[PATCH mptcp-next v2 0/2] BPF redundant scheduler, part 3

[PATCH mptcp-next v2 0/2] BPF redundant scheduler, part 3

[Geliang Tang]

v2:
- drop retrans_redundant flag.
- call __mptcp_retrans() directly.
- depends on "BPF redundant scheduler, part 2" v23.

v1:
- The DSS issue has been fixed in this version, and all tests
(mptcp_connect.sh, mptcp_join.sh, simult_flows.sh and BPF test_progs)
passed.
- No need to set already_sent to 0, drop this.
- Add retrans_redundant flag.
- depends on "BPF redundant scheduler, part 2" v22.

Geliang Tang (2):
  mptcp: add use_lock for mptcp_retrans
  mptcp: retrans for redundant sends

 net/mptcp/protocol.c | 51 +++++++++++++++++++++++++++++++++++---------
 1 file changed, 41 insertions(+), 10 deletions(-)

-- 
2.35.3

[PATCH mptcp-next v2 1/2] mptcp: add use_lock for mptcp_retrans

[Geliang Tang]

This patch adds a new parameter named 'ues_lock' for __mptcp_retrans()
to set whether to use socket locks in this function.

Signed-off-by: Geliang Tang <geliang.tang@suse.com>
---
 net/mptcp/protocol.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 5505c1439b3e..9f0237b5a4b4 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -2523,7 +2523,7 @@ static void mptcp_check_fastclose(struct mptcp_sock *msk)
 	sk_error_report(sk);
 }
 
-static void __mptcp_retrans(struct sock *sk)
+static void __mptcp_retrans(struct sock *sk, bool use_lock)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	struct mptcp_subflow_context *subflow;
@@ -2566,7 +2566,8 @@ static void __mptcp_retrans(struct sock *sk)
 
 			ssk = mptcp_subflow_tcp_sock(subflow);
 
-			lock_sock(ssk);
+			if (use_lock)
+				lock_sock(ssk);
 
 			/* limit retransmission to the bytes already sent on some subflows */
 			info.sent = 0;
@@ -2588,7 +2589,8 @@ static void __mptcp_retrans(struct sock *sk)
 				WRITE_ONCE(msk->allow_infinite_fallback, false);
 			}
 
-			release_sock(ssk);
+			if (use_lock)
+				release_sock(ssk);
 
 			msk->last_snd = ssk;
 		}
@@ -2694,7 +2696,7 @@ static void mptcp_worker(struct work_struct *work)
 		__mptcp_close_subflow(msk);
 
 	if (test_and_clear_bit(MPTCP_WORK_RTX, &msk->flags))
-		__mptcp_retrans(sk);
+		__mptcp_retrans(sk, true);
 
 	fail_tout = msk->first ? READ_ONCE(mptcp_subflow_ctx(msk->first)->fail_tout) : 0;
 	if (fail_tout && time_after(jiffies, fail_tout))
@@ -3308,7 +3310,7 @@ static void mptcp_release_cb(struct sock *sk)
 		if (flags & BIT(MPTCP_PUSH_PENDING))
 			__mptcp_push_pending(sk, 0);
 		if (flags & BIT(MPTCP_RETRANSMIT))
-			__mptcp_retrans(sk);
+			__mptcp_retrans(sk, true);
 
 		cond_resched();
 		spin_lock_bh(&sk->sk_lock.slock);
-- 
2.35.3

[PATCH mptcp-next v2 2/2] mptcp: retrans for redundant sends

[Geliang Tang]

Redundant sends need to work more like the MPTCP retransmit code path.
When the scheduler selects multiple subflows, the first subflow to send
is a "normal" transmit, and any other subflows would act like a retransmit
when accessing the dfrags.

Signed-off-by: Geliang Tang <geliang.tang@suse.com>
---
 net/mptcp/protocol.c | 39 ++++++++++++++++++++++++++++++++++-----
 1 file changed, 34 insertions(+), 5 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 9f0237b5a4b4..a8cb12af1d51 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -45,6 +45,7 @@ static struct percpu_counter mptcp_sockets_allocated ____cacheline_aligned_in_sm
 
 static void __mptcp_destroy_sock(struct sock *sk);
 static void __mptcp_check_send_data_fin(struct sock *sk);
+static void __mptcp_retrans(struct sock *sk, bool use_lock);
 
 DEFINE_PER_CPU(struct mptcp_delegated_action, mptcp_delegated_actions);
 static struct net_device mptcp_napi_dev;
@@ -998,7 +999,7 @@ static void __mptcp_clean_una(struct sock *sk)
 
 		if (unlikely(dfrag == msk->first_pending)) {
 			/* in recovery mode can see ack after the current snd head */
-			if (WARN_ON_ONCE(!msk->recovery))
+			if (!msk->recovery)
 				break;
 
 			WRITE_ONCE(msk->first_pending, mptcp_send_next(sk));
@@ -1013,7 +1014,7 @@ static void __mptcp_clean_una(struct sock *sk)
 
 		/* prevent wrap around in recovery mode */
 		if (unlikely(delta > dfrag->already_sent)) {
-			if (WARN_ON_ONCE(!msk->recovery))
+			if (!msk->recovery)
 				goto out;
 			if (WARN_ON_ONCE(delta > dfrag->data_len))
 				goto out;
@@ -1473,7 +1474,8 @@ struct sock *mptcp_subflow_get_send(struct mptcp_sock *msk)
 
 static void mptcp_push_release(struct sock *ssk, struct mptcp_sendmsg_info *info)
 {
-	tcp_push(ssk, 0, info->mss_now, tcp_sk(ssk)->nonagle, info->size_goal);
+	if (info->mss_now)
+		tcp_push(ssk, 0, info->mss_now, tcp_sk(ssk)->nonagle, info->size_goal);
 	release_sock(ssk);
 }
 
@@ -1559,10 +1561,15 @@ void __mptcp_push_pending(struct sock *sk, unsigned int flags)
 				.flags = flags,
 	};
 	bool do_check_data_fin = false;
+	struct mptcp_data_frag *head;
 	int push_count = 1;
 
+	head = mptcp_send_head(sk);
+	if (!head)
+		goto out;
+
 	while (mptcp_send_head(sk) && (push_count > 0)) {
-		int ret = 0;
+		int ret = 0, i = 0;
 
 		if (mptcp_sched_get_send(msk))
 			break;
@@ -1571,6 +1578,13 @@ void __mptcp_push_pending(struct sock *sk, unsigned int flags)
 
 		mptcp_for_each_subflow(msk, subflow) {
 			if (READ_ONCE(subflow->scheduled)) {
+				if (i > 0) {
+					WRITE_ONCE(msk->first_pending, head);
+					mptcp_push_release(ssk, &info);
+					__mptcp_retrans(sk, true);
+					goto out;
+				}
+
 				mptcp_subflow_set_scheduled(subflow, false);
 
 				prev_ssk = ssk;
@@ -1599,6 +1613,7 @@ void __mptcp_push_pending(struct sock *sk, unsigned int flags)
 						push_count--;
 					continue;
 				}
+				i++;
 				do_check_data_fin = true;
 				msk->last_snd = ssk;
 			}
@@ -1609,6 +1624,7 @@ void __mptcp_push_pending(struct sock *sk, unsigned int flags)
 	if (ssk)
 		mptcp_push_release(ssk, &info);
 
+out:
 	/* ensure the rtx timer is running */
 	if (!mptcp_timer_pending(sk))
 		mptcp_reset_timer(sk);
@@ -1623,14 +1639,19 @@ static void __mptcp_subflow_push_pending(struct sock *sk, struct sock *ssk, bool
 	struct mptcp_sendmsg_info info = {
 		.data_lock_held = true,
 	};
+	struct mptcp_data_frag *head;
 	struct sock *xmit_ssk;
 	bool push = true;
 	int copied = 0;
 
+	head = mptcp_send_head(sk);
+	if (!head)
+		goto out;
+
 	info.flags = 0;
 	while (mptcp_send_head(sk) && push) {
 		bool delegate = false;
-		int ret = 0;
+		int ret = 0, i = 0;
 
 		/* check for a different subflow usage only after
 		 * spooling the first chunk of data
@@ -1650,6 +1671,12 @@ static void __mptcp_subflow_push_pending(struct sock *sk, struct sock *ssk, bool
 
 		mptcp_for_each_subflow(msk, subflow) {
 			if (READ_ONCE(subflow->scheduled)) {
+				if (i > 0) {
+					WRITE_ONCE(msk->first_pending, head);
+					__mptcp_retrans(sk, false);
+					goto out;
+				}
+
 				mptcp_subflow_set_scheduled(subflow, false);
 
 				xmit_ssk = mptcp_subflow_tcp_sock(subflow);
@@ -1660,6 +1687,7 @@ static void __mptcp_subflow_push_pending(struct sock *sk, struct sock *ssk, bool
 						goto out;
 					mptcp_subflow_delegate(subflow,
 							       MPTCP_DELEGATE_SEND);
+					i++;
 					msk->last_snd = ssk;
 					delegate = true;
 					push = false;
@@ -1671,6 +1699,7 @@ static void __mptcp_subflow_push_pending(struct sock *sk, struct sock *ssk, bool
 					push = false;
 					continue;
 				}
+				i++;
 				copied += ret;
 				msk->last_snd = ssk;
 			}
-- 
2.35.3