Skip to contentSkip to conversation

mptcp netlink event support

PATCH Superseded Florian Westphal 3 participants 20 messages Jan 14, 2021 – Jan 20, 2021

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Florian Westphal

author
#0 lore raw

cover mptcp netlink event support

This series adds netlink event support.
The format is largely identical to the one used by the out-of-tree kernel, with the exception of the genl name.

Out of tree kernel uses "mptcp", while this used the existing "mptcp_pm".

The initial two patches are preparation patches not related to the rest, so they could be applied independently.

The other patches make little sense without the actual netlink patch.

Comments welcome. I'll send a iproute2 patch to help with debugging in a bit.

Florian Westphal (7):
mptcp: pm: add lockdep assertions
mptcp: move pm netlink work into pm_netlink
mptcp: split __mptcp_close_ssk helper
mptcp: pass subflow socket to a few helpers
mptcp: avoid lock_fast usage in accept path
genetlink: add CAP_NET_ADMIN test for multicast bind
mptcp: add netlink event support

 include/net/genetlink.h    |   1 +
 include/uapi/linux/mptcp.h |  70 +++++++++
 net/mptcp/options.c        |   2 +-
 net/mptcp/pm.c             |  26 +++-
 net/mptcp/pm_netlink.c     | 296 ++++++++++++++++++++++++++++++++++++-
 net/mptcp/protocol.c       |  65 ++++----
 net/mptcp/protocol.h       |  24 +--
 net/mptcp/subflow.c        |   2 +-
 net/netlink/genetlink.c    |  32 ++++
 9 files changed, 452 insertions(+), 66 deletions(-)
Signature · 1 line
2.26.2
author
#1 lore raw

PATCH 1/7 mptcp: pm: add lockdep assertions

Add a few assertions to make sure functions are called with the needed locks held.
Two functions gain might_sleep annotations because they contain conditional calls to functions that sleep.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 net/mptcp/pm.c         |  2 ++
 net/mptcp/pm_netlink.c | 13 +++++++++++++
 net/mptcp/protocol.c   |  4 ++++
 net/mptcp/protocol.h   |  5 +++++
 4 files changed, 24 insertions(+)
net/mptcp/pm.c +2 -0
index 0a6ebd0642ec..192ebd424cd6 100644
@@ -20,6 +20,8 @@ int mptcp_pm_announce_addr(struct mptcp_sock *msk,
 
 	pr_debug("msk=%p, local_id=%d", msk, addr->id);
 
+	lockdep_assert_held(&msk->pm.lock);
+
 	if (add_addr) {
 		pr_warn("addr_signal error, add_addr=%d", add_addr);
 		return -EINVAL;
net/mptcp/pm_netlink.c +13 -0
index 9b1f6298bbdb..fabe431f256e 100644
@@ -138,6 +138,8 @@ select_local_address(const struct pm_nl_pernet *pernet,
 {
 	struct mptcp_pm_addr_entry *entry, *ret = NULL;
 
+	msk_owned_by_me(msk);
+
 	rcu_read_lock();
 	__mptcp_flush_join_list(msk);
 	list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
@@ -195,6 +197,8 @@ lookup_anno_list_by_saddr(struct mptcp_sock *msk,
 {
 	struct mptcp_pm_add_entry *entry;
 
+	lockdep_assert_held(&msk->pm.lock);
+
 	list_for_each_entry(entry, &msk->pm.anno_list, list) {
 		if (addresses_equal(&entry->addr, addr, false))
 			return entry;
@@ -270,6 +274,8 @@ static bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 	struct sock *sk = (struct sock *)msk;
 	struct net *net = sock_net(sk);
 
+	lockdep_assert_held(&msk->pm.lock);
+
 	if (lookup_anno_list_by_saddr(msk, &entry->addr))
 		return false;
 
@@ -412,6 +418,9 @@ void mptcp_pm_nl_add_addr_send_ack(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow;
 
+	msk_owned_by_me(msk);
+	lockdep_assert_held(&msk->pm.lock);
+
 	if (!mptcp_pm_should_add_signal_ipv6(msk) &&
 	    !mptcp_pm_should_add_signal_port(msk))
 		return;
@@ -484,6 +493,8 @@ void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk)
 
 	pr_debug("address rm_id %d", msk->pm.rm_id);
 
+	msk_owned_by_me(msk);
+
 	if (!msk->pm.rm_id)
 		return;
 
@@ -519,6 +530,8 @@ void mptcp_pm_nl_rm_subflow_received(struct mptcp_sock *msk, u8 rm_id)
 
 	pr_debug("subflow rm_id %d", rm_id);
 
+	msk_owned_by_me(msk);
+
 	if (!rm_id)
 		return;
 
net/mptcp/protocol.c +4 -0
index d538f5a6810c..61d138a571b0 100644
@@ -2170,6 +2170,8 @@ static void __mptcp_close_subflow(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow, *tmp;
 
+	might_sleep();
+
 	list_for_each_entry_safe(subflow, tmp, &msk->conn_list, node) {
 		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 
@@ -2517,6 +2519,8 @@ static void __mptcp_destroy_sock(struct sock *sk)
 
 	pr_debug("msk=%p", msk);
 
+	might_sleep();
+
 	/* dispose the ancillatory tcp socket, if any */
 	if (msk->subflow) {
 		iput(SOCK_INODE(msk->subflow));
net/mptcp/protocol.h +5 -0
index 871534df6140..dfd96b790f43 100644
@@ -291,6 +291,11 @@ struct mptcp_sock {
 #define mptcp_for_each_subflow(__msk, __subflow)			\
 	list_for_each_entry(__subflow, &((__msk)->conn_list), node)
 
+static inline void msk_owned_by_me(const struct mptcp_sock *msk)
+{
+	sock_owned_by_me((const struct sock *)msk);
+}
+
 static inline struct mptcp_sock *mptcp_sk(const struct sock *sk)
 {
 	return (struct mptcp_sock *)sk;
Signature · 1 line
2.26.2
#2 lore raw

[MPTCP] Re: [PATCH mptcp 1/7] mptcp: pm: add lockdep assertions

Quoted 5 lines · Florian Westphal

Add a few assertions to make sure functions are called with the needed locks held.
Two functions gain might_sleep annotations because they contain conditional calls to functions that sleep.

Quoted 8 lines
Signed-off-by Florian Westphal <fw(a)strlen.de>

---

net/mptcp/pm.c         |  2 ++
net/mptcp/pm_netlink.c | 13 +++++++++++++
net/mptcp/protocol.c   |  4 ++++
net/mptcp/protocol.h   |  5 +++++
4 files changed, 24 insertions(+)

These annotations look good - I think it would be good to apply this in the export branch now, like you noted in the cover letter.

--
Mat Martineau
Intel

author
#3 lore raw

PATCH 2/7 mptcp: move pm netlink work into pm_netlink

In reply to Florian Westphal

allows to make some functions static and avoids aquire of the pm spinlock in protocol.c.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 net/mptcp/pm_netlink.c | 45 +++++++++++++++++++++++++++++++++++++-----
 net/mptcp/protocol.c   | 33 +------------------------------
 net/mptcp/protocol.h   |  6 +-----
 3 files changed, 42 insertions(+), 42 deletions(-)
net/mptcp/pm_netlink.c +40 -5
index fabe431f256e..214b429bf3b8 100644
@@ -55,6 +55,8 @@ struct pm_nl_pernet {
 #define MPTCP_PM_ADDR_MAX	8
 #define ADD_ADDR_RETRANS_MAX	3
 
+static void mptcp_pm_nl_add_addr_send_ack(struct mptcp_sock *msk);
+
 static bool addresses_equal(const struct mptcp_addr_info *a,
 			    struct mptcp_addr_info *b, bool use_port)
 {
@@ -369,17 +371,17 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 	}
 }
 
-void mptcp_pm_nl_fully_established(struct mptcp_sock *msk)
+static void mptcp_pm_nl_fully_established(struct mptcp_sock *msk)
 {
 	mptcp_pm_create_subflow_or_signal_addr(msk);
 }
 
-void mptcp_pm_nl_subflow_established(struct mptcp_sock *msk)
+static void mptcp_pm_nl_subflow_established(struct mptcp_sock *msk)
 {
 	mptcp_pm_create_subflow_or_signal_addr(msk);
 }
 
-void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
+static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
 {
 	struct sock *sk = (struct sock *)msk;
 	struct mptcp_addr_info remote;
@@ -414,7 +416,7 @@ void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
 	mptcp_pm_nl_add_addr_send_ack(msk);
 }
 
-void mptcp_pm_nl_add_addr_send_ack(struct mptcp_sock *msk)
+static void mptcp_pm_nl_add_addr_send_ack(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow;
 
@@ -486,7 +488,7 @@ int mptcp_pm_nl_mp_prio_send_ack(struct mptcp_sock *msk,
 	return -EINVAL;
 }
 
-void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk)
+static void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow, *tmp;
 	struct sock *sk = (struct sock *)msk;
@@ -523,6 +525,39 @@ void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk)
 	}
 }
 
+void mptcp_pm_nl_work(struct mptcp_sock *msk)
+{
+	struct mptcp_pm_data *pm = &msk->pm;
+
+	msk_owned_by_me(msk);
+
+	spin_lock_bh(&msk->pm.lock);
+
+	pr_debug("msk=%p status=%x", msk, pm->status);
+	if (pm->status & BIT(MPTCP_PM_ADD_ADDR_RECEIVED)) {
+		pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_RECEIVED);
+		mptcp_pm_nl_add_addr_received(msk);
+	}
+	if (pm->status & BIT(MPTCP_PM_ADD_ADDR_SEND_ACK)) {
+		pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_SEND_ACK);
+		mptcp_pm_nl_add_addr_send_ack(msk);
+	}
+	if (pm->status & BIT(MPTCP_PM_RM_ADDR_RECEIVED)) {
+		pm->status &= ~BIT(MPTCP_PM_RM_ADDR_RECEIVED);
+		mptcp_pm_nl_rm_addr_received(msk);
+	}
+	if (pm->status & BIT(MPTCP_PM_ESTABLISHED)) {
+		pm->status &= ~BIT(MPTCP_PM_ESTABLISHED);
+		mptcp_pm_nl_fully_established(msk);
+	}
+	if (pm->status & BIT(MPTCP_PM_SUBFLOW_ESTABLISHED)) {
+		pm->status &= ~BIT(MPTCP_PM_SUBFLOW_ESTABLISHED);
+		mptcp_pm_nl_subflow_established(msk);
+	}
+
+	spin_unlock_bh(&msk->pm.lock);
+}
+
 void mptcp_pm_nl_rm_subflow_received(struct mptcp_sock *msk, u8 rm_id)
 {
 	struct mptcp_subflow_context *subflow, *tmp;
net/mptcp/protocol.c +1 -32
index 61d138a571b0..b1e152196057 100644
@@ -2135,37 +2135,6 @@ static unsigned int mptcp_sync_mss(struct sock *sk, u32 pmtu)
 	return 0;
 }
 
-static void pm_work(struct mptcp_sock *msk)
-{
-	struct mptcp_pm_data *pm = &msk->pm;
-
-	spin_lock_bh(&msk->pm.lock);
-
-	pr_debug("msk=%p status=%x", msk, pm->status);
-	if (pm->status & BIT(MPTCP_PM_ADD_ADDR_RECEIVED)) {
-		pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_RECEIVED);
-		mptcp_pm_nl_add_addr_received(msk);
-	}
-	if (pm->status & BIT(MPTCP_PM_ADD_ADDR_SEND_ACK)) {
-		pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_SEND_ACK);
-		mptcp_pm_nl_add_addr_send_ack(msk);
-	}
-	if (pm->status & BIT(MPTCP_PM_RM_ADDR_RECEIVED)) {
-		pm->status &= ~BIT(MPTCP_PM_RM_ADDR_RECEIVED);
-		mptcp_pm_nl_rm_addr_received(msk);
-	}
-	if (pm->status & BIT(MPTCP_PM_ESTABLISHED)) {
-		pm->status &= ~BIT(MPTCP_PM_ESTABLISHED);
-		mptcp_pm_nl_fully_established(msk);
-	}
-	if (pm->status & BIT(MPTCP_PM_SUBFLOW_ESTABLISHED)) {
-		pm->status &= ~BIT(MPTCP_PM_SUBFLOW_ESTABLISHED);
-		mptcp_pm_nl_subflow_established(msk);
-	}
-
-	spin_unlock_bh(&msk->pm.lock);
-}
-
 static void __mptcp_close_subflow(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow, *tmp;
@@ -2255,7 +2224,7 @@ static void mptcp_worker(struct work_struct *work)
 		__mptcp_close_subflow(msk);
 
 	if (msk->pm.status)
-		pm_work(msk);
+		mptcp_pm_nl_work(msk);
 
 	if (test_and_clear_bit(MPTCP_WORK_EOF, &msk->flags))
 		mptcp_check_for_eof(msk);
net/mptcp/protocol.h +1 -5
index dfd96b790f43..914722b646b3 100644
@@ -644,13 +644,9 @@ int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc);
 
 void __init mptcp_pm_nl_init(void);
 void mptcp_pm_nl_data_init(struct mptcp_sock *msk);
-void mptcp_pm_nl_fully_established(struct mptcp_sock *msk);
-void mptcp_pm_nl_subflow_established(struct mptcp_sock *msk);
-void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk);
-void mptcp_pm_nl_add_addr_send_ack(struct mptcp_sock *msk);
-void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk);
 void mptcp_pm_nl_rm_subflow_received(struct mptcp_sock *msk, u8 rm_id);
 int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct sock_common *skc);
+void mptcp_pm_nl_work(struct mptcp_sock *msk);
 
 static inline struct mptcp_ext *mptcp_get_ext(struct sk_buff *skb)
 {
Signature · 1 line
2.26.2
#4 lore raw

[MPTCP] Re: [PATCH mptcp 2/7] mptcp: move pm netlink work into pm_netlink

Quoted 10 lines · Florian Westphal

allows to make some functions static and avoids aquire of the pm spinlock in protocol.c.

Signed-off-by Florian Westphal <fw(a)strlen.de>

---

net/mptcp/pm_netlink.c | 45 +++++++++++++++++++++++++++++++++++++-----
net/mptcp/protocol.c   | 33 +------------------------------
net/mptcp/protocol.h   |  6 +-----
3 files changed, 42 insertions(+), 42 deletions(-)

Also ok with me to apply this to the export branch.

--
Mat Martineau
Intel

#5 lore raw

[MPTCP] Re: [PATCH mptcp 2/7] mptcp: move pm netlink work into pm_netlink

Hi Florian, Mat,

Quoted 13 lines · Mat Martineau
Quoted 10 lines · Florian Westphal

allows to make some functions static and avoids aquire of the pm spinlock in protocol.c.

Signed-off-by Florian Westphal <fw(a)strlen.de>

---
net/mptcp/pm_netlink.c | 45 +++++++++++++++++++++++++++++++++++++----- net/mptcp/protocol.c   | 33 +------------------------------ net/mptcp/protocol.h   |  6 +-----

3 files changed, 42 insertions(+), 42 deletions(-)
Quoted 1 line

Also ok with me to apply this to the export branch.

Thank you for the patches and the reviews!

Just added the two first patches in our tree with Mat's RvB tag (and I fixed a typo in the commit message reported by codespell):

- 168b68baccfa: mptcp: pm: add lockdep assertions
- 6df8f68967a5: mptcp: move pm netlink work into pm_netlink
- Results: a1d71a555f36..44215ba71300

Tests + export have been queued!

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
author
#6 lore raw

PATCH 3/7 mptcp: split __mptcp_close_ssk helper

In reply to Florian Westphal

Prepare for subflow close events:

When mptcp connection is torn down its enough to send the mptcp socket close notification rather than a subflow close event for all of the subflows followed by the mptcp close event.

This splits the helper: mptcp_close_ssk() will emit the close notification, __mptcp_close_ssk will not.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 net/mptcp/pm_netlink.c |  4 ++--
 net/mptcp/protocol.c   | 12 +++++++++---
 net/mptcp/protocol.h   |  4 ++--
 3 files changed, 13 insertions(+), 7 deletions(-)
net/mptcp/pm_netlink.c +2 -2
index 214b429bf3b8..8d0628590ca8 100644
@@ -512,7 +512,7 @@ static void mptcp_pm_nl_rm_addr_received(struct mptcp_sock *msk)
 
 		spin_unlock_bh(&msk->pm.lock);
 		mptcp_subflow_shutdown(sk, ssk, how);
-		__mptcp_close_ssk(sk, ssk, subflow);
+		mptcp_close_ssk(sk, ssk, subflow);
 		spin_lock_bh(&msk->pm.lock);
 
 		msk->pm.add_addr_accepted--;
@@ -582,7 +582,7 @@ void mptcp_pm_nl_rm_subflow_received(struct mptcp_sock *msk, u8 rm_id)
 
 		spin_unlock_bh(&msk->pm.lock);
 		mptcp_subflow_shutdown(sk, ssk, how);
-		__mptcp_close_ssk(sk, ssk, subflow);
+		mptcp_close_ssk(sk, ssk, subflow);
 		spin_lock_bh(&msk->pm.lock);
 
 		msk->pm.local_addr_used--;
net/mptcp/protocol.c +9 -3
index b1e152196057..ecf440b7ec40 100644
@@ -2097,8 +2097,8 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk)
  * so we need to use tcp_close() after detaching them from the mptcp
  * parent socket.
  */
-void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
-		       struct mptcp_subflow_context *subflow)
+static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
+			      struct mptcp_subflow_context *subflow)
 {
 	list_del(&subflow->node);
 
@@ -2130,6 +2130,12 @@ void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 	sock_put(ssk);
 }
 
+void mptcp_close_ssk(struct sock *sk, struct sock *ssk,
+		     struct mptcp_subflow_context *subflow)
+{
+	__mptcp_close_ssk(sk, ssk, subflow);
+}
+
 static unsigned int mptcp_sync_mss(struct sock *sk, u32 pmtu)
 {
 	return 0;
@@ -2147,7 +2153,7 @@ static void __mptcp_close_subflow(struct mptcp_sock *msk)
 		if (inet_sk_state_load(ssk) != TCP_CLOSE)
 			continue;
 
-		__mptcp_close_ssk((struct sock *)msk, ssk, subflow);
+		mptcp_close_ssk((struct sock *)msk, ssk, subflow);
 	}
 }
 
net/mptcp/protocol.h +2 -2
index 914722b646b3..ac83cc50c981 100644
@@ -475,8 +475,8 @@ void mptcp_subflow_fully_established(struct mptcp_subflow_context *subflow,
 bool mptcp_subflow_data_available(struct sock *sk);
 void __init mptcp_subflow_init(void);
 void mptcp_subflow_shutdown(struct sock *sk, struct sock *ssk, int how);
-void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
-		       struct mptcp_subflow_context *subflow);
+void mptcp_close_ssk(struct sock *sk, struct sock *ssk,
+		     struct mptcp_subflow_context *subflow);
 void mptcp_subflow_reset(struct sock *ssk);
 void mptcp_sock_graft(struct sock *sk, struct socket *parent);
 
Signature · 1 line
2.26.2
author
#7 lore raw

PATCH 4/7 mptcp: pass subflow socket to a few helpers

In reply to Florian Westphal

Pass the first/initial subflow to the existing functions so they can pass this on to the notification handler in the next patch.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 net/mptcp/options.c  | 2 +-
 net/mptcp/pm.c       | 4 ++--
 net/mptcp/protocol.c | 4 ++--
 net/mptcp/protocol.h | 4 ++--
 net/mptcp/subflow.c  | 2 +-
 5 files changed, 8 insertions(+), 8 deletions(-)
net/mptcp/options.c +1 -1
index c9643344a8d7..d42e61ca8c1e 100644
@@ -866,7 +866,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,
 		clear_3rdack_retransmission(ssk);
 		mptcp_pm_subflow_established(msk, subflow);
 	} else {
-		mptcp_pm_fully_established(msk);
+		mptcp_pm_fully_established(msk, ssk, GFP_ATOMIC);
 	}
 	return true;
 
net/mptcp/pm.c +2 -2
index 192ebd424cd6..309c44f13436 100644
@@ -68,7 +68,7 @@ int mptcp_pm_remove_subflow(struct mptcp_sock *msk, u8 local_id)
 
 /* path manager event handlers */
 
-void mptcp_pm_new_connection(struct mptcp_sock *msk, int server_side)
+void mptcp_pm_new_connection(struct mptcp_sock *msk, const struct sock *ssk, int server_side)
 {
 	struct mptcp_pm_data *pm = &msk->pm;
 
@@ -116,7 +116,7 @@ static bool mptcp_pm_schedule_work(struct mptcp_sock *msk,
 	return true;
 }
 
-void mptcp_pm_fully_established(struct mptcp_sock *msk)
+void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp)
 {
 	struct mptcp_pm_data *pm = &msk->pm;
 
net/mptcp/protocol.c +2 -2
index ecf440b7ec40..30986c4008c0 100644
@@ -2993,7 +2993,7 @@ void mptcp_finish_connect(struct sock *ssk)
 	WRITE_ONCE(msk->can_ack, 1);
 	WRITE_ONCE(msk->snd_una, msk->write_seq);
 
-	mptcp_pm_new_connection(msk, 0);
+	mptcp_pm_new_connection(msk, ssk, 0);
 
 	mptcp_rcv_space_init(msk, ssk);
 }
@@ -3237,7 +3237,7 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 		list_add(&subflow->node, &msk->conn_list);
 		sock_hold(msk->first);
 		if (mptcp_is_fully_established(newsk))
-			mptcp_pm_fully_established(msk);
+			mptcp_pm_fully_established(msk, msk->first, GFP_KERNEL);
 
 		mptcp_copy_inaddrs(newsk, msk->first);
 		mptcp_rcv_space_init(msk, msk->first);
net/mptcp/protocol.h +2 -2
index ac83cc50c981..419aff080621 100644
@@ -571,8 +571,8 @@ void mptcp_crypto_hmac_sha(u64 key1, u64 key2, u8 *msg, int len, void *hmac);
 
 void __init mptcp_pm_init(void);
 void mptcp_pm_data_init(struct mptcp_sock *msk);
-void mptcp_pm_new_connection(struct mptcp_sock *msk, int server_side);
-void mptcp_pm_fully_established(struct mptcp_sock *msk);
+void mptcp_pm_new_connection(struct mptcp_sock *msk, const struct sock *ssk, int server_side);
+void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp);
 bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk);
 void mptcp_pm_connection_closed(struct mptcp_sock *msk);
 void mptcp_pm_subflow_established(struct mptcp_sock *msk,
net/mptcp/subflow.c +1 -1
index 31cc362a4638..77b84d3d47ea 100644
@@ -629,7 +629,7 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 			 * created mptcp socket
 			 */
 			new_msk->sk_destruct = mptcp_sock_destruct;
-			mptcp_pm_new_connection(mptcp_sk(new_msk), 1);
+			mptcp_pm_new_connection(mptcp_sk(new_msk), child, 1);
 			mptcp_token_accept(subflow_req, mptcp_sk(new_msk));
 			ctx->conn = new_msk;
 			new_msk = NULL;
Signature · 1 line
2.26.2
#8 lore raw

[MPTCP] Re: [PATCH mptcp 4/7] mptcp: pass subflow socket to a few helpers

Hi Florian,

Quoted 2 lines · Florian Westphal

Pass the first/initial subflow to the existing functions so they can pass this on to the notification handler in the next patch.

(tiny detail: I don't know if it is important but we don't do that "in the next patch" but in the last one of the series. But I am sure it is not an issue.)

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
author
#9 lore raw

PATCH 5/7 mptcp: avoid lock_fast usage in accept path

In reply to Florian Westphal

Once event support is added this may need to allocate memory while msk lock is held with softirqs disabled.

Not using lock_fast also allows to do the allocation with GFP_KERNEL.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 net/mptcp/protocol.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
net/mptcp/protocol.c +2 -3
index 30986c4008c0..d31f92503abf 100644
@@ -3225,9 +3225,8 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 		struct mptcp_sock *msk = mptcp_sk(newsock->sk);
 		struct mptcp_subflow_context *subflow;
 		struct sock *newsk = newsock->sk;
-		bool slowpath;
 
-		slowpath = lock_sock_fast(newsk);
+		lock_sock(newsk);
 
 		/* PM/worker can now acquire the first subflow socket
 		 * lock without racing with listener queue cleanup,
@@ -3253,7 +3252,7 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 			if (!ssk->sk_socket)
 				mptcp_sock_graft(ssk, newsock);
 		}
-		unlock_sock_fast(newsk, slowpath);
+		release_sock(newsk);
 	}
 
 	if (inet_csk_listen_poll(ssock->sk))
Signature · 1 line
2.26.2
author
#10 lore raw

PATCH 6/7 genetlink: add CAP_NET_ADMIN test for multicast bind

In reply to Florian Westphal

genetlink sets NL_CFG_F_NONROOT_RECV for its netlink socket so anyone can subscribe to multicast messages.

rtnetlink doesn't allow this unconditionally, rtnetlink_bind() restricts bind requests to CAP_NET_ADMIN for a few groups.

This allows to set GENL_UNS_ADMIN_PERM flag on genl mcast groups to mandate CAP_NET_ADMIN.

This will be used by the upcoming mptcp netlink event facility which exposes the token (mptcp connection identifier) to userspace.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 include/net/genetlink.h |  1 +
 net/netlink/genetlink.c | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 33 insertions(+)
include/net/genetlink.h +1 -0
index e55ec1597ce7..7cb3fa8310ed 100644
@@ -14,6 +14,7 @@
  */
 struct genl_multicast_group {
 	char			name[GENL_NAMSIZ];
+	u8			flags;
 };
 
 struct genl_ops;
net/netlink/genetlink.c +32 -0
index c992424e4d63..9af4d50e197f 100644
@@ -1360,11 +1360,43 @@ static struct genl_family genl_ctrl __ro_after_init = {
 	.netnsok = true,
 };
 
+static int genl_bind(struct net *net, int group)
+{
+	const struct genl_family *family;
+	unsigned int id;
+	int ret = 0;
+
+	genl_lock_all();
+
+	idr_for_each_entry(&genl_fam_idr, family, id) {
+		const struct genl_multicast_group *grp;
+		int i;
+
+		if (family->n_mcgrps == 0)
+			continue;
+
+		i = group - family->mcgrp_offset;
+		if (i < 0 || i >= family->n_mcgrps)
+			continue;
+
+		grp = &family->mcgrps[i];
+		if ((grp->flags & GENL_UNS_ADMIN_PERM) &&
+		     !ns_capable(net->user_ns, CAP_NET_ADMIN))
+			ret = -EPERM;
+
+		break;
+	}
+
+	genl_unlock_all();
+	return ret;
+}
+
 static int __net_init genl_pernet_init(struct net *net)
 {
 	struct netlink_kernel_cfg cfg = {
 		.input		= genl_rcv,
 		.flags		= NL_CFG_F_NONROOT_RECV,
+		.bind		= genl_bind,
 	};
 
 	/* we'll bump the group number right afterwards */
Signature · 1 line
2.26.2
author
#11 lore raw

PATCH 7/7 mptcp: add netlink event support

In reply to Florian Westphal

Allow userspace (mptcpd) to subscribe to mptcp genl multicast events. This implementation reuses the same event API as the mptcp kernel fork to ease integration of existing tools, e.g. mptcpd.

Supported events include:
1. start and close of an mptcp connection
2. start and close of subflows (joins)
3. announce and withdrawals of addresses
4. subflow priority (backup/non-backup) change.

Signed-off-by Florian Westphal <fw(a)strlen.de>
---
 include/uapi/linux/mptcp.h |  70 +++++++++++
 net/mptcp/pm.c             |  20 +++-
 net/mptcp/pm_netlink.c     | 234 ++++++++++++++++++++++++++++++++++++-
 net/mptcp/protocol.c       |   9 +-
 net/mptcp/protocol.h       |   5 +
 5 files changed, 331 insertions(+), 7 deletions(-)
include/uapi/linux/mptcp.h +70 -0
index 3674a451a18c..7ee275a2cfdd 100644
@@ -36,6 +36,7 @@ enum {
 /* netlink interface */
 #define MPTCP_PM_NAME		"mptcp_pm"
 #define MPTCP_PM_CMD_GRP_NAME	"mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME    "mptcp_events"
 #define MPTCP_PM_VER		0x1
 
 /*
@@ -104,4 +105,73 @@ struct mptcp_info {
 	__u64	mptcpi_rcv_nxt;
 };
 
+/*
+ * MPTCP_EVENT_CREATED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                         sport, dport
+ *    A new MPTCP connection has been created. It is the good time to allocate
+ *    memory and send ADD_ADDR if needed. Depending on the traffic-patterns
+ *    it can take a long time until the MPTCP_EVENT_ESTABLISHED is sent.
+ *
+ * MPTCP_EVENT_ESTABLISHED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *			    sport, dport
+ *    A MPTCP connection is established (can start new subflows).
+ *
+ * MPTCP_EVENT_CLOSED: token
+ *    A MPTCP connection has stopped.
+ *
+ * MPTCP_EVENT_ANNOUNCED: token, rem_id, family, daddr4 | daddr6 [, dport]
+ *    A new address has been announced by the peer.
+ *
+ * MPTCP_EVENT_REMOVED: token, rem_id
+ *    An address has been lost by the peer.
+ *
+ * MPTCP_EVENT_SUB_ESTABLISHED: token, family, saddr4 | saddr6,
+ *                              daddr4 | daddr6, sport, dport, backup,
+ *                              if_idx
+ *
+ * MPTCP_EVENT_SUB_CLOSED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                         sport, dport, backup, if_idx
+ *    A subflow has been closed.
+ *
+ * MPTCP_EVENT_SUB_PRIORITY: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                              sport, dport, backup, if_idx
+ */
+enum mptcp_event_type {
+	MPTCP_EVENT_UNSPEC = 0,
+	MPTCP_EVENT_CREATED = 1,
+	MPTCP_EVENT_ESTABLISHED = 2,
+	MPTCP_EVENT_CLOSED = 3,
+
+	MPTCP_EVENT_ANNOUNCED = 6,
+	MPTCP_EVENT_REMOVED = 7,
+
+	MPTCP_EVENT_SUB_ESTABLISHED = 10,
+	MPTCP_EVENT_SUB_CLOSED = 11,
+
+	MPTCP_EVENT_SUB_PRIORITY = 13,
+};
+
+enum mptcp_event_attr {
+	MPTCP_ATTR_UNSPEC = 0,
+
+	MPTCP_ATTR_TOKEN,	/* u32 */
+	MPTCP_ATTR_FAMILY,	/* u16 */
+	MPTCP_ATTR_LOC_ID,	/* u8 */
+	MPTCP_ATTR_REM_ID,	/* u8 */
+	MPTCP_ATTR_SADDR4,	/* be32 */
+	MPTCP_ATTR_SADDR6,	/* struct in6_addr */
+	MPTCP_ATTR_DADDR4,	/* be32 */
+	MPTCP_ATTR_DADDR6,	/* struct in6_addr */
+	MPTCP_ATTR_SPORT,	/* be16 */
+	MPTCP_ATTR_DPORT,	/* be16 */
+	MPTCP_ATTR_BACKUP,	/* u8 */
+	MPTCP_ATTR_ERROR,	/* u8 */
+	MPTCP_ATTR_FLAGS,	/* u16 */
+	MPTCP_ATTR_TIMEOUT,	/* u32 */
+	MPTCP_ATTR_IF_IDX,	/* s32 */
+
+	__MPTCP_ATTR_AFTER_LAST
+};
+#define MPTCP_ATTR_MAX (__MPTCP_ATTR_AFTER_LAST - 1)
+
 #endif /* _UAPI_MPTCP_H */
net/mptcp/pm.c +15 -5
index 309c44f13436..dc612d5e2752 100644
@@ -75,6 +75,7 @@ void mptcp_pm_new_connection(struct mptcp_sock *msk, const struct sock *ssk, int
 	pr_debug("msk=%p, token=%u side=%d", msk, msk->token, server_side);
 
 	WRITE_ONCE(pm->server_side, server_side);
+	mptcp_event(MPTCP_EVENT_CREATED, msk, ssk, GFP_ATOMIC);
 }
 
 bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk)
@@ -119,13 +120,10 @@ static bool mptcp_pm_schedule_work(struct mptcp_sock *msk,
 void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp)
 {
 	struct mptcp_pm_data *pm = &msk->pm;
+	bool announce = false;
 
 	pr_debug("msk=%p", msk);
 
-	/* try to avoid acquiring the lock below */
-	if (!READ_ONCE(pm->work_pending))
-		return;
-
 	spin_lock_bh(&pm->lock);
 
 	/* mptcp_pm_fully_established() can be invoked by multiple
@@ -135,9 +133,15 @@ void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk,
 	if (READ_ONCE(pm->work_pending) &&
 	    !(msk->pm.status & BIT(MPTCP_PM_ALREADY_ESTABLISHED)))
 		mptcp_pm_schedule_work(msk, MPTCP_PM_ESTABLISHED);
-	msk->pm.status |= BIT(MPTCP_PM_ALREADY_ESTABLISHED);
 
+	if ((msk->pm.status & BIT(MPTCP_PM_ALREADY_ESTABLISHED)) == 0)
+		announce = true;
+
+	msk->pm.status |= BIT(MPTCP_PM_ALREADY_ESTABLISHED);
 	spin_unlock_bh(&pm->lock);
+
+	if (announce)
+		mptcp_event(MPTCP_EVENT_ESTABLISHED, msk, ssk, gfp);
 }
 
 void mptcp_pm_connection_closed(struct mptcp_sock *msk)
@@ -176,6 +180,8 @@ void mptcp_pm_add_addr_received(struct mptcp_sock *msk,
 	pr_debug("msk=%p remote_id=%d accept=%d", msk, addr->id,
 		 READ_ONCE(pm->accept_addr));
 
+	mptcp_event_addr_announced(msk, addr);
+
 	spin_lock_bh(&pm->lock);
 
 	if (!READ_ONCE(pm->accept_addr)) {
@@ -203,6 +209,8 @@ void mptcp_pm_rm_addr_received(struct mptcp_sock *msk, u8 rm_id)
 
 	pr_debug("msk=%p remote_id=%d", msk, rm_id);
 
+	mptcp_event_addr_removed(msk, rm_id);
+
 	spin_lock_bh(&pm->lock);
 	mptcp_pm_schedule_work(msk, MPTCP_PM_RM_ADDR_RECEIVED);
 	pm->rm_id = rm_id;
@@ -215,6 +223,8 @@ void mptcp_pm_mp_prio_received(struct sock *sk, u8 bkup)
 
 	pr_debug("subflow->backup=%d, bkup=%d\n", subflow->backup, bkup);
 	subflow->backup = bkup;
+
+	mptcp_event(MPTCP_EVENT_SUB_PRIORITY, mptcp_sk(subflow->conn), sk, GFP_ATOMIC);
 }
 
 /* path manager helpers */
net/mptcp/pm_netlink.c +233 -1

249 lines not on this page · raw message

net/mptcp/protocol.c +8 -1
index d31f92503abf..ef0b54f19d96 100644
@@ -2133,6 +2133,7 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 void mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 		     struct mptcp_subflow_context *subflow)
 {
+	mptcp_event(MPTCP_EVENT_SUB_CLOSED, mptcp_sk(sk), ssk, GFP_KERNEL);
 	__mptcp_close_ssk(sk, ssk, subflow);
 }
 
@@ -2570,6 +2571,10 @@ static void mptcp_close(struct sock *sk, long timeout)
 	release_sock(sk);
 	if (do_cancel_work)
 		mptcp_cancel_work(sk);
+
+	if (mptcp_sk(sk)->token)
+		mptcp_event(MPTCP_EVENT_CLOSED, mptcp_sk(sk), NULL, GFP_KERNEL);
+
 	sock_put(sk);
 }
 
@@ -3022,7 +3027,7 @@ bool mptcp_finish_join(struct sock *ssk)
 		return false;
 
 	if (!msk->pm.server_side)
-		return true;
+		goto out;
 
 	if (!mptcp_pm_allow_new_subflow(msk))
 		return false;
@@ -3049,6 +3054,8 @@ bool mptcp_finish_join(struct sock *ssk)
 	if (parent_sock && !ssk->sk_socket)
 		mptcp_sock_graft(ssk, parent_sock);
 	subflow->map_seq = READ_ONCE(msk->ack_seq);
+out:
+	mptcp_event(MPTCP_EVENT_SUB_ESTABLISHED, msk, ssk, GFP_ATOMIC);
 	return true;
 }
 
net/mptcp/protocol.h +5 -0
index 419aff080621..191485b8fdfe 100644
@@ -10,6 +10,7 @@
 #include <linux/random.h>
 #include <net/tcp.h>
 #include <net/inet_connection_sock.h>
+#include <uapi/linux/mptcp.h>
 
 #define MPTCP_SUPPORTED_VERSION	1
 
@@ -597,6 +598,10 @@ int mptcp_pm_announce_addr(struct mptcp_sock *msk,
 int mptcp_pm_remove_addr(struct mptcp_sock *msk, u8 local_id);
 int mptcp_pm_remove_subflow(struct mptcp_sock *msk, u8 local_id);
 
+void mptcp_event(enum mptcp_event_type type, const struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp);
+void mptcp_event_addr_announced(const struct mptcp_sock *msk, const struct mptcp_addr_info *info);
+void mptcp_event_addr_removed(const struct mptcp_sock *msk, u8 id);
+
 static inline bool mptcp_pm_should_add_signal(struct mptcp_sock *msk)
 {
 	return READ_ONCE(msk->pm.addr_signal) & BIT(MPTCP_ADD_ADDR_SIGNAL);
Signature · 1 line
2.26.2
#12 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Hi Florian,

Quoted 9 lines · Florian Westphal

Allow userspace (mptcpd) to subscribe to mptcp genl multicast events. This implementation reuses the same event API as the mptcp kernel fork to ease integration of existing tools, e.g. mptcpd.

Supported events include:
1. start and close of an mptcp connection
2. start and close of subflows (joins)
3. announce and withdrawals of addresses
4. subflow priority (backup/non-backup) change.

Thank you for looking at that, that's really nice! :)

The previous patches look really good to me but I would not mind a second look for the previous patches modifying how the locks are used:
I'm not sure I have a good global view to give a proper review on these ones.

My comments for this patch are below if you don't mind looking:

Quoted 29 lines
include/uapi/linux/mptcp.h +14 -0
index 3674a451a18c..7ee275a2cfdd 100644
@@ -36,6 +36,7 @@ enum {
  /* netlink interface */
  #define MPTCP_PM_NAME		"mptcp_pm"
  #define MPTCP_PM_CMD_GRP_NAME	"mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME    "mptcp_events"
  #define MPTCP_PM_VER		0x1
 
  /*
@@ -104,4 +105,73 @@ struct mptcp_info {
  	__u64	mptcpi_rcv_nxt;
  };
 
+/*
+ * MPTCP_EVENT_CREATED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                         sport, dport
+ *    A new MPTCP connection has been created. It is the good time to allocate
+ *    memory and send ADD_ADDR if needed. Depending on the traffic-patterns
+ *    it can take a long time until the MPTCP_EVENT_ESTABLISHED is sent.
+ *
+ * MPTCP_EVENT_ESTABLISHED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *			    sport, dport
+ *    A MPTCP connection is established (can start new subflows).
+ *
+ * MPTCP_EVENT_CLOSED: token
+ *    A MPTCP connection has stopped.

include/uapi/linux/mptcp.h struct mptcp_info { · line 130

Quoted 10 lines
include/uapi/linux/mptcp.h +10 -0
+ *
+ * MPTCP_EVENT_ANNOUNCED: token, rem_id, family, daddr4 | daddr6 [, dport]
+ *    A new address has been announced by the peer.
+ *
+ * MPTCP_EVENT_REMOVED: token, rem_id
+ *    An address has been lost by the peer.
+ *
+ * MPTCP_EVENT_SUB_ESTABLISHED: token, family, saddr4 | saddr6,
+ *                              daddr4 | daddr6, sport, dport, backup,
+ *                              if_idx

Even if it is quite obvious, should we add a comment for this one like for the others, e.g.

A new subflow has been established.

(at least to explain that "sub" stands for "subflow" :) )

include/uapi/linux/mptcp.h struct mptcp_info { · line 134

Quoted 4 lines
include/uapi/linux/mptcp.h +4 -0
+ *
+ * MPTCP_EVENT_SUB_CLOSED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                         sport, dport, backup, if_idx
+ *    A subflow has been closed.

I noticed that the error (copy of sk_err) is not in the arguments. Is it simply because we don't need them in the first version or because there is a technical issue?

A PM might want to know why a subflow has been closed: was it closed by a RST, timeout, etc.?

No issue for me to add it later (if it is possible) ;-)

include/uapi/linux/mptcp.h struct mptcp_info { · line 138

Quoted 4 lines
include/uapi/linux/mptcp.h +4 -0
+ *
+ * MPTCP_EVENT_SUB_PRIORITY: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                              sport, dport, backup, if_idx
+ */

May we add a short description here too?

The priority of a subflow has changed.

(...)

Quoted 4 lines
net/mptcp/pm_netlink.c +0 -0
index 8d0628590ca8..8c193e009886 100644

(...)

Quoted 1 line

+static int mptcp_event_sub_established(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)

(tiny detail: is it OK to have such long line? Same for the functions below. Just to avoid checkpatch warnings but it is maybe not catched by checkpatch?)

(...)

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1361

Quoted 3 lines
net/mptcp/pm_netlink.c +3 -0
+static int mptcp_event_sub_closed(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)
+{
+	if (mptcp_event_sub_established(skb, msk, ssk))

That looks strange to call mptcp_event_sub_established() in mptcp_event_sub_closed() but I understand we want to pack the same info :)

(...)

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1380

Quoted 4 lines
net/mptcp/pm_netlink.c +4 -0
+void mptcp_event_addr_removed(const struct mptcp_sock *msk, uint8_t id)
+{
+	struct net *net = sock_net((const struct sock *)msk);
+        struct nlmsghdr *nlh;

(detail: the indentation is not correct)

(...)

net/mptcp/pm_netlink.c

Quoted 1 line
net/mptcp/pm_netlink.c +1 -0
+	return mptcp_nl_mcast_send(net, skb, GFP_ATOMIC);

Just out of curiosity, is it always OK to have a "return f()" here while the current function returns a "void"? I am sure it is OK because the called function also returns "void" but no issues with some compilers (and some options) nor static analytic tools?

(no need to change of course, just I would not have thought about writing it like that)

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1410

Quoted 9 lines
net/mptcp/pm_netlink.c +9 -0
+
+nla_put_failure:
+	kfree_skb(skb);
+}
+
+void mptcp_event_addr_announced(const struct mptcp_sock *msk, const struct mptcp_addr_info *info)
+{
+	struct net *net = sock_net((const struct sock *)msk);
+        struct nlmsghdr *nlh;

(detail: the indentation is not correct)

Quoted 7 lines
net/mptcp/pm_netlink.c +7 -0
+	struct sk_buff *skb;
+
+	if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
+		return;
+
+	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
+	if (!skb)

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1427

Quoted 10 lines
net/mptcp/pm_netlink.c +10 -0
+		return;
+
+	nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0, MPTCP_EVENT_ANNOUNCED);
+	if (!nlh)
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_REM_ID, info->id))

I think it's a u8 for rem ID.
Same above in mptcp_event_addr_removed()

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1435

Quoted 8 lines
net/mptcp/pm_netlink.c +8 -0
+		goto nla_put_failure;
+
+	if (nla_put_be16(skb, MPTCP_ATTR_DPORT, info->port))
+		goto nla_put_failure;
+
+	switch (info->family) {
+	case AF_INET:
+		if (nla_put_in_addr(skb, MPTCP_ATTR_DADDR4, info->addr.s_addr))

(I should not review that late, I blocked here a few seconds thinking you were using the "source" addr for DADDR :) )

Quoted 14 lines
net/mptcp/pm_netlink.c +14 -0
+			goto nla_put_failure;
+		break;
+#if IS_ENABLED(CONFIG_MPTCP_IPV6)
+	case AF_INET6:
+		if (nla_put_in6_addr(skb, MPTCP_ATTR_DADDR6, &info->addr6))
+			goto nla_put_failure;
+		break;
+#endif
+	default:
+		WARN_ON_ONCE(1);
+		goto nla_put_failure;
+	}
+
+	genlmsg_end(skb, nlh);

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1459

Quoted 10 lines
net/mptcp/pm_netlink.c +10 -0
+	return mptcp_nl_mcast_send(net, skb, GFP_ATOMIC);
+
+nla_put_failure:
+	kfree_skb(skb);
+}
+
+void mptcp_event(enum mptcp_event_type type, const struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp)
+{
+	struct net *net = sock_net((const struct sock *)msk);
+        struct nlmsghdr *nlh;

(detail: the indentation is not correct)

Quoted 7 lines
net/mptcp/pm_netlink.c +7 -0
+	struct sk_buff *skb;
+
+	if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
+		return;
+
+	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, gfp);
+	if (!skb)

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1476

Quoted 10 lines
net/mptcp/pm_netlink.c +10 -0
+		return;
+
+	nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0, type);
+	if (!nlh)
+		goto nla_put_failure;
+
+	switch (type) {
+	case MPTCP_EVENT_UNSPEC:
+		WARN_ON_ONCE(1);
+		break;

Probably useless but should we move this to the end and add "default"?

Quoted 14 lines
net/mptcp/pm_netlink.c +14 -0
+	case MPTCP_EVENT_CREATED:
+	case MPTCP_EVENT_ESTABLISHED:
+		if (mptcp_event_created(skb, msk, ssk) < 0)
+			goto nla_put_failure;
+		break;
+	case MPTCP_EVENT_CLOSED:
+		if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token) < 0)
+			goto nla_put_failure;
+		break;
+	case MPTCP_EVENT_ANNOUNCED:
+	case MPTCP_EVENT_REMOVED:
+		/* call mptcp_event_addr_announced()/removed instead */
+		WARN_ON_ONCE(1);
+		break;

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1500

Quoted 10 lines
net/mptcp/pm_netlink.c +10 -0
+	case MPTCP_EVENT_SUB_ESTABLISHED:
+	case MPTCP_EVENT_SUB_PRIORITY:
+		if (mptcp_event_sub_established(skb, msk, ssk) < 0)
+			goto nla_put_failure;
+		break;
+	case MPTCP_EVENT_SUB_CLOSED:
+		if (mptcp_event_sub_closed(skb, msk, ssk) < 0)
+			goto nla_put_failure;
+		break;
+	}

(...)

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
author
#13 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Quoted 23 lines · Matthieu Baerts
Quoted 14 lines

+ * MPTCP_EVENT_CREATED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ * sport, dport

+ *    A new MPTCP connection has been created. It is the good time to allocate
+ *    memory and send ADD_ADDR if needed. Depending on the traffic-patterns
+ *    it can take a long time until the MPTCP_EVENT_ESTABLISHED is sent.

+ *
+ * MPTCP_EVENT_ESTABLISHED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ * sport, dport
+ * A MPTCP connection is established (can start new subflows).
+ *
+ * MPTCP_EVENT_CLOSED: token
+ * A MPTCP connection has stopped.
+ *
+ * MPTCP_EVENT_ANNOUNCED: token, rem_id, family, daddr4 | daddr6 [, dport]

Quoted 8 lines

+ * A new address has been announced by the peer.
+ *
+ * MPTCP_EVENT_REMOVED: token, rem_id
+ * An address has been lost by the peer.
+ *
+ * MPTCP_EVENT_SUB_ESTABLISHED: token, family, saddr4 | saddr6,
+ * daddr4 | daddr6, sport, dport, backup,
+ * if_idx

Quoted 6 lines

Even if it is quite obvious, should we add a comment for this one like for the others, e.g.

A new subflow has been established.

(at least to explain that "sub" stands for "subflow" :) )

Done.

Quoted 8 lines
Quoted 4 lines

+ *
+ * MPTCP_EVENT_SUB_CLOSED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ * sport, dport, backup, if_idx
+ * A subflow has been closed.

I noticed that the error (copy of sk_err) is not in the arguments. Is it simply because we don't need them in the first version or because there is a technical issue?

No, i've added it now for v2. It exports ssk->sk_err in the subflow announcements.

Quoted 7 lines
Quoted 3 lines

+ * MPTCP_EVENT_SUB_PRIORITY: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ * sport, dport, backup, if_idx
+ */

May we add a short description here too?

The priority of a subflow has changed.

Sure, I took the one from mptc kernel.

Quoted 5 lines
Quoted 1 line

+static int mptcp_event_sub_established(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)

(tiny detail: is it OK to have such long line? Same for the functions below. Just to avoid checkpatch warnings but it is maybe not catched by checkpatch?)

No, checkpatch barfs here. I've added linebreaks everywhere.

Quoted 6 lines

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1361

Quoted 3 lines
net/mptcp/pm_netlink.c +3 -0
+static int mptcp_event_sub_closed(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)
+{
+	if (mptcp_event_sub_established(skb, msk, ssk))

That looks strange to call mptcp_event_sub_established() in mptcp_event_sub_closed() but I understand we want to pack the same info :)

I've renamed the function for next iteration.

Quoted 4 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+	struct net *net = sock_net((const struct sock *)msk);
+        struct nlmsghdr *nlh;

(detail: the indentation is not correct)

Right, I've fixed this up.

Quoted 6 lines

Review

Quoted 1 line
(quoted diff) +1 -0
+	return mptcp_nl_mcast_send(net, skb, GFP_ATOMIC);

Just out of curiosity, is it always OK to have a "return f()" here while the current function returns a "void"? I am sure it is OK because the called function also returns "void" but no issues with some compilers (and some options) nor static analytic tools?

Its ok, but actually comes from an older iteration where return type was 'int'. But since there is no way to deal with a netlink error in a sensible way here i later changed it to void.

I've altered this to the more common
foo();
return;

Quoted 4 lines

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1427

Quoted 1 line
net/mptcp/pm_netlink.c +1 -0
+	if (nla_put_u32(skb, MPTCP_ATTR_REM_ID, info->id))

I think it's a u8 for rem ID.
Same above in mptcp_event_addr_removed()

Indeed, fixed it up.

Quoted 9 lines

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1435

Quoted 6 lines
net/mptcp/pm_netlink.c +6 -0
+	if (nla_put_be16(skb, MPTCP_ATTR_DPORT, info->port))
+		goto nla_put_failure;
+
+	switch (info->family) {
+	case AF_INET:
+		if (nla_put_in_addr(skb, MPTCP_ATTR_DADDR4, info->addr.s_addr))

(I should not review that late, I blocked here a few seconds thinking you were using the "source" addr for DADDR :) )

Right, it would be better to use __be32 instead of struct inaddr but that would result in more (useless) code churn so I left that as-is.

Quoted 5 lines

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1476

Quoted 3 lines
net/mptcp/pm_netlink.c +3 -0
+	case MPTCP_EVENT_UNSPEC:
+		WARN_ON_ONCE(1);
+		break;

Probably useless but should we move this to the end and add "default"?

gcc emits a warning if you add a new name to an enum and a switch statement doesn't handle that.

This is why there is no default branch but all the other 'should not happen' values are listed verbatim with a warn-on.

#14 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Hi Florian,

Thank you for your reply and explanations!
Sounds good to me!

Quoted 11 lines · Florian Westphal
Quoted 7 lines · Matthieu Baerts
Quoted 3 lines

+ * MPTCP_EVENT_SUB_CLOSED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ * sport, dport, backup, if_idx
+ * A subflow has been closed.

I noticed that the error (copy of sk_err) is not in the arguments. Is it simply because we don't need them in the first version or because there is a technical issue?

No, i've added it now for v2. It exports ssk->sk_err in the subflow announcements.

Great, thanks!

Quoted 10 lines
Quoted 9 lines

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1435

Quoted 6 lines
net/mptcp/pm_netlink.c +6 -0
+	if (nla_put_be16(skb, MPTCP_ATTR_DPORT, info->port))
+		goto nla_put_failure;
+
+	switch (info->family) {
+	case AF_INET:
+		if (nla_put_in_addr(skb, MPTCP_ATTR_DADDR4, info->addr.s_addr))

(I should not review that late, I blocked here a few seconds thinking you were using the "source" addr for DADDR :) )

Quoted 2 lines

Right, it would be better to use __be32 instead of struct inaddr but that would result in more (useless) code churn so I left that as-is.

All good, this code is good like that :)
(I only blocked yesterday because we hadd "DADDR" followed by s_addr :) )

Quoted 11 lines
Quoted 5 lines

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1476

Quoted 3 lines
net/mptcp/pm_netlink.c +3 -0
+	case MPTCP_EVENT_UNSPEC:
+		WARN_ON_ONCE(1);
+		break;

Probably useless but should we move this to the end and add "default"?

gcc emits a warning if you add a new name to an enum and a switch statement doesn't handle that.

This is why there is no default branch but all the other 'should not happen' values are listed verbatim with a warn-on.

That's right, I forgot about that!

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
#15 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

In reply to Florian Westphal
Quoted 11 lines · Florian Westphal

Allow userspace (mptcpd) to subscribe to mptcp genl multicast events. This implementation reuses the same event API as the mptcp kernel fork to ease integration of existing tools, e.g. mptcpd.

Supported events include:
1. start and close of an mptcp connection
2. start and close of subflows (joins)
3. announce and withdrawals of addresses
4. subflow priority (backup/non-backup) change.

Signed-off-by Florian Westphal <fw(a)strlen.de>

Thanks for the patch series, Florian. I have a few minor comments below, and will work with Ossama to get mptcpd listening to the events on the upstream kernel and really test things out.

Quoted 13 lines

---

include/uapi/linux/mptcp.h |  70 +++++++++++
net/mptcp/pm.c             |  20 +++-
net/mptcp/pm_netlink.c     | 234 ++++++++++++++++++++++++++++++++++++-
net/mptcp/protocol.c       |   9 +-
net/mptcp/protocol.h       |   5 +
5 files changed, 331 insertions(+), 7 deletions(-)
include/uapi/linux/mptcp.h +0 -0
index 3674a451a18c..7ee275a2cfdd 100644

include/uapi/linux/mptcp.h enum { · line 39

Quoted 4 lines
include/uapi/linux/mptcp.h +1 -0
@@ -36,6 +36,7 @@ enum {
 /* netlink interface */
 #define MPTCP_PM_NAME		"mptcp_pm"
 #define MPTCP_PM_CMD_GRP_NAME	"mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME    "mptcp_events"

Suggest "mptcp_pm_events" instead.

Quoted 280 lines
include/uapi/linux/mptcp.h +69 -0
 #define MPTCP_PM_VER		0x1
 
 /*
@@ -104,4 +105,73 @@ struct mptcp_info {
 	__u64	mptcpi_rcv_nxt;
 };
 
+/*
+ * MPTCP_EVENT_CREATED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                         sport, dport
+ *    A new MPTCP connection has been created. It is the good time to allocate
+ *    memory and send ADD_ADDR if needed. Depending on the traffic-patterns
+ *    it can take a long time until the MPTCP_EVENT_ESTABLISHED is sent.
+ *
+ * MPTCP_EVENT_ESTABLISHED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *			    sport, dport
+ *    A MPTCP connection is established (can start new subflows).
+ *
+ * MPTCP_EVENT_CLOSED: token
+ *    A MPTCP connection has stopped.
+ *
+ * MPTCP_EVENT_ANNOUNCED: token, rem_id, family, daddr4 | daddr6 [, dport]
+ *    A new address has been announced by the peer.
+ *
+ * MPTCP_EVENT_REMOVED: token, rem_id
+ *    An address has been lost by the peer.
+ *
+ * MPTCP_EVENT_SUB_ESTABLISHED: token, family, saddr4 | saddr6,
+ *                              daddr4 | daddr6, sport, dport, backup,
+ *                              if_idx
+ *
+ * MPTCP_EVENT_SUB_CLOSED: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                         sport, dport, backup, if_idx
+ *    A subflow has been closed.
+ *
+ * MPTCP_EVENT_SUB_PRIORITY: token, family, saddr4 | saddr6, daddr4 | daddr6,
+ *                              sport, dport, backup, if_idx
+ */
+enum mptcp_event_type {
+	MPTCP_EVENT_UNSPEC = 0,
+	MPTCP_EVENT_CREATED = 1,
+	MPTCP_EVENT_ESTABLISHED = 2,
+	MPTCP_EVENT_CLOSED = 3,
+
+	MPTCP_EVENT_ANNOUNCED = 6,
+	MPTCP_EVENT_REMOVED = 7,
+
+	MPTCP_EVENT_SUB_ESTABLISHED = 10,
+	MPTCP_EVENT_SUB_CLOSED = 11,
+
+	MPTCP_EVENT_SUB_PRIORITY = 13,
+};
+
+enum mptcp_event_attr {
+	MPTCP_ATTR_UNSPEC = 0,
+
+	MPTCP_ATTR_TOKEN,	/* u32 */
+	MPTCP_ATTR_FAMILY,	/* u16 */
+	MPTCP_ATTR_LOC_ID,	/* u8 */
+	MPTCP_ATTR_REM_ID,	/* u8 */
+	MPTCP_ATTR_SADDR4,	/* be32 */
+	MPTCP_ATTR_SADDR6,	/* struct in6_addr */
+	MPTCP_ATTR_DADDR4,	/* be32 */
+	MPTCP_ATTR_DADDR6,	/* struct in6_addr */
+	MPTCP_ATTR_SPORT,	/* be16 */
+	MPTCP_ATTR_DPORT,	/* be16 */
+	MPTCP_ATTR_BACKUP,	/* u8 */
+	MPTCP_ATTR_ERROR,	/* u8 */
+	MPTCP_ATTR_FLAGS,	/* u16 */
+	MPTCP_ATTR_TIMEOUT,	/* u32 */
+	MPTCP_ATTR_IF_IDX,	/* s32 */
+
+	__MPTCP_ATTR_AFTER_LAST
+};
+#define MPTCP_ATTR_MAX (__MPTCP_ATTR_AFTER_LAST - 1)
+
 #endif /* _UAPI_MPTCP_H */
net/mptcp/pm.c +15 -5
index 309c44f13436..dc612d5e2752 100644
@@ -75,6 +75,7 @@ void mptcp_pm_new_connection(struct mptcp_sock *msk, const struct sock *ssk, int
 	pr_debug("msk=%p, token=%u side=%d", msk, msk->token, server_side);
 
 	WRITE_ONCE(pm->server_side, server_side);
+	mptcp_event(MPTCP_EVENT_CREATED, msk, ssk, GFP_ATOMIC);
 }
 
 bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk)
@@ -119,13 +120,10 @@ static bool mptcp_pm_schedule_work(struct mptcp_sock *msk,
 void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp)
 {
 	struct mptcp_pm_data *pm = &msk->pm;
+	bool announce = false;
 
 	pr_debug("msk=%p", msk);
 
-	/* try to avoid acquiring the lock below */
-	if (!READ_ONCE(pm->work_pending))
-		return;
-
 	spin_lock_bh(&pm->lock);
 
 	/* mptcp_pm_fully_established() can be invoked by multiple
@@ -135,9 +133,15 @@ void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk,
 	if (READ_ONCE(pm->work_pending) &&
 	    !(msk->pm.status & BIT(MPTCP_PM_ALREADY_ESTABLISHED)))
 		mptcp_pm_schedule_work(msk, MPTCP_PM_ESTABLISHED);
-	msk->pm.status |= BIT(MPTCP_PM_ALREADY_ESTABLISHED);
 
+	if ((msk->pm.status & BIT(MPTCP_PM_ALREADY_ESTABLISHED)) == 0)
+		announce = true;
+
+	msk->pm.status |= BIT(MPTCP_PM_ALREADY_ESTABLISHED);
 	spin_unlock_bh(&pm->lock);
+
+	if (announce)
+		mptcp_event(MPTCP_EVENT_ESTABLISHED, msk, ssk, gfp);
 }
 
 void mptcp_pm_connection_closed(struct mptcp_sock *msk)
@@ -176,6 +180,8 @@ void mptcp_pm_add_addr_received(struct mptcp_sock *msk,
 	pr_debug("msk=%p remote_id=%d accept=%d", msk, addr->id,
 		 READ_ONCE(pm->accept_addr));
 
+	mptcp_event_addr_announced(msk, addr);
+
 	spin_lock_bh(&pm->lock);
 
 	if (!READ_ONCE(pm->accept_addr)) {
@@ -203,6 +209,8 @@ void mptcp_pm_rm_addr_received(struct mptcp_sock *msk, u8 rm_id)
 
 	pr_debug("msk=%p remote_id=%d", msk, rm_id);
 
+	mptcp_event_addr_removed(msk, rm_id);
+
 	spin_lock_bh(&pm->lock);
 	mptcp_pm_schedule_work(msk, MPTCP_PM_RM_ADDR_RECEIVED);
 	pm->rm_id = rm_id;
@@ -215,6 +223,8 @@ void mptcp_pm_mp_prio_received(struct sock *sk, u8 bkup)
 
 	pr_debug("subflow->backup=%d, bkup=%d\n", subflow->backup, bkup);
 	subflow->backup = bkup;
+
+	mptcp_event(MPTCP_EVENT_SUB_PRIORITY, mptcp_sk(subflow->conn), sk, GFP_ATOMIC);
 }
 
 /* path manager helpers */
net/mptcp/pm_netlink.c +5 -1
index 8d0628590ca8..8c193e009886 100644
@@ -731,10 +731,14 @@ void mptcp_pm_nl_data_init(struct mptcp_sock *msk)
 	WRITE_ONCE(pm->accept_subflow, subflows);
 }
 
-#define MPTCP_PM_CMD_GRP_OFFSET	0
+#define MPTCP_PM_CMD_GRP_OFFSET       0
+#define MPTCP_PM_EV_GRP_OFFSET        1
 
 static const struct genl_multicast_group mptcp_pm_mcgrps[] = {
 	[MPTCP_PM_CMD_GRP_OFFSET]	= { .name = MPTCP_PM_CMD_GRP_NAME, },
+	[MPTCP_PM_EV_GRP_OFFSET]        = { .name = MPTCP_PM_EV_GRP_NAME,
+					    .flags = GENL_UNS_ADMIN_PERM,
+					  },
 };
 

static const struct nla_policy

net/mptcp/pm_netlink.c +108 -0
@@ -1274,6 +1278,234 @@ static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info)
 	return 0;
 }
 
+static void mptcp_nl_mcast_send(struct net *net, struct sk_buff *nlskb, gfp_t gfp)
+{
+	genlmsg_multicast_netns(&mptcp_genl_family, net,
+				nlskb, 0, MPTCP_PM_EV_GRP_OFFSET, gfp);
+}
+
+static int mptcp_event_add_subflow(struct sk_buff *skb, const struct sock *ssk)
+{
+	const struct inet_sock *issk = inet_sk(ssk);
+	const struct mptcp_subflow_context *sf;
+
+	if (nla_put_u16(skb, MPTCP_ATTR_FAMILY, ssk->sk_family))
+		return -EMSGSIZE;
+
+	switch (ssk->sk_family) {
+	case AF_INET:
+		if (nla_put_in_addr(skb, MPTCP_ATTR_SADDR4, issk->inet_saddr))
+			return -EMSGSIZE;
+		if (nla_put_in_addr(skb, MPTCP_ATTR_DADDR4, issk->inet_daddr))
+			return -EMSGSIZE;
+		break;
+#if IS_ENABLED(CONFIG_MPTCP_IPV6)
+	case AF_INET6: {
+		const struct ipv6_pinfo *np = inet6_sk(ssk);
+
+		if (nla_put_in6_addr(skb, MPTCP_ATTR_SADDR6, &np->saddr))
+			return -EMSGSIZE;
+		if (nla_put_in6_addr(skb, MPTCP_ATTR_DADDR6, &ssk->sk_v6_daddr))
+			return -EMSGSIZE;
+		break;
+	}
+#endif
+	default:
+		WARN_ON_ONCE(1);
+		return -EMSGSIZE;
+	}
+
+	if (nla_put_be16(skb, MPTCP_ATTR_SPORT, issk->inet_sport))
+		return -EMSGSIZE;
+	if (nla_put_be16(skb, MPTCP_ATTR_DPORT, issk->inet_dport))
+		return -EMSGSIZE;
+
+	sf = mptcp_subflow_ctx(ssk);
+	if (WARN_ON_ONCE(!sf))
+		return -EINVAL;
+
+	if (nla_put_u8(skb, MPTCP_ATTR_LOC_ID, sf->local_id))
+		return -EMSGSIZE;
+
+	if (nla_put_u8(skb, MPTCP_ATTR_REM_ID, sf->remote_id))
+		return -EMSGSIZE;
+
+	return 0;
+}
+
+static int mptcp_event_sub_established(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)
+{
+	const struct mptcp_subflow_context *sf;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
+		return -EMSGSIZE;
+
+	if (mptcp_event_add_subflow(skb, ssk))
+		return -EMSGSIZE;
+
+	sf = mptcp_subflow_ctx(ssk);
+	if (WARN_ON_ONCE(!sf))
+		return -EINVAL;
+
+	if (nla_put_u8(skb, MPTCP_ATTR_BACKUP, sf->backup))
+		return -EMSGSIZE;
+
+	if (ssk->sk_bound_dev_if && nla_put_s32(skb, MPTCP_ATTR_IF_IDX, ssk->sk_bound_dev_if))
+		return -EMSGSIZE;
+
+	return 0;
+}
+
+static int mptcp_event_sub_closed(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)
+{
+	if (mptcp_event_sub_established(skb, msk, ssk))
+		return -EMSGSIZE;
+
+	return 0;
+}
+
+static int mptcp_event_created(struct sk_buff *skb, const struct mptcp_sock *msk, const struct sock *ssk)
+{
+	int err = nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token);
+
+	if (err)
+		return err;
+
+	return mptcp_event_add_subflow(skb, ssk);
+}
+
+void mptcp_event_addr_removed(const struct mptcp_sock *msk, uint8_t id)
+{
+	struct net *net = sock_net((const struct sock *)msk);
+        struct nlmsghdr *nlh;
+	struct sk_buff *skb;
+
+	if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
+		return;
+
+	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
+	if (!skb)
+		return;

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1398

Quoted 10 lines
net/mptcp/pm_netlink.c +10 -0
+
+	nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0, MPTCP_EVENT_REMOVED);
+	if (!nlh)
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_REM_ID, id))
+		goto nla_put_failure;

Matthieu already noted this, but MPTCP_ATTR_REM_ID is a u8.

Quoted 20 lines
net/mptcp/pm_netlink.c +20 -0
+
+	genlmsg_end(skb, nlh);
+	return mptcp_nl_mcast_send(net, skb, GFP_ATOMIC);
+
+nla_put_failure:
+	kfree_skb(skb);
+}
+
+void mptcp_event_addr_announced(const struct mptcp_sock *msk, const struct mptcp_addr_info *info)
+{
+	struct net *net = sock_net((const struct sock *)msk);
+        struct nlmsghdr *nlh;
+	struct sk_buff *skb;
+
+	if (!genl_has_listeners(&mptcp_genl_family, net, MPTCP_PM_EV_GRP_OFFSET))
+		return;
+
+	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
+	if (!skb)
+		return;

net/mptcp/pm_netlink.c static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info) · line 1428

Quoted 10 lines
net/mptcp/pm_netlink.c +10 -0
+
+	nlh = genlmsg_put(skb, 0, 0, &mptcp_genl_family, 0, MPTCP_EVENT_ANNOUNCED);
+	if (!nlh)
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_TOKEN, msk->token))
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, MPTCP_ATTR_REM_ID, info->id))
+		goto nla_put_failure;

Another REM_ID attribute that should be a u8.

--
Mat Martineau
Intel

author
#16 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Quoted 12 lines · Mat Martineau
Quoted 10 lines · Florian Westphal

Allow userspace (mptcpd) to subscribe to mptcp genl multicast events. This implementation reuses the same event API as the mptcp kernel fork to ease integration of existing tools, e.g. mptcpd.

Supported events include:
1. start and close of an mptcp connection
2. start and close of subflows (joins)
3. announce and withdrawals of addresses
4. subflow priority (backup/non-backup) change.

Quoted 5 lines
Quoted 1 line
Signed-off-by Florian Westphal <fw(a)strlen.de>

Thanks for the patch series, Florian. I have a few minor comments below, and will work with Ossama to get mptcpd listening to the events on the upstream kernel and really test things out.

Thanks, that would help a lot.

Quoted 5 lines
Quoted 3 lines

#define MPTCP_PM_NAME "mptcp_pm"
#define MPTCP_PM_CMD_GRP_NAME "mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME "mptcp_events"

Suggest "mptcp_pm_events" instead.

I don't mind changing it but it would mean one more change needed on mptcpd side vs. the out of tree kernel.

#17 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Hi Florian,

Quoted 10 lines · Florian Westphal
Quoted 6 lines · Mat Martineau
Quoted 3 lines · Florian Westphal

#define MPTCP_PM_NAME "mptcp_pm"
#define MPTCP_PM_CMD_GRP_NAME "mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME "mptcp_events"

Suggest "mptcp_pm_events" instead.

I don't mind changing it but it would mean one more change needed on mptcpd side vs. the out of tree kernel.

Instinctively, I would prefer having different names just in case we need to do something different from one implementation to another. In userspace, we would only have to try one or the other but at least we will know which one we are connected to, "just in case" we need different behaviours for one or the other.

Of course, if we are sure we are doing the same thing is both, that's fine to have the same name. I'm maybe feeling less confident just because there is no RFC for these events :)
(arf, I'm being influenced by the IETF :-O But I'm also fine with "mptcp_events" if we prefer!)

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
author
#18 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Quoted 11 lines · Matthieu Baerts

Hi Florian,

Quoted 8 lines · Florian Westphal
Quoted 6 lines · Mat Martineau
Quoted 3 lines · Florian Westphal

#define MPTCP_PM_NAME "mptcp_pm"
#define MPTCP_PM_CMD_GRP_NAME "mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME "mptcp_events"

Suggest "mptcp_pm_events" instead.

Quoted 8 lines
Quoted 2 lines

I don't mind changing it but it would mean one more change needed on mptcpd side vs. the out of tree kernel.

Instinctively, I would prefer having different names just in case we need to do something different from one implementation to another. In userspace, we would only have to try one or the other but at least we will know which one we are connected to, "just in case" we need different behaviours for one or the other.

mptcp kernel uses "mptcp" as the genetlink name, not "mptcp_pm", so it already knows if its running on mptcp.org or upstream kernel.

#19 lore raw

[MPTCP] Re: [PATCH mptcp 7/7] mptcp: add netlink event support

Quoted 15 lines · Florian Westphal
Quoted 14 lines · Matthieu Baerts

Hi Florian,

Quoted 10 lines · Florian Westphal
Quoted 6 lines · Mat Martineau
Quoted 3 lines · Florian Westphal

#define MPTCP_PM_NAME "mptcp_pm"
#define MPTCP_PM_CMD_GRP_NAME "mptcp_pm_cmds"
+#define MPTCP_PM_EV_GRP_NAME "mptcp_events"

Suggest "mptcp_pm_events" instead.

I don't mind changing it but it would mean one more change needed on mptcpd side vs. the out of tree kernel.

Quoted 8 lines
Quoted 5 lines

Instinctively, I would prefer having different names just in case we need to do something different from one implementation to another. In userspace, we would only have to try one or the other but at least we will know which one we are connected to, "just in case" we need different behaviours for one or the other.

mptcp kernel uses "mptcp" as the genetlink name, not "mptcp_pm", so it already knows if its running on mptcp.org or upstream kernel.

Good point :)

Because the new "define" (MPTCP_PM_EV_GRP_NAME) has not the same name as in mptcp.org (MPTCP_GENL_EV_GRP_NAME) -- which makes sense -- I'm fine with either names, either prefixed with "mptcp_pm_" like the others here or the same as in mptcp.org. I guess the best is to wait for Ossama's point of view :)

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net