[PATCH mptcp-next v6 0/9] The infinite mapping support

[PATCH mptcp-next v6 0/9] The infinite mapping support

[Geliang Tang]

v6:
 - use allow_infinite_fallback instead of last_retrans_seq.
 - rename mptcp_is_data_contiguous to mptcp_allow_infinite_fallback.
 - rename last_fully_acked_dss_start_seq to last_ack_dss_start.

v5:
 - move last_retrans_seq from msk to mptcp_subflow_context

v4:
 - update patch 1 and patch 2

v3:
 - drop MPTCP_INFINITE_DONE flag
 - drop MAPPING_INFINITE
 - add mptcp_is_data_contiguous helper
 - add the fallback check
 - The u32 target testcase has not been completed yet.

v2:
 - add MPTCP_INFINITE_DONE flag
 - add MAPPING_INFINITE mapping status
 - add start_seq in the msk

v1:
 - add noncontiguous flag
 - add the mibs check
 - tag: export/20210904T080009

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/216

Geliang Tang (9):
  mptcp: don't send RST for single subflow
  mptcp: add the fallback check
  mptcp: track and update contiguous data status
  mptcp: add last_ack_dss_start in the msk
  mptcp: infinite mapping sending
  mptcp: infinite mapping receiving
  mptcp: add mib for infinite map sending
  selftests: mptcp: add infinite map mibs check
  DO-NOT-MERGE: mptcp: mp_fail test

 include/net/mptcp.h                           |  3 +-
 net/mptcp/mib.c                               |  1 +
 net/mptcp/mib.h                               |  1 +
 net/mptcp/options.c                           |  2 +-
 net/mptcp/pm.c                                |  5 ++
 net/mptcp/protocol.c                          | 34 +++++++++++
 net/mptcp/protocol.h                          | 23 +++++++-
 net/mptcp/subflow.c                           | 54 +++++++++---------
 .../testing/selftests/net/mptcp/mptcp_join.sh | 56 +++++++++++++++++++
 9 files changed, 149 insertions(+), 30 deletions(-)

-- 
2.31.1

[PATCH mptcp-next v6 1/9] mptcp: don't send RST for single subflow

[Geliang Tang]

When a bad checksum is detected and a single subflow is in use, don't
send RST + MP_FAIL, send data_ack + MP_FAIL instead.

Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/subflow.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 6172f380dfb7..92b45a7c997e 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1167,14 +1167,14 @@ static bool subflow_check_data_avail(struct sock *ssk)
 	/* RFC 8684 section 3.7. */
 	if (subflow->send_mp_fail) {
 		if (mptcp_has_another_subflow(ssk)) {
+			ssk->sk_err = EBADMSG;
+			tcp_set_state(ssk, TCP_CLOSE);
+			subflow->reset_transient = 0;
+			subflow->reset_reason = MPTCP_RST_EMIDDLEBOX;
+			tcp_send_active_reset(ssk, GFP_ATOMIC);
 			while ((skb = skb_peek(&ssk->sk_receive_queue)))
 				sk_eat_skb(ssk, skb);
 		}
-		ssk->sk_err = EBADMSG;
-		tcp_set_state(ssk, TCP_CLOSE);
-		subflow->reset_transient = 0;
-		subflow->reset_reason = MPTCP_RST_EMIDDLEBOX;
-		tcp_send_active_reset(ssk, GFP_ATOMIC);
 		WRITE_ONCE(subflow->data_avail, 0);
 		return true;
 	}
-- 
2.31.1

[PATCH mptcp-next v6 2/9] mptcp: add the fallback check

[Geliang Tang]

This patch added the fallback check in subflow_check_data_avail. Only do
the fallback when the msk isn't fallen back yet.

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/subflow.c | 45 ++++++++++++++++++++++++---------------------
 1 file changed, 24 insertions(+), 21 deletions(-)

diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 92b45a7c997e..87a9ffebcc42 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1164,35 +1164,38 @@ static bool subflow_check_data_avail(struct sock *ssk)
 	return false;
 
 fallback:
-	/* RFC 8684 section 3.7. */
-	if (subflow->send_mp_fail) {
-		if (mptcp_has_another_subflow(ssk)) {
+	if (!__mptcp_check_fallback(msk)) {
+		/* RFC 8684 section 3.7. */
+		if (subflow->send_mp_fail) {
+			if (mptcp_has_another_subflow(ssk)) {
+				ssk->sk_err = EBADMSG;
+				tcp_set_state(ssk, TCP_CLOSE);
+				subflow->reset_transient = 0;
+				subflow->reset_reason = MPTCP_RST_EMIDDLEBOX;
+				tcp_send_active_reset(ssk, GFP_ATOMIC);
+				while ((skb = skb_peek(&ssk->sk_receive_queue)))
+					sk_eat_skb(ssk, skb);
+			}
+			WRITE_ONCE(subflow->data_avail, 0);
+			return true;
+		}
+
+		if (subflow->mp_join || subflow->fully_established) {
+			/* fatal protocol error, close the socket.
+			 * subflow_error_report() will introduce the appropriate barriers
+			 */
 			ssk->sk_err = EBADMSG;
 			tcp_set_state(ssk, TCP_CLOSE);
 			subflow->reset_transient = 0;
-			subflow->reset_reason = MPTCP_RST_EMIDDLEBOX;
+			subflow->reset_reason = MPTCP_RST_EMPTCP;
 			tcp_send_active_reset(ssk, GFP_ATOMIC);
-			while ((skb = skb_peek(&ssk->sk_receive_queue)))
-				sk_eat_skb(ssk, skb);
+			WRITE_ONCE(subflow->data_avail, 0);
+			return false;
 		}
-		WRITE_ONCE(subflow->data_avail, 0);
-		return true;
-	}
 
-	if (subflow->mp_join || subflow->fully_established) {
-		/* fatal protocol error, close the socket.
-		 * subflow_error_report() will introduce the appropriate barriers
-		 */
-		ssk->sk_err = EBADMSG;
-		tcp_set_state(ssk, TCP_CLOSE);
-		subflow->reset_transient = 0;
-		subflow->reset_reason = MPTCP_RST_EMPTCP;
-		tcp_send_active_reset(ssk, GFP_ATOMIC);
-		WRITE_ONCE(subflow->data_avail, 0);
-		return false;
+		__mptcp_do_fallback(msk);
 	}
 
-	__mptcp_do_fallback(msk);
 	skb = skb_peek(&ssk->sk_receive_queue);
 	subflow->map_valid = 1;
 	subflow->map_seq = READ_ONCE(msk->ack_seq);
-- 
2.31.1

[PATCH mptcp-next v6 3/9] mptcp: track and update contiguous data status

[Geliang Tang]

This patch added a new member allow_infinite_fallback in mptcp_sock,
which gets initialized to 'true' when the connection begins and is set
to 'false' on any retransmit or successful MP_JOIN.

Rename the helper function mptcp_is_data_contiguous() to
mptcp_allow_infinite_fallback(). In it, only do infinite mapping fallback
if there is a single subflow AND there have been no retransmissions AND
there have never been any MP_JOINs.

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/protocol.c |  2 ++
 net/mptcp/protocol.h | 10 +++++++---
 net/mptcp/subflow.c  |  2 +-
 3 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index f8ad049d6941..48979cb82126 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -2464,6 +2464,7 @@ static void __mptcp_retrans(struct sock *sk)
 		dfrag->already_sent = max(dfrag->already_sent, info.sent);
 		tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
 			 info.size_goal);
+		WRITE_ONCE(msk->allow_infinite_fallback, false);
 	}
 
 	release_sock(ssk);
@@ -2542,6 +2543,7 @@ static int __mptcp_init_sock(struct sock *sk)
 	msk->first = NULL;
 	inet_csk(sk)->icsk_sync_mss = mptcp_sync_mss;
 	WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
+	WRITE_ONCE(msk->allow_infinite_fallback, true);
 	msk->recovery = false;
 
 	mptcp_pm_data_init(msk);
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index 7379ab580a7e..4807e486e762 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -249,6 +249,7 @@ struct mptcp_sock {
 	bool		rcv_fastclose;
 	bool		use_64bit_ack; /* Set when we received a 64-bit DSN */
 	bool		csum_enabled;
+	bool		allow_infinite_fallback;
 	spinlock_t	join_list_lock;
 	struct work_struct work;
 	struct sk_buff  *ooo_last_skb;
@@ -612,17 +613,20 @@ static inline void mptcp_subflow_tcp_fallback(struct sock *sk,
 	inet_csk(sk)->icsk_af_ops = ctx->icsk_af_ops;
 }
 
-static inline bool mptcp_has_another_subflow(struct sock *ssk)
+static inline bool mptcp_allow_infinite_fallback(struct sock *ssk)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk), *tmp;
 	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
 
 	mptcp_for_each_subflow(msk, tmp) {
+		if (tmp->mp_join)
+			return false;
+
 		if (tmp != subflow)
-			return true;
+			return false;
 	}
 
-	return false;
+	return READ_ONCE(msk->allow_infinite_fallback);
 }
 
 void __init mptcp_proto_init(void);
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 87a9ffebcc42..28ed7dc6e170 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1167,7 +1167,7 @@ static bool subflow_check_data_avail(struct sock *ssk)
 	if (!__mptcp_check_fallback(msk)) {
 		/* RFC 8684 section 3.7. */
 		if (subflow->send_mp_fail) {
-			if (mptcp_has_another_subflow(ssk)) {
+			if (!mptcp_allow_infinite_fallback(ssk)) {
 				ssk->sk_err = EBADMSG;
 				tcp_set_state(ssk, TCP_CLOSE);
 				subflow->reset_transient = 0;
-- 
2.31.1

[PATCH mptcp-next v6 4/9] mptcp: add last_ack_dss_start in the msk

[Geliang Tang]

This patch added a new member named last_ack_dss_start to the msk to
keep track of the beginning of the last fully-acked data segment. This
would be updated in __mptcp_clean_una.

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/protocol.c | 3 +++
 net/mptcp/protocol.h | 1 +
 2 files changed, 4 insertions(+)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 48979cb82126..334bbce69fcb 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1071,6 +1071,7 @@ static void __mptcp_clean_una(struct sock *sk)
 			WRITE_ONCE(msk->first_pending, mptcp_send_next(sk));
 		}
 
+		msk->last_ack_dss_start = dfrag->data_seq;
 		dfrag_clear(sk, dfrag);
 		cleaned = true;
 	}
@@ -2891,6 +2892,7 @@ struct sock *mptcp_sk_clone(const struct sock *sk,
 	msk->snd_una = msk->write_seq;
 	msk->wnd_end = msk->snd_nxt + req->rsk_rcv_wnd;
 	msk->setsockopt_seq = mptcp_sk(sk)->setsockopt_seq;
+	msk->last_ack_dss_start = subflow_req->idsn - 1;
 
 	if (mp_opt->suboptions & OPTIONS_MPTCP_MPC) {
 		msk->can_ack = true;
@@ -3147,6 +3149,7 @@ void mptcp_finish_connect(struct sock *ssk)
 	WRITE_ONCE(msk->rcv_wnd_sent, ack_seq);
 	WRITE_ONCE(msk->can_ack, 1);
 	WRITE_ONCE(msk->snd_una, msk->write_seq);
+	WRITE_ONCE(msk->last_ack_dss_start, subflow->idsn - 1);
 
 	mptcp_pm_new_connection(msk, ssk, 0);
 
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index 4807e486e762..7927acf53f06 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -224,6 +224,7 @@ struct mptcp_sock {
 	u64		remote_key;
 	u64		write_seq;
 	u64		snd_nxt;
+	u64		last_ack_dss_start;
 	u64		ack_seq;
 	u64		rcv_wnd_sent;
 	u64		rcv_data_fin_seq;
-- 
2.31.1

[PATCH mptcp-next v6 5/9] mptcp: infinite mapping sending

[Geliang Tang]

This patch added the infinite mapping sending logic.

Added a new flag send_infinite_map in struct mptcp_subflow_context. Set
it true when a single contiguous subflow is in use in
mptcp_pm_mp_fail_received.

In mptcp_sendmsg_frag, if this flag is true, call the new function
mptcp_update_infinite_map to set the infinite mapping.

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 include/net/mptcp.h  |  3 ++-
 net/mptcp/options.c  |  2 +-
 net/mptcp/pm.c       |  5 +++++
 net/mptcp/protocol.c | 19 +++++++++++++++++++
 net/mptcp/protocol.h | 12 ++++++++++++
 5 files changed, 39 insertions(+), 2 deletions(-)

diff --git a/include/net/mptcp.h b/include/net/mptcp.h
index f83fa48408b3..29e930540ea2 100644
--- a/include/net/mptcp.h
+++ b/include/net/mptcp.h
@@ -35,7 +35,8 @@ struct mptcp_ext {
 			frozen:1,
 			reset_transient:1;
 	u8		reset_reason:4,
-			csum_reqd:1;
+			csum_reqd:1,
+			infinite_map:1;
 };
 
 #define MPTCP_RM_IDS_MAX	8
diff --git a/net/mptcp/options.c b/net/mptcp/options.c
index 422f4acfb3e6..f4591421ed22 100644
--- a/net/mptcp/options.c
+++ b/net/mptcp/options.c
@@ -816,7 +816,7 @@ bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
 
 	opts->suboptions = 0;
 
-	if (unlikely(__mptcp_check_fallback(msk)))
+	if (unlikely(__mptcp_check_fallback(msk) && !mptcp_check_infinite_map(skb)))
 		return false;
 
 	if (unlikely(skb && TCP_SKB_CB(skb)->tcp_flags & TCPHDR_RST)) {
diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c
index 6ab386ff3294..5b99c0c9c17e 100644
--- a/net/mptcp/pm.c
+++ b/net/mptcp/pm.c
@@ -251,7 +251,12 @@ void mptcp_pm_mp_prio_received(struct sock *sk, u8 bkup)
 
 void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
 {
+	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
+
 	pr_debug("fail_seq=%llu", fail_seq);
+
+	if (mptcp_allow_infinite_fallback(sk))
+		subflow->send_infinite_map = 1;
 }
 
 /* path manager helpers */
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 334bbce69fcb..5b73493c75c5 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1274,6 +1274,23 @@ static void mptcp_update_data_checksum(struct sk_buff *skb, int added)
 	mpext->csum = csum_fold(csum_block_add(csum, skb_checksum(skb, offset, added, 0), offset));
 }
 
+static void mptcp_update_infinite_map(struct mptcp_sock *msk, struct sock *ssk,
+				      struct mptcp_ext *mpext)
+{
+	if (!mpext)
+		return;
+
+	mpext->infinite_map = 1;
+	mpext->data_seq = READ_ONCE(msk->last_ack_dss_start);
+	mpext->subflow_seq = 0;
+	mpext->data_len = 0;
+	mpext->csum = 0;
+
+	mptcp_subflow_ctx(ssk)->send_infinite_map = 0;
+	pr_fallback(msk);
+	__mptcp_do_fallback(msk);
+}
+
 static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 			      struct mptcp_data_frag *dfrag,
 			      struct mptcp_sendmsg_info *info)
@@ -1406,6 +1423,8 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 out:
 	if (READ_ONCE(msk->csum_enabled))
 		mptcp_update_data_checksum(skb, copy);
+	if (mptcp_subflow_ctx(ssk)->send_infinite_map)
+		mptcp_update_infinite_map(msk, ssk, mpext);
 	mptcp_subflow_ctx(ssk)->rel_write_seq += copy;
 	return copy;
 }
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index 7927acf53f06..9da4dd3b2e2d 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -433,6 +433,7 @@ struct mptcp_subflow_context {
 		backup : 1,
 		send_mp_prio : 1,
 		send_mp_fail : 1,
+		send_infinite_map : 1,
 		rx_eof : 1,
 		can_ack : 1,        /* only after processing the remote a key */
 		disposable : 1,	    /* ctx can be free at ulp release time */
@@ -872,6 +873,17 @@ static inline void mptcp_do_fallback(struct sock *sk)
 
 #define pr_fallback(a) pr_debug("%s:fallback to TCP (msk=%p)", __func__, a)
 
+static inline bool mptcp_check_infinite_map(struct sk_buff *skb)
+{
+	struct mptcp_ext *mpext;
+
+	mpext = skb ? mptcp_get_ext(skb) : NULL;
+	if (mpext && mpext->infinite_map)
+		return true;
+
+	return false;
+}
+
 static inline bool subflow_simultaneous_connect(struct sock *sk)
 {
 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
-- 
2.31.1

[PATCH mptcp-next v6 6/9] mptcp: infinite mapping receiving

[Geliang Tang]

This patch added the infinite mapping receiving logic. When the infinite
mapping is received, set the map_data_len of the subflow to 0.

In subflow_check_data_avail, only reset the subflow when the map_data_len
of the subflow is non-zero.

Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/subflow.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 28ed7dc6e170..101bfa277a1c 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -968,6 +968,7 @@ static enum mapping_status get_mapping_status(struct sock *ssk,
 	data_len = mpext->data_len;
 	if (data_len == 0) {
 		MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_INFINITEMAPRX);
+		subflow->map_data_len = 0;
 		return MAPPING_INVALID;
 	}
 
@@ -1180,7 +1181,7 @@ static bool subflow_check_data_avail(struct sock *ssk)
 			return true;
 		}
 
-		if (subflow->mp_join || subflow->fully_established) {
+		if ((subflow->mp_join || subflow->fully_established) && subflow->map_data_len) {
 			/* fatal protocol error, close the socket.
 			 * subflow_error_report() will introduce the appropriate barriers
 			 */
-- 
2.31.1

[PATCH mptcp-next v6 7/9] mptcp: add mib for infinite map sending

[Geliang Tang]

This patch added a new mib named MPTCP_MIB_INFINITEMAPTX, increase it
when a infinite mapping has been sent out.

Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/mib.c      | 1 +
 net/mptcp/mib.h      | 1 +
 net/mptcp/protocol.c | 1 +
 3 files changed, 3 insertions(+)

diff --git a/net/mptcp/mib.c b/net/mptcp/mib.c
index b21ff9be04c6..ab55afdcae22 100644
--- a/net/mptcp/mib.c
+++ b/net/mptcp/mib.c
@@ -24,6 +24,7 @@ static const struct snmp_mib mptcp_snmp_list[] = {
 	SNMP_MIB_ITEM("MPJoinAckRx", MPTCP_MIB_JOINACKRX),
 	SNMP_MIB_ITEM("MPJoinAckHMacFailure", MPTCP_MIB_JOINACKMAC),
 	SNMP_MIB_ITEM("DSSNotMatching", MPTCP_MIB_DSSNOMATCH),
+	SNMP_MIB_ITEM("InfiniteMapTx", MPTCP_MIB_INFINITEMAPTX),
 	SNMP_MIB_ITEM("InfiniteMapRx", MPTCP_MIB_INFINITEMAPRX),
 	SNMP_MIB_ITEM("DSSNoMatchTCP", MPTCP_MIB_DSSTCPMISMATCH),
 	SNMP_MIB_ITEM("DataCsumErr", MPTCP_MIB_DATACSUMERR),
diff --git a/net/mptcp/mib.h b/net/mptcp/mib.h
index ecd3d8b117e0..7901f1338d15 100644
--- a/net/mptcp/mib.h
+++ b/net/mptcp/mib.h
@@ -17,6 +17,7 @@ enum linux_mptcp_mib_field {
 	MPTCP_MIB_JOINACKRX,		/* Received an ACK + MP_JOIN */
 	MPTCP_MIB_JOINACKMAC,		/* HMAC was wrong on ACK + MP_JOIN */
 	MPTCP_MIB_DSSNOMATCH,		/* Received a new mapping that did not match the previous one */
+	MPTCP_MIB_INFINITEMAPTX,	/* Sent an infinite mapping */
 	MPTCP_MIB_INFINITEMAPRX,	/* Received an infinite mapping */
 	MPTCP_MIB_DSSTCPMISMATCH,	/* DSS-mapping did not map with TCP's sequence numbers */
 	MPTCP_MIB_DATACSUMERR,		/* The data checksum fail */
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 5b73493c75c5..5122c4ea4350 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1286,6 +1286,7 @@ static void mptcp_update_infinite_map(struct mptcp_sock *msk, struct sock *ssk,
 	mpext->data_len = 0;
 	mpext->csum = 0;
 
+	MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_INFINITEMAPTX);
 	mptcp_subflow_ctx(ssk)->send_infinite_map = 0;
 	pr_fallback(msk);
 	__mptcp_do_fallback(msk);
-- 
2.31.1

[PATCH mptcp-next v6 8/9] selftests: mptcp: add infinite map mibs check

[Geliang Tang]

This patch added a function chk_infi_nr to check the mibs for the
infinite mapping.

Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 .../testing/selftests/net/mptcp/mptcp_join.sh | 38 +++++++++++++++++++
 1 file changed, 38 insertions(+)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index 255793c5ac4f..fe0c8f3164a7 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -615,6 +615,43 @@ chk_fail_nr()
 	fi
 }
 
+chk_infi_nr()
+{
+	local mp_infi_nr_tx=$1
+	local mp_infi_nr_rx=$2
+	local count
+	local dump_stats
+
+	printf "%-39s %s" " " "itx"
+	count=`ip netns exec $ns1 nstat -as | grep InfiniteMapTx | awk '{print $2}'`
+	[ -z "$count" ] && count=0
+	if [ "$count" != "$mp_infi_nr_tx" ]; then
+		echo "[fail] got $count infinite map[s] TX expected $mp_infi_nr_tx"
+		ret=1
+		dump_stats=1
+	else
+		echo -n "[ ok ]"
+	fi
+
+	echo -n " - irx   "
+	count=`ip netns exec $ns2 nstat -as | grep InfiniteMapRx | awk '{print $2}'`
+	[ -z "$count" ] && count=0
+	if [ "$count" != "$mp_infi_nr_rx" ]; then
+		echo "[fail] got $count infinite map[s] RX expected $mp_infi_nr_rx"
+		ret=1
+		dump_stats=1
+	else
+		echo "[ ok ]"
+	fi
+
+	if [ "${dump_stats}" = 1 ]; then
+		echo Server ns stats
+		ip netns exec $ns1 nstat -as | grep MPTcp
+		echo Client ns stats
+		ip netns exec $ns2 nstat -as | grep MPTcp
+	fi
+}
+
 chk_join_nr()
 {
 	local msg="$1"
@@ -665,6 +702,7 @@ chk_join_nr()
 	if [ $checksum -eq 1 ]; then
 		chk_csum_nr
 		chk_fail_nr 0 0
+		chk_infi_nr 0 0
 	fi
 }
 
-- 
2.31.1

[PATCH mptcp-next v6 9/9] DO-NOT-MERGE: mptcp: mp_fail test

[Geliang Tang]

./mptcp_join.sh -Cf

Signed-off-by: Geliang Tang <geliangtang@gmail.com>
---
 net/mptcp/protocol.c                           |  9 +++++++++
 .../testing/selftests/net/mptcp/mptcp_join.sh  | 18 ++++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 5122c4ea4350..d3b9f6894eb1 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1292,6 +1292,8 @@ static void mptcp_update_infinite_map(struct mptcp_sock *msk, struct sock *ssk,
 	__mptcp_do_fallback(msk);
 }
 
+static int j;
+
 static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 			      struct mptcp_data_frag *dfrag,
 			      struct mptcp_sendmsg_info *info)
@@ -1426,6 +1428,13 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 		mptcp_update_data_checksum(skb, copy);
 	if (mptcp_subflow_ctx(ssk)->send_infinite_map)
 		mptcp_update_infinite_map(msk, ssk, mpext);
+
+	pr_debug("%s j=%d", __func__, j++);
+	if (j == 20)
+		skb->data_len = 1;
+	if (j > 40)
+		j = 0;
+
 	mptcp_subflow_ctx(ssk)->rel_write_seq += copy;
 	return copy;
 }
diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index fe0c8f3164a7..38663f6373b8 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -977,6 +977,24 @@ chk_link_usage()
 
 subflows_tests()
 {
+	# 1 subflow
+	reset
+	ip netns exec $ns1 ./pm_nl_ctl limits 0 2
+	ip netns exec $ns2 ./pm_nl_ctl limits 0 2
+	run_tests $ns1 $ns2 10.0.1.1 0 0 0 slow
+	chk_join_nr "1 subflow" 0 0 0
+
+	exit
+
+	# multiple subflows
+	reset
+	ip netns exec $ns1 ./pm_nl_ctl limits 0 2
+	ip netns exec $ns2 ./pm_nl_ctl limits 0 2
+	ip netns exec $ns2 ./pm_nl_ctl add 10.0.3.2 flags subflow
+	ip netns exec $ns2 ./pm_nl_ctl add 10.0.2.2 flags subflow
+	run_tests $ns1 $ns2 10.0.1.1 0 0 0 slow
+	chk_join_nr "multiple subflows" 2 2 2
+
 	reset
 	run_tests $ns1 $ns2 10.0.1.1
 	chk_join_nr "no JOIN" "0" "0" "0"
-- 
2.31.1

Re: [PATCH mptcp-next v6 3/9] mptcp: track and update contiguous data status

[Mat Martineau]

On Wed, 29 Sep 2021, Geliang Tang wrote:

> This patch added a new member allow_infinite_fallback in mptcp_sock,
> which gets initialized to 'true' when the connection begins and is set
> to 'false' on any retransmit or successful MP_JOIN.
>
> Rename the helper function mptcp_is_data_contiguous() to
> mptcp_allow_infinite_fallback(). In it, only do infinite mapping fallback
> if there is a single subflow AND there have been no retransmissions AND
> there have never been any MP_JOINs.
>
> Suggested-by: Paolo Abeni <pabeni@redhat.com>
> Signed-off-by: Geliang Tang <geliangtang@gmail.com>
> ---
> net/mptcp/protocol.c |  2 ++
> net/mptcp/protocol.h | 10 +++++++---
> net/mptcp/subflow.c  |  2 +-
> 3 files changed, 10 insertions(+), 4 deletions(-)
>
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index f8ad049d6941..48979cb82126 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -2464,6 +2464,7 @@ static void __mptcp_retrans(struct sock *sk)
> 		dfrag->already_sent = max(dfrag->already_sent, info.sent);
> 		tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle,
> 			 info.size_goal);
> +		WRITE_ONCE(msk->allow_infinite_fallback, false);
> 	}
>
> 	release_sock(ssk);
> @@ -2542,6 +2543,7 @@ static int __mptcp_init_sock(struct sock *sk)
> 	msk->first = NULL;
> 	inet_csk(sk)->icsk_sync_mss = mptcp_sync_mss;
> 	WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
> +	WRITE_ONCE(msk->allow_infinite_fallback, true);
> 	msk->recovery = false;
>
> 	mptcp_pm_data_init(msk);
> diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
> index 7379ab580a7e..4807e486e762 100644
> --- a/net/mptcp/protocol.h
> +++ b/net/mptcp/protocol.h
> @@ -249,6 +249,7 @@ struct mptcp_sock {
> 	bool		rcv_fastclose;
> 	bool		use_64bit_ack; /* Set when we received a 64-bit DSN */
> 	bool		csum_enabled;
> +	bool		allow_infinite_fallback;
> 	spinlock_t	join_list_lock;
> 	struct work_struct work;
> 	struct sk_buff  *ooo_last_skb;
> @@ -612,17 +613,20 @@ static inline void mptcp_subflow_tcp_fallback(struct sock *sk,
> 	inet_csk(sk)->icsk_af_ops = ctx->icsk_af_ops;
> }
>
> -static inline bool mptcp_has_another_subflow(struct sock *ssk)
> +static inline bool mptcp_allow_infinite_fallback(struct sock *ssk)
> {
> 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk), *tmp;
> 	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
>
> 	mptcp_for_each_subflow(msk, tmp) {
> +		if (tmp->mp_join)
> +			return false;
> +
> 		if (tmp != subflow)
> -			return true;
> +			return false;

Sorry I missed your patch-1 v5 reply yesterday! I think it's better for 
now to implement the simpler check discussed in the meeting last week and 
in the v5 review, which is to set allow_infinite_fallback = false when a 
join happens (in mptcp_finish_join() and __mptcp_subflow_connect()). Then 
there's no need to iterate over the subflows here.

We don't have a good answer yet for how to guarantee a subflow and the 
MPTCP-level stream are in a correct state to fallback *after* 
retransmissions or joins, so I think it's important to get this 
single-subflow infinite mapping working and tests implemented first. There 
must be absolute certainty that fallback will not corrupt the application 
data stream.

Mat

> 	}
>
> -	return false;
> +	return READ_ONCE(msk->allow_infinite_fallback);
> }
>
> void __init mptcp_proto_init(void);
> diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
> index 87a9ffebcc42..28ed7dc6e170 100644
> --- a/net/mptcp/subflow.c
> +++ b/net/mptcp/subflow.c
> @@ -1167,7 +1167,7 @@ static bool subflow_check_data_avail(struct sock *ssk)
> 	if (!__mptcp_check_fallback(msk)) {
> 		/* RFC 8684 section 3.7. */
> 		if (subflow->send_mp_fail) {
> -			if (mptcp_has_another_subflow(ssk)) {
> +			if (!mptcp_allow_infinite_fallback(ssk)) {
> 				ssk->sk_err = EBADMSG;
> 				tcp_set_state(ssk, TCP_CLOSE);
> 				subflow->reset_transient = 0;
> -- 
> 2.31.1
>
>
>

--
Mat Martineau
Intel

Re: [PATCH mptcp-next v6 4/9] mptcp: add last_ack_dss_start in the msk

[Mat Martineau]

On Wed, 29 Sep 2021, Geliang Tang wrote:

> This patch added a new member named last_ack_dss_start to the msk to
> keep track of the beginning of the last fully-acked data segment. This
> would be updated in __mptcp_clean_una.
>

Can drop this patch, see patch 5 reply for more details.


Mat


> Suggested-by: Paolo Abeni <pabeni@redhat.com>
> Signed-off-by: Geliang Tang <geliangtang@gmail.com>
> ---
> net/mptcp/protocol.c | 3 +++
> net/mptcp/protocol.h | 1 +
> 2 files changed, 4 insertions(+)
>
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index 48979cb82126..334bbce69fcb 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -1071,6 +1071,7 @@ static void __mptcp_clean_una(struct sock *sk)
> 			WRITE_ONCE(msk->first_pending, mptcp_send_next(sk));
> 		}
>
> +		msk->last_ack_dss_start = dfrag->data_seq;
> 		dfrag_clear(sk, dfrag);
> 		cleaned = true;
> 	}
> @@ -2891,6 +2892,7 @@ struct sock *mptcp_sk_clone(const struct sock *sk,
> 	msk->snd_una = msk->write_seq;
> 	msk->wnd_end = msk->snd_nxt + req->rsk_rcv_wnd;
> 	msk->setsockopt_seq = mptcp_sk(sk)->setsockopt_seq;
> +	msk->last_ack_dss_start = subflow_req->idsn - 1;
>
> 	if (mp_opt->suboptions & OPTIONS_MPTCP_MPC) {
> 		msk->can_ack = true;
> @@ -3147,6 +3149,7 @@ void mptcp_finish_connect(struct sock *ssk)
> 	WRITE_ONCE(msk->rcv_wnd_sent, ack_seq);
> 	WRITE_ONCE(msk->can_ack, 1);
> 	WRITE_ONCE(msk->snd_una, msk->write_seq);
> +	WRITE_ONCE(msk->last_ack_dss_start, subflow->idsn - 1);
>
> 	mptcp_pm_new_connection(msk, ssk, 0);
>
> diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
> index 4807e486e762..7927acf53f06 100644
> --- a/net/mptcp/protocol.h
> +++ b/net/mptcp/protocol.h
> @@ -224,6 +224,7 @@ struct mptcp_sock {
> 	u64		remote_key;
> 	u64		write_seq;
> 	u64		snd_nxt;
> +	u64		last_ack_dss_start;
> 	u64		ack_seq;
> 	u64		rcv_wnd_sent;
> 	u64		rcv_data_fin_seq;
> -- 
> 2.31.1
>
>
>

--
Mat Martineau
Intel

Re: [PATCH mptcp-next v6 5/9] mptcp: infinite mapping sending

[Mat Martineau]

On Wed, 29 Sep 2021, Geliang Tang wrote:

> This patch added the infinite mapping sending logic.
>
> Added a new flag send_infinite_map in struct mptcp_subflow_context. Set
> it true when a single contiguous subflow is in use in
> mptcp_pm_mp_fail_received.
>
> In mptcp_sendmsg_frag, if this flag is true, call the new function
> mptcp_update_infinite_map to set the infinite mapping.
>
> Suggested-by: Paolo Abeni <pabeni@redhat.com>
> Signed-off-by: Geliang Tang <geliangtang@gmail.com>
> ---
> include/net/mptcp.h  |  3 ++-
> net/mptcp/options.c  |  2 +-
> net/mptcp/pm.c       |  5 +++++
> net/mptcp/protocol.c | 19 +++++++++++++++++++
> net/mptcp/protocol.h | 12 ++++++++++++
> 5 files changed, 39 insertions(+), 2 deletions(-)
>
> diff --git a/include/net/mptcp.h b/include/net/mptcp.h
> index f83fa48408b3..29e930540ea2 100644
> --- a/include/net/mptcp.h
> +++ b/include/net/mptcp.h
> @@ -35,7 +35,8 @@ struct mptcp_ext {
> 			frozen:1,
> 			reset_transient:1;
> 	u8		reset_reason:4,
> -			csum_reqd:1;
> +			csum_reqd:1,
> +			infinite_map:1;
> };
>
> #define MPTCP_RM_IDS_MAX	8
> diff --git a/net/mptcp/options.c b/net/mptcp/options.c
> index 422f4acfb3e6..f4591421ed22 100644
> --- a/net/mptcp/options.c
> +++ b/net/mptcp/options.c
> @@ -816,7 +816,7 @@ bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
>
> 	opts->suboptions = 0;
>
> -	if (unlikely(__mptcp_check_fallback(msk)))
> +	if (unlikely(__mptcp_check_fallback(msk) && !mptcp_check_infinite_map(skb)))
> 		return false;
>
> 	if (unlikely(skb && TCP_SKB_CB(skb)->tcp_flags & TCPHDR_RST)) {
> diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c
> index 6ab386ff3294..5b99c0c9c17e 100644
> --- a/net/mptcp/pm.c
> +++ b/net/mptcp/pm.c
> @@ -251,7 +251,12 @@ void mptcp_pm_mp_prio_received(struct sock *sk, u8 bkup)
>
> void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
> {
> +	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
> +
> 	pr_debug("fail_seq=%llu", fail_seq);
> +
> +	if (mptcp_allow_infinite_fallback(sk))
> +		subflow->send_infinite_map = 1;
> }
>
> /* path manager helpers */
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index 334bbce69fcb..5b73493c75c5 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -1274,6 +1274,23 @@ static void mptcp_update_data_checksum(struct sk_buff *skb, int added)
> 	mpext->csum = csum_fold(csum_block_add(csum, skb_checksum(skb, offset, added, 0), offset));
> }
>
> +static void mptcp_update_infinite_map(struct mptcp_sock *msk, struct sock *ssk,
> +				      struct mptcp_ext *mpext)
> +{
> +	if (!mpext)
> +		return;
> +
> +	mpext->infinite_map = 1;
> +	mpext->data_seq = READ_ONCE(msk->last_ack_dss_start);
> +	mpext->subflow_seq = 0;

To match up with Christoph's reply from a couple of days ago (if I'm 
reading it right), I think you just delete the above 2 lines so data_seq 
and subflow_seq are unmodified from the values they were set to in 
mptcp_sendmsg_frag().

Mat


> +	mpext->data_len = 0;
> +	mpext->csum = 0;
> +
> +	mptcp_subflow_ctx(ssk)->send_infinite_map = 0;
> +	pr_fallback(msk);
> +	__mptcp_do_fallback(msk);
> +}
> +
> static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
> 			      struct mptcp_data_frag *dfrag,
> 			      struct mptcp_sendmsg_info *info)
> @@ -1406,6 +1423,8 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
> out:
> 	if (READ_ONCE(msk->csum_enabled))
> 		mptcp_update_data_checksum(skb, copy);
> +	if (mptcp_subflow_ctx(ssk)->send_infinite_map)
> +		mptcp_update_infinite_map(msk, ssk, mpext);
> 	mptcp_subflow_ctx(ssk)->rel_write_seq += copy;
> 	return copy;
> }
> diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
> index 7927acf53f06..9da4dd3b2e2d 100644
> --- a/net/mptcp/protocol.h
> +++ b/net/mptcp/protocol.h
> @@ -433,6 +433,7 @@ struct mptcp_subflow_context {
> 		backup : 1,
> 		send_mp_prio : 1,
> 		send_mp_fail : 1,
> +		send_infinite_map : 1,
> 		rx_eof : 1,
> 		can_ack : 1,        /* only after processing the remote a key */
> 		disposable : 1,	    /* ctx can be free at ulp release time */
> @@ -872,6 +873,17 @@ static inline void mptcp_do_fallback(struct sock *sk)
>
> #define pr_fallback(a) pr_debug("%s:fallback to TCP (msk=%p)", __func__, a)
>
> +static inline bool mptcp_check_infinite_map(struct sk_buff *skb)
> +{
> +	struct mptcp_ext *mpext;
> +
> +	mpext = skb ? mptcp_get_ext(skb) : NULL;
> +	if (mpext && mpext->infinite_map)
> +		return true;
> +
> +	return false;
> +}
> +
> static inline bool subflow_simultaneous_connect(struct sock *sk)
> {
> 	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
> -- 
> 2.31.1
>
>
>

--
Mat Martineau
Intel