
 	}
 
-	if (copied)
+	if (copied) {
+		msk->tx_pending_data += copied;
 		mptcp_push_pending(sk, msg->msg_flags);
+	}
 
 out:
 	release_sock(sk);
@@ -1427,11 +1701,10 @@ static int __mptcp_recvmsg_mskq(struct mptcp_sock *msk,
 				struct msghdr *msg,
 				size_t len)
 {
-	struct sock *sk = (struct sock *)msk;
 	struct sk_buff *skb;
 	int copied = 0;
 
-	while ((skb = skb_peek(&sk->sk_receive_queue)) != NULL) {
+	while ((skb = skb_peek(&msk->receive_queue)) != NULL) {
 		u32 offset = MPTCP_SKB_CB(skb)->offset;
 		u32 data_len = skb->len - offset;
 		u32 count = min_t(size_t, len - copied, data_len);
@@ -1451,7 +1724,10 @@ static int __mptcp_recvmsg_mskq(struct mptcp_sock *msk,
 			break;
 		}
 
-		__skb_unlink(skb, &sk->sk_receive_queue);
+		/* we will bulk release the skb memory later */
+		skb->destructor = NULL;
+		msk->rmem_released += skb->truesize;
+		__skb_unlink(skb, &msk->receive_queue);
 		__kfree_skb(skb);
 
 		if (copied >= len)
@@ -1559,25 +1835,47 @@ new_measure:
 	msk->rcvq_space.time = mstamp;
 }
 
+static void __mptcp_update_rmem(struct sock *sk)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	if (!msk->rmem_released)
+		return;
+
+	atomic_sub(msk->rmem_released, &sk->sk_rmem_alloc);
+	sk_mem_uncharge(sk, msk->rmem_released);
+	msk->rmem_released = 0;
+}
+
+static void __mptcp_splice_receive_queue(struct sock *sk)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+
+	skb_queue_splice_tail_init(&sk->sk_receive_queue, &msk->receive_queue);
+}
+
 static bool __mptcp_move_skbs(struct mptcp_sock *msk, unsigned int rcv)
 {
+	struct sock *sk = (struct sock *)msk;
 	unsigned int moved = 0;
-	bool done;
-
-	/* avoid looping forever below on racing close */
-	if (((struct sock *)msk)->sk_state == TCP_CLOSE)
-		return false;
+	bool ret, done;
 
 	__mptcp_flush_join_list(msk);
 	do {
 		struct sock *ssk = mptcp_subflow_recv_lookup(msk);
 		bool slowpath;
 
-		if (!ssk)
+		/* we can have data pending in the subflows only if the msk
+		 * receive buffer was full at subflow_data_ready() time,
+		 * that is an unlikely slow path.
+		 */
+		if (likely(!ssk))
 			break;
 
 		slowpath = lock_sock_fast(ssk);
+		mptcp_data_lock(sk);
 		done = __mptcp_move_skbs_from_subflow(msk, ssk, &moved);
+		mptcp_data_unlock(sk);
 		if (moved && rcv) {
 			WRITE_ONCE(msk->rmem_pending, min(rcv, moved));
 			tcp_cleanup_rbuf(ssk, 1);
@@ -1586,11 +1884,19 @@ static bool __mptcp_move_skbs(struct mptcp_sock *msk, unsigned int rcv)
 		unlock_sock_fast(ssk, slowpath);
 	} while (!done);
 
-	if (mptcp_ofo_queue(msk) || moved > 0) {
-		mptcp_check_data_fin((struct sock *)msk);
-		return true;
+	/* acquire the data lock only if some input data is pending */
+	ret = moved > 0;
+	if (!RB_EMPTY_ROOT(&msk->out_of_order_queue) ||
+	    !skb_queue_empty_lockless(&sk->sk_receive_queue)) {
+		mptcp_data_lock(sk);
+		__mptcp_update_rmem(sk);
+		ret |= __mptcp_ofo_queue(msk);
+		__mptcp_splice_receive_queue(sk);
+		mptcp_data_unlock(sk);
 	}
-	return false;
+	if (ret)
+		mptcp_check_data_fin((struct sock *)msk);
+	return !skb_queue_empty(&msk->receive_queue);
 }
 
 static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
@@ -1604,7 +1910,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 	if (msg->msg_flags & ~(MSG_WAITALL | MSG_DONTWAIT))
 		return -EOPNOTSUPP;
 
-	lock_sock(sk);
+	mptcp_lock_sock(sk, __mptcp_splice_receive_queue(sk));
 	if (unlikely(sk->sk_state == TCP_LISTEN)) {
 		copied = -ENOTCONN;
 		goto out_err;
@@ -1614,7 +1920,6 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 
 	len = min_t(size_t, len, INT_MAX);
 	target = sock_rcvlowat(sk, flags & MSG_WAITALL, len);
-	__mptcp_flush_join_list(msk);
 
 	for (;;) {
 		int bytes_read, old_space;
@@ -1628,7 +1933,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 
 		copied += bytes_read;
 
-		if (skb_queue_empty(&sk->sk_receive_queue) &&
+		if (skb_queue_empty(&msk->receive_queue) &&
 		    __mptcp_move_skbs(msk, len - copied))
 			continue;
 
@@ -1659,8 +1964,14 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 			if (test_and_clear_bit(MPTCP_WORK_EOF, &msk->flags))
 				mptcp_check_for_eof(msk);
 
-			if (sk->sk_shutdown & RCV_SHUTDOWN)
+			if (sk->sk_shutdown & RCV_SHUTDOWN) {
+				/* race breaker: the shutdown could be after the
+				 * previous receive queue check
+				 */
+				if (__mptcp_move_skbs(msk, len - copied))
+					continue;
 				break;
+			}
 
 			if (sk->sk_state == TCP_CLOSE) {
 				copied = -ENOTCONN;
@@ -1682,7 +1993,8 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 		mptcp_wait_data(sk, &timeo);
 	}
 
-	if (skb_queue_empty(&sk->sk_receive_queue)) {
+	if (skb_queue_empty_lockless(&sk->sk_receive_queue) &&
+	    skb_queue_empty(&msk->receive_queue)) {
 		/* entire backlog drained, clear DATA_READY. */
 		clear_bit(MPTCP_DATA_READY, &msk->flags);
 
@@ -1698,7 +2010,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 out_err:
 	pr_debug("msk=%p data_ready=%d rx queue empty=%d copied=%d",
 		 msk, test_bit(MPTCP_DATA_READY, &msk->flags),
-		 skb_queue_empty(&sk->sk_receive_queue), copied);
+		 skb_queue_empty_lockless(&sk->sk_receive_queue), copied);
 	mptcp_rcv_space_adjust(msk, copied);
 
 	release_sock(sk);
@@ -1709,12 +2021,8 @@ static void mptcp_retransmit_handler(struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 
-	if (atomic64_read(&msk->snd_una) == READ_ONCE(msk->snd_nxt)) {
-		mptcp_stop_timer(sk);
-	} else {
-		set_bit(MPTCP_WORK_RTX, &msk->flags);
-		mptcp_schedule_work(sk);
-	}
+	set_bit(MPTCP_WORK_RTX, &msk->flags);
+	mptcp_schedule_work(sk);
 }
 
 static void mptcp_retransmit_timer(struct timer_list *t)
@@ -1915,21 +2223,18 @@ static void mptcp_worker(struct work_struct *work)
 	if (unlikely(state == TCP_CLOSE))
 		goto unlock;
 
-	mptcp_clean_una_wakeup(sk);
 	mptcp_check_data_fin_ack(sk);
 	__mptcp_flush_join_list(msk);
 	if (test_and_clear_bit(MPTCP_WORK_CLOSE_SUBFLOW, &msk->flags))
 		__mptcp_close_subflow(msk);
 
-	if (mptcp_send_head(sk))
-		mptcp_push_pending(sk, 0);
-
 	if (msk->pm.status)
 		pm_work(msk);
 
 	if (test_and_clear_bit(MPTCP_WORK_EOF, &msk->flags))
 		mptcp_check_for_eof(msk);
 
+	__mptcp_check_send_data_fin(sk);
 	mptcp_check_data_fin(sk);
 
 	/* if the msk data is completely acked, or the socket timedout,
@@ -1951,9 +2256,6 @@ static void mptcp_worker(struct work_struct *work)
 	if (!dfrag)
 		goto unlock;
 
-	if (!mptcp_ext_cache_refill(msk))
-		goto reset_unlock;
-
 	ssk = mptcp_subflow_get_retrans(msk);
 	if (!ssk)
 		goto reset_unlock;
@@ -1964,6 +2266,9 @@ static void mptcp_worker(struct work_struct *work)
 	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;
@@ -1971,9 +2276,6 @@ static void mptcp_worker(struct work_struct *work)
 		MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_RETRANSSEGS);
 		copied += ret;
 		info.sent += ret;
-
-		if (!mptcp_ext_cache_refill(msk))
-			break;
 	}
 	if (copied)
 		tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
@@ -2001,8 +2303,14 @@ static int __mptcp_init_sock(struct sock *sk)
 	INIT_LIST_HEAD(&msk->join_list);
 	INIT_LIST_HEAD(&msk->rtx_queue);
 	INIT_WORK(&msk->work, mptcp_worker);
+	__skb_queue_head_init(&msk->receive_queue);
+	__skb_queue_head_init(&msk->skb_tx_cache);
 	msk->out_of_order_queue = RB_ROOT;
 	msk->first_pending = NULL;
+	msk->wmem_reserved = 0;
+	msk->rmem_released = 0;
+	msk->tx_pending_data = 0;
+	msk->size_goal_cache = TCP_BASE_MSS;
 
 	msk->ack_hint = NULL;
 	msk->first = NULL;
@@ -2046,12 +2354,15 @@ static void __mptcp_clear_xmit(struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	struct mptcp_data_frag *dtmp, *dfrag;
-
-	sk_stop_timer(sk, &msk->sk.icsk_retransmit_timer);
+	struct sk_buff *skb;
 
 	WRITE_ONCE(msk->first_pending, NULL);
 	list_for_each_entry_safe(dfrag, dtmp, &msk->rtx_queue, list)
 		dfrag_clear(sk, dfrag);
+	while ((skb = __skb_dequeue(&msk->skb_tx_cache)) != NULL) {
+		sk->sk_forward_alloc += skb->truesize;
+		kfree_skb(skb);
+	}
 }
 
 static void mptcp_cancel_work(struct sock *sk)
@@ -2186,7 +2497,7 @@ static void __mptcp_destroy_sock(struct sock *sk)
 	spin_unlock_bh(&msk->join_list_lock);
 	list_splice_init(&msk->conn_list, &conn_list);
 
-	__mptcp_clear_xmit(sk);
+	sk_stop_timer(sk, &msk->sk.icsk_retransmit_timer);
 	sk_stop_timer(sk, &sk->sk_timer);
 	msk->pm.status = 0;
 
@@ -2197,6 +2508,8 @@ static void __mptcp_destroy_sock(struct sock *sk)
 
 	sk->sk_prot->destroy(sk);
 
+	WARN_ON_ONCE(msk->wmem_reserved);
+	WARN_ON_ONCE(msk->rmem_released);
 	sk_stream_kill_queues(sk);
 	xfrm_sk_free_policy(sk);
 	sk_refcnt_debug_release(sk);
@@ -2326,8 +2639,8 @@ struct sock *mptcp_sk_clone(const struct sock *sk,
 
 	msk->write_seq = subflow_req->idsn + 1;
 	msk->snd_nxt = msk->write_seq;
-	atomic64_set(&msk->snd_una, msk->write_seq);
-	atomic64_set(&msk->wnd_end, msk->snd_nxt + req->rsk_rcv_wnd);
+	msk->snd_una = msk->write_seq;
+	msk->wnd_end = msk->snd_nxt + req->rsk_rcv_wnd;
 
 	if (mp_opt->mp_capable) {
 		msk->can_ack = true;
@@ -2363,7 +2676,7 @@ void mptcp_rcv_space_init(struct mptcp_sock *msk, const struct sock *ssk)
 	if (msk->rcvq_space.space == 0)
 		msk->rcvq_space.space = TCP_INIT_CWND * TCP_MSS_DEFAULT;
 
-	atomic64_set(&msk->wnd_end, msk->snd_nxt + tcp_sk(ssk)->snd_wnd);
+	WRITE_ONCE(msk->wnd_end, msk->snd_nxt + tcp_sk(ssk)->snd_wnd);
 }
 
 static struct sock *mptcp_accept(struct sock *sk, int flags, int *err,
@@ -2414,6 +2727,13 @@ static struct sock *mptcp_accept(struct sock *sk, int flags, int *err,
 
 void mptcp_destroy_common(struct mptcp_sock *msk)
 {
+	struct sock *sk = (struct sock *)msk;
+
+	__mptcp_clear_xmit(sk);
+
+	/* move to sk_receive_queue, sk_stream_kill_queues will purge it */
+	skb_queue_splice_tail_init(&msk->receive_queue, &sk->sk_receive_queue);
+
 	skb_rbtree_purge(&msk->out_of_order_queue);
 	mptcp_token_destroy(msk);
 	mptcp_pm_free_anno_list(msk);
@@ -2423,9 +2743,6 @@ static void mptcp_destroy(struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 
-	if (msk->cached_ext)
-		__skb_ext_put(msk->cached_ext);
-
 	mptcp_destroy_common(msk);
 	sk_sockets_allocated_dec(sk);
 }
@@ -2540,15 +2857,58 @@ static int mptcp_getsockopt(struct sock *sk, int level, int optname,
 	return -EOPNOTSUPP;
 }
 
+void __mptcp_data_acked(struct sock *sk)
+{
+	if (!sock_owned_by_user(sk))
+		__mptcp_clean_una(sk);
+	else
+		set_bit(MPTCP_CLEAN_UNA, &mptcp_sk(sk)->flags);
+
+	if (mptcp_pending_data_fin_ack(sk))
+		mptcp_schedule_work(sk);
+}
+
+void __mptcp_wnd_updated(struct sock *sk, struct sock *ssk)
+{
+	if (!mptcp_send_head(sk))
+		return;
+
+	if (!sock_owned_by_user(sk))
+		__mptcp_subflow_push_pending(sk, ssk);
+	else
+		set_bit(MPTCP_PUSH_PENDING, &mptcp_sk(sk)->flags);
+}
+
 #define MPTCP_DEFERRED_ALL (TCPF_WRITE_TIMER_DEFERRED)
 
-/* this is very alike tcp_release_cb() but we must handle differently a
- * different set of events
- */
+/* processes deferred events and flush wmem */
 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
+		 *
+		 * 1) can't be invoked in atomic scope
+		 * 2) must avoid ABBA deadlock with msk socket spinlock: the RX
+		 *    datapath acquires the msk socket spinlock while helding
+		 *    the subflow socket lock
+		 */
+
+		spin_unlock_bh(&sk->sk_lock.slock);
+		mptcp_push_pending(sk, 0);
+		spin_lock_bh(&sk->sk_lock.slock);
+	}
+
+	/* clear any wmem reservation and errors */
+	__mptcp_update_wmem(sk);
+	__mptcp_update_rmem(sk);
+
 	do {
 		flags = sk->sk_tsq_flags;
 		if (!(flags & MPTCP_DEFERRED_ALL))
@@ -2619,7 +2979,7 @@ void mptcp_finish_connect(struct sock *ssk)
 	WRITE_ONCE(msk->ack_seq, ack_seq);
 	WRITE_ONCE(msk->rcv_wnd_sent, ack_seq);
 	WRITE_ONCE(msk->can_ack, 1);
-	atomic64_set(&msk->snd_una, msk->write_seq);
+	WRITE_ONCE(msk->snd_una, msk->write_seq);
 
 	mptcp_pm_new_connection(msk, 0);
 
@@ -2880,24 +3240,9 @@ static __poll_t mptcp_check_readable(struct mptcp_sock *msk)
 	       0;
 }
 
-static bool __mptcp_check_writeable(struct mptcp_sock *msk)
-{
-	struct sock *sk = (struct sock *)msk;
-	bool mptcp_writable;
-
-	mptcp_clean_una(sk);
-	mptcp_writable = sk_stream_is_writeable(sk);
-	if (!mptcp_writable)
-		mptcp_nospace(msk);
-
-	return mptcp_writable;
-}
-
 static __poll_t mptcp_check_writeable(struct mptcp_sock *msk)
 {
 	struct sock *sk = (struct sock *)msk;
-	__poll_t ret = 0;
-	bool slow;
 
 	if (unlikely(sk->sk_shutdown & SEND_SHUTDOWN))
 		return 0;
@@ -2905,12 +3250,12 @@ static __poll_t mptcp_check_writeable(struct mptcp_sock *msk)
 	if (sk_stream_is_writeable(sk))
 		return EPOLLOUT | EPOLLWRNORM;
 
-	slow = lock_sock_fast(sk);
-	if (__mptcp_check_writeable(msk))
-		ret = EPOLLOUT | EPOLLWRNORM;
+	set_bit(MPTCP_NOSPACE, &msk->flags);
+	smp_mb__after_atomic(); /* msk->flags is changed by write_space cb */
+	if (sk_stream_is_writeable(sk))
+		return EPOLLOUT | EPOLLWRNORM;
 
-	unlock_sock_fast(sk, slow);
-	return ret;
+	return 0;
 }
 
 static __poll_t mptcp_poll(struct file *file, struct socket *sock,
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index 82d5626323b1..fc56e730fb35 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -91,6 +91,8 @@
 #define MPTCP_WORK_EOF		3
 #define MPTCP_FALLBACK_DONE	4
 #define MPTCP_WORK_CLOSE_SUBFLOW 5
+#define MPTCP_PUSH_PENDING	6
+#define MPTCP_CLEAN_UNA		7
 
 static inline bool before64(__u64 seq1, __u64 seq2)
 {
@@ -218,14 +220,16 @@ struct mptcp_sock {
 	u64		ack_seq;
 	u64		rcv_wnd_sent;
 	u64		rcv_data_fin_seq;
+	int		wmem_reserved;
 	struct sock	*last_snd;
 	int		snd_burst;
 	int		old_wspace;
-	atomic64_t	snd_una;
-	atomic64_t	wnd_end;
+	u64		snd_una;
+	u64		wnd_end;
 	unsigned long	timer_ival;
 	u32		token;
 	int		rmem_pending;
+	int		rmem_released;
 	unsigned long	flags;
 	bool		can_ack;
 	bool		fully_established;
@@ -237,11 +241,14 @@ struct mptcp_sock {
 	struct work_struct work;
 	struct sk_buff  *ooo_last_skb;
 	struct rb_root  out_of_order_queue;
+	struct sk_buff_head receive_queue;
+	struct sk_buff_head skb_tx_cache;	/* this is wmem accounted */
+	int		tx_pending_data;
+	int		size_goal_cache;
 	struct list_head conn_list;
 	struct list_head rtx_queue;
 	struct mptcp_data_frag *first_pending;
 	struct list_head join_list;
-	struct skb_ext	*cached_ext;	/* for the next sendmsg */
 	struct socket	*subflow; /* outgoing connect/listener/!mp_capable */
 	struct sock	*first;
 	struct mptcp_pm_data	pm;
@@ -253,6 +260,22 @@ struct mptcp_sock {
 	} rcvq_space;
 };
 
+#define mptcp_lock_sock(___sk, cb) do {					\
+	struct sock *__sk = (___sk); /* silence macro reuse warning */	\
+	might_sleep();							\
+	spin_lock_bh(&__sk->sk_lock.slock);				\
+	if (__sk->sk_lock.owned)					\
+		__lock_sock(__sk);					\
+	cb;								\
+	__sk->sk_lock.owned = 1;					\
+	spin_unlock(&__sk->sk_lock.slock);				\
+	mutex_acquire(&__sk->sk_lock.dep_map, 0, 0, _RET_IP_);		\
+	local_bh_enable();						\
+} while (0)
+
+#define mptcp_data_lock(sk) spin_lock_bh(&(sk)->sk_lock.slock)
+#define mptcp_data_unlock(sk) spin_unlock_bh(&(sk)->sk_lock.slock)
+
 #define mptcp_for_each_subflow(__msk, __subflow)			\
 	list_for_each_entry(__subflow, &((__msk)->conn_list), node)
 
@@ -300,7 +323,7 @@ static inline struct mptcp_data_frag *mptcp_rtx_tail(const struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 
-	if (!before64(msk->snd_nxt, atomic64_read(&msk->snd_una)))
+	if (!before64(msk->snd_nxt, READ_ONCE(msk->snd_una)))
 		return NULL;
 
 	return list_last_entry(&msk->rtx_queue, struct mptcp_data_frag, list);
@@ -474,7 +497,8 @@ void mptcp_rcv_space_init(struct mptcp_sock *msk, const struct sock *ssk);
 void mptcp_data_ready(struct sock *sk, struct sock *ssk);
 bool mptcp_finish_join(struct sock *sk);
 bool mptcp_schedule_work(struct sock *sk);
-void mptcp_data_acked(struct sock *sk);
+void __mptcp_wnd_updated(struct sock *sk, struct sock *ssk);
+void __mptcp_data_acked(struct sock *sk);
 void mptcp_subflow_eof(struct sock *sk);
 bool mptcp_update_rcv_data_fin(struct mptcp_sock *msk, u64 data_fin_seq, bool use_64bit);
 void __mptcp_flush_join_list(struct mptcp_sock *msk);
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 2e5c3f4da3a4..023c856424b7 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -995,19 +995,9 @@ static void subflow_data_ready(struct sock *sk)
 		mptcp_data_ready(parent, sk);
 }
 
-static void subflow_write_space(struct sock *sk)
+static void subflow_write_space(struct sock *ssk)
 {
-	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
-	struct socket *sock = READ_ONCE(sk->sk_socket);
-	struct sock *parent = subflow->conn;
-
-	if (!sk_stream_is_writeable(sk))
-		return;
-
-	if (sock && sk_stream_is_writeable(parent))
-		clear_bit(SOCK_NOSPACE, &sock->flags);
-
-	sk_stream_write_space(parent);
+	/* we take action in __mptcp_clean_una() */
 }
 
 static struct inet_connection_sock_af_ops *