[MPTCP][PATCH v7 mptcp-next 0/6] fullmesh path manager support

[MPTCP][PATCH v7 mptcp-next 0/6] fullmesh path manager support

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

v7:
 - add a new patch to drop flags and ifindex arguments of
   __mptcp_subflow_connect().
 - add more comments for the testcase.

v6:
 - drop lookup_subflow_by_addrs()
 - pass three arrays instead of 'entries' in fill_local_adresses_vec()
 - drop the 'remote' argument of fill_local_addresses_vec
 - drop the 'local' argument of fill_remote_addresses_vec
 - fix the pm.subflows.
 - add limit testcase.

v5:
 - patch 1, add a new helper lookup_address_in_vec.
 - patch 2, update pm.subflows in the non-fullmesh case.
 - patch 4, add more tests.
 - tag: export/20210727T054640

v4:
 - add new helpers, fill_local/remote_addresses_vec
 - add max_subflows checks
 - add 'local' into the local addresses array only when no fullmesh
   entry found.
 - add signal,fullmesh check

v3:
 - the in-kernel fullmesh path manager has been dropped from this
   patchset, only keep the fullmesh flag support code.

v2:
 - Implement the fullmesh mode as an extension to the netlink PM, not a
   standalone PM as Paolo suggested.
 - drop duplicate code.
 - add a new per endpoint flag MPTCP_PM_ADDR_FLAG_FULLMESH.

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/193

Geliang Tang (6):
  mptcp: drop flags and ifindex arguments
  mptcp: remote addresses fullmesh
  mptcp: local addresses fullmesh
  selftests: mptcp: set and print the fullmesh flag
  selftests: mptcp: add fullmesh testcases
  selftests: mptcp: delete uncontinuous removing ids

 include/uapi/linux/mptcp.h                    |   1 +
 net/mptcp/pm_netlink.c                        | 155 ++++++++++++++++--
 net/mptcp/protocol.h                          |   5 +-
 net/mptcp/subflow.c                           |   7 +-
 .../testing/selftests/net/mptcp/mptcp_join.sh |  76 ++++++++-
 tools/testing/selftests/net/mptcp/pm_nl_ctl.c |  16 +-
 6 files changed, 233 insertions(+), 27 deletions(-)

-- 
2.31.1

[MPTCP][PATCH v7 mptcp-next 1/6] mptcp: drop flags and ifindex arguments

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

This patch added a new helper mptcp_pm_get_flags_and_ifindex_by_id(),
and used it in __mptcp_subflow_connect() to get the flags and ifindex
values.

Then the two arguments flags and ifindex of __mptcp_subflow_connect()
can be dropped.

Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
---
 net/mptcp/pm_netlink.c | 25 ++++++++++++++++++++++---
 net/mptcp/protocol.h   |  5 +++--
 net/mptcp/subflow.c    |  7 +++++--
 3 files changed, 30 insertions(+), 7 deletions(-)

diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
index ba0e1d71504d..94c68d6093de 100644
--- a/net/mptcp/pm_netlink.c
+++ b/net/mptcp/pm_netlink.c
@@ -462,8 +462,7 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 			check_work_pending(msk);
 			remote_address((struct sock_common *)sk, &remote);
 			spin_unlock_bh(&msk->pm.lock);
-			__mptcp_subflow_connect(sk, &local->addr, &remote,
-						local->flags, local->ifindex);
+			__mptcp_subflow_connect(sk, &local->addr, &remote);
 			spin_lock_bh(&msk->pm.lock);
 			return;
 		}
@@ -518,7 +517,7 @@ static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
 	local.family = remote.family;
 
 	spin_unlock_bh(&msk->pm.lock);
-	__mptcp_subflow_connect(sk, &local, &remote, 0, 0);
+	__mptcp_subflow_connect(sk, &local, &remote);
 	spin_lock_bh(&msk->pm.lock);
 
 add_addr_echo:
@@ -1103,6 +1102,26 @@ __lookup_addr_by_id(struct pm_nl_pernet *pernet, unsigned int id)
 	return NULL;
 }
 
+int mptcp_pm_get_flags_and_ifindex_by_id(struct net *net, unsigned int id,
+					 u8 *flags, int *ifindex)
+{
+	struct mptcp_pm_addr_entry *entry;
+
+	rcu_read_lock();
+	entry = __lookup_addr_by_id(net_generic(net, pm_nl_pernet_id), id);
+	rcu_read_unlock();
+
+	if (entry) {
+		*flags = entry->flags;
+		*ifindex = entry->ifindex;
+	} else {
+		*flags = 0;
+		*ifindex = 0;
+	}
+
+	return 0;
+}
+
 static bool remove_anno_list_by_saddr(struct mptcp_sock *msk,
 				      struct mptcp_addr_info *addr)
 {
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index e8a36ff52af6..d276ce16f126 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -576,8 +576,7 @@ struct socket *__mptcp_nmpc_socket(const struct mptcp_sock *msk);
 
 /* called with sk socket lock held */
 int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc,
-			    const struct mptcp_addr_info *remote,
-			    u8 flags, int ifindex);
+			    const struct mptcp_addr_info *remote);
 int mptcp_subflow_create_socket(struct sock *sk, struct socket **new_sock);
 void mptcp_info2sockaddr(const struct mptcp_addr_info *info,
 			 struct sockaddr_storage *addr,
@@ -732,6 +731,8 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 struct mptcp_pm_add_entry *
 mptcp_lookup_anno_list_by_saddr(struct mptcp_sock *msk,
 				struct mptcp_addr_info *addr);
+int mptcp_pm_get_flags_and_ifindex_by_id(struct net *net, unsigned int id,
+					 u8 *flags, int *ifindex);
 
 int mptcp_pm_announce_addr(struct mptcp_sock *msk,
 			   const struct mptcp_addr_info *addr,
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 1151926d335b..8c43aa14897a 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1355,8 +1355,7 @@ void mptcp_info2sockaddr(const struct mptcp_addr_info *info,
 }
 
 int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc,
-			    const struct mptcp_addr_info *remote,
-			    u8 flags, int ifindex)
+			    const struct mptcp_addr_info *remote)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	struct mptcp_subflow_context *subflow;
@@ -1367,6 +1366,8 @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc,
 	struct sock *ssk;
 	u32 remote_token;
 	int addrlen;
+	int ifindex;
+	u8 flags;
 	int err;
 
 	if (!mptcp_is_fully_established(sk))
@@ -1390,6 +1391,8 @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc,
 		local_id = err;
 	}
 
+	mptcp_pm_get_flags_and_ifindex_by_id(sock_net(sk), local_id,
+					     &flags, &ifindex);
 	subflow->remote_key = msk->remote_key;
 	subflow->local_key = msk->local_key;
 	subflow->token = msk->token;
-- 
2.31.1

[MPTCP][PATCH v7 mptcp-next 2/6] mptcp: remote addresses fullmesh

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

This patch added and managed a new per endpoint flag, named
MPTCP_PM_ADDR_FLAG_FULLMESH.

In mptcp_pm_create_subflow_or_signal_addr(), if such flag is set, instead
of:
        remote_address((struct sock_common *)sk, &remote);
fill a temporary allocated array of all known remote address. After
releaseing the pm lock loop on such array and create a subflow for each
remote address from the given local.

Note that the we could still use an array even for non 'fullmesh'
endpoint: with a single entry corresponding to the primary MPC subflow
remote address.

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
---
 include/uapi/linux/mptcp.h |  1 +
 net/mptcp/pm_netlink.c     | 61 +++++++++++++++++++++++++++++++++++---
 2 files changed, 58 insertions(+), 4 deletions(-)

diff --git a/include/uapi/linux/mptcp.h b/include/uapi/linux/mptcp.h
index 7b05f7102321..f66038b9551f 100644
--- a/include/uapi/linux/mptcp.h
+++ b/include/uapi/linux/mptcp.h
@@ -73,6 +73,7 @@ enum {
 #define MPTCP_PM_ADDR_FLAG_SIGNAL			(1 << 0)
 #define MPTCP_PM_ADDR_FLAG_SUBFLOW			(1 << 1)
 #define MPTCP_PM_ADDR_FLAG_BACKUP			(1 << 2)
+#define MPTCP_PM_ADDR_FLAG_FULLMESH			(1 << 3)
 
 enum {
 	MPTCP_PM_CMD_UNSPEC,
diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
index 94c68d6093de..77cab67e732d 100644
--- a/net/mptcp/pm_netlink.c
+++ b/net/mptcp/pm_netlink.c
@@ -410,6 +410,57 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk)
 	}
 }
 
+static bool lookup_address_in_vec(struct mptcp_addr_info *addrs, unsigned int nr,
+				  struct mptcp_addr_info *addr)
+{
+	int i;
+
+	for (i = 0; i < nr; i++) {
+		if (addresses_equal(&addrs[i], addr, addr->port))
+			return true;
+	}
+
+	return false;
+}
+
+/* Fill all the remote addresses into the array addrs[],
+ * and return the array size.
+ */
+static unsigned int fill_remote_addresses_vec(struct mptcp_sock *msk, bool fullmesh,
+					      struct mptcp_addr_info *addrs)
+{
+	struct sock *sk = (struct sock *)msk, *ssk;
+	struct mptcp_subflow_context *subflow;
+	struct mptcp_addr_info remote = { 0 };
+	struct pm_nl_pernet *pernet;
+	unsigned int subflows_max;
+	int i = 0;
+
+	pernet = net_generic(sock_net(sk), pm_nl_pernet_id);
+	subflows_max = mptcp_pm_get_subflows_max(msk);
+
+	/* Non-fullmesh endpoint, fill in the single entry
+	 * corresponding to the primary MPC subflow remote address
+	 */
+	if (!fullmesh) {
+		remote_address((struct sock_common *)sk, &remote);
+		msk->pm.subflows++;
+		addrs[i++] = remote;
+	} else {
+		mptcp_for_each_subflow(msk, subflow) {
+			ssk = mptcp_subflow_tcp_sock(subflow);
+			remote_address((struct sock_common *)ssk, &remote);
+			if (!lookup_address_in_vec(addrs, i, &remote) &&
+			    msk->pm.subflows < subflows_max) {
+				msk->pm.subflows++;
+				addrs[i++] = remote;
+			}
+		}
+	}
+
+	return i;
+}
+
 static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 {
 	struct sock *sk = (struct sock *)msk;
@@ -455,14 +506,16 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 	    !READ_ONCE(msk->pm.remote_deny_join_id0)) {
 		local = select_local_address(pernet, msk);
 		if (local) {
-			struct mptcp_addr_info remote = { 0 };
+			bool fullmesh = !!(local->flags & MPTCP_PM_ADDR_FLAG_FULLMESH);
+			struct mptcp_addr_info addrs[MPTCP_PM_ADDR_MAX];
+			int i, nr;
 
 			msk->pm.local_addr_used++;
-			msk->pm.subflows++;
 			check_work_pending(msk);
-			remote_address((struct sock_common *)sk, &remote);
+			nr = fill_remote_addresses_vec(msk, fullmesh, addrs);
 			spin_unlock_bh(&msk->pm.lock);
-			__mptcp_subflow_connect(sk, &local->addr, &remote);
+			for (i = 0; i < nr; i++)
+				__mptcp_subflow_connect(sk, &local->addr, &addrs[i]);
 			spin_lock_bh(&msk->pm.lock);
 			return;
 		}
-- 
2.31.1

[MPTCP][PATCH v7 mptcp-next 3/6] mptcp: local addresses fullmesh

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

In mptcp_pm_nl_add_addr_received(), fill a temporary allocate array of
all local address corresponding to the fullmesh endpoint. If such array
is empty, keep the current behavior.

Elsewhere loop on such array and create a subflow for each local address
towards the given remote address

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
---
 net/mptcp/pm_netlink.c | 73 ++++++++++++++++++++++++++++++++++++------
 1 file changed, 63 insertions(+), 10 deletions(-)

diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
index 77cab67e732d..b0be7c9477d7 100644
--- a/net/mptcp/pm_netlink.c
+++ b/net/mptcp/pm_netlink.c
@@ -536,13 +536,67 @@ static void mptcp_pm_nl_subflow_established(struct mptcp_sock *msk)
 	mptcp_pm_create_subflow_or_signal_addr(msk);
 }
 
+/* Fill all the local addresses into the array addrs[],
+ * and return the array size.
+ */
+static unsigned int fill_local_addresses_vec(struct mptcp_sock *msk,
+					     struct mptcp_addr_info *addrs)
+{
+	struct sock *sk = (struct sock *)msk;
+	struct mptcp_pm_addr_entry *entry;
+	struct mptcp_addr_info local;
+	struct pm_nl_pernet *pernet;
+	unsigned int subflows_max;
+	int i = 0;
+
+	pernet = net_generic(sock_net(sk), pm_nl_pernet_id);
+	subflows_max = mptcp_pm_get_subflows_max(msk);
+
+	rcu_read_lock();
+	__mptcp_flush_join_list(msk);
+	list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) {
+		if (!(entry->flags & MPTCP_PM_ADDR_FLAG_FULLMESH))
+			continue;
+
+		if (entry->addr.family != sk->sk_family) {
+#if IS_ENABLED(CONFIG_MPTCP_IPV6)
+			if ((entry->addr.family == AF_INET &&
+			     !ipv6_addr_v4mapped(&sk->sk_v6_daddr)) ||
+			    (sk->sk_family == AF_INET &&
+			     !ipv6_addr_v4mapped(&entry->addr.addr6)))
+#endif
+				continue;
+		}
+
+		if (msk->pm.subflows < subflows_max) {
+			msk->pm.subflows++;
+			addrs[i++] = entry->addr;
+		}
+	}
+	rcu_read_unlock();
+
+	/* If the array is empty, fill in the single
+	 * 'IPADDRANY' local address
+	 */
+	if (!i) {
+		memset(&local, 0, sizeof(local));
+		local.family = msk->pm.remote.family;
+
+		msk->pm.subflows++;
+		addrs[i++] = local;
+	}
+
+	return i;
+}
+
 static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
 {
+	struct mptcp_addr_info addrs[MPTCP_PM_ADDR_MAX];
 	struct sock *sk = (struct sock *)msk;
 	unsigned int add_addr_accept_max;
 	struct mptcp_addr_info remote;
-	struct mptcp_addr_info local;
 	unsigned int subflows_max;
+	int i, nr;
 
 	add_addr_accept_max = mptcp_pm_get_add_addr_accept_max(msk);
 	subflows_max = mptcp_pm_get_subflows_max(msk);
@@ -554,23 +608,22 @@ static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
 	if (lookup_subflow_by_daddr(&msk->conn_list, &msk->pm.remote))
 		goto add_addr_echo;
 
-	msk->pm.add_addr_accepted++;
-	msk->pm.subflows++;
-	if (msk->pm.add_addr_accepted >= add_addr_accept_max ||
-	    msk->pm.subflows >= subflows_max)
-		WRITE_ONCE(msk->pm.accept_addr, false);
-
 	/* connect to the specified remote address, using whatever
 	 * local address the routing configuration will pick.
 	 */
 	remote = msk->pm.remote;
 	if (!remote.port)
 		remote.port = sk->sk_dport;
-	memset(&local, 0, sizeof(local));
-	local.family = remote.family;
+	nr = fill_local_addresses_vec(msk, addrs);
+
+	msk->pm.add_addr_accepted++;
+	if (msk->pm.add_addr_accepted >= add_addr_accept_max ||
+	    msk->pm.subflows >= subflows_max)
+		WRITE_ONCE(msk->pm.accept_addr, false);
 
 	spin_unlock_bh(&msk->pm.lock);
-	__mptcp_subflow_connect(sk, &local, &remote);
+	for (i = 0; i < nr; i++)
+		__mptcp_subflow_connect(sk, &addrs[i], &remote);
 	spin_lock_bh(&msk->pm.lock);
 
 add_addr_echo:
-- 
2.31.1

[MPTCP][PATCH v7 mptcp-next 4/6] selftests: mptcp: set and print the fullmesh flag

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

This patch dealt with the MPTCP_PM_ADDR_FLAG_FULLMESH flag in add_addr()
and print_addr(), to set and print out the fullmesh flag.

Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
---
 tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/net/mptcp/pm_nl_ctl.c b/tools/testing/selftests/net/mptcp/pm_nl_ctl.c
index 115decfdc1ef..354784512748 100644
--- a/tools/testing/selftests/net/mptcp/pm_nl_ctl.c
+++ b/tools/testing/selftests/net/mptcp/pm_nl_ctl.c
@@ -25,7 +25,7 @@
 static void syntax(char *argv[])
 {
 	fprintf(stderr, "%s add|get|set|del|flush|dump|accept [<args>]\n", argv[0]);
-	fprintf(stderr, "\tadd [flags signal|subflow|backup] [id <nr>] [dev <name>] <ip>\n");
+	fprintf(stderr, "\tadd [flags signal|subflow|backup|fullmesh] [id <nr>] [dev <name>] <ip>\n");
 	fprintf(stderr, "\tdel <id> [<ip>]\n");
 	fprintf(stderr, "\tget <id>\n");
 	fprintf(stderr, "\tset <ip> [flags backup|nobackup]\n");
@@ -236,11 +236,18 @@ int add_addr(int fd, int pm_family, int argc, char *argv[])
 					flags |= MPTCP_PM_ADDR_FLAG_SIGNAL;
 				else if (!strcmp(tok, "backup"))
 					flags |= MPTCP_PM_ADDR_FLAG_BACKUP;
+				else if (!strcmp(tok, "fullmesh"))
+					flags |= MPTCP_PM_ADDR_FLAG_FULLMESH;
 				else
 					error(1, errno,
 					      "unknown flag %s", argv[arg]);
 			}
 
+			if (flags & MPTCP_PM_ADDR_FLAG_SIGNAL &&
+			    flags & MPTCP_PM_ADDR_FLAG_FULLMESH) {
+				error(1, errno, "error flag fullmesh");
+			}
+
 			rta = (void *)(data + off);
 			rta->rta_type = MPTCP_PM_ADDR_ATTR_FLAGS;
 			rta->rta_len = RTA_LENGTH(4);
@@ -422,6 +429,13 @@ static void print_addr(struct rtattr *attrs, int len)
 					printf(",");
 			}
 
+			if (flags & MPTCP_PM_ADDR_FLAG_FULLMESH) {
+				printf("fullmesh");
+				flags &= ~MPTCP_PM_ADDR_FLAG_FULLMESH;
+				if (flags)
+					printf(",");
+			}
+
 			/* bump unknown flags, if any */
 			if (flags)
 				printf("0x%x", flags);
-- 
2.31.1

[MPTCP][PATCH v7 mptcp-next 5/6] selftests: mptcp: add fullmesh testcases

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

This patch added the testcases for the fullmesh address flag of the path
manager.

Reuse the above 10 address numbers for the fullmesh test.

Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
---
 .../testing/selftests/net/mptcp/mptcp_join.sh | 66 ++++++++++++++++++-
 1 file changed, 63 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index 937e861e9490..cb5f9a43baa1 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -367,7 +367,13 @@ do_transfer()
 	fi
 
 	if [ $addr_nr_ns2 -gt 0 ]; then
-		let add_nr_ns2=addr_nr_ns2
+		if [ $addr_nr_ns2 -gt 10 ]; then
+			let add_nr_ns2=addr_nr_ns2-10
+			flags=subflow,fullmesh
+		else
+			let add_nr_ns2=addr_nr_ns2
+			flags=subflow
+		fi
 		counter=3
 		sleep 1
 		while [ $add_nr_ns2 -gt 0 ]; do
@@ -377,7 +383,7 @@ do_transfer()
 			else
 				addr="10.0.$counter.2"
 			fi
-			ip netns exec $ns2 ./pm_nl_ctl add $addr flags subflow
+			ip netns exec $ns2 ./pm_nl_ctl add $addr flags $flags
 			let counter+=1
 			let add_nr_ns2-=1
 		done
@@ -1697,6 +1703,55 @@ deny_join_id0_tests()
 	chk_join_nr "subflow and address allow join id0 2" 1 1 1
 }
 
+fullmesh_tests()
+{
+	# fullmesh 1
+	# 2 fullmesh addrs in ns2, added before the connection,
+	# 1 non-fullmesh addr in ns1, added during the connection.
+	reset
+	ip netns exec $ns1 ./pm_nl_ctl limits 0 4
+	ip netns exec $ns2 ./pm_nl_ctl limits 1 4
+	ip netns exec $ns2 ./pm_nl_ctl add 10.0.2.2 flags subflow,fullmesh
+	ip netns exec $ns2 ./pm_nl_ctl add 10.0.3.2 flags subflow,fullmesh
+	run_tests $ns1 $ns2 10.0.1.1 0 1 0 slow
+	chk_join_nr "fullmesh test 2x1" 4 4 4
+	chk_add_nr 1 1
+
+	# fullmesh 2
+	# 1 non-fullmesh addr in ns1, added before the connection,
+	# 1 fullmesh addr in ns2, added during the connection.
+	reset
+	ip netns exec $ns1 ./pm_nl_ctl limits 1 3
+	ip netns exec $ns2 ./pm_nl_ctl limits 1 3
+	ip netns exec $ns1 ./pm_nl_ctl add 10.0.2.1 flags signal
+	run_tests $ns1 $ns2 10.0.1.1 0 0 11 slow # 11 here meams creating 1 fullmesh subflow
+	chk_join_nr "fullmesh test 1x1" 3 3 3
+	chk_add_nr 1 1
+
+	# fullmesh 3
+	# 1 non-fullmesh addr in ns1, added before the connection,
+	# 2 fullmesh addrs in ns2, added during the connection.
+	reset
+	ip netns exec $ns1 ./pm_nl_ctl limits 2 5
+	ip netns exec $ns2 ./pm_nl_ctl limits 1 5
+	ip netns exec $ns1 ./pm_nl_ctl add 10.0.2.1 flags signal
+	run_tests $ns1 $ns2 10.0.1.1 0 0 12 slow # 12 here meams creating 2 fullmesh subflows
+	chk_join_nr "fullmesh test 1x2" 5 5 5
+	chk_add_nr 1 1
+
+	# fullmesh 4
+	# 1 non-fullmesh addr in ns1, added before the connection,
+	# 2 fullmesh addrs in ns2, added during the connection,
+	# limit max_subflows to 4.
+	reset
+	ip netns exec $ns1 ./pm_nl_ctl limits 2 4
+	ip netns exec $ns2 ./pm_nl_ctl limits 1 4
+	ip netns exec $ns1 ./pm_nl_ctl add 10.0.2.1 flags signal
+	run_tests $ns1 $ns2 10.0.1.1 0 0 12 slow
+	chk_join_nr "fullmesh test 1x2, limited" 4 4 4
+	chk_add_nr 1 1
+}
+
 all_tests()
 {
 	subflows_tests
@@ -1712,6 +1767,7 @@ all_tests()
 	syncookies_tests
 	checksum_tests
 	deny_join_id0_tests
+	fullmesh_tests
 }
 
 usage()
@@ -1730,6 +1786,7 @@ usage()
 	echo "  -k syncookies_tests"
 	echo "  -S checksum_tests"
 	echo "  -d deny_join_id0_tests"
+	echo "  -m fullmesh_tests"
 	echo "  -c capture pcap files"
 	echo "  -C enable data checksum"
 	echo "  -h help"
@@ -1765,7 +1822,7 @@ if [ $do_all_tests -eq 1 ]; then
 	exit $ret
 fi
 
-while getopts 'fsltra64bpkdchCS' opt; do
+while getopts 'fsltra64bpkdmchCS' opt; do
 	case $opt in
 		f)
 			subflows_tests
@@ -1806,6 +1863,9 @@ while getopts 'fsltra64bpkdchCS' opt; do
 		d)
 			deny_join_id0_tests
 			;;
+		m)
+			fullmesh_tests
+			;;
 		c)
 			;;
 		C)
-- 
2.31.1

[MPTCP][PATCH v7 mptcp-next 6/6] selftests: mptcp: delete uncontinuous removing ids

[Geliang Tang]

From: Geliang Tang <geliangtang@xiaomi.com>

The removing addresses testcases can only deal with the continuous ids.
This patch added the uncontinuous removing ids support.

Fixes: f87744ad42446 ("selftests: mptcp: set addr id for removing testcases")
Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
---
 tools/testing/selftests/net/mptcp/mptcp_join.sh | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index cb5f9a43baa1..dab7fbd6813a 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -344,17 +344,18 @@ do_transfer()
 		let rm_nr_ns1=-addr_nr_ns1
 		if [ $rm_nr_ns1 -lt 8 ]; then
 			counter=1
+			pos=1
 			dump=(`ip netns exec ${listener_ns} ./pm_nl_ctl dump`)
 			if [ ${#dump[@]} -gt 0 ]; then
-				id=${dump[1]}
 				sleep 1
 
 				while [ $counter -le $rm_nr_ns1 ]
 				do
+					id=${dump[$pos]}
 					ip netns exec ${listener_ns} ./pm_nl_ctl del $id
 					sleep 1
 					let counter+=1
-					let id+=1
+					let pos+=5
 				done
 			fi
 		elif [ $rm_nr_ns1 -eq 8 ]; then
@@ -392,17 +393,18 @@ do_transfer()
 		let rm_nr_ns2=-addr_nr_ns2
 		if [ $rm_nr_ns2 -lt 8 ]; then
 			counter=1
+			pos=1
 			dump=(`ip netns exec ${connector_ns} ./pm_nl_ctl dump`)
 			if [ ${#dump[@]} -gt 0 ]; then
-				id=${dump[1]}
 				sleep 1
 
 				while [ $counter -le $rm_nr_ns2 ]
 				do
+					id=${dump[$pos]}
 					ip netns exec ${connector_ns} ./pm_nl_ctl del $id
 					sleep 1
 					let counter+=1
-					let id+=1
+					let pos+=5
 				done
 			fi
 		elif [ $rm_nr_ns2 -eq 8 ]; then
-- 
2.31.1

Re: [MPTCP][PATCH v7 mptcp-next 1/6] mptcp: drop flags and ifindex arguments

[Paolo Abeni]

On Thu, 2021-07-29 at 15:20 +0800, Geliang Tang wrote:
> From: Geliang Tang <geliangtang@xiaomi.com>
> 
> This patch added a new helper mptcp_pm_get_flags_and_ifindex_by_id(),
> and used it in __mptcp_subflow_connect() to get the flags and ifindex
> values.
> 
> Then the two arguments flags and ifindex of __mptcp_subflow_connect()
> can be dropped.
> 
> Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
> ---
>  net/mptcp/pm_netlink.c | 25 ++++++++++++++++++++++---
>  net/mptcp/protocol.h   |  5 +++--
>  net/mptcp/subflow.c    |  7 +++++--
>  3 files changed, 30 insertions(+), 7 deletions(-)
> 
> diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
> index ba0e1d71504d..94c68d6093de 100644
> --- a/net/mptcp/pm_netlink.c
> +++ b/net/mptcp/pm_netlink.c
> @@ -462,8 +462,7 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
>  			check_work_pending(msk);
>  			remote_address((struct sock_common *)sk, &remote);
>  			spin_unlock_bh(&msk->pm.lock);
> -			__mptcp_subflow_connect(sk, &local->addr, &remote,
> -						local->flags, local->ifindex);
> +			__mptcp_subflow_connect(sk, &local->addr, &remote);
>  			spin_lock_bh(&msk->pm.lock);
>  			return;
>  		}
> @@ -518,7 +517,7 @@ static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
>  	local.family = remote.family;
>  
>  	spin_unlock_bh(&msk->pm.lock);
> -	__mptcp_subflow_connect(sk, &local, &remote, 0, 0);
> +	__mptcp_subflow_connect(sk, &local, &remote);
>  	spin_lock_bh(&msk->pm.lock);
>  
>  add_addr_echo:
> @@ -1103,6 +1102,26 @@ __lookup_addr_by_id(struct pm_nl_pernet *pernet, unsigned int id)
>  	return NULL;
>  }
>  
> +int mptcp_pm_get_flags_and_ifindex_by_id(struct net *net, unsigned int id,
> +					 u8 *flags, int *ifindex)
> +{
> +	struct mptcp_pm_addr_entry *entry;
> +
> +	rcu_read_lock();
> +	entry = __lookup_addr_by_id(net_generic(net, pm_nl_pernet_id), id);
> +	rcu_read_unlock();
> +
> +	if (entry) {
> +		*flags = entry->flags;
> +		*ifindex = entry->ifindex;
> +	} else {
> +		*flags = 0;
> +		*ifindex = 0;
> +	}

You need to extend the RCU section after the entry reference above, or
on preempt-enabled build, the entry could be deleted in between the rcu
 unlock and the later access.

I'm wondering if there are issues with the 0 id ?!?

/P

Re: [MPTCP][PATCH v7 mptcp-next 1/6] mptcp: drop flags and ifindex arguments

[Geliang Tang]

Paolo Abeni <pabeni@redhat.com> 于2021年7月29日周四 下午7:12写道：
>
> On Thu, 2021-07-29 at 15:20 +0800, Geliang Tang wrote:
> > From: Geliang Tang <geliangtang@xiaomi.com>
> >
> > This patch added a new helper mptcp_pm_get_flags_and_ifindex_by_id(),
> > and used it in __mptcp_subflow_connect() to get the flags and ifindex
> > values.
> >
> > Then the two arguments flags and ifindex of __mptcp_subflow_connect()
> > can be dropped.
> >
> > Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
> > ---
> >  net/mptcp/pm_netlink.c | 25 ++++++++++++++++++++++---
> >  net/mptcp/protocol.h   |  5 +++--
> >  net/mptcp/subflow.c    |  7 +++++--
> >  3 files changed, 30 insertions(+), 7 deletions(-)
> >
> > diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
> > index ba0e1d71504d..94c68d6093de 100644
> > --- a/net/mptcp/pm_netlink.c
> > +++ b/net/mptcp/pm_netlink.c
> > @@ -462,8 +462,7 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
> >                       check_work_pending(msk);
> >                       remote_address((struct sock_common *)sk, &remote);
> >                       spin_unlock_bh(&msk->pm.lock);
> > -                     __mptcp_subflow_connect(sk, &local->addr, &remote,
> > -                                             local->flags, local->ifindex);
> > +                     __mptcp_subflow_connect(sk, &local->addr, &remote);
> >                       spin_lock_bh(&msk->pm.lock);
> >                       return;
> >               }
> > @@ -518,7 +517,7 @@ static void mptcp_pm_nl_add_addr_received(struct mptcp_sock *msk)
> >       local.family = remote.family;
> >
> >       spin_unlock_bh(&msk->pm.lock);
> > -     __mptcp_subflow_connect(sk, &local, &remote, 0, 0);
> > +     __mptcp_subflow_connect(sk, &local, &remote);
> >       spin_lock_bh(&msk->pm.lock);
> >
> >  add_addr_echo:
> > @@ -1103,6 +1102,26 @@ __lookup_addr_by_id(struct pm_nl_pernet *pernet, unsigned int id)
> >       return NULL;
> >  }
> >
> > +int mptcp_pm_get_flags_and_ifindex_by_id(struct net *net, unsigned int id,
> > +                                      u8 *flags, int *ifindex)
> > +{
> > +     struct mptcp_pm_addr_entry *entry;
> > +
> > +     rcu_read_lock();
> > +     entry = __lookup_addr_by_id(net_generic(net, pm_nl_pernet_id), id);
> > +     rcu_read_unlock();
> > +
> > +     if (entry) {
> > +             *flags = entry->flags;
> > +             *ifindex = entry->ifindex;
> > +     } else {
> > +             *flags = 0;
> > +             *ifindex = 0;
> > +     }
>
> You need to extend the RCU section after the entry reference above, or
> on preempt-enabled build, the entry could be deleted in between the rcu
>  unlock and the later access.

Thanks Paolo, I'll send a squash-to patch to fix this.

>
> I'm wondering if there are issues with the 0 id ?!?

The 0 id address isn't on the local_addr_list, we should set it's
flags and ifindex all to 0.

I'll add a non-zero check in the squash-to patch too.

Thanks,
-Geliang

>
> /P
>

Re: [MPTCP][PATCH v7 mptcp-next 0/6] fullmesh path manager support

[Mat Martineau]

On Thu, 29 Jul 2021, Geliang Tang wrote:

> From: Geliang Tang <geliangtang@xiaomi.com>
>
> v7:
> - add a new patch to drop flags and ifindex arguments of
>   __mptcp_subflow_connect().
> - add more comments for the testcase.
>

Ok, in addition to the review I did for v6 I've also looked at the subflow 
topologies generated by the test cases. Those look good to me.

I think this is good for the export branch (with the squash-to patch).
Thanks Geliang.

Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>


> v6:
> - drop lookup_subflow_by_addrs()
> - pass three arrays instead of 'entries' in fill_local_adresses_vec()
> - drop the 'remote' argument of fill_local_addresses_vec
> - drop the 'local' argument of fill_remote_addresses_vec
> - fix the pm.subflows.
> - add limit testcase.
>
> v5:
> - patch 1, add a new helper lookup_address_in_vec.
> - patch 2, update pm.subflows in the non-fullmesh case.
> - patch 4, add more tests.
> - tag: export/20210727T054640
>
> v4:
> - add new helpers, fill_local/remote_addresses_vec
> - add max_subflows checks
> - add 'local' into the local addresses array only when no fullmesh
>   entry found.
> - add signal,fullmesh check
>
> v3:
> - the in-kernel fullmesh path manager has been dropped from this
>   patchset, only keep the fullmesh flag support code.
>
> v2:
> - Implement the fullmesh mode as an extension to the netlink PM, not a
>   standalone PM as Paolo suggested.
> - drop duplicate code.
> - add a new per endpoint flag MPTCP_PM_ADDR_FLAG_FULLMESH.
>
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/193
>
> Geliang Tang (6):
>  mptcp: drop flags and ifindex arguments
>  mptcp: remote addresses fullmesh
>  mptcp: local addresses fullmesh
>  selftests: mptcp: set and print the fullmesh flag
>  selftests: mptcp: add fullmesh testcases
>  selftests: mptcp: delete uncontinuous removing ids
>
> include/uapi/linux/mptcp.h                    |   1 +
> net/mptcp/pm_netlink.c                        | 155 ++++++++++++++++--
> net/mptcp/protocol.h                          |   5 +-
> net/mptcp/subflow.c                           |   7 +-
> .../testing/selftests/net/mptcp/mptcp_join.sh |  76 ++++++++-
> tools/testing/selftests/net/mptcp/pm_nl_ctl.c |  16 +-
> 6 files changed, 233 insertions(+), 27 deletions(-)
>
> -- 
> 2.31.1
>
>
>

--
Mat Martineau
Intel

Re: [MPTCP][PATCH v7 mptcp-next 0/6] fullmesh path manager support

[Matthieu Baerts]

Hi Geliang, Mat, Paolo,

On 29/07/2021 09:20, Geliang Tang wrote:
> From: Geliang Tang <geliangtang@xiaomi.com>
> 
> v7:
>  - add a new patch to drop flags and ifindex arguments of
>    __mptcp_subflow_connect().
>  - add more comments for the testcase.

(...)

> 
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/193
> 
> Geliang Tang (6):
>   mptcp: drop flags and ifindex arguments
>   mptcp: remote addresses fullmesh
>   mptcp: local addresses fullmesh
>   selftests: mptcp: set and print the fullmesh flag
>   selftests: mptcp: add fullmesh testcases
>   selftests: mptcp: delete uncontinuous removing ids

Thank you for the patches and the reviews!

Now in our tree with Mat's RvB tag:

- 8235745392a1: mptcp: drop flags and ifindex arguments
- 31e4fdaf9c0b: mptcp: remote addresses fullmesh
- 6ad65bd12896: mptcp: local addresses fullmesh
- 3faae7f56182: selftests: mptcp: set and print the fullmesh flag
- ee6f6d4a5b48: selftests: mptcp: add fullmesh testcases
- 68e72848611e: selftests: mptcp: delete uncontinuous removing ids
- Results: d5c73e4f859d..dcbca66e0658

And the squash-to patch:

- 22876224b527: "squashed" in "mptcp: drop flags and ifindex arguments"
- Results: dcbca66e0658..b1d5eb590503

Builds and tests are now in progress:

https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20210803T155939
https://github.com/multipath-tcp/mptcp_net-next/actions/workflows/build-validation.yml?query=branch:export/20210803T155939

Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net