Skip to contentSkip to conversation

Implement outgoing MP_JOIN

PATCH In review Peter Krystad 4 participants 28 messages Aug 7, 2019 – Aug 23, 2019

Subsystems (3)
lore
Share & tools
lore

Peter Krystad

author
#0 lore raw

v3 cover Implement outgoing MP_JOIN

This is several accumulated fixes that are required for MP_JOIN (secondary subflows) plus the outgoing code.

Both outgoing and incoming MP_JOIN now inter-operate with the multipath.org implementation although there is still a deadlock when incoming is fully established. Since the incoming direction did not work at all before this is an improvement. I have not yet tested running our implmentation against itself.

The final patch in the set is an "interim path manager", this is throw-away code that I include for testing purposes. For an incoming connection it is necessary for the server to advertise a secondary address and for my testing I have been hard-coding the IP address of my secondary interface when I build the kernel. Obviously not the final solution but we need to discuss what approach to use for the self tests. I also think we may not want to publish this patch on our net-next branch.

v2:
- move mptcp_options_received formatting to options.c
- resolve conflicts with Davide's crypto commit
- make helper for setting request fields
- fix locking in token_lookup()
- move token_lookup() to "Implement path manager interface commands:
(where it belongs)
- split interim path manager out of final commit, will submit seperately
- remove use of local_id in pm_create_subflow
- restrict announcing addresses to server side only
- add one more commit with get_local_id fix for incoming MP_JOIN

v3:
- Revise nested if logic in options.c
- set subflow->request_cksum correctly (do not support)
- lock mptcp_sock in subflow_connect() and take ref
- return EINPROGRESS from subflow_connect() in normal case
- cleanup socket when bind()/connect() fails in subflow_connect()
- take reference to mptcp_sock in token_join_request() and token_join_valid()
- rename token_lookup() to token_lookup_get()
- track remote token seperately in subflow (previously local token field was being overwritten, preventing cleanup)
- take long-lived reference to mptcp_sock in token_new_subflow

Peter Krystad (10):
Extend path manager interface
Fix locking in mptcp_finish_join.
Fix parsing MP_JOIN third ACK.
Re-factor and fixes for crypto_hmac_sha1()
Rename mptcp_attach_dss()
Fix return value if no DSS option is created.
Move setting request_ fields of subflow to MPTCP layer helper
mptcp: Add handling of outgoing MP_JOIN requests Use pm_get_local_id() to get the real local address id.
mptcp: Implement path manager interface commands

 include/net/mptcp.h  |  12 ++--
 net/ipv4/tcp_input.c |   4 +-
 net/mptcp/crypto.c   |  23 +++----
 net/mptcp/options.c  |  98 +++++++++++++++++++++++++++---
 net/mptcp/pm.c       | 140 +++++++++++++++++++++++++++++++++++++++----
 net/mptcp/protocol.c |  21 +++++--
 net/mptcp/protocol.h |  58 ++++++++++++++----
 net/mptcp/subflow.c  |  77 +++++++++++++++++++++---
 net/mptcp/token.c    | 111 +++++++++++++++++++++++++++-------
 9 files changed, 455 insertions(+), 89 deletions(-)
Signature · 1 line
2.17.2
author
#1 lore raw

PATCH v3 1/10 Extend path manager interface

A little more functionality and some comments clarifying which routines are for a path manager to call and which are callbacks.

squashto: Add path manager interface

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/options.c  | 40 +++++++++++++++++++--
 net/mptcp/pm.c       | 84 +++++++++++++++++++++++++++++++++++++-------
 net/mptcp/protocol.c |  5 +--
 net/mptcp/protocol.h | 38 +++++++++++++++-----
 4 files changed, 140 insertions(+), 27 deletions(-)
net/mptcp/options.c +37 -3
index 58215f19829a..ab3fce071c17 100644
@@ -411,11 +411,29 @@ static bool mptcp_established_options_addr(struct sock *sk,
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
 	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
+	struct sockaddr_storage saddr;
+	u8 id;
 
-	if (subflow->fourth_ack)
-		return pm_addr_signal(msk, size, remaining, opts);
+	if (!msk)
+		return false;
 
-	return false;
+	if (!msk->pm.fully_established || !msk->addr_signal)
+		return false;
+
+	if (pm_addr_signal(msk, &id, &saddr))
+		return false;
+
+	if (saddr.ss_family == AF_INET && remaining < TCPOLEN_MPTCP_ADD_ADDR)
+		return false;
+
+	opts->suboptions |= OPTION_MPTCP_ADD_ADDR;
+	opts->addr_id = id;
+	opts->addr.s_addr = ((struct sockaddr_in *)&saddr)->sin_addr.s_addr;
+	*size = TCPOLEN_MPTCP_ADD_ADDR;
+
+	msk->addr_signal = 0;
+
+	return true;
 }
 
 bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
@@ -479,11 +497,23 @@ bool mptcp_synack_options(const struct request_sock *req, unsigned int *size,
 void mptcp_attach_dss(struct sock *sk, struct sk_buff *skb,
 		      struct tcp_options_received *opt_rx)
 {
+	struct subflow_context *subflow = subflow_ctx(sk);
+	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
 	struct mptcp_options_received *mp_opt;
 	struct mptcp_ext *mpext;
 
 	mp_opt = &opt_rx->mptcp;
 
+	if (msk && mp_opt->add_addr) {
+		if (mp_opt->family == MPTCP_ADDR_IPVERSION_4)
+			pm_add_addr(msk, &mp_opt->addr, mp_opt->addr_id);
+#if IS_ENABLED(CONFIG_IPV6)
+		else if (mp_opt->family == MPTCP_ADDR_IPVERSION_6)
+			pm_add_addr6(msk, &mp_opt->addr6, mp_opt->addr_id);
+#endif
+		mp_opt->add_addr = 0;
+	}
+
 	if (!mp_opt->dss)
 		return;
 
@@ -510,6 +540,10 @@ void mptcp_attach_dss(struct sock *sk, struct sk_buff *skb,
 	}
 
 	mpext->data_fin = mp_opt->data_fin;
+
+	if (msk)
+		pm_fully_established(msk);
+
 }
 
 void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
net/mptcp/pm.c +71 -13

111 lines not on this page · raw message

net/mptcp/protocol.c +3 -2
index 4d2844bff36f..5501c405c6e0 100644
@@ -663,7 +663,7 @@ static struct sock *mptcp_accept(struct sock *sk, int flags, int *err,
 		token_update_accept(new_sock->sk, new_mptcp_sock);
 		msk->subflow = NULL;
 
-		pm_new_connection(msk);
+		pm_new_connection(msk, 1);
 
 		crypto_key_sha1(msk->remote_key, NULL, &ack_seq);
 		msk->write_seq = subflow->idsn + 1;
@@ -798,7 +798,7 @@ void mptcp_finish_connect(struct sock *sk, int mp_capable)
 		msk->token = subflow->token;
 		pr_debug("msk=%p, token=%u", msk, msk->token);
 
-		pm_new_connection(msk);
+		pm_new_connection(msk, 0);
 
 		crypto_key_sha1(msk->remote_key, NULL, &ack_seq);
 		msk->write_seq = subflow->idsn + 1;
@@ -1090,6 +1090,7 @@ void __init mptcp_init(void)
 
 	token_init();
 	subflow_init();
+	pm_init();
 
 	if (proto_register(&mptcp_prot, 1) != 0)
 		panic("Failed to register MPTCP proto.\n");
net/mptcp/protocol.h +29 -9
index 4c098acf4170..71c86e31e19c 100644
@@ -78,15 +78,32 @@ static inline u32 mptcp_option(u8 subopt, u8 len, u8 nib, u8 field)
 		     ((nib & 0xF) << 8) | field);
 }
 
-struct pm_data {
-	u8 addr_id;
-	sa_family_t family;
+struct mptcp_pm_data {
+	u8	local_valid;
+	u8	local_id;
+	sa_family_t local_family;
+	union {
+		struct in_addr local_addr;
+#if IS_ENABLED(CONFIG_IPV6)
+		struct in6_addr local_addr6;
+#endif
+	};
+	u8	remote_valid;
+	u8	remote_id;
+	sa_family_t remote_family;
 	union {
-		struct in_addr addr;
+		struct in_addr remote_addr;
 #if IS_ENABLED(CONFIG_IPV6)
-		struct in6_addr addr6;
+		struct in6_addr remote_addr6;
 #endif
 	};
+	u8	server_side : 1,
+		fully_established : 1;
+
+	/* for interim path manager */
+	struct	work_struct addr_work;
+	struct	work_struct subflow_work;
+	u32	token;
 };
 
 /* MPTCP connection sock */
@@ -100,7 +117,7 @@ struct mptcp_sock {
 	u32		token;
 	struct list_head conn_list;
 	struct socket	*subflow; /* outgoing connect/listener/!mp_capable */
-	struct pm_data	pm;
+	struct mptcp_pm_data	pm;
 	u8		addr_signal;
 };
 
@@ -223,7 +240,8 @@ static inline void crypto_key_gen_sha1(u64 *key, u32 *token, u64 *idsn)
 	crypto_key_sha1(*key, token, idsn);
 }
 
-void pm_new_connection(struct mptcp_sock *msk);
+void pm_init(void);
+void pm_new_connection(struct mptcp_sock *msk, int server_side);
 void pm_fully_established(struct mptcp_sock *msk);
 void pm_connection_closed(struct mptcp_sock *msk);
 void pm_subflow_established(struct mptcp_sock *msk, u8 id);
@@ -231,8 +249,10 @@ void pm_subflow_closed(struct mptcp_sock *msk, u8 id);
 void pm_add_addr(struct mptcp_sock *msk, const struct in_addr *addr, u8 id);
 void pm_add_addr6(struct mptcp_sock *msk, const struct in6_addr *addr, u8 id);
 void pm_rm_addr(struct mptcp_sock *msk, u8 id);
-bool pm_addr_signal(struct mptcp_sock *msk, unsigned int *size,
-		    unsigned int remaining, struct mptcp_out_options *opts);
+int pm_addr_signal(struct mptcp_sock *msk, u8 *id,
+		   struct sockaddr_storage *saddr);
+int pm_get_local_id(struct request_sock *req, struct sock *sk,
+		    const struct sk_buff *skb);
 
 static inline struct mptcp_ext *mptcp_get_ext(struct sk_buff *skb)
 {
Signature · 1 line
2.17.2
author
#5 lore raw

PATCH v3 2/10 Fix locking in mptcp_finish_join.

In reply to Peter Krystad

Lock was being taken on the subflow sock, not mptcp_sock.

squashto: Add handling of incoming MP_JOIN requests

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/protocol.c | 9 +++++----
 net/mptcp/protocol.h | 2 +-
 net/mptcp/subflow.c  | 2 +-
 3 files changed, 7 insertions(+), 6 deletions(-)
net/mptcp/protocol.c +5 -4
index 5501c405c6e0..8a841d489b52 100644
@@ -816,18 +816,19 @@ void mptcp_finish_connect(struct sock *sk, int mp_capable)
 	inet_sk_state_store(sk, TCP_ESTABLISHED);
 }
 
-void mptcp_finish_join(struct sock *conn, struct sock *sk)
+void mptcp_finish_join(struct sock *sk)
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
-	struct mptcp_sock *msk = mptcp_sk(conn);
+	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
 
 	pr_debug("msk=%p, subflow=%p", msk, subflow);
 
 	local_bh_disable();
-	bh_lock_sock_nested(sk);
+	bh_lock_sock_nested(subflow->conn);
 	list_add_tail(&subflow->node, &msk->conn_list);
-	bh_unlock_sock(sk);
+	bh_unlock_sock(subflow->conn);
 	local_bh_enable();
+	inet_sk_state_store(sk, TCP_ESTABLISHED);
 }
 
 bool mptcp_sk_is_subflow(const struct sock *sk)
net/mptcp/protocol.h +1 -1
index 71c86e31e19c..dedd3524e4a2 100644
@@ -210,7 +210,7 @@ void mptcp_get_options(const struct sk_buff *skb,
 		       struct tcp_options_received *opt_rx);
 
 void mptcp_finish_connect(struct sock *sk, int mp_capable);
-void mptcp_finish_join(struct sock *conn, struct sock *sk);
+void mptcp_finish_join(struct sock *sk);
 
 void token_init(void);
 void token_new_request(struct request_sock *req, const struct sk_buff *skb);
net/mptcp/subflow.c +1 -1
index d91d817b8779..9d91014c0d1b 100644
@@ -165,7 +165,7 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 			if (token_new_join(child))
 				goto close_child;
 			else
-				mptcp_finish_join(ctx->conn, child);
+				mptcp_finish_join(child);
 		}
 	}
 
Signature · 1 line
2.17.2
author
#7 lore raw

PATCH v3 3/10 Fix parsing MP_JOIN third ACK.

In reply to Peter Krystad

There is a reserved byte of zero's in there.

squashto: Add handling of incoming MP_JOIN requests

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/options.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/mptcp/options.c +1 -1
index ab3fce071c17..a1920574161d 100644
@@ -92,7 +92,7 @@ void mptcp_parse_option(const unsigned char *ptr, int opsize,
 				 mp_opt->backup, mp_opt->join_id,
 				 mp_opt->thmac, mp_opt->nonce);
 		} else if (opsize == TCPOLEN_MPTCP_MPJ_ACK) {
-			ptr++;
+			ptr += 2;
 			memcpy(mp_opt->hmac, ptr, MPTCPOPT_HMAC_LEN);
 			pr_debug("MP_JOIN hmac");
 		} else {
Signature · 1 line
2.17.2
author
#9 lore raw

PATCH v3 4/10 Re-factor and fixes for crypto_hmac_sha1()

In reply to Peter Krystad

Re-factor to use parameters specific to MPTCP use case similar to crypto_key_sha1() and get rid of var args. Also fix endianness issues.

squash to: Add key generation and token tree
and: Add handling of incoming MP_JOIN requests

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/crypto.c   | 23 +++++++++--------------
 net/mptcp/protocol.h |  4 ++--
 net/mptcp/token.c    | 19 ++++++++-----------
 3 files changed, 19 insertions(+), 27 deletions(-)
net/mptcp/crypto.c +9 -14
index 5b764798c536..8b67b11e626c 100644
@@ -51,23 +51,23 @@ void crypto_key_sha1(u64 key, u32 *token, u64 *idsn)
 		*idsn = ((u64)mptcp_hashed_key[3] << 32) + mptcp_hashed_key[4];
 }
 
-void crypto_hmac_sha1(u64 key1, u64 key2, u32 *hash_out,
-		      int arg_num, ...)
+void crypto_hmac_sha1(u64 key1, u64 key2, u32 nonce1, u32 nonce2, u32 *hash_out)
 {
 	u32 workspace[SHA_WORKSPACE_WORDS];
 	u8 input[128]; /* 2 512-bit blocks */
 	int i;
 	int index;
-	int length;
-	u8 *msg;
-	va_list list;
 	u8 key_1[8];
 	u8 key_2[8];
+	u8 nonce_1[4];
+	u8 nonce_2[4];
 
 	memset(workspace, 0, sizeof(workspace));
 
 	put_unaligned_be64(key1, key_1);
 	put_unaligned_be64(key2, key_2);
+	put_unaligned_be32(nonce1, nonce_1);
+	put_unaligned_be32(nonce2, nonce_2);
 
 	/* Generate key xored with ipad */
 	memset(input, 0x36, 64);
@@ -76,16 +76,11 @@ void crypto_hmac_sha1(u64 key1, u64 key2, u32 *hash_out,
 	for (i = 0; i < 8; i++)
 		input[i + 8] ^= key_2[i];
 
-	va_start(list, arg_num);
 	index = 64;
-	for (i = 0; i < arg_num; i++) {
-		length = va_arg(list, int);
-		msg = va_arg(list, u8 *);
-		WARN_ON(index + length > 125); /* Message is too long */
-		memcpy(&input[index], msg, length);
-		index += length;
-	}
-	va_end(list);
+	memcpy(&input[index], nonce_1, 4);
+	index = 68;
+	memcpy(&input[index], nonce_2, 4);
+	index = 72;
 
 	input[index] = 0x80; /* Padding: First bit after message = 1 */
 	memset(&input[index + 1], 0, (126 - index));
net/mptcp/protocol.h +2 -2
index dedd3524e4a2..1d1e89109e98 100644
@@ -226,8 +226,6 @@ void token_release(u32 token);
 void token_destroy(u32 token);
 
 void crypto_key_sha1(u64 key, u32 *token, u64 *idsn);
-void crypto_hmac_sha1(u64 key1, u64 key2, u32 *hash_out, int arg_num, ...);
-
 static inline void crypto_key_gen_sha1(u64 *key, u32 *token, u64 *idsn)
 {
 	/* we might consider a faster version that computes the key as a
@@ -239,6 +237,8 @@ static inline void crypto_key_gen_sha1(u64 *key, u32 *token, u64 *idsn)
 	get_random_bytes(key, sizeof(u64));
 	crypto_key_sha1(*key, token, idsn);
 }
+void crypto_hmac_sha1(u64 key1, u64 key2, u32 nonce1, u32 nonce2,
+		      u32 *hash_out);
 
 void pm_init(void);
 void pm_new_connection(struct mptcp_sock *msk, int server_side);
net/mptcp/token.c +8 -11
index e5e0c9689e3a..ef03ef19af98 100644
@@ -84,12 +84,11 @@ static void new_req_join(struct request_sock *req, struct sock *sk,
 	u8 hmac[MPTCPOPT_HMAC_LEN];
 
 	get_random_bytes(&subflow_req->local_nonce, sizeof(u32));
-	crypto_hmac_sha1(msk->local_key,
-			 msk->remote_key,
-			 (u32 *)hmac, 2,
-			 4, (u8 *)&subflow_req->local_nonce,
-			 4, (u8 *)&subflow_req->remote_nonce);
-	subflow_req->thmac = *(u64 *)hmac;
+	crypto_hmac_sha1(msk->local_key, msk->remote_key,
+			 subflow_req->local_nonce, subflow_req->remote_nonce,
+			 (u32 *)hmac);
+
+	subflow_req->thmac = get_unaligned_be64(hmac);
 	pr_debug("local_nonce=%u, thmac=%llu", subflow_req->local_nonce,
 		 subflow_req->thmac);
 }
@@ -101,11 +100,9 @@ static int new_join_valid(struct request_sock *req, struct sock *sk,
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	u8 hmac[MPTCPOPT_HMAC_LEN];
 
-	crypto_hmac_sha1(msk->remote_key,
-			 msk->local_key,
-			 (u32 *)hmac, 2,
-			 4, (u8 *)&subflow_req->remote_nonce,
-			 4, (u8 *)&subflow_req->local_nonce);
+	crypto_hmac_sha1(msk->remote_key, msk->local_key,
+			 subflow_req->remote_nonce, subflow_req->local_nonce,
+			 (u32 *)hmac);
 
 	return memcmp(hmac, (char *)rx_opt->mptcp.hmac, MPTCPOPT_HMAC_LEN);
 }
Signature · 1 line
2.17.2
author
#11 lore raw

PATCH v3 5/10 Rename mptcp_attach_dss()

In reply to Peter Krystad

Other incoming MPTCP options will be handled in this routine.

squashto: Implement MPTCP receive path

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 include/net/mptcp.h  | 10 +++++-----
 net/ipv4/tcp_input.c |  4 ++--
 net/mptcp/options.c  |  4 ++--
 3 files changed, 9 insertions(+), 9 deletions(-)
include/net/mptcp.h +5 -5
index 6cd650838ddf..bb2dd193c0c5 100644
@@ -68,9 +68,8 @@ bool mptcp_synack_options(const struct request_sock *req, unsigned int *size,
 bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
 			       unsigned int *size, unsigned int remaining,
 			       struct mptcp_out_options *opts);
-
-void mptcp_attach_dss(struct sock *sk, struct sk_buff *skb,
-		      struct tcp_options_received *opt_rx);
+void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb,
+			    struct tcp_options_received *opt_rx);
 
 static inline bool mptcp_skb_ext_exist(const struct sk_buff *skb)
 {
@@ -128,8 +127,9 @@ static inline bool mptcp_established_options(struct sock *sk,
 	return false;
 }
 
-static inline void mptcp_attach_dss(struct sock *sk, struct sk_buff *skb,
-				    struct tcp_options_received *opt_rx)
+static inline void mptcp_incoming_options(struct sock *sk,
+					  struct sk_buff *skb,
+					  struct tcp_options_received *opt_rx)
 {
 }
 
net/ipv4/tcp_input.c +2 -2
index b92985b4762c..ca2fea882281 100644
@@ -5668,9 +5668,9 @@ void tcp_rcv_established(struct sock *sk, struct sk_buff *skb)
 	/* Process urgent data. */
 	tcp_urg(sk, skb, th);
 
-	/* Prepare MPTCP sequence data */
+	/* Process MPTCP options */
 	if (sk_is_mptcp(sk))
-		mptcp_attach_dss(sk, skb, &tp->rx_opt);
+		mptcp_incoming_options(sk, skb, &tp->rx_opt);
 
 	/* step 7: process the segment text */
 	tcp_data_queue(sk, skb);
net/mptcp/options.c +2 -2
index a1920574161d..1f54c543304e 100644
@@ -494,8 +494,8 @@ bool mptcp_synack_options(const struct request_sock *req, unsigned int *size,
 	return false;
 }
 
-void mptcp_attach_dss(struct sock *sk, struct sk_buff *skb,
-		      struct tcp_options_received *opt_rx)
+void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb,
+			    struct tcp_options_received *opt_rx)
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
 	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
Signature · 1 line
2.17.2
author
#13 lore raw

PATCH v3 6/10 Fix return value if no DSS option is created.

In reply to Peter Krystad

Strictly speaking mptcp_established_options_dss() should return false if no DSS option is created.

squashto: Write MPTCP DSS headers to outgoing data packets

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/options.c | 3 +++
 1 file changed, 3 insertions(+)
net/mptcp/options.c +3 -0
index 1f54c543304e..01f59ad629e5 100644
@@ -400,6 +400,9 @@ static bool mptcp_established_options_dss(struct sock *sk, struct sk_buff *skb,
 		WARN(1, "MPTCP: Ack dropped");
 	}
 
+	if (!dss_size)
+		return false;
+
 	*size = ALIGN(dss_size, 4);
 	return true;
 }
Signature · 1 line
2.17.2
author
#15 lore raw

PATCH v3 7/10 Move setting request_ fields of subflow to MPTCP layer helper

In reply to Peter Krystad

For secondary subflows different request_ fields will be added that will be set differently based on subflow type (initial vs secondary).

squashto: Associate MPTCP context with TCP socket

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/protocol.c | 6 ++++++
 net/mptcp/protocol.h | 2 +-
 net/mptcp/subflow.c  | 7 ++-----
 3 files changed, 9 insertions(+), 6 deletions(-)
net/mptcp/protocol.c +6 -0
index 8a841d489b52..fce5eea729ba 100644
@@ -860,6 +860,7 @@ static struct proto mptcp_prot = {
 static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk)
 {
 	struct sock *sk = (struct sock *)msk;
+	struct subflow_context *subflow;
 	struct socket *ssock;
 	int err;
 
@@ -875,6 +876,11 @@ static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk)
 	}
 
 	msk->subflow = ssock;
+	subflow = subflow_ctx(msk->subflow->sk);
+	subflow->request_mptcp = 1; /* @@ if MPTCP enabled */
+	subflow->request_cksum = 0; /* checksum not supported */
+	subflow->request_version = 0; /* only v0 supported */
+
 	sock_hold(ssock->sk);
 
 release:
net/mptcp/protocol.h +1 -1
index 1d1e89109e98..9774ff5a5481 100644
@@ -168,10 +168,10 @@ struct subflow_context {
 	u16	map_data_len;
 	u16	request_mptcp : 1,  /* send MP_CAPABLE */
 		request_cksum : 1,
+		request_version : 4,
 		mp_capable : 1,     /* remote is MPTCP capable */
 		mp_join : 1,        /* remote is JOINing */
 		fourth_ack : 1,     /* send initial DSS */
-		version : 4,
 		conn_finished : 1,
 		use_checksum : 1,
 		map_valid : 1,
net/mptcp/subflow.c +2 -5
index 9d91014c0d1b..678fca99c23a 100644
@@ -62,8 +62,8 @@ static void subflow_v4_init_req(struct request_sock *req,
 
 	if (rx_opt.mptcp.mp_capable && listener->request_mptcp) {
 		subflow_req->mp_capable = 1;
-		if (rx_opt.mptcp.version >= listener->version)
-			subflow_req->version = listener->version;
+		if (rx_opt.mptcp.version >= listener->request_version)
+			subflow_req->version = listener->request_version;
 		else
 			subflow_req->version = rx_opt.mptcp.version;
 		if ((rx_opt.mptcp.flags & MPTCP_CAP_CHECKSUM_REQD) ||
@@ -218,9 +218,6 @@ int subflow_create_socket(struct sock *sk, struct socket **new_sock)
 
 	*new_sock = sf;
 	subflow->conn = sk;
-	subflow->request_mptcp = 1; // @@ if MPTCP enabled
-	subflow->request_cksum = 1; // @@ if checksum enabled
-	subflow->version = 0;
 
 	return 0;
 }
Signature · 1 line
2.17.2
author
#17 lore raw

PATCH v3 8/10 mptcp: Add handling of outgoing MP_JOIN requests

In reply to Peter Krystad

Subflow creation may be initiated by the path manager when the primary connection is fully established and a remote address has been received via ADD_ADDR.

Create an in-kernel sock and use kernel_connect() to initiate connection. When a valid SYN-ACK is received the new sock is added to the tail of the mptcp sock conn_list where it will not interfere with data flow on the original connection.

Data flow and connection failover not addressed by this commit.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 include/net/mptcp.h  |  2 ++
 net/mptcp/options.c  | 51 ++++++++++++++++++++++++++++++---
 net/mptcp/protocol.c |  1 +
 net/mptcp/protocol.h | 11 +++++++
 net/mptcp/subflow.c  | 68 +++++++++++++++++++++++++++++++++++++++++++-
 net/mptcp/token.c    | 48 +++++++++++++++++++++++++++++++
 6 files changed, 176 insertions(+), 5 deletions(-)
include/net/mptcp.h +2 -0
index bb2dd193c0c5..50cd1b31ebdd 100644
@@ -40,6 +40,8 @@ struct mptcp_out_options {
 	u8 backup;
 	u32 nonce;
 	u64 thmac;
+	u32 token;
+	u8 hmac[20];
 	struct mptcp_ext ext_copy;
 #endif
 };
net/mptcp/options.c +47 -4

105 lines not on this page · raw message

net/mptcp/protocol.c +1 -0
index fce5eea729ba..80a8c7e5a3e7 100644
@@ -797,6 +797,7 @@ void mptcp_finish_connect(struct sock *sk, int mp_capable)
 		msk->local_key = subflow->local_key;
 		msk->token = subflow->token;
 		pr_debug("msk=%p, token=%u", msk, msk->token);
+		msk->dport = ntohs(inet_sk(msk->subflow->sk)->inet_dport);
 
 		pm_new_connection(msk, 0);
 
net/mptcp/protocol.h +11 -0
index 9774ff5a5481..243d896d9e42 100644
@@ -49,8 +49,10 @@
 #define TCPOLEN_MPTCP_ADD_ADDR6		20
 #define TCPOLEN_MPTCP_RM_ADDR		4
 
+/* MPTCP MP_JOIN flags */
 #define MPTCPOPT_BACKUP		BIT(0)
 #define MPTCPOPT_HMAC_LEN	20
+#define MPTCPOPT_THMAC_LEN	8
 
 /* MPTCP MP_CAPABLE flags */
 #define MPTCP_VERSION_MASK	(0x0F)
@@ -115,6 +117,7 @@ struct mptcp_sock {
 	u64		write_seq;
 	u64		ack_seq;
 	u32		token;
+	u16		dport;
 	struct list_head conn_list;
 	struct socket	*subflow; /* outgoing connect/listener/!mp_capable */
 	struct mptcp_pm_data	pm;
@@ -167,7 +170,9 @@ struct subflow_context {
 	u32	ssn_offset;
 	u16	map_data_len;
 	u16	request_mptcp : 1,  /* send MP_CAPABLE */
+		request_join : 1,   /* send MP_JOIN */
 		request_cksum : 1,
+		request_bkup : 1,
 		request_version : 4,
 		mp_capable : 1,     /* remote is MPTCP capable */
 		mp_join : 1,        /* remote is JOINing */
@@ -179,6 +184,8 @@ struct subflow_context {
 	u32	remote_nonce;
 	u64	thmac;
 	u32	local_nonce;
+	u32	remote_token;
+	u8	hmac[MPTCPOPT_HMAC_LEN];
 	u8	local_id;
 	u8	remote_id;
 
@@ -202,6 +209,8 @@ mptcp_subflow_tcp_socket(const struct subflow_context *subflow)
 }
 
 void subflow_init(void);
+int subflow_connect(struct sock *sk, struct sockaddr_in *local,
+		    struct sockaddr_in *remote, u8 remote_id);
 int subflow_create_socket(struct sock *sk, struct socket **new_sock);
 
 extern const struct inet_connection_sock_af_ops ipv4_specific;
@@ -215,10 +224,12 @@ void mptcp_finish_join(struct sock *sk);
 void token_init(void);
 void token_new_request(struct request_sock *req, const struct sk_buff *skb);
 int token_join_request(struct request_sock *req, const struct sk_buff *skb);
+int token_join_response(struct sock *sk);
 int token_join_valid(struct request_sock *req,
 		     struct tcp_options_received *rx_opt);
 void token_destroy_request(u32 token);
 void token_new_connect(struct sock *sk);
+void token_new_subflow(struct sock *sk);
 void token_new_accept(struct sock *sk);
 int token_new_join(struct sock *sk);
 void token_update_accept(struct sock *sk, struct sock *conn);
net/mptcp/subflow.c +67 -1

92 lines not on this page · raw message

net/mptcp/token.c +48 -0

66 lines not on this page · raw message

Signature · 1 line
2.17.2
author
#22 lore raw

PATCH v3 9/10 Use pm_get_local_id() to get the real local address id.

In reply to Peter Krystad

Also, take reference to mptcp sock while using it

squashto: Add handling of incoming MP_JOIN requests

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/token.c | 31 ++++++++++++++++++++++---------
 1 file changed, 22 insertions(+), 9 deletions(-)
net/mptcp/token.c +22 -9
index d7cf40b40beb..80acb6c6657f 100644
@@ -222,14 +222,22 @@ int token_join_request(struct request_sock *req, const struct sk_buff *skb)
 	pr_debug("subflow_req=%p, token=%u", subflow_req, subflow_req->token);
 	spin_lock_bh(&token_tree_lock);
 	conn = lookup_token(subflow_req->token);
+	if (conn)
+		sock_hold(conn);
 	spin_unlock_bh(&token_tree_lock);
-	if (conn) {
-		// @@ get real local address id for this skb->saddr
-		subflow_req->local_id = 0;
-		new_req_join(req, conn, skb);
-		return 0;
+
+	if (!conn)
+		return -1;
+
+	if (pm_get_local_id(req, conn, skb)) {
+		sock_put(conn);
+		return -1;
 	}
-	return -1;
+
+	new_req_join(req, conn, skb);
+
+	sock_put(conn);
+	return 0;
 }
 
 /* validate received truncated hmac and create hmac for third ACK */
@@ -247,15 +255,20 @@ int token_join_valid(struct request_sock *req,
 {
 	struct subflow_request_sock *subflow_req = subflow_rsk(req);
 	struct sock *conn;
+	int err;
 
 	pr_debug("subflow_req=%p, token=%u", subflow_req, subflow_req->token);
 	spin_lock_bh(&token_tree_lock);
 	conn = lookup_token(subflow_req->token);
-	spin_unlock_bh(&token_tree_lock);
 	if (conn)
-		return new_join_valid(req, conn, rx_opt);
+		sock_hold(conn);
+	spin_unlock_bh(&token_tree_lock);
+	if (!conn)
+		return -1;
 
-	return -1;
+	err = new_join_valid(req, conn, rx_opt);
+	sock_put(conn);
+	return err;
 }
 
 /* create new local key, idsn, and token for subflow */
Signature · 1 line
2.17.2
author
#24 lore raw

PATCH v3 10/10 mptcp: Implement path manager interface commands

In reply to Peter Krystad

Use the addr_signal flag to indicate to the subflow layer that a local address may be announced, and call subflow_connect() to initiate a secondary subflow.

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/pm.c       | 62 +++++++++++++++++++++++++++++++++++++++++---
 net/mptcp/protocol.h |  1 +
 net/mptcp/token.c    | 13 ++++++++++
 3 files changed, 73 insertions(+), 3 deletions(-)
net/mptcp/pm.c +59 -3
index e81475a337b7..c98b92512adf 100644
@@ -13,17 +13,73 @@
 int pm_announce_addr(u32 token, sa_family_t family, u8 local_id,
 		     struct in_addr *addr)
 {
-	return -ENOTSUPP;
+	struct mptcp_sock *msk = mptcp_sk(token_lookup_get(token));
+	int err = 0;
+
+	if (!msk)
+		return -EINVAL;
+
+	if (msk->pm.local_valid) {
+		err = -EBADR;
+		goto announce_put;
+	}
+
+	pr_debug("msk=%p, local_id=%d, addr=%x", msk, local_id, addr->s_addr);
+	msk->pm.local_valid = 1;
+	msk->pm.local_id = local_id;
+	msk->pm.local_family = family;
+	msk->pm.local_addr.s_addr = addr->s_addr;
+	msk->addr_signal = 1;
+
+announce_put:
+	sock_put((struct sock *)msk);
+	return err;
 }
 
 int pm_remove_addr(u32 token, u8 local_id)
 {
-	return -ENOTSUPP;
+	struct mptcp_sock *msk = mptcp_sk(token_lookup_get(token));
+
+	if (!msk)
+		return -EINVAL;
+
+	pr_debug("msk=%p", msk);
+	msk->pm.local_valid = 0;
+
+	sock_put((struct sock *)msk);
+	return 0;
 }
 
 int pm_create_subflow(u32 token, u8 remote_id)
 {
-	return -ENOTSUPP;
+	struct mptcp_sock *msk = mptcp_sk(token_lookup_get(token));
+	struct sockaddr_in remote;
+	struct sockaddr_in local;
+	int err;
+
+	if (!msk)
+		return -EINVAL;
+
+	pr_debug("msk=%p", msk);
+
+	if (!msk->pm.remote_valid || remote_id != msk->pm.remote_id) {
+		err = -EBADR;
+		goto create_put;
+	}
+
+	local.sin_family = AF_INET;
+	local.sin_port = 0;
+	local.sin_addr.s_addr = INADDR_ANY;
+
+	remote.sin_family = msk->pm.remote_family;
+	remote.sin_port = htons(msk->dport);
+	remote.sin_addr.s_addr = htonl(msk->pm.remote_addr.s_addr);
+
+	err = subflow_connect((struct sock *)msk, &local, &remote, remote_id);
+
+create_put:
+	sock_put((struct sock *)msk);
+	return err;
 }
 
 int pm_remove_subflow(u32 token, u8 remote_id)
net/mptcp/protocol.h +1 -0
index 243d896d9e42..bc1fbd3827a6 100644
@@ -233,6 +233,7 @@ void token_new_subflow(struct sock *sk);
 void token_new_accept(struct sock *sk);
 int token_new_join(struct sock *sk);
 void token_update_accept(struct sock *sk, struct sock *conn);
+struct sock *token_lookup_get(u32 token);
 void token_release(u32 token);
 void token_destroy(u32 token);
 
net/mptcp/token.c +13 -0
index 80acb6c6657f..b54f617d4c26 100644
@@ -348,6 +348,19 @@ int token_new_join(struct sock *sk)
 	return -1;
 }
 
+struct sock *token_lookup_get(u32 token)
+{
+	struct sock *conn;
+
+	spin_lock_bh(&token_tree_lock);
+	conn = lookup_token(token);
+	if (conn)
+		sock_hold(conn);
+
+	spin_unlock_bh(&token_tree_lock);
+	return conn;
+}
+
 void token_destroy_request(u32 token)
 {
 	pr_debug("token=%u", token);
Signature · 1 line
2.17.2
#26 lore raw
In reply to Peter Krystad

Hi all,

Quoted 31 lines · Peter Krystad

This is several accumulated fixes that are required for MP_JOIN (secondary subflows) plus the outgoing code.

Both outgoing and incoming MP_JOIN now inter-operate with the multipath.org implementation although there is still a deadlock when incoming is fully established. Since the incoming direction did not work at all before this is an improvement. I have not yet tested running our implmentation against itself.

The final patch in the set is an "interim path manager", this is throw-away code that I include for testing purposes. For an incoming connection it is necessary for the server to advertise a secondary address and for my testing I have been hard-coding the IP address of my secondary interface when I build the kernel. Obviously not the final solution but we need to discuss what approach to use for the self tests. I also think we may not want to publish this patch on our net-next branch.

v2:
- move mptcp_options_received formatting to options.c
- resolve conflicts with Davide's crypto commit
- make helper for setting request fields
- fix locking in token_lookup()
- move token_lookup() to "Implement path manager interface commands:
(where it belongs)
- split interim path manager out of final commit, will submit seperately
- remove use of local_id in pm_create_subflow
- restrict announcing addresses to server side only
- add one more commit with get_local_id fix for incoming MP_JOIN

Quoted 12 lines

v3:
- Revise nested if logic in options.c
- set subflow->request_cksum correctly (do not support)
- lock mptcp_sock in subflow_connect() and take ref
- return EINPROGRESS from subflow_connect() in normal case
- cleanup socket when bind()/connect() fails in subflow_connect()
- take reference to mptcp_sock in token_join_request() and token_join_valid()
- rename token_lookup() to token_lookup_get()
- track remote token seperately in subflow (previously local token field was being overwritten, preventing cleanup)
- take long-lived reference to mptcp_sock in token_new_subflow

I went through the whole series and it looks like all my comments has been addressed, thank you Peter!

LGMT!

Paolo

#18 lore raw

Re: [MPTCP] [PATCH v3 08/10] mptcp: Add handling of outgoing MP_JOIN requests

In reply to Peter Krystad
Quoted 312 lines · Peter Krystad

Subflow creation may be initiated by the path manager when the primary connection is fully established and a remote address has been received via ADD_ADDR.

Create an in-kernel sock and use kernel_connect() to initiate connection. When a valid SYN-ACK is received the new sock is added to the tail of the mptcp sock conn_list where it will not interfere with data flow on the original connection.

Data flow and connection failover not addressed by this commit.

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

---

include/net/mptcp.h  |  2 ++
net/mptcp/options.c  | 51 ++++++++++++++++++++++++++++++---
net/mptcp/protocol.c |  1 +
net/mptcp/protocol.h | 11 +++++++
net/mptcp/subflow.c  | 68 +++++++++++++++++++++++++++++++++++++++++++-
net/mptcp/token.c    | 48 +++++++++++++++++++++++++++++++
6 files changed, 176 insertions(+), 5 deletions(-)
include/net/mptcp.h +2 -0
index bb2dd193c0c5..50cd1b31ebdd 100644
@@ -40,6 +40,8 @@ struct mptcp_out_options {
 	u8 backup;
 	u32 nonce;
 	u64 thmac;
+	u32 token;
+	u8 hmac[20];
 	struct mptcp_ext ext_copy;
 #endif
 };
net/mptcp/options.c +47 -4
index 01f59ad629e5..be7afa5aac6a 100644
@@ -300,6 +300,16 @@ bool mptcp_syn_options(struct sock *sk, unsigned int *size,
 		opts->sndr_key = subflow->local_key;
 		*size = TCPOLEN_MPTCP_MPC_SYN;
 		return true;
+	} else if (subflow->request_join) {
+		pr_debug("remote_token=%u, nonce=%u", subflow->remote_token,
+			 subflow->local_nonce);
+		opts->suboptions = OPTION_MPTCP_MPJ_SYN;
+		opts->join_id = subflow->remote_id;
+		opts->token = subflow->remote_token;
+		opts->nonce = subflow->local_nonce;
+		opts->backup = subflow->request_bkup;
+		*size = TCPOLEN_MPTCP_MPJ_SYN;
+		return true;
 	}
 	return false;
 }
@@ -309,10 +319,17 @@ void mptcp_rcv_synsent(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct subflow_context *subflow = subflow_ctx(sk);
 
-	pr_debug("subflow=%p", subflow);
 	if (subflow->request_mptcp && tp->rx_opt.mptcp.mp_capable) {
 		subflow->mp_capable = 1;
 		subflow->remote_key = tp->rx_opt.mptcp.sndr_key;
+		pr_debug("subflow=%p, remote_key=%llu", subflow,
+			 subflow->remote_key);
+	} else if (subflow->request_join && tp->rx_opt.mptcp.mp_join) {
+		subflow->mp_join = 1;
+		subflow->thmac = tp->rx_opt.mptcp.thmac;
+		subflow->remote_nonce = tp->rx_opt.mptcp.nonce;
+		pr_debug("subflow=%p, thmac=%llu, remote_nonce=%u", subflow,
+			 subflow->thmac, subflow->remote_nonce);
 	}
 }
 
@@ -322,7 +339,8 @@ static bool mptcp_established_options_mp(struct sock *sk, unsigned int *size,
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
 
-	if (!subflow->fourth_ack && remaining >= TCPOLEN_MPTCP_MPC_ACK) {
+	if (subflow->mp_capable && !subflow->fourth_ack &&
+	    remaining >= TCPOLEN_MPTCP_MPC_ACK) {
 		opts->suboptions = OPTION_MPTCP_MPC_ACK;
 		opts->sndr_key = subflow->local_key;
 		opts->rcvr_key = subflow->remote_key;
@@ -331,6 +349,14 @@ static bool mptcp_established_options_mp(struct sock *sk, unsigned int *size,
 		pr_debug("subflow=%p, local_key=%llu, remote_key=%llu",
 			 subflow, subflow->local_key, subflow->remote_key);
 		return true;
+	} else if (subflow->mp_join && !subflow->fourth_ack &&
+		   remaining >= TCPOLEN_MPTCP_MPJ_ACK) {
+		opts->suboptions = OPTION_MPTCP_MPJ_ACK;
+		memcpy(opts->hmac, subflow->hmac, MPTCPOPT_HMAC_LEN);
+		*size = TCPOLEN_MPTCP_MPJ_ACK;
+		subflow->fourth_ack = 1;
+		pr_debug("subflow=%p", subflow);
+		return true;
 	}
 	return false;
 }
@@ -443,10 +469,11 @@ bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
 			       unsigned int *size, unsigned int remaining,
 			       struct mptcp_out_options *opts)
 {
+	struct subflow_context *subflow = subflow_ctx(sk);
 	unsigned int opt_size = 0;
 	bool ret = false;
 
-	if (!subflow_ctx(sk)->mp_capable)
+	if (!subflow->mp_capable && !subflow->mp_join)
 		return false;
 
 	opts->suboptions = 0;
@@ -546,7 +573,6 @@ void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb,
 
 	if (msk)
 		pm_fully_established(msk);
-
 }
 
 void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
@@ -595,6 +621,16 @@ void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
 				      0, opts->addr_id);
 	}
 
+	if (OPTION_MPTCP_MPJ_SYN & opts->suboptions) {
+		*ptr++ = mptcp_option(MPTCPOPT_MP_JOIN,
+				      TCPOLEN_MPTCP_MPJ_SYN,
+				      opts->backup, opts->join_id);
+		put_unaligned_be32(opts->token, ptr);
+		ptr += 1;
+		put_unaligned_be32(opts->nonce, ptr);
+		ptr += 1;
+	}
+
 	if (OPTION_MPTCP_MPJ_SYNACK & opts->suboptions) {
 		*ptr++ = mptcp_option(MPTCPOPT_MP_JOIN,
 				      TCPOLEN_MPTCP_MPJ_SYNACK,
@@ -605,6 +641,13 @@ void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
 		ptr += 1;
 	}
 
+	if (OPTION_MPTCP_MPJ_ACK & opts->suboptions) {
+		*ptr++ = mptcp_option(MPTCPOPT_MP_JOIN,
+				      TCPOLEN_MPTCP_MPJ_ACK, 0, 0);
+		memcpy(ptr, opts->hmac, MPTCPOPT_HMAC_LEN);
+		ptr += 5;
+	}
+
 	if (opts->ext_copy.use_ack || opts->ext_copy.use_map) {
 		struct mptcp_ext *mpext = &opts->ext_copy;
 		u8 len = TCPOLEN_MPTCP_DSS_BASE;
net/mptcp/protocol.c +1 -0
index fce5eea729ba..80a8c7e5a3e7 100644
@@ -797,6 +797,7 @@ void mptcp_finish_connect(struct sock *sk, int mp_capable)
 		msk->local_key = subflow->local_key;
 		msk->token = subflow->token;
 		pr_debug("msk=%p, token=%u", msk, msk->token);
+		msk->dport = ntohs(inet_sk(msk->subflow->sk)->inet_dport);
 
 		pm_new_connection(msk, 0);
 
net/mptcp/protocol.h +11 -0
index 9774ff5a5481..243d896d9e42 100644
@@ -49,8 +49,10 @@
 #define TCPOLEN_MPTCP_ADD_ADDR6		20
 #define TCPOLEN_MPTCP_RM_ADDR		4
 
+/* MPTCP MP_JOIN flags */
 #define MPTCPOPT_BACKUP		BIT(0)
 #define MPTCPOPT_HMAC_LEN	20
+#define MPTCPOPT_THMAC_LEN	8
 
 /* MPTCP MP_CAPABLE flags */
 #define MPTCP_VERSION_MASK	(0x0F)
@@ -115,6 +117,7 @@ struct mptcp_sock {
 	u64		write_seq;
 	u64		ack_seq;
 	u32		token;
+	u16		dport;
 	struct list_head conn_list;
 	struct socket	*subflow; /* outgoing connect/listener/!mp_capable */
 	struct mptcp_pm_data	pm;
@@ -167,7 +170,9 @@ struct subflow_context {
 	u32	ssn_offset;
 	u16	map_data_len;
 	u16	request_mptcp : 1,  /* send MP_CAPABLE */
+		request_join : 1,   /* send MP_JOIN */
 		request_cksum : 1,
+		request_bkup : 1,
 		request_version : 4,
 		mp_capable : 1,     /* remote is MPTCP capable */
 		mp_join : 1,        /* remote is JOINing */
@@ -179,6 +184,8 @@ struct subflow_context {
 	u32	remote_nonce;
 	u64	thmac;
 	u32	local_nonce;
+	u32	remote_token;
+	u8	hmac[MPTCPOPT_HMAC_LEN];
 	u8	local_id;
 	u8	remote_id;
 
@@ -202,6 +209,8 @@ mptcp_subflow_tcp_socket(const struct subflow_context *subflow)
 }
 
 void subflow_init(void);
+int subflow_connect(struct sock *sk, struct sockaddr_in *local,
+		    struct sockaddr_in *remote, u8 remote_id);
 int subflow_create_socket(struct sock *sk, struct socket **new_sock);
 
 extern const struct inet_connection_sock_af_ops ipv4_specific;
@@ -215,10 +224,12 @@ void mptcp_finish_join(struct sock *sk);
 void token_init(void);
 void token_new_request(struct request_sock *req, const struct sk_buff *skb);
 int token_join_request(struct request_sock *req, const struct sk_buff *skb);
+int token_join_response(struct sock *sk);
 int token_join_valid(struct request_sock *req,
 		     struct tcp_options_received *rx_opt);
 void token_destroy_request(u32 token);
 void token_new_connect(struct sock *sk);
+void token_new_subflow(struct sock *sk);
 void token_new_accept(struct sock *sk);
 int token_new_join(struct sock *sk);
 void token_update_accept(struct sock *sk, struct sock *conn);
net/mptcp/subflow.c +54 -1
index 678fca99c23a..448c9e7f567a 100644
@@ -22,6 +22,9 @@ static int subflow_rebuild_header(struct sock *sk)
 	if (subflow->request_mptcp && !subflow->token) {
 		pr_debug("subflow=%p", sk);
 		token_new_connect(sk);
+	} else if (subflow->request_join && !subflow->local_nonce) {
+		pr_debug("subflow=%p", sk);
+		token_new_subflow(sk);
 	}
 
 	return inet_sk_rebuild_header(sk);
@@ -95,7 +98,10 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 
 	inet_sk_rx_dst_set(sk, skb);
 
-	if (subflow->conn && !subflow->conn_finished) {
+	if (!subflow->conn)
+		return;
+
+	if (subflow->mp_capable && !subflow->conn_finished) {
 		pr_debug("subflow=%p, remote_key=%llu", subflow_ctx(sk),
 			 subflow->remote_key);
 		mptcp_finish_connect(subflow->conn, subflow->mp_capable);
@@ -105,6 +111,17 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 			pr_debug("synack seq=%u", TCP_SKB_CB(skb)->seq);
 			subflow->ssn_offset = TCP_SKB_CB(skb)->seq;
 		}
+	} else if (subflow->mp_join && !subflow->conn_finished) {
+		pr_debug("subflow=%p, thmac=%llu, remote_nonce=%u",
+			 subflow_ctx(sk), subflow->thmac,
+			 subflow->remote_nonce);
+		if (token_join_response(sk)) {
+			subflow->mp_join = 0;
+			// @@ need to trigger RST
+		} else {
+			mptcp_finish_join(sk);
+			subflow->conn_finished = 1;
+		}
 	}
 }
 
@@ -195,6 +212,55 @@ static void subflow_data_ready(struct sock *sk)
 	}
 }
 
+int subflow_connect(struct sock *sk, struct sockaddr_in *local,
+		    struct sockaddr_in *remote, u8 remote_id)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+	struct subflow_context *subflow;
+	struct socket *sf;
+	u32 remote_token;
+	int err;
+
+	lock_sock(sk);
+	err = subflow_create_socket(sk, &sf);
+	if (err) {
+		release_sock(sk);
+		return err;
+	}
+
+	subflow = subflow_ctx(sf->sk);
+	subflow->remote_key = msk->remote_key;
+	subflow->local_key = msk->local_key;
+	subflow->token = msk->token;
+
+	sock_hold(sf->sk);
+	release_sock(sk);
+
+	err = kernel_bind(sf, (struct sockaddr *)local,
+			  sizeof(struct sockaddr_in));
+	if (err)
+		goto failed;
+
+	crypto_key_sha1(subflow->remote_key, &remote_token, NULL);
+	pr_debug("msk=%p remote_token=%u", msk, remote_token);
+	subflow->remote_token = remote_token;
+	subflow->remote_id = remote_id;
+	subflow->request_join = 1;
+	subflow->request_bkup = 1;
+

net/mptcp/subflow.c static void subflow_data_ready(struct sock *sk) · line 260

Quoted 10 lines
net/mptcp/subflow.c +10 -0
+	err = kernel_connect(sf, (struct sockaddr *)remote,
+			     sizeof(struct sockaddr_in), O_NONBLOCK);
+	if (err && err != -EINPROGRESS)
+		goto failed;
+
+	sock_put(sf->sk);
+	return err;
+
+failed:
+	sock_release(sf);

Should there be a sock_put(sf->sk) on this path too? Looks like sock_release() will free the subflow context but not the struct sock itself.

Mat

Quoted 87 lines
net/mptcp/subflow.c +3 -0
+	return err;
+}
+
 int subflow_create_socket(struct sock *sk, struct socket **new_sock)
 {
 	struct subflow_context *subflow;
net/mptcp/token.c +48 -0
index ef03ef19af98..d7cf40b40beb 100644
@@ -93,6 +93,28 @@ static void new_req_join(struct request_sock *req, struct sock *sk,
 		 subflow_req->thmac);
 }
 
+static int new_rsp_join(struct sock *sk)
+{
+	struct subflow_context *subflow = subflow_ctx(sk);
+	u8 hmac[MPTCPOPT_HMAC_LEN];
+	u64 thmac;
+
+	crypto_hmac_sha1(subflow->remote_key, subflow->local_key,
+			 subflow->remote_nonce, subflow->local_nonce,
+			 (u32 *)hmac);
+
+	thmac = get_unaligned_be64(hmac);
+	pr_debug("thmac=%llu", thmac);
+	if (thmac != subflow->thmac)
+		return -1;
+
+	crypto_hmac_sha1(subflow->local_key, subflow->remote_key,
+			 subflow->local_nonce, subflow->remote_nonce,
+			 (u32 *)subflow->hmac);
+
+	return 0;
+}
+
 static int new_join_valid(struct request_sock *req, struct sock *sk,
 			  struct tcp_options_received *rx_opt)
 {
@@ -210,6 +232,15 @@ int token_join_request(struct request_sock *req, const struct sk_buff *skb)
 	return -1;
 }
 
+/* validate received truncated hmac and create hmac for third ACK */
+int token_join_response(struct sock *sk)
+{
+	struct subflow_context *subflow = subflow_ctx(sk);
+
+	pr_debug("subflow=%p, token=%u", subflow, subflow->token);
+	return new_rsp_join(sk);
+}
+
 /* validate hmac received in third ACK */
 int token_join_valid(struct request_sock *req,
 		     struct tcp_options_received *rx_opt)
@@ -247,6 +278,23 @@ void token_new_connect(struct sock *sk)
 	spin_unlock_bh(&token_tree_lock);
 }
 
+/* take reference to connection and create nonce for secondary subflow */
+void token_new_subflow(struct sock *sk)
+{
+	struct subflow_context *subflow = subflow_ctx(sk);
+	struct sock *conn;
+
+	pr_debug("subflow=%p", subflow);
+
+	spin_lock_bh(&token_tree_lock);
+	conn = lookup_token(subflow->token);
+	if (conn)
+		sock_hold(conn);
+	spin_unlock_bh(&token_tree_lock);
+
+	get_random_bytes(&subflow->local_nonce, sizeof(u32));
+}
+
 void token_new_accept(struct sock *sk)
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
Signature · 6 lines
2.17.2

_______________________________________________
mptcp mailing list
mptcp(a)lists.01.org
https://lists.01.org/mailman/listinfo/mptcp

--
Mat Martineau
Intel

#2 lore raw

Re: [MPTCP] [PATCH v3 01/10] Extend path manager interface

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 5 lines · Peter Krystad

A little more functionality and some comments clarifying which routines are for a path manager to call and which are callbacks.

squashto: Add path manager interface

Thank you for the patch and the reviews!

I had to split this patch in 2 parts:
- c8625a37c692: part 1: "squashed" in "mptcp: Add path manager interface"
- 4412b8e8e6c9: conflict in t/mptcp-Add-ADD_ADDR-handling
- 0757aec1b6f2: part 2: "squashed" in "mptcp: Add ADD_ADDR handling"
- b1477322bbb7..711ae3fdd44f: results

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 02/10] Fix locking in mptcp_finish_join.

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 3 lines · Peter Krystad

Lock was being taken on the subflow sock, not mptcp_sock.

squashto: Add handling of incoming MP_JOIN requests

Thank you for the patch and the reviews!

- 590cbe28643f: "squashed" in "mptcp: Add handling of incoming MP_JOIN requests"
- 711ae3fdd44f..34fd270ae102: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 03/10] Fix parsing MP_JOIN third ACK.

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 3 lines · Peter Krystad

There is a reserved byte of zero's in there.

squashto: Add handling of incoming MP_JOIN requests

Thank you for the patch and the reviews!

- 1437af3a308a: "squashed" in "mptcp: Add handling of incoming MP_JOIN requests"
- 34fd270ae102..6c6c0f074bf5: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 04/10] Re-factor and fixes for crypto_hmac_sha1()

In reply to Peter Krystad

Hi Peter, Mat, Paolo,

Quoted 5 lines · Peter Krystad

Re-factor to use parameters specific to MPTCP use case similar to crypto_key_sha1() and get rid of var args. Also fix endianness issues.

squash to: Add key generation and token tree
and: Add handling of incoming MP_JOIN requests

Thank you for the patch and the reviews!

- bc31e4f64971: part 1: "squashed" in "mptcp: Add key generation and token tree"
- f282f6b36b49: part 2: "squahsed" in "mptcp: Add handling of incoming MP_JOIN requests"
- f7938dd15948..009b132dc29d: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 05/10] Rename mptcp_attach_dss()

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 3 lines · Peter Krystad

Other incoming MPTCP options will be handled in this routine.

squashto: Implement MPTCP receive path

Thank you for the patch and the reviews!

- 3776155bf643: "squashed" in "mptcp: Implement MPTCP receive path"
- 06256d7cf989..b1477322bbb7: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 06/10] Fix return value if no DSS option is created.

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 4 lines · Peter Krystad

Strictly speaking mptcp_established_options_dss() should return false if no DSS option is created.

squashto: Write MPTCP DSS headers to outgoing data packets

Thank you for the patch and the reviews!

- 7b039830202c: "squashed" in "mptcp: Write MPTCP DSS headers to outgoing data packets"
- 009b132dc29d..06256d7cf989: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 07/10] Move setting request_ fields of subflow to MPTCP layer helper

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 5 lines · Peter Krystad

For secondary subflows different request_ fields will be added that will be set differently based on subflow type (initial vs secondary).

squashto: Associate MPTCP context with TCP socket

Thank you for the patch and the reviews!

I had to split it in 3 parts:
- a831a5f64a0d: part 1: "squashed" in "mptcp: Associate MPTCP context with TCP socket"
- 4dff4a3c37ec: conflict in t/mptcp-Handle-MP_CAPABLE-options-for-outgoing-connections
- 7a108cfff657: part 2: "squashed" in "mptcp: Create SUBFLOW socket for incoming connections"
- ff3b994f4b4c: conflict in t/mptcp-Write-MPTCP-DSS-headers-to-outgoing-data-packets
- 987b6d7366be: conflict in t/mptcp-Add-handling-of-incoming-MP_JOIN-requests
- b99045fe03e0: part 3: "squashed" in "mptcp: harmonize locking on all socket operations."
- 64fb32edc6b2: signed-off
- b81ff8d90de3..f7938dd15948: result.

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 08/10] mptcp: Add handling of outgoing MP_JOIN requests

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 11 lines · Peter Krystad

Subflow creation may be initiated by the path manager when the primary connection is fully established and a remote address has been received via ADD_ADDR.

Create an in-kernel sock and use kernel_connect() to initiate connection. When a valid SYN-ACK is received the new sock is added to the tail of the mptcp sock conn_list where it will not interfere with data flow on the original connection.

Data flow and connection failover not addressed by this commit.

Thank you for the patch and the reviews!

Added at the end of the series:
- 31bf10533d43: "new commit" in
t/mptcp-Add-handling-of-outgoing-MP_JOIN-requests
- 417bdddfeedd..f7d9e8ce4717: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 09/10] Use pm_get_local_id() to get the real local address id.

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 3 lines · Peter Krystad

Also, take reference to mptcp sock while using it

squashto: Add handling of incoming MP_JOIN requests

Thank you for the patch and the reviews!

- 05f13c173cb5: "squashed" in "mptcp: Add handling of incoming MP_JOIN requests"
- 6c6c0f074bf5..417bdddfeedd: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 10/10] mptcp: Implement path manager interface commands

In reply to Peter Krystad

Hi Peter, Paolo, Mat,

Quoted 3 lines · Peter Krystad

Use the addr_signal flag to indicate to the subflow layer that a local address may be announced, and call subflow_connect() to initiate a secondary subflow.

Thank you for the patch and the reviews!

Added at the end of the series:
- 9ad3c22dcab0: "new commit" in
t/mptcp-Implement-path-manager-interface-commands
- f7d9e8ce4717..2038a22e9690: result

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 08/10] mptcp: Add handling of outgoing MP_JOIN requests

In reply to Mat Martineau

Hi Mat, Peter,

Quoted 1 line · Mat Martineau

On Wed, 7 Aug 2019, Peter Krystad wrote:

(...)

Quoted 4 lines
Quoted 4 lines
net/mptcp/subflow.c +0 -0
index 678fca99c23a..448c9e7f567a 100644

(...)

Quoted 46 lines
Quoted 46 lines
net/mptcp/subflow.c +42 -0
@@ -195,6 +212,55 @@ static void subflow_data_ready(struct sock *sk)
     }
 }
 
+int subflow_connect(struct sock *sk, struct sockaddr_in *local,
+            struct sockaddr_in *remote, u8 remote_id)
+{
+    struct mptcp_sock *msk = mptcp_sk(sk);
+    struct subflow_context *subflow;
+    struct socket *sf;
+    u32 remote_token;
+    int err;
+
+    lock_sock(sk);
+    err = subflow_create_socket(sk, &sf);
+    if (err) {
+        release_sock(sk);
+        return err;
+    }
+
+    subflow = subflow_ctx(sf->sk);
+    subflow->remote_key = msk->remote_key;
+    subflow->local_key = msk->local_key;
+    subflow->token = msk->token;
+
+    sock_hold(sf->sk);
+    release_sock(sk);
+
+    err = kernel_bind(sf, (struct sockaddr *)local,
+              sizeof(struct sockaddr_in));
+    if (err)
+        goto failed;
+
+    crypto_key_sha1(subflow->remote_key, &remote_token, NULL);
+    pr_debug("msk=%p remote_token=%u", msk, remote_token);
+    subflow->remote_token = remote_token;
+    subflow->remote_id = remote_id;
+    subflow->request_join = 1;
+    subflow->request_bkup = 1;
+
+    err = kernel_connect(sf, (struct sockaddr *)remote,
+                 sizeof(struct sockaddr_in), O_NONBLOCK);
+    if (err && err != -EINPROGRESS)
+        goto failed;
+
+    sock_put(sf->sk);
Quoted 8 lines

net/mptcp/subflow.c static void subflow_data_ready(struct sock *sk) · line 260

Quoted 4 lines
net/mptcp/subflow.c +4 -0
+    return err;
+
+failed:
+    sock_release(sf);

Should there be a sock_put(sf->sk) on this path too? Looks like sock_release() will free the subflow context but not the struct sock itself.

Please note that I just applied this patch series. But do not hesitate to create a new patch to fix this if needed!

Cheers,
Matt

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

Hi Peter, Paolo, Mat,

Quoted 46 lines · Paolo Abeni

Hi all,

Quoted 31 lines · Peter Krystad

This is several accumulated fixes that are required for MP_JOIN (secondary subflows) plus the outgoing code.

Both outgoing and incoming MP_JOIN now inter-operate with the multipath.org implementation although there is still a deadlock when incoming is fully established. Since the incoming direction did not work at all before this is an improvement. I have not yet tested running our implmentation against itself.

The final patch in the set is an "interim path manager", this is throw-away code that I include for testing purposes. For an incoming connection it is necessary for the server to advertise a secondary address and for my testing I have been hard-coding the IP address of my secondary interface when I build the kernel. Obviously not the final solution but we need to discuss what approach to use for the self tests. I also think we may not want to publish this patch on our net-next branch.

v2:
- move mptcp_options_received formatting to options.c
- resolve conflicts with Davide's crypto commit
- make helper for setting request fields
- fix locking in token_lookup()
- move token_lookup() to "Implement path manager interface commands:
(where it belongs)
- split interim path manager out of final commit, will submit seperately
- remove use of local_id in pm_create_subflow
- restrict announcing addresses to server side only
- add one more commit with get_local_id fix for incoming MP_JOIN

Quoted 12 lines

v3:
- Revise nested if logic in options.c
- set subflow->request_cksum correctly (do not support)
- lock mptcp_sock in subflow_connect() and take ref
- return EINPROGRESS from subflow_connect() in normal case
- cleanup socket when bind()/connect() fails in subflow_connect()
- take reference to mptcp_sock in token_join_request() and token_join_valid()
- rename token_lookup() to token_lookup_get()
- track remote token seperately in subflow (previously local token field was being overwritten, preventing cleanup)
- take long-lived reference to mptcp_sock in token_new_subflow

Quoted 2 lines

I went through the whole series and it looks like all my comments has been addressed, thank you Peter!

Thank you for the patches and the reviews!

I just applied these 10 patches. Please check that everything looks good to you, especially the patches that I had to split in multiple ones and squash in different commits. Do not hesitate to tell me if the squashes were not in the right commits or if I have to change stuffs.

Note that I just saw Mat's email on the patch number 8. Feel free to continue the discussions there and create a new patch if needed.

The compilation is still OK. Tests are in progress. I will tell you tomorrow if there is an issue with them.

I think I applied all accepted patches on the list. Do not hesitate to tell me if other are also good to be applied ;)

Cheers,
Matt

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

Re: [MPTCP] [PATCH v3 08/10] mptcp: Add handling of outgoing MP_JOIN requests

In reply to Mat Martineau
Quoted 320 lines · Mat Martineau
Quoted 318 lines · Peter Krystad

Subflow creation may be initiated by the path manager when the primary connection is fully established and a remote address has been received via ADD_ADDR.

Create an in-kernel sock and use kernel_connect() to initiate connection. When a valid SYN-ACK is received the new sock is added to the tail of the mptcp sock conn_list where it will not interfere with data flow on the original connection.

Data flow and connection failover not addressed by this commit.

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

---

include/net/mptcp.h  |  2 ++
net/mptcp/options.c  | 51 ++++++++++++++++++++++++++++++---
net/mptcp/protocol.c |  1 +
net/mptcp/protocol.h | 11 +++++++
net/mptcp/subflow.c  | 68 +++++++++++++++++++++++++++++++++++++++++++-
net/mptcp/token.c    | 48 +++++++++++++++++++++++++++++++
6 files changed, 176 insertions(+), 5 deletions(-)
include/net/mptcp.h +2 -0
index bb2dd193c0c5..50cd1b31ebdd 100644
@@ -40,6 +40,8 @@ struct mptcp_out_options {
 	u8 backup;
 	u32 nonce;
 	u64 thmac;
+	u32 token;
+	u8 hmac[20];
 	struct mptcp_ext ext_copy;
 #endif
 };
net/mptcp/options.c +47 -4
index 01f59ad629e5..be7afa5aac6a 100644
@@ -300,6 +300,16 @@ bool mptcp_syn_options(struct sock *sk, unsigned int *size,
 		opts->sndr_key = subflow->local_key;
 		*size = TCPOLEN_MPTCP_MPC_SYN;
 		return true;
+	} else if (subflow->request_join) {
+		pr_debug("remote_token=%u, nonce=%u", subflow->remote_token,
+			 subflow->local_nonce);
+		opts->suboptions = OPTION_MPTCP_MPJ_SYN;
+		opts->join_id = subflow->remote_id;
+		opts->token = subflow->remote_token;
+		opts->nonce = subflow->local_nonce;
+		opts->backup = subflow->request_bkup;
+		*size = TCPOLEN_MPTCP_MPJ_SYN;
+		return true;
 	}
 	return false;
 }
@@ -309,10 +319,17 @@ void mptcp_rcv_synsent(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct subflow_context *subflow = subflow_ctx(sk);
 
-	pr_debug("subflow=%p", subflow);
 	if (subflow->request_mptcp && tp->rx_opt.mptcp.mp_capable) {
 		subflow->mp_capable = 1;
 		subflow->remote_key = tp->rx_opt.mptcp.sndr_key;
+		pr_debug("subflow=%p, remote_key=%llu", subflow,
+			 subflow->remote_key);
+	} else if (subflow->request_join && tp->rx_opt.mptcp.mp_join) {
+		subflow->mp_join = 1;
+		subflow->thmac = tp->rx_opt.mptcp.thmac;
+		subflow->remote_nonce = tp->rx_opt.mptcp.nonce;
+		pr_debug("subflow=%p, thmac=%llu, remote_nonce=%u", subflow,
+			 subflow->thmac, subflow->remote_nonce);
 	}
 }
 
@@ -322,7 +339,8 @@ static bool mptcp_established_options_mp(struct sock *sk, unsigned int *size,
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
 
-	if (!subflow->fourth_ack && remaining >= TCPOLEN_MPTCP_MPC_ACK) {
+	if (subflow->mp_capable && !subflow->fourth_ack &&
+	    remaining >= TCPOLEN_MPTCP_MPC_ACK) {
 		opts->suboptions = OPTION_MPTCP_MPC_ACK;
 		opts->sndr_key = subflow->local_key;
 		opts->rcvr_key = subflow->remote_key;
@@ -331,6 +349,14 @@ static bool mptcp_established_options_mp(struct sock *sk, unsigned int *size,
 		pr_debug("subflow=%p, local_key=%llu, remote_key=%llu",
 			 subflow, subflow->local_key, subflow->remote_key);
 		return true;
+	} else if (subflow->mp_join && !subflow->fourth_ack &&
+		   remaining >= TCPOLEN_MPTCP_MPJ_ACK) {
+		opts->suboptions = OPTION_MPTCP_MPJ_ACK;
+		memcpy(opts->hmac, subflow->hmac, MPTCPOPT_HMAC_LEN);
+		*size = TCPOLEN_MPTCP_MPJ_ACK;
+		subflow->fourth_ack = 1;
+		pr_debug("subflow=%p", subflow);
+		return true;
 	}
 	return false;
 }
@@ -443,10 +469,11 @@ bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
 			       unsigned int *size, unsigned int remaining,
 			       struct mptcp_out_options *opts)
 {
+	struct subflow_context *subflow = subflow_ctx(sk);
 	unsigned int opt_size = 0;
 	bool ret = false;
 
-	if (!subflow_ctx(sk)->mp_capable)
+	if (!subflow->mp_capable && !subflow->mp_join)
 		return false;
 
 	opts->suboptions = 0;
@@ -546,7 +573,6 @@ void mptcp_incoming_options(struct sock *sk, struct sk_buff *skb,
 
 	if (msk)
 		pm_fully_established(msk);
-
 }
 
 void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
@@ -595,6 +621,16 @@ void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
 				      0, opts->addr_id);
 	}
 
+	if (OPTION_MPTCP_MPJ_SYN & opts->suboptions) {
+		*ptr++ = mptcp_option(MPTCPOPT_MP_JOIN,
+				      TCPOLEN_MPTCP_MPJ_SYN,
+				      opts->backup, opts->join_id);
+		put_unaligned_be32(opts->token, ptr);
+		ptr += 1;
+		put_unaligned_be32(opts->nonce, ptr);
+		ptr += 1;
+	}
+
 	if (OPTION_MPTCP_MPJ_SYNACK & opts->suboptions) {
 		*ptr++ = mptcp_option(MPTCPOPT_MP_JOIN,
 				      TCPOLEN_MPTCP_MPJ_SYNACK,
@@ -605,6 +641,13 @@ void mptcp_write_options(__be32 *ptr, struct mptcp_out_options *opts)
 		ptr += 1;
 	}
 
+	if (OPTION_MPTCP_MPJ_ACK & opts->suboptions) {
+		*ptr++ = mptcp_option(MPTCPOPT_MP_JOIN,
+				      TCPOLEN_MPTCP_MPJ_ACK, 0, 0);
+		memcpy(ptr, opts->hmac, MPTCPOPT_HMAC_LEN);
+		ptr += 5;
+	}
+
 	if (opts->ext_copy.use_ack || opts->ext_copy.use_map) {
 		struct mptcp_ext *mpext = &opts->ext_copy;
 		u8 len = TCPOLEN_MPTCP_DSS_BASE;
net/mptcp/protocol.c +1 -0
index fce5eea729ba..80a8c7e5a3e7 100644
@@ -797,6 +797,7 @@ void mptcp_finish_connect(struct sock *sk, int mp_capable)
 		msk->local_key = subflow->local_key;
 		msk->token = subflow->token;
 		pr_debug("msk=%p, token=%u", msk, msk->token);
+		msk->dport = ntohs(inet_sk(msk->subflow->sk)->inet_dport);
 
 		pm_new_connection(msk, 0);
 
net/mptcp/protocol.h +11 -0
index 9774ff5a5481..243d896d9e42 100644
@@ -49,8 +49,10 @@
 #define TCPOLEN_MPTCP_ADD_ADDR6		20
 #define TCPOLEN_MPTCP_RM_ADDR		4
 
+/* MPTCP MP_JOIN flags */
 #define MPTCPOPT_BACKUP		BIT(0)
 #define MPTCPOPT_HMAC_LEN	20
+#define MPTCPOPT_THMAC_LEN	8
 
 /* MPTCP MP_CAPABLE flags */
 #define MPTCP_VERSION_MASK	(0x0F)
@@ -115,6 +117,7 @@ struct mptcp_sock {
 	u64		write_seq;
 	u64		ack_seq;
 	u32		token;
+	u16		dport;
 	struct list_head conn_list;
 	struct socket	*subflow; /* outgoing connect/listener/!mp_capable */
 	struct mptcp_pm_data	pm;
@@ -167,7 +170,9 @@ struct subflow_context {
 	u32	ssn_offset;
 	u16	map_data_len;
 	u16	request_mptcp : 1,  /* send MP_CAPABLE */
+		request_join : 1,   /* send MP_JOIN */
 		request_cksum : 1,
+		request_bkup : 1,
 		request_version : 4,
 		mp_capable : 1,     /* remote is MPTCP capable */
 		mp_join : 1,        /* remote is JOINing */
@@ -179,6 +184,8 @@ struct subflow_context {
 	u32	remote_nonce;
 	u64	thmac;
 	u32	local_nonce;
+	u32	remote_token;
+	u8	hmac[MPTCPOPT_HMAC_LEN];
 	u8	local_id;
 	u8	remote_id;
 
@@ -202,6 +209,8 @@ mptcp_subflow_tcp_socket(const struct subflow_context *subflow)
 }
 
 void subflow_init(void);
+int subflow_connect(struct sock *sk, struct sockaddr_in *local,
+		    struct sockaddr_in *remote, u8 remote_id);
 int subflow_create_socket(struct sock *sk, struct socket **new_sock);
 
 extern const struct inet_connection_sock_af_ops ipv4_specific;
@@ -215,10 +224,12 @@ void mptcp_finish_join(struct sock *sk);
 void token_init(void);
 void token_new_request(struct request_sock *req, const struct sk_buff *skb);
 int token_join_request(struct request_sock *req, const struct sk_buff *skb);
+int token_join_response(struct sock *sk);
 int token_join_valid(struct request_sock *req,
 		     struct tcp_options_received *rx_opt);
 void token_destroy_request(u32 token);
 void token_new_connect(struct sock *sk);
+void token_new_subflow(struct sock *sk);
 void token_new_accept(struct sock *sk);
 int token_new_join(struct sock *sk);
 void token_update_accept(struct sock *sk, struct sock *conn);
net/mptcp/subflow.c +60 -1
index 678fca99c23a..448c9e7f567a 100644
@@ -22,6 +22,9 @@ static int subflow_rebuild_header(struct sock *sk)
 	if (subflow->request_mptcp && !subflow->token) {
 		pr_debug("subflow=%p", sk);
 		token_new_connect(sk);
+	} else if (subflow->request_join && !subflow->local_nonce) {
+		pr_debug("subflow=%p", sk);
+		token_new_subflow(sk);
 	}
 
 	return inet_sk_rebuild_header(sk);
@@ -95,7 +98,10 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 
 	inet_sk_rx_dst_set(sk, skb);
 
-	if (subflow->conn && !subflow->conn_finished) {
+	if (!subflow->conn)
+		return;
+
+	if (subflow->mp_capable && !subflow->conn_finished) {
 		pr_debug("subflow=%p, remote_key=%llu", subflow_ctx(sk),
 			 subflow->remote_key);
 		mptcp_finish_connect(subflow->conn, subflow->mp_capable);
@@ -105,6 +111,17 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 			pr_debug("synack seq=%u", TCP_SKB_CB(skb)->seq);
 			subflow->ssn_offset = TCP_SKB_CB(skb)->seq;
 		}
+	} else if (subflow->mp_join && !subflow->conn_finished) {
+		pr_debug("subflow=%p, thmac=%llu, remote_nonce=%u",
+			 subflow_ctx(sk), subflow->thmac,
+			 subflow->remote_nonce);
+		if (token_join_response(sk)) {
+			subflow->mp_join = 0;
+			// @@ need to trigger RST
+		} else {
+			mptcp_finish_join(sk);
+			subflow->conn_finished = 1;
+		}
 	}
 }
 
@@ -195,6 +212,55 @@ static void subflow_data_ready(struct sock *sk)
 	}
 }
 
+int subflow_connect(struct sock *sk, struct sockaddr_in *local,
+		    struct sockaddr_in *remote, u8 remote_id)
+{
+	struct mptcp_sock *msk = mptcp_sk(sk);
+	struct subflow_context *subflow;
+	struct socket *sf;
+	u32 remote_token;
+	int err;
+
+	lock_sock(sk);
+	err = subflow_create_socket(sk, &sf);
+	if (err) {
+		release_sock(sk);
+		return err;
+	}
+
+	subflow = subflow_ctx(sf->sk);
+	subflow->remote_key = msk->remote_key;
+	subflow->local_key = msk->local_key;
+	subflow->token = msk->token;
+
+	sock_hold(sf->sk);
+	release_sock(sk);
+
+	err = kernel_bind(sf, (struct sockaddr *)local,
+			  sizeof(struct sockaddr_in));
+	if (err)
+		goto failed;
+
+	crypto_key_sha1(subflow->remote_key, &remote_token, NULL);
+	pr_debug("msk=%p remote_token=%u", msk, remote_token);
+	subflow->remote_token = remote_token;
+	subflow->remote_id = remote_id;
+	subflow->request_join = 1;
+	subflow->request_bkup = 1;
+
+	err = kernel_connect(sf, (struct sockaddr *)remote,
+			     sizeof(struct sockaddr_in), O_NONBLOCK);
+	if (err && err != -EINPROGRESS)
+		goto failed;
+
+	sock_put(sf->sk);
Quoted 9 lines

net/mptcp/subflow.c static void subflow_data_ready(struct sock *sk) · line 260

Quoted 4 lines
net/mptcp/subflow.c +4 -0
+	return err;
+
+failed:
+	sock_release(sf);

Should there be a sock_put(sf->sk) on this path too? Looks like sock_release() will free the subflow context but not the struct sock itself.

Yes. Since this is already merged I'll submit a separate patch.

Peter.

Quoted 94 lines

Mat

Quoted 87 lines
net/mptcp/subflow.c +3 -0
+	return err;
+}
+
 int subflow_create_socket(struct sock *sk, struct socket **new_sock)
 {
 	struct subflow_context *subflow;
net/mptcp/token.c +48 -0
index ef03ef19af98..d7cf40b40beb 100644
@@ -93,6 +93,28 @@ static void new_req_join(struct request_sock *req, struct sock *sk,
 		 subflow_req->thmac);
 }
 
+static int new_rsp_join(struct sock *sk)
+{
+	struct subflow_context *subflow = subflow_ctx(sk);
+	u8 hmac[MPTCPOPT_HMAC_LEN];
+	u64 thmac;
+
+	crypto_hmac_sha1(subflow->remote_key, subflow->local_key,
+			 subflow->remote_nonce, subflow->local_nonce,
+			 (u32 *)hmac);
+
+	thmac = get_unaligned_be64(hmac);
+	pr_debug("thmac=%llu", thmac);
+	if (thmac != subflow->thmac)
+		return -1;
+
+	crypto_hmac_sha1(subflow->local_key, subflow->remote_key,
+			 subflow->local_nonce, subflow->remote_nonce,
+			 (u32 *)subflow->hmac);
+
+	return 0;
+}
+
 static int new_join_valid(struct request_sock *req, struct sock *sk,
 			  struct tcp_options_received *rx_opt)
 {
@@ -210,6 +232,15 @@ int token_join_request(struct request_sock *req, const struct sk_buff *skb)
 	return -1;
 }
 
+/* validate received truncated hmac and create hmac for third ACK */
+int token_join_response(struct sock *sk)
+{
+	struct subflow_context *subflow = subflow_ctx(sk);
+
+	pr_debug("subflow=%p, token=%u", subflow, subflow->token);
+	return new_rsp_join(sk);
+}
+
 /* validate hmac received in third ACK */
 int token_join_valid(struct request_sock *req,
 		     struct tcp_options_received *rx_opt)
@@ -247,6 +278,23 @@ void token_new_connect(struct sock *sk)
 	spin_unlock_bh(&token_tree_lock);
 }
 
+/* take reference to connection and create nonce for secondary subflow */
+void token_new_subflow(struct sock *sk)
+{
+	struct subflow_context *subflow = subflow_ctx(sk);
+	struct sock *conn;
+
+	pr_debug("subflow=%p", subflow);
+
+	spin_lock_bh(&token_tree_lock);
+	conn = lookup_token(subflow->token);
+	if (conn)
+		sock_hold(conn);
+	spin_unlock_bh(&token_tree_lock);
+
+	get_random_bytes(&subflow->local_nonce, sizeof(u32));
+}
+
 void token_new_accept(struct sock *sk)
 {
 	struct subflow_context *subflow = subflow_ctx(sk);
Signature · 6 lines
2.17.2

_______________________________________________
mptcp mailing list
mptcp(a)lists.01.org
https://lists.01.org/mailman/listinfo/mptcp

--
Mat Martineau
Intel

author
#3 lore raw

Re: [MPTCP] [PATCH v3 01/10] Extend path manager interface

In reply to Matthieu Baerts

Hi Matthieu -

It seems a hunk went missing when this was split and squashed. The following piece of code from this patch does not appear in mptcp_incoming_options() in the current export branch as I would expect to.

+ if (msk && mp_opt->add_addr) {
+ if (mp_opt->family == MPTCP_ADDR_IPVERSION_4)
+ pm_add_addr(msk, &mp_opt->addr, mp_opt->addr_id);
+#if IS_ENABLED(CONFIG_IPV6)
+ else if (mp_opt->family == MPTCP_ADDR_IPVERSION_6)
+ pm_add_addr6(msk, &mp_opt->addr6, mp_opt->addr_id);
+#endif
+ mp_opt->add_addr = 0;
+ }
+

This causes the received ADD_ADDR option to not be passed up to the path manager. I'll send a patch.

Peter.

Quoted 20 lines · Matthieu Baerts

Hi Peter, Paolo, Mat,

Quoted 5 lines · Peter Krystad

A little more functionality and some comments clarifying which routines are for a path manager to call and which are callbacks.

squashto: Add path manager interface

Thank you for the patch and the reviews!

I had to split this patch in 2 parts:
- c8625a37c692: part 1: "squashed" in "mptcp: Add path manager interface"
- 4412b8e8e6c9: conflict in t/mptcp-Add-ADD_ADDR-handling
- 0757aec1b6f2: part 2: "squashed" in "mptcp: Add ADD_ADDR handling"
- b1477322bbb7..711ae3fdd44f: results

Cheers,
Matt

#4 lore raw

Re: [MPTCP] [PATCH v3 01/10] Extend path manager interface

Hi Peter,

Quoted 18 lines · Peter Krystad

Hi Matthieu -

It seems a hunk went missing when this was split and squashed. The following piece of code from this patch does not appear in mptcp_incoming_options() in the current export branch as I would expect to.

net/mptcp/options.c +10 -0
+       if (msk && mp_opt->add_addr) {
+               if (mp_opt->family == MPTCP_ADDR_IPVERSION_4)
+                       pm_add_addr(msk, &mp_opt->addr, mp_opt->addr_id);
+#if IS_ENABLED(CONFIG_IPV6)
+               else if (mp_opt->family == MPTCP_ADDR_IPVERSION_6)
+                       pm_add_addr6(msk, &mp_opt->addr6, mp_opt->addr_id);
+#endif
+               mp_opt->add_addr = 0;
+       }
+
Quoted 2 lines

This causes the received ADD_ADDR option to not be passed up to the path manager. I'll send a patch.

Oh sorry, I thought I carefully checked that everything was there. I will apply this missing part ASAP!

Cheers,
Matt

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