Skip to contentSkip to conversation

squash and cleanup

PATCH In review Paolo Abeni 3 participants 9 messages Mar 20, 2020

Subsystems (2)
lore
Share & tools
lore

Paolo Abeni

author
#0 lore raw

cover squash and cleanup

This series contains a "squashable" version of:

subflow: place further subflows on new 'join_list'

and a bunch of minor cleanup (unneeded check after not so recent net-next changes)

Rebased tree available at:

https://github.com/pabeni/mptcp/tree/mptcp_net-next_part3_14

this supersede the squashed branch posted previously.

Self-tests loop running. Will update here with results.

 options.c  |   11 +++--------
 protocol.c |   31 ++++++++++++++++++++++++++++++-
 protocol.h |    2 ++
 subflow.c  |    5 ++++-
 4 files changed, 39 insertions(+), 10 deletions(-)
author
#1 lore raw

PATCH 1/5 Squash-to: "mptcp: Add ADD_ADDR handling"

subflow->conn is always valid for is_mptcp socket, drop unneeded check

Signed-off-by Paolo Abeni <pabeni(a)redhat.com>
---
 net/mptcp/options.c | 3 ---
 1 file changed, 3 deletions(-)
net/mptcp/options.c +0 -3
index 9b90bdc419ac..633bda691fff 100644
@@ -486,9 +486,6 @@ static bool mptcp_established_options_addr(struct sock *sk,
 	struct sockaddr_storage saddr;
 	u8 id;
 
-	if (!msk)
-		return false;
-
 	id = 0;
 	memset(&saddr, 0, sizeof(saddr));
 
Signature · 1 line
2.21.1
author
#2 lore raw

PATCH 2/5 Squash-to: "mptcp: Add path manager interface"

In reply to Paolo Abeni

Same as previous patch

Signed-off-by Paolo Abeni <pabeni(a)redhat.com>
---
 net/mptcp/options.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
net/mptcp/options.c +1 -2
index fdab1869bf10..b21ea3147b91 100644
@@ -692,8 +692,7 @@ void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb,
 
 	mpext->data_fin = mp_opt->data_fin;
 
-	if (msk)
-		mptcp_pm_fully_established(msk);
+	mptcp_pm_fully_established(msk);
 }
 
 void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
Signature · 1 line
2.21.1
author
#3 lore raw

PATCH 3/5 Squash-to: "mptcp: Add handling of incoming MP_JOIN requests"

In reply to Paolo Abeni

Drop another unneeded check, and remote the paired, obsoleted comment.

Signed-off-by Paolo Abeni <pabeni(a)redhat.com>
---
 net/mptcp/options.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
net/mptcp/options.c +1 -2
index 3d6000f98e69..df8184aba3e1 100644
@@ -661,8 +661,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *sk,
 	subflow->can_ack = 1;
 
 fully_established:
-	/* msk can be null for MPC subflow on passive socket */
-	if (subflow->pm_notified || !msk)
+	if (likely(subflow->pm_notified))
 		return true;
 
 	subflow->pm_notified = 1;
Signature · 1 line
2.21.1
author
#4 lore raw

PATCH 4/5 Squash-to: "mptcp: Add handling of outgoing MP_JOIN requests"

In reply to Paolo Abeni

This is just a rebase of existing later patch "subflow: place further subflows on new 'join_list'".

The original should be update accordingly changing:

"""
When a valid SYN-ACK is received the
new sock is added to the tail of the mptcp sock conn_list where it will not interfere with data flow on the original connection.
"""

With something alike:

"""
Passive sockets can't acquire the mptcp socket lock at subflow creation time, so an additional list protected by a new spinlock is used to track the MPJ subflows.

Such list is spliced into conn_list tail every time the msk socket lock is acquired, so that it will not interfere with data flow on the original connection.
"""

Signed-off-by Paolo Abeni <pabeni(a)redhat.com>
---
 net/mptcp/protocol.c | 31 ++++++++++++++++++++++++++++++-
 net/mptcp/protocol.h |  2 ++
 net/mptcp/subflow.c  |  5 ++++-
 3 files changed, 36 insertions(+), 2 deletions(-)
net/mptcp/protocol.c +30 -1
index a9a7fac6fb5a..84c28b4326ff 100644
@@ -241,6 +241,16 @@ void mptcp_data_ready(struct sock *sk, struct sock *ssk)
 	sk->sk_data_ready(sk);
 }
 
+static void __mptcp_flush_join_list(struct mptcp_sock *msk)
+{
+	if (likely(list_empty(&msk->join_list)))
+		return;
+
+	spin_lock_bh(&msk->join_list_lock);
+	list_splice_tail_init(&msk->join_list, &msk->conn_list);
+	spin_unlock_bh(&msk->join_list_lock);
+}
+
 static bool mptcp_ext_cache_refill(struct mptcp_sock *msk)
 {
 	if (!msk->cached_ext)
@@ -462,6 +472,7 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		return ret >= 0 ? ret + copied : (copied ? copied : ret);
 	}
 
+	__mptcp_flush_join_list(msk);
 	ssk = mptcp_subflow_get_send(msk);
 	while (!sk_stream_memory_free(sk) || !ssk) {
 		ret = sk_stream_wait_memory(sk, &timeo);
@@ -603,6 +614,7 @@ 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);
 
 	while (len > (size_t)copied) {
 		int bytes_read;
@@ -718,6 +730,7 @@ static void mptcp_worker(struct work_struct *work)
 	struct sock *sk = &msk->sk.icsk_inet.sk;
 
 	lock_sock(sk);
+	__mptcp_flush_join_list(msk);
 	__mptcp_move_skbs(msk);
 	release_sock(sk);
 	sock_put(sk);
@@ -727,7 +740,10 @@ static int __mptcp_init_sock(struct sock *sk)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 
+	spin_lock_init(&msk->join_list_lock);
+
 	INIT_LIST_HEAD(&msk->conn_list);
+	INIT_LIST_HEAD(&msk->join_list);
 	__set_bit(MPTCP_SEND_SPACE, &msk->flags);
 	INIT_WORK(&msk->work, mptcp_worker);
 
@@ -800,6 +816,8 @@ static void mptcp_close(struct sock *sk, long timeout)
 	mptcp_token_destroy(msk->token);
 	inet_sk_state_store(sk, TCP_CLOSE);
 
+	__mptcp_flush_join_list(msk);
+
 	list_splice_init(&msk->conn_list, &conn_list);
 
 	data_fin_tx_seq = msk->write_seq;
@@ -1111,6 +1129,7 @@ bool mptcp_finish_join(struct sock *sk)
 	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
 	struct sock *parent = (void *)msk;
 	struct socket *parent_sock;
+	bool ret;
 
 	pr_debug("msk=%p, subflow=%p", msk, subflow);
 
@@ -1126,7 +1145,15 @@ bool mptcp_finish_join(struct sock *sk)
 	if (parent_sock && !sk->sk_socket)
 		mptcp_sock_graft(sk, parent_sock);
 
-	return mptcp_pm_allow_new_subflow(msk);
+	ret = mptcp_pm_allow_new_subflow(msk);
+	if (ret) {
+		/* active connections are already on conn_list */
+		spin_lock_bh(&msk->join_list_lock);
+		if (!WARN_ON_ONCE(!list_empty(&subflow->node)))
+			list_add_tail(&subflow->node, &msk->join_list);
+		spin_unlock_bh(&msk->join_list_lock);
+	}
+	return ret;
 }
 
 bool mptcp_sk_is_subflow(const struct sock *sk)
@@ -1315,6 +1342,7 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 		/* set ssk->sk_socket of accept()ed flows to mptcp socket.
 		 * This is needed so NOSPACE flag can be set from tcp stack.
 		 */
+		__mptcp_flush_join_list(msk);
 		list_for_each_entry(subflow, &msk->conn_list, node) {
 			struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 
@@ -1396,6 +1424,7 @@ static int mptcp_shutdown(struct socket *sock, int how)
 			sock->state = SS_CONNECTED;
 	}
 
+	__mptcp_flush_join_list(msk);
 	mptcp_for_each_subflow(msk, subflow) {
 		struct sock *tcp_sk = mptcp_subflow_tcp_sock(subflow);
 
net/mptcp/protocol.h +2 -0
index abc9da787dd7..df134ac91274 100644
@@ -150,8 +150,10 @@ struct mptcp_sock {
 	u32		token;
 	unsigned long	flags;
 	bool		can_ack;
+	spinlock_t	join_list_lock;
 	struct work_struct work;
 	struct list_head conn_list;
+	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;
net/mptcp/subflow.c +4 -1
index 740a6b43fad7..bf58b599a820 100644
@@ -865,10 +865,13 @@ int __mptcp_subflow_connect(struct sock *sk, int ifindex,
 	if (err && err != -EINPROGRESS)
 		goto failed;
 
+	spin_lock_bh(&msk->join_list_lock);
+	list_add_tail(&subflow->node, &msk->join_list);
+	spin_unlock_bh(&msk->join_list_lock);
+
 	return err;
 
 failed:
-	list_del_init(&subflow->node);
 	sock_release(sf);
 	return err;
 }
Signature · 1 line
2.21.1
author
#5 lore raw

PATCH 5/5 Squash-to: "mptcp: update per unacked sequence on pkt reception"

In reply to Paolo Abeni

Drop another unneeded check

Signed-off-by Paolo Abeni <pabeni(a)redhat.com>
---
 net/mptcp/options.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/mptcp/options.c +1 -1
index 353f9e74876d..4b86f1bbd6f1 100644
@@ -839,7 +839,7 @@ void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb,
 	/* we can't wait for recvmsg() to update the ack_seq, otherwise
 	 * monodirectional flows will stuck
 	 */
-	if (msk && mp_opt->use_ack)
+	if (mp_opt->use_ack)
 		update_una(msk, mp_opt);
 
 	mpext = skb_ext_add(skb, SKB_EXT_MPTCP);
Signature · 1 line
2.21.1
#6 lore raw
In reply to Paolo Abeni
Quoted 13 lines · Paolo Abeni

This series contains a "squashable" version of:

subflow: place further subflows on new 'join_list'

and a bunch of minor cleanup (unneeded check after not so recent net-next changes)

Rebased tree available at:

https://github.com/pabeni/mptcp/tree/mptcp_net-next_part3_14

this supersede the squashed branch posted previously.

Quoted 7 lines

Self-tests loop running. Will update here with results.

options.c  |   11 +++--------
protocol.c |   31 ++++++++++++++++++++++++++++++-
protocol.h |    2 ++
subflow.c  |    5 ++++-
4 files changed, 39 insertions(+), 10 deletions(-)

Changes look good to me - thanks Paolo!

--
Mat Martineau
Intel

author
#8 lore raw
In reply to Paolo Abeni
Quoted 7 lines · Paolo Abeni

This series contains a "squashable" version of:

subflow: place further subflows on new 'join_list'

and a bunch of minor cleanup (unneeded check after not so recent net-next changes)

Quoted 7 lines

Rebased tree available at:

https://github.com/pabeni/mptcp/tree/mptcp_net-next_part3_14

this supersede the squashed branch posted previously.

Self-tests loop running. Will update here with results.

Went throught ~700 self-tests loop iteration without issues.

/P

#7 lore raw
In reply to Mat Martineau

Hi Paolo, Mat

@Paolo: I have a comment for you below

Quoted 18 lines · Mat Martineau
Quoted 15 lines · Paolo Abeni

This series contains a "squashable" version of:

subflow: place further subflows on new 'join_list'

and a bunch of minor cleanup (unneeded check after not so recent net-next changes)

Rebased tree available at:

https://github.com/pabeni/mptcp/tree/mptcp_net-next_part3_14

this supersede the squashed branch posted previously.

Self-tests loop running. Will update here with results.

Quoted 7 lines
Quoted 5 lines

options.c  |   11 +++--------
protocol.c |   31 ++++++++++++++++++++++++++++++-
protocol.h |    2 ++
subflow.c  |    5 ++++-

4 files changed, 39 insertions(+), 10 deletions(-)

Changes look good to me - thanks Paolo!

Thank you for the patches, the tests and the reviews!

- d9c29116606a: "squashed" patch 1/5 in "mptcp: Add ADD_ADDR handling"
- 348e235ba3d5: conflict in t/mptcp-Add-path-manager-interface
- c011ac35dcba..dffdcd105251: result

- e7ac600fc440: "squashed" patch 2/5 in "mptcp: Add path manager interface"
- 7c7660a7e9d8: conflict in t/mptcp-Add-handling-of-incoming-MP_JOIN-requests
- dffdcd105251..83e0a152c29b: result (empty, as expected)

- fe80488ecbae: "squashed" patch 3/5 in "mptcp: Add handling of incoming MP_JOIN requests"
- 83e0a152c29b..082785f12357: result

- 5079f6bd9a9b: "squashed" patch 4/5 in "mptcp: Add handling of outgoing MP_JOIN requests"
- 49a712650dca: tg: add a description from parent commit
- 6303bcfe7f27: conflict in t/mptcp-queue-data-for-mptcp-level-retransmission
- 9f05cafd0fc3: conflict in t/mptcp-introduce-MPTCP-retransmission-timer
- 66c69c29abdc: conflict in t/mptcp-implement-and-use-MPTCP-level-retransmission
- 082785f12357..9345f86134a1: result

*Paolo* : the result is not empty because the commit to squash was not empty. I guess in your previous message (squashing of "subflow: place further subflows on new 'join_list'") the diff was inverted, right?

Additional commit:
- 1c46c069c95d: mptcp:worker: flush the join list sooner
- c85ce15b5be0..2bfb96355343: result

- 7585b5b02c27: "squashed" patch 5/5 in "mptcp: update per unacked sequence on pkt reception"
- 9345f86134a1..c85ce15b5be0: result

- 4ae633d18908: tg: depend on t/sendmsg-truncate-source-buffer-if-mptcp-sndbuf-size-was-set-from-userspace
- 2bfb96355343..49bd5a76f678: result (empty as expected)

Tests + export are going to be launched soon.

Cheers
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium