Skip to contentSkip to conversation

IPv6 support

PATCH Superseded Peter Krystad 3 participants 15 messages Oct 30, 2019 – Nov 5, 2019

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Peter Krystad

author
#0 lore raw

cover IPv6 support

Implemented support for IPv6. selftests pass with ipv6=true, but they take twice as long to run...

With re-factoring the addition of IPv6 is very clean but I did not split files yet as it will be hard to squash any changes after the such a split.

Still to do: multiple subflows, handling IPv4-mapped connections.

Peter Krystad (6):
mptcp: Reference af_ops routines through struct ipv4_specific
mptcp: Re-factor create socket/subflow routines to be af-agnostic
mptcp: Re-factor mptcp_stream_ops routines to be af-agnostic
mptcp: Re-factor subflow routines to be af-agnostic
mptcp: Export low-level routines for IPv6
mptcp: Add IPv6 support

 include/net/mptcp.h  |   3 +
 include/net/tcp.h    |   3 +
 net/ipv6/tcp_ipv6.c  |  16 +++-
 net/mptcp/ctrl.c     |  11 +++
 net/mptcp/protocol.c | 174 +++++++++++++++++++++++++++++++--------
 net/mptcp/protocol.h |   9 +-
 net/mptcp/subflow.c  | 191 +++++++++++++++++++++++++++++++++++++------
 7 files changed, 342 insertions(+), 65 deletions(-)
Signature · 1 line
2.17.2
author
#1 lore raw

PATCH 1/6 mptcp: Reference af_ops routines through struct ipv4_specific

Make this look similar to how it will have to be done for IPv6, as the IPv6 routines are declared static and thus not visible.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/subflow.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
net/mptcp/subflow.c +6 -5
index 7bae7c35ea6b..e6bb68e1c292 100644
@@ -25,6 +25,8 @@ static inline void SUBFLOW_REQ_INC_STATS(struct request_sock *req,
 	MPTCP_INC_STATS(sock_net(req_to_sk(req)), field);
 }
 
+static struct inet_connection_sock_af_ops subflow_specific;
+
 static int subflow_rebuild_header(struct sock *sk)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
@@ -45,7 +47,7 @@ static int subflow_rebuild_header(struct sock *sk)
 	if (err)
 		return err;
 
-	return inet_sk_rebuild_header(sk);
+	return ipv4_specific.rebuild_header(sk);
 }
 
 static void subflow_req_destructor(struct request_sock *req)
@@ -169,7 +171,7 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
 
-	inet_sk_rx_dst_set(sk, skb);
+	ipv4_specific.sk_rx_dst_set(sk, skb);
 
 	if (!subflow->conn)
 		return;
@@ -276,7 +278,8 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 			return NULL;
 	}
 
-	child = tcp_v4_syn_recv_sock(sk, skb, req, dst, req_unhash, own_req);
+	child = ipv4_specific.syn_recv_sock(sk, skb, req, dst, req_unhash,
+					    own_req);
 
 	if (child && *own_req) {
 		struct mptcp_subflow_context *ctx = mptcp_subflow_ctx(child);
@@ -309,8 +312,6 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 	return NULL;
 }
 
-static struct inet_connection_sock_af_ops subflow_specific;
-
 enum mapping_status {
 	MAPPING_OK,
 	MAPPING_INVALID,
Signature · 1 line
2.17.2
#2 lore raw

[MPTCP] Re: [PATCH 1/6] mptcp: Reference af_ops routines through struct ipv4_specific

Quoted 4 lines · Peter Krystad

Make this look similar to how it will have to be done for IPv6, as the IPv6 routines are declared static and thus not visible.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>

I think:

- 'subflow_finish_connect()' and 'inet_connection_sock_af_ops subflow_specific;' changes should be squashed into "mptcp: Handle MP_CAPABLE options for outgoing connections"

- 'subflow_syn_recv_sock()' into "mptcp: Create SUBFLOW socket for incoming connections"

- 'inet_sk_rebuild_head()' into "mptcp: Add key generation and token tree"

right?

/P

author
#3 lore raw

[MPTCP] Re: [PATCH 1/6] mptcp: Reference af_ops routines through struct ipv4_specific

Quoted 15 lines · Paolo Abeni
Quoted 4 lines · Peter Krystad

Make this look similar to how it will have to be done for IPv6, as the IPv6 routines are declared static and thus not visible.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>

I think:

- 'subflow_finish_connect()' and 'inet_connection_sock_af_ops subflow_specific;' changes should be squashed into "mptcp: Handle MP_CAPABLE options for outgoing connections"

- 'subflow_syn_recv_sock()' into "mptcp: Create SUBFLOW socket for incoming connections"

Quoted 7 lines

- 'inet_sk_rebuild_head()' into "mptcp: Add key generation and token tree"

right?

/P

In the v2 patchset I am going to modify this change. We will save a pointer to the original af_icsk_ops structure and call the inet_xxx routines through it. This will save a lot of code duplication for IPv6.

The change will squash to the same commits you identify above.

Peter.

author
#4 lore raw

PATCH 2/6 mptcp: Re-factor create socket/subflow routines to be af-agnostic

In reply to Peter Krystad

Re-factor mptcp_subflow_create_socket() and mptcp_socket_create_get() to work for both IPv4 and IPv6.

squashto: Associate MPTCP context with TCP socket

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/protocol.c | 11 ++++++-----
 net/mptcp/protocol.h |  3 ++-
 net/mptcp/subflow.c  |  7 ++++---
 3 files changed, 12 insertions(+), 9 deletions(-)
net/mptcp/protocol.c +6 -5
index 3af5f121af9e..ee108e246dcf 100644
@@ -1072,7 +1072,8 @@ static struct proto mptcp_prot = {
 	.no_autobind	= 1,
 };
 
-static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk)
+static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk,
+					      sa_family_t family)
 {
 	struct mptcp_subflow_context *subflow;
 	struct sock *sk = (struct sock *)msk;
@@ -1084,7 +1085,7 @@ static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk)
 	if (ssock)
 		goto release;
 
-	err = mptcp_subflow_create_socket(sk, &ssock);
+	err = mptcp_subflow_create_socket(sk, family, &ssock);
 	if (err) {
 		ssock = ERR_PTR(err);
 		goto release;
@@ -1112,7 +1113,7 @@ static int mptcp_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 	if (uaddr->sa_family != AF_INET) // @@ allow only IPv4 for now
 		return err;
 
-	ssock = mptcp_socket_create_get(msk);
+	ssock = mptcp_socket_create_get(msk, AF_INET);
 	if (IS_ERR(ssock))
 		return PTR_ERR(ssock);
 
@@ -1131,7 +1132,7 @@ static int mptcp_stream_connect(struct socket *sock, struct sockaddr *uaddr,
 	if (uaddr->sa_family != AF_INET) // @@ allow only IPv4 for now
 		return err;
 
-	ssock = mptcp_socket_create_get(msk);
+	ssock = mptcp_socket_create_get(msk, AF_INET);
 	if (IS_ERR(ssock))
 		return PTR_ERR(ssock);
 
@@ -1193,7 +1194,7 @@ static int mptcp_listen(struct socket *sock, int backlog)
 
 	pr_debug("msk=%p", msk);
 
-	ssock = mptcp_socket_create_get(msk);
+	ssock = mptcp_socket_create_get(msk, AF_INET);
 	if (IS_ERR(ssock))
 		return PTR_ERR(ssock);
 
net/mptcp/protocol.h +2 -1
index 8fe5f9383d38..2635d68e58a0 100644
@@ -266,7 +266,8 @@ bool mptcp_subflow_data_available(struct sock *sk);
 void mptcp_subflow_init(void);
 int mptcp_subflow_connect(struct sock *sk, struct sockaddr *local,
 			  struct sockaddr *remote, u8 remote_id);
-int mptcp_subflow_create_socket(struct sock *sk, struct socket **new_sock);
+int mptcp_subflow_create_socket(struct sock *sk, sa_family_t family,
+				struct socket **new_sock);
 
 extern const struct inet_connection_sock_af_ops ipv4_specific;
 
net/mptcp/subflow.c +4 -3
index e6bb68e1c292..13e207458e0b 100644
@@ -632,7 +632,7 @@ int mptcp_subflow_connect(struct sock *sk, struct sockaddr *local,
 	int err;
 
 	lock_sock(sk);
-	err = mptcp_subflow_create_socket(sk, &sf);
+	err = mptcp_subflow_create_socket(sk, local->sa_family, &sf);
 	if (err) {
 		release_sock(sk);
 		return err;
@@ -675,14 +675,15 @@ int mptcp_subflow_connect(struct sock *sk, struct sockaddr *local,
 	return err;
 }
 
-int mptcp_subflow_create_socket(struct sock *sk, struct socket **new_sock)
+int mptcp_subflow_create_socket(struct sock *sk, sa_family_t family,
+				struct socket **new_sock)
 {
 	struct mptcp_subflow_context *subflow;
 	struct net *net = sock_net(sk);
 	struct socket *sf;
 	int err;
 
-	err = sock_create_kern(net, PF_INET, SOCK_STREAM, IPPROTO_TCP, &sf);
+	err = sock_create_kern(net, family, SOCK_STREAM, IPPROTO_TCP, &sf);
 	if (err)
 		return err;
 
Signature · 1 line
2.17.2
author
#5 lore raw

PATCH 3/6 mptcp: Re-factor mptcp_stream_ops routines to be af-agnostic

In reply to Peter Krystad

Split mptcp_bind/getname/listen into two parts: af-specific part and functional part.

squashto: Associate MPTCP context with TCP socket (mptcp_bind) and Create SUBFLOW socket for incoming connections (rest)

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/protocol.c | 84 +++++++++++++++++++++++++++++---------------
 1 file changed, 56 insertions(+), 28 deletions(-)
net/mptcp/protocol.c +56 -28
index ee108e246dcf..9659afc9b444 100644
@@ -1104,24 +1104,32 @@ static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk,
 	return ssock;
 }
 
-static int mptcp_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
+static int mptcp_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len,
+		      sa_family_t family, const struct proto_ops *ops)
 {
 	struct mptcp_sock *msk = mptcp_sk(sock->sk);
 	struct socket *ssock;
-	int err = -ENOTSUPP;
-
-	if (uaddr->sa_family != AF_INET) // @@ allow only IPv4 for now
-		return err;
+	int err;
 
-	ssock = mptcp_socket_create_get(msk, AF_INET);
+	ssock = mptcp_socket_create_get(msk, family);
 	if (IS_ERR(ssock))
 		return PTR_ERR(ssock);
 
-	err = inet_bind(ssock, uaddr, addr_len);
+	err = ops->bind(ssock, uaddr, addr_len);
 	sock_put(ssock->sk);
 	return err;
 }
 
+static int mptcp_v4_bind(struct socket *sock, struct sockaddr *uaddr,
+			 int addr_len)
+{
+	int err;
+
+	err = mptcp_bind(sock, uaddr, addr_len, AF_INET, &inet_stream_ops);
+
+	return err;
+}
+
 static int mptcp_stream_connect(struct socket *sock, struct sockaddr *uaddr,
 				int addr_len, int flags)
 {
@@ -1142,31 +1150,19 @@ static int mptcp_stream_connect(struct socket *sock, struct sockaddr *uaddr,
 }
 
 static int mptcp_getname(struct socket *sock, struct sockaddr *uaddr,
-			 int peer)
+			 int peer, const struct proto_ops *ops)
 {
 	struct mptcp_sock *msk = mptcp_sk(sock->sk);
 	struct socket *ssock;
 	struct sock *ssk;
 	int ret;
 
-	if (sock->sk->sk_prot == &tcp_prot) {
-		/* we are being invoked from __sys_accept4, after
-		 * mptcp_accept() has just accepted a non-mp-capable
-		 * flow: sk is a tcp_sk, not an mptcp one.
-		 *
-		 * Hand the socket over to tcp so all further socket ops
-		 * bypass mptcp.
-		 */
-		sock->ops = &inet_stream_ops;
-		return inet_getname(sock, uaddr, peer);
-	}
-
 	lock_sock(sock->sk);
 	ssock = __mptcp_fallback_get_ref(msk);
 	if (ssock) {
 		release_sock(sock->sk);
 		pr_debug("subflow=%p", ssock->sk);
-		ret = inet_getname(ssock, uaddr, peer);
+		ret = ops->getname(ssock, uaddr, peer);
 		sock_put(ssock->sk);
 		return ret;
 	}
@@ -1181,12 +1177,35 @@ static int mptcp_getname(struct socket *sock, struct sockaddr *uaddr,
 		return -ENOTCONN;
 	}
 
-	ret = inet_getname(ssk->sk_socket, uaddr, peer);
+	ret = ops->getname(ssk->sk_socket, uaddr, peer);
 	release_sock(sock->sk);
 	return ret;
 }
 
-static int mptcp_listen(struct socket *sock, int backlog)
+static int mptcp_v4_getname(struct socket *sock, struct sockaddr *uaddr,
+			    int peer)
+{
+	int ret;
+
+	if (sock->sk->sk_prot == &tcp_prot) {
+		/* we are being invoked from __sys_accept4 after
+		 * mptcp_accept() has accepted a non-mp-capable
+		 * subflow: sk is a tcp_sk, not mptcp.
+		 *
+		 * Hand the socket over to tcp so all further
+		 * socket ops bypass mptcp.
+		 */
+		sock->ops = &inet_stream_ops;
+		return sock->ops->getname(sock, uaddr, peer);
+	}
+
+	ret = mptcp_getname(sock, uaddr, peer, &inet_stream_ops);
+
+	return ret;
+}
+
+static int mptcp_listen(struct socket *sock, int backlog, sa_family_t family,
+			const struct proto_ops *ops)
 {
 	struct mptcp_sock *msk = mptcp_sk(sock->sk);
 	struct socket *ssock;
@@ -1194,15 +1213,24 @@ static int mptcp_listen(struct socket *sock, int backlog)
 
 	pr_debug("msk=%p", msk);
 
-	ssock = mptcp_socket_create_get(msk, AF_INET);
+	ssock = mptcp_socket_create_get(msk, family);
 	if (IS_ERR(ssock))
 		return PTR_ERR(ssock);
 
-	err = inet_listen(ssock, backlog);
+	err = ops->listen(ssock, backlog);
 	sock_put(ssock->sk);
 	return err;
 }
 
+static int mptcp_v4_listen(struct socket *sock, int backlog)
+{
+	int ret;
+
+	ret = mptcp_listen(sock, backlog, AF_INET, &inet_stream_ops);
+
+	return ret;
+}
+
 static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 			       int flags, bool kern)
 {
@@ -1298,12 +1326,12 @@ void mptcp_proto_init(void)
 {
 	mptcp_prot.h.hashinfo = tcp_prot.h.hashinfo;
 	mptcp_stream_ops = inet_stream_ops;
-	mptcp_stream_ops.bind = mptcp_bind;
+	mptcp_stream_ops.bind = mptcp_v4_bind;
 	mptcp_stream_ops.connect = mptcp_stream_connect;
 	mptcp_stream_ops.poll = mptcp_poll;
 	mptcp_stream_ops.accept = mptcp_stream_accept;
-	mptcp_stream_ops.getname = mptcp_getname;
-	mptcp_stream_ops.listen = mptcp_listen;
+	mptcp_stream_ops.getname = mptcp_v4_getname;
+	mptcp_stream_ops.listen = mptcp_v4_listen;
 	mptcp_stream_ops.shutdown = mptcp_shutdown;
 
 	if (percpu_counter_init(&mptcp_sockets_allocated, 0, GFP_KERNEL))
Signature · 1 line
2.17.2
author
#6 lore raw

PATCH 4/6 mptcp: Re-factor subflow routines to be af-agnostic

In reply to Peter Krystad

Re-factor subflow_rebuild_header, init_req, conn_request, syn_recv_sock, and finish_connect to separate af-specific and functional parts.

squash unfortunately is complicated:
subflow_v4_finish_connect: Handle MP_CAPABLE options for outgoing...
subflow_v4_rebuild_header: Add key generation and token tree subflow_v4_conn_request,
subflow_v4_syn_recv_sock,
subflow_v4_init_req: Create SUBFLOW socket for incoming connections

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/subflow.c | 76 +++++++++++++++++++++++++++++++++------------
 1 file changed, 56 insertions(+), 20 deletions(-)
net/mptcp/subflow.c +56 -20
index 13e207458e0b..758b4b36644a 100644
@@ -25,7 +25,7 @@ static inline void SUBFLOW_REQ_INC_STATS(struct request_sock *req,
 	MPTCP_INC_STATS(sock_net(req_to_sk(req)), field);
 }
 
-static struct inet_connection_sock_af_ops subflow_specific;
+static struct inet_connection_sock_af_ops subflow_v4_specific;
 
 static int subflow_rebuild_header(struct sock *sk)
 {
@@ -44,6 +44,14 @@ static int subflow_rebuild_header(struct sock *sk)
 		} while (!subflow->local_nonce);
 	}
 
+	return err;
+}
+
+static int subflow_v4_rebuild_header(struct sock *sk)
+{
+	int err;
+
+	err = subflow_rebuild_header(sk);
 	if (err)
 		return err;
 
@@ -92,19 +100,14 @@ static bool subflow_token_join_request(struct request_sock *req,
 	return true;
 }
 
-static void subflow_v4_init_req(struct request_sock *req,
-				const struct sock *sk_listener,
-				struct sk_buff *skb)
+static void subflow_init_req(struct request_sock *req,
+			     const struct sock *sk_listener,
+			     struct sk_buff *skb)
 {
 	struct mptcp_subflow_context *listener = mptcp_subflow_ctx(sk_listener);
 	struct mptcp_subflow_request_sock *subflow_req = mptcp_subflow_rsk(req);
 	struct tcp_options_received rx_opt;
 
-	tcp_rsk(req)->is_mptcp = 1;
-	pr_debug("subflow_req=%p, listener=%p", subflow_req, listener);
-
-	tcp_request_sock_ipv4_ops.init_req(req, sk_listener, skb);
-
 	memset(&rx_opt.mptcp, 0, sizeof(rx_opt.mptcp));
 	mptcp_get_options(skb, &rx_opt);
 
@@ -148,6 +151,17 @@ static void subflow_v4_init_req(struct request_sock *req,
 	}
 }
 
+static void subflow_v4_init_req(struct request_sock *req,
+				const struct sock *sk_listener,
+				struct sk_buff *skb)
+{
+	tcp_rsk(req)->is_mptcp = 1;
+
+	tcp_request_sock_ipv4_ops.init_req(req, sk_listener, skb);
+
+	subflow_init_req(req, sk_listener, skb);
+}
+
 /* validate received truncated hmac and create hmac for third ACK */
 static bool subflow_thmac_valid(struct mptcp_subflow_context *subflow)
 {
@@ -171,8 +185,6 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
 
-	ipv4_specific.sk_rx_dst_set(sk, skb);
-
 	if (!subflow->conn)
 		return;
 
@@ -207,10 +219,18 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 	}
 }
 
+static void subflow_v4_finish_connect(struct sock *sk,
+				      const struct sk_buff *skb)
+{
+	ipv4_specific.sk_rx_dst_set(sk, skb);
+
+	subflow_finish_connect(sk, skb);
+}
+
 static struct request_sock_ops subflow_request_sock_ops;
 static struct tcp_request_sock_ops subflow_request_sock_ipv4_ops;
 
-static int subflow_conn_request(struct sock *sk, struct sk_buff *skb)
+static int subflow_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
 
@@ -259,7 +279,8 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 					  struct request_sock *req,
 					  struct dst_entry *dst,
 					  struct request_sock *req_unhash,
-					  bool *own_req)
+					  bool *own_req,
+					  const struct inet_connection_sock_af_ops *af_ops)
 {
 	struct mptcp_subflow_context *listener = mptcp_subflow_ctx(sk);
 	struct mptcp_subflow_request_sock *subflow_req;
@@ -278,7 +299,7 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 			return NULL;
 	}
 
-	child = ipv4_specific.syn_recv_sock(sk, skb, req, dst, req_unhash,
+	child = af_ops->syn_recv_sock(sk, skb, req, dst, req_unhash,
 					    own_req);
 
 	if (child && *own_req) {
@@ -312,6 +333,21 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 	return NULL;
 }
 
+static struct sock *subflow_v4_syn_recv_sock(const struct sock *sk,
+					     struct sk_buff *skb,
+					     struct request_sock *req,
+					     struct dst_entry *dst,
+					     struct request_sock *req_unhash,
+					     bool *own_req)
+{
+	struct sock *child;
+
+	child = subflow_syn_recv_sock(sk, skb, req, dst, req_unhash, own_req,
+				      &ipv4_specific);
+
+	return child;
+}
+
 enum mapping_status {
 	MAPPING_OK,
 	MAPPING_INVALID,
@@ -770,7 +806,7 @@ static int subflow_ulp_init(struct sock *sk)
 	pr_debug("subflow=%p", ctx);
 
 	tp->is_mptcp = 1;
-	icsk->icsk_af_ops = &subflow_specific;
+	icsk->icsk_af_ops = &subflow_v4_specific;
 	ctx->tcp_sk_data_ready = sk->sk_data_ready;
 	sk->sk_data_ready = subflow_data_ready;
 	sk->sk_write_space = subflow_write_space;
@@ -862,11 +898,11 @@ void mptcp_subflow_init(void)
 	subflow_request_sock_ipv4_ops = tcp_request_sock_ipv4_ops;
 	subflow_request_sock_ipv4_ops.init_req = subflow_v4_init_req;
 
-	subflow_specific = ipv4_specific;
-	subflow_specific.conn_request = subflow_conn_request;
-	subflow_specific.syn_recv_sock = subflow_syn_recv_sock;
-	subflow_specific.sk_rx_dst_set = subflow_finish_connect;
-	subflow_specific.rebuild_header = subflow_rebuild_header;
+	subflow_v4_specific = ipv4_specific;
+	subflow_v4_specific.conn_request = subflow_v4_conn_request;
+	subflow_v4_specific.syn_recv_sock = subflow_v4_syn_recv_sock;
+	subflow_v4_specific.sk_rx_dst_set = subflow_v4_finish_connect;
+	subflow_v4_specific.rebuild_header = subflow_v4_rebuild_header;
 
 	mptcp_diag_subflow_init(&subflow_ulp_ops);
 
Signature · 1 line
2.17.2
author
#7 lore raw

PATCH 5/6 mptcp: Export low-level routines for IPv6

In reply to Peter Krystad

ipv6_specific and tcp_request_sock_ipv6_ops will be referenced during TCP subflow creation.

squashto: tcp: Export TCP functions and ops

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 include/net/tcp.h    | 3 +++
 net/ipv6/tcp_ipv6.c  | 9 +++++----
 net/mptcp/protocol.h | 3 +++
 3 files changed, 11 insertions(+), 4 deletions(-)
include/net/tcp.h +3 -0
index f1a0bdb7d1b1..74840bb2a4f3 100644
@@ -1990,6 +1990,9 @@ struct tcp_request_sock_ops {
 };
 
 extern const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops;
+#if IS_ENABLED(CONFIG_IPV6)
+extern const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops;
+#endif
 
 #ifdef CONFIG_SYN_COOKIES
 static inline __u32 cookie_init_sequence(const struct tcp_request_sock_ops *ops,
net/ipv6/tcp_ipv6.c +5 -4
index 4804b6dc5e65..50f6f21bbdb1 100644
@@ -75,7 +75,7 @@ static void	tcp_v6_reqsk_send_ack(const struct sock *sk, struct sk_buff *skb,
 static int	tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb);
 
 static const struct inet_connection_sock_af_ops ipv6_mapped;
-static const struct inet_connection_sock_af_ops ipv6_specific;
+const struct inet_connection_sock_af_ops ipv6_specific;
 #ifdef CONFIG_TCP_MD5SIG
 static const struct tcp_sock_af_ops tcp_sock_ipv6_specific;
 static const struct tcp_sock_af_ops tcp_sock_ipv6_mapped_specific;
@@ -785,7 +785,7 @@ struct request_sock_ops tcp6_request_sock_ops __read_mostly = {
 	.syn_ack_timeout =	tcp_syn_ack_timeout,
 };
 
-static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
+const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
 	.mss_clamp	=	IPV6_MIN_MTU - sizeof(struct tcphdr) -
 				sizeof(struct ipv6hdr),
 #ifdef CONFIG_TCP_MD5SIG
@@ -1111,7 +1111,8 @@ static void tcp_v6_restore_cb(struct sk_buff *skb)
 		sizeof(struct inet6_skb_parm));
 }
 
-static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
+static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk,
+					 struct sk_buff *skb,
 					 struct request_sock *req,
 					 struct dst_entry *dst,
 					 struct request_sock *req_unhash,
@@ -1739,7 +1740,7 @@ static struct timewait_sock_ops tcp6_timewait_sock_ops = {
 	.twsk_destructor = tcp_twsk_destructor,
 };
 
-static const struct inet_connection_sock_af_ops ipv6_specific = {
+const struct inet_connection_sock_af_ops ipv6_specific = {
 	.queue_xmit	   = inet6_csk_xmit,
 	.send_check	   = tcp_v6_send_check,
 	.rebuild_header	   = inet6_sk_rebuild_header,
net/mptcp/protocol.h +3 -0
index 2635d68e58a0..71967bac03c0 100644
@@ -270,6 +270,9 @@ int mptcp_subflow_create_socket(struct sock *sk, sa_family_t family,
 				struct socket **new_sock);
 
 extern const struct inet_connection_sock_af_ops ipv4_specific;
+#if IS_ENABLED(CONFIG_IPV6)
+extern const struct inet_connection_sock_af_ops ipv6_specific;
+#endif
 
 void mptcp_proto_init(void);
 
Signature · 1 line
2.17.2
#8 lore raw

[MPTCP] Re: [PATCH 5/6] mptcp: Export low-level routines for IPv6

Hi,

net/ipv6/tcp_ipv6.c static void tcp_v6_restore_cb(struct sk_buff *skb) · line 1118

Quoted 10 lines · Peter Krystad
net/ipv6/tcp_ipv6.c +2 -1
@@ -1111,7 +1111,8 @@ static void tcp_v6_restore_cb(struct sk_buff *skb)
 		sizeof(struct inet6_skb_parm));
 }
 
-static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
+static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk,
+					 struct sk_buff *skb,
 					 struct request_sock *req,
 					 struct dst_entry *dst,
 					 struct request_sock *req_unhash,

Minor nit: it looks like this chunk is not needed ?

Cheers,

Paolo

author
#9 lore raw

[MPTCP] Re: [PATCH 5/6] mptcp: Export low-level routines for IPv6

Quoted 7 lines · Paolo Abeni

Hi,

Quoted 4 lines · Peter Krystad
net/ipv6/tcp_ipv6.c +0 -0
@@ -1111,7 +1111,8 @@ static void tcp_v6_restore_cb(struct sk_buff *skb)
 		sizeof(struct inet6_skb_parm));
 }
 
Quoted 8 lines

net/ipv6/tcp_ipv6.c static void tcp_v6_restore_cb(struct sk_buff *skb) · line 1118

Quoted 6 lines
net/ipv6/tcp_ipv6.c +2 -1
-static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
+static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk,
+					 struct sk_buff *skb,
 					 struct request_sock *req,
 					 struct dst_entry *dst,
 					 struct request_sock *req_unhash,

Minor nit: it looks like this chunk is not needed ?

It seemed like since I was in the neighborhood I would fix this over-80 char line. If that's not appropriate we can drop this chunk.

Peter.

Quoted 4 lines

Cheers,

Paolo

author
#10 lore raw

PATCH 6/6 mptcp: Add IPv6 support

In reply to Peter Krystad

Mimic hooks and option handling for IPv6 as was done for IPv4 MPTCP sockets.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 include/net/mptcp.h  |   3 ++
 net/ipv6/tcp_ipv6.c  |   7 +++
 net/mptcp/ctrl.c     |  11 +++++
 net/mptcp/protocol.c |  85 ++++++++++++++++++++++++++++++++--
 net/mptcp/protocol.h |   3 ++
 net/mptcp/subflow.c  | 107 ++++++++++++++++++++++++++++++++++++++++++-
 6 files changed, 209 insertions(+), 7 deletions(-)
include/net/mptcp.h +3 -0
index eba39a881767..41225c50aebb 100644
@@ -53,6 +53,9 @@ struct mptcp_out_options {
 #ifdef CONFIG_MPTCP
 
 void mptcp_init(void);
+#if IS_ENABLED(CONFIG_IPV6)
+int mptcpv6_init(void);
+#endif
 
 static inline bool sk_is_mptcp(const struct sock *sk)
 {
net/ipv6/tcp_ipv6.c +7 -0
index 50f6f21bbdb1..10d1c033d458 100644
@@ -2109,9 +2109,16 @@ int __init tcpv6_init(void)
 	ret = register_pernet_subsys(&tcpv6_net_ops);
 	if (ret)
 		goto out_tcpv6_protosw;
+
+	ret = mptcpv6_init();
+	if (ret)
+		goto out_tcpv6_pernet_subsys;
+
 out:
 	return ret;
 
+out_tcpv6_pernet_subsys:
+	unregister_pernet_subsys(&tcpv6_net_ops);
 out_tcpv6_protosw:
 	inet6_unregister_protosw(&tcpv6_protosw);
 out_tcpv6_protocol:
net/mptcp/ctrl.c +11 -0
index 6023c9f722ea..a8c5652e0203 100644
@@ -117,3 +117,14 @@ void __init mptcp_init(void)
 	if (register_pernet_subsys(&mptcp_pernet_ops) < 0)
 		panic("Failed to register MPTCP pernet subsystem.\n");
 }
+
+#if IS_ENABLED(CONFIG_IPV6)
+int __init mptcpv6_init(void)
+{
+	int err;
+
+	err = mptcp_proto_v6_init();
+
+	return err;
+}
+#endif
net/mptcp/protocol.c +80 -5

118 lines not on this page · raw message

net/mptcp/protocol.h +3 -0
index 71967bac03c0..0951f6ec9954 100644
@@ -275,6 +275,9 @@ extern const struct inet_connection_sock_af_ops ipv6_specific;
 #endif
 
 void mptcp_proto_init(void);
+#if IS_ENABLED(CONFIG_IPV6)
+int mptcp_proto_v6_init(void);
+#endif
 
 struct mptcp_read_arg {
 	struct msghdr *msg;
net/mptcp/subflow.c +105 -2

163 lines not on this page · raw message

Signature · 1 line
2.17.2
#11 lore raw

[MPTCP] Re: [PATCH 6/6] mptcp: Add IPv6 support

hello Peter!

Quoted 18 lines · Peter Krystad

Mimic hooks and option handling for IPv6 as was done for IPv4 MPTCP sockets.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 include/net/mptcp.h  |   3 ++
 net/ipv6/tcp_ipv6.c  |   7 +++
 net/mptcp/ctrl.c     |  11 +++++
 net/mptcp/protocol.c |  85 ++++++++++++++++++++++++++++++++--
 net/mptcp/protocol.h |   3 ++
 net/mptcp/subflow.c  | 107 ++++++++++++++++++++++++++++++++++++++++++-
 6 files changed, 209 insertions(+), 7 deletions(-)
include/net/mptcp.h +0 -0
index eba39a881767..41225c50aebb 100644

include/net/mptcp.h struct mptcp_out_options { · line 58

Quoted 6 lines
include/net/mptcp.h +3 -0
@@ -53,6 +53,9 @@ struct mptcp_out_options {
 #ifdef CONFIG_MPTCP
 
 void mptcp_init(void);
+#if IS_ENABLED(CONFIG_IPV6)
+int mptcpv6_init(void);
+#endif

I didn't see a stub version of mptcpv6_init(),

Quoted 7 lines
include/net/mptcp.h +0 -0
 static inline bool sk_is_mptcp(const struct sock *sk)
 {
net/ipv6/tcp_ipv6.c +0 -0
index 50f6f21bbdb1..10d1c033d458 100644

net/ipv6/tcp_ipv6.c int __init tcpv6_init(void) · line 2118

Quoted 10 lines
net/ipv6/tcp_ipv6.c +5 -0
@@ -2109,9 +2109,16 @@ int __init tcpv6_init(void)
 	ret = register_pernet_subsys(&tcpv6_net_ops);
 	if (ret)
 		goto out_tcpv6_protosw;
+
+	ret = mptcpv6_init();
+	if (ret)
+		goto out_tcpv6_pernet_subsys;
+
 out:
 	return ret;

so probably the above hunk will cause build problems when CONFIG_MPTCP is unset. Am I missing something?

thanks!

Signature · 1 line
davide
author
#12 lore raw

[MPTCP] Re: [PATCH 6/6] mptcp: Add IPv6 support

Quoted 42 lines · Davide Caratti

hello Peter!

Quoted 18 lines · Peter Krystad

Mimic hooks and option handling for IPv6 as was done for IPv4 MPTCP sockets.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 include/net/mptcp.h  |   3 ++
 net/ipv6/tcp_ipv6.c  |   7 +++
 net/mptcp/ctrl.c     |  11 +++++
 net/mptcp/protocol.c |  85 ++++++++++++++++++++++++++++++++--
 net/mptcp/protocol.h |   3 ++
 net/mptcp/subflow.c  | 107 ++++++++++++++++++++++++++++++++++++++++++-
 6 files changed, 209 insertions(+), 7 deletions(-)
include/net/mptcp.h +0 -0
index eba39a881767..41225c50aebb 100644

include/net/mptcp.h struct mptcp_out_options { · line 58

Quoted 6 lines
include/net/mptcp.h +3 -0
@@ -53,6 +53,9 @@ struct mptcp_out_options {
 #ifdef CONFIG_MPTCP
 
 void mptcp_init(void);
+#if IS_ENABLED(CONFIG_IPV6)
+int mptcpv6_init(void);
+#endif

I didn't see a stub version of mptcpv6_init(),

Quoted 12 lines
include/net/mptcp.h +0 -0
 static inline bool sk_is_mptcp(const struct sock *sk)
 {
net/ipv6/tcp_ipv6.c +2 -0
index 50f6f21bbdb1..10d1c033d458 100644
@@ -2109,9 +2109,16 @@ int __init tcpv6_init(void)
 	ret = register_pernet_subsys(&tcpv6_net_ops);
 	if (ret)
 		goto out_tcpv6_protosw;
+
+	ret = mptcpv6_init();
Quoted 8 lines

net/ipv6/tcp_ipv6.c int __init tcpv6_init(void) · line 2118

Quoted 5 lines
net/ipv6/tcp_ipv6.c +3 -0
+	if (ret)
+		goto out_tcpv6_pernet_subsys;
+
 out:
 	return ret;

so probably the above hunk will cause build problems when CONFIG_MPTCP is unset. Am I missing something?

Nope, you are right. I will fix this in v2.

Peter.

Quoted 2 lines

thanks!

#13 lore raw
In reply to Peter Krystad
Quoted 1 line · Peter Krystad

Implemented support for IPv6. selftests pass with ipv6=true,

Nice!

Quoted 1 line

but they take twice as long to run...

Yep, that is expected, right? ;)

Quoted 3 lines

With re-factoring the addition of IPv6 is very clean but I did not split files yet as it will be hard to squash any changes after the such a split.

We still need the Kconfig boilerplate, right? something alike:

---

net/mptcp/Kconfig +0 -0
index f21190a4f7e9..3fa6d91f57ee 100644
@@ -9,3 +9,8 @@ config MPTCP
          uses the TCP protocol, and TCP options carry header

information for
MPTCP.

+config MPTCP_IPV6

+       bool "IPv6 support form Multipath TCP"
+       depends on MPTCP
+       select IPV6
+       default y

---

and than replace IS_ENABLED(IPV6) with CONFIG_MPTCP_IPV6.

Additinal question: what happens if ipv6 is disabled at boot time (ipv6.disable=1 on the kernel command line)?

Thanks,

Paolo

author
#14 lore raw

Hi Paolo -

Quoted 8 lines · Paolo Abeni
Quoted 1 line · Peter Krystad

Implemented support for IPv6. selftests pass with ipv6=true,

Nice!

Quoted 1 line

but they take twice as long to run...

Yep, that is expected, right? ;)

Right.

Quoted 24 lines
Quoted 4 lines

With re-factoring the addition of IPv6 is very clean but I did not split files yet as it will be hard to squash any changes after the such a split.

We still need the Kconfig boilerplate, right? something alike:

---

net/mptcp/Kconfig +0 -0
index f21190a4f7e9..3fa6d91f57ee 100644
@@ -9,3 +9,8 @@ config MPTCP
          uses the TCP protocol, and TCP options carry header

information for
MPTCP.

net/mptcp/Kconfig +5 -1
+config MPTCP_IPV6
+       bool "IPv6 support form Multipath TCP"
+       depends on MPTCP
+       select IPV6
+       default y
---
Quoted 1 line

and than replace IS_ENABLED(IPV6) with CONFIG_MPTCP_IPV6.

Indeed we do, I will submit a patchset that does that will need to be merged before the v2 IPv6 patchset (coming tomorrow).

Quoted 2 lines

Additinal question: what happens if ipv6 is disabled at boot time (ipv6.disable=1 on the kernel command line)?

I believe the initialization in inet6_init() will not run and it will not be possible to create a socket in the AF_INET6 domain.

Peter.

Quoted 4 lines

Thanks,

Paolo