[PATCH v5 mptcp-next 0/6] mptcp: add some more diag info

[PATCH v5 mptcp-next 0/6] mptcp: add some more diag info

[Paolo Abeni]

This is a follow up to the topic discussed in recent pubblic mtg.

Introduces unique id for accurate subflow stats tracking and
aggregate mptcp counters, plus some minimal self-tests.

The tests themself do not take in account support for running on
older kernel.

This is on top of "mptcp: a bunch of data race fixes".

There should be non trivial conflicts with:

"mptcp: use get_retrans wrapper".

v4 -> v5:
 - changed again binary layout for MPTCP_FULL_INFO structs (Matttbe)
 - fixed 32bit build issue
 - reordered the patches, less controversial first

v3 -> v4:
 - change binary layout for MPTCP_FULL_INFO structs (Florian)

v2 -> v3:
 - address Matttbe comments on patch 1, 2 and 5, see the indivdual
   patches changelog for the details

v1 -> v2:
 - introduce MPTCP_FULL_INFO instead of overloading a tcp_info field
 - add related self-tests
 - fix a couple of subflow_id initialization bugs

Paolo Abeni (6):
  mptcp: move snd_una update earlier for fallback socket.
  mptcp: track some aggregate data counters.
  selftests: mptcp: explicitly tests aggregate counters
  mptcp: add subflow unique id
  mptcp: introduce MPTCP_FULL_INFO getsockopt
  selftests: mptcp: add MPTCP_FULL_INFO testcase

 include/uapi/linux/mptcp.h                    |  29 ++++
 net/mptcp/options.c                           |  14 +-
 net/mptcp/protocol.c                          |  24 ++-
 net/mptcp/protocol.h                          |   9 +-
 net/mptcp/sockopt.c                           | 155 +++++++++++++++++-
 net/mptcp/subflow.c                           |   2 +
 .../selftests/net/mptcp/mptcp_sockopt.c       | 118 ++++++++++++-
 7 files changed, 332 insertions(+), 19 deletions(-)

-- 
2.40.1

[PATCH v5 mptcp-next 1/6] mptcp: move snd_una update earlier for fallback socket.

[Paolo Abeni]

That will avoid an unneeded conditional in both the fast-path
and in the fallback case and will simplify a bit the next
patch.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
 net/mptcp/options.c  | 6 ++++++
 net/mptcp/protocol.c | 6 ------
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/net/mptcp/options.c b/net/mptcp/options.c
index 8a8083207be4..4bdcd2b326bd 100644
--- a/net/mptcp/options.c
+++ b/net/mptcp/options.c
@@ -1119,6 +1119,12 @@ bool mptcp_incoming_options(struct sock *sk, struct sk_buff *skb)
 		mptcp_data_lock(subflow->conn);
 		if (sk_stream_memory_free(sk))
 			__mptcp_check_push(subflow->conn, sk);
+
+		/* on fallback we just need to ignore the msk-level snd_una, as
+		 * this is really plain TCP
+		 */
+		msk->snd_una = READ_ONCE(msk->snd_nxt);
+
 		__mptcp_data_acked(subflow->conn);
 		mptcp_data_unlock(subflow->conn);
 		return true;
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 763f709fd5f5..7ecd2f250909 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1004,12 +1004,6 @@ static void __mptcp_clean_una(struct sock *sk)
 	struct mptcp_data_frag *dtmp, *dfrag;
 	u64 snd_una;
 
-	/* on fallback we just need to ignore snd_una, as this is really
-	 * plain TCP
-	 */
-	if (__mptcp_check_fallback(msk))
-		msk->snd_una = READ_ONCE(msk->snd_nxt);
-
 	snd_una = msk->snd_una;
 	list_for_each_entry_safe(dfrag, dtmp, &msk->rtx_queue, list) {
 		if (after64(dfrag->data_seq + dfrag->data_len, snd_una))
-- 
2.40.1

[PATCH v5 mptcp-next 2/6] mptcp: track some aggregate data counters.

[Paolo Abeni]

Currently there are no data transfer counters accounting for all
the subflows used by a given MPTCP socket. The user-space can compute
such figures aggregating the subflow info, but that is inaccurate
if any subflow is closed before the MPTCP socket itself.

Add the new counters in the MPTCP socket itself and expose them
via the existing diag and sockopt. While touching mptcp_diag_fill_info(),
acquire the relevant locks before fetching the msk data, to ensure
better data consistency

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
 include/uapi/linux/mptcp.h |  5 +++++
 net/mptcp/options.c        | 10 ++++++++--
 net/mptcp/protocol.c       | 12 +++++++++++-
 net/mptcp/protocol.h       |  4 ++++
 net/mptcp/sockopt.c        | 22 +++++++++++++++++-----
 5 files changed, 45 insertions(+), 8 deletions(-)

diff --git a/include/uapi/linux/mptcp.h b/include/uapi/linux/mptcp.h
index 32af2d278cb4..a124be6ebbba 100644
--- a/include/uapi/linux/mptcp.h
+++ b/include/uapi/linux/mptcp.h
@@ -123,6 +123,11 @@ struct mptcp_info {
 	__u8	mptcpi_local_addr_used;
 	__u8	mptcpi_local_addr_max;
 	__u8	mptcpi_csum_enabled;
+	__u32	mptcpi_retransmits;
+	__u64	mptcpi_bytes_retrans;
+	__u64	mptcpi_bytes_sent;
+	__u64	mptcpi_bytes_received;
+	__u64	mptcpi_bytes_acked;
 };
 
 /*
diff --git a/net/mptcp/options.c b/net/mptcp/options.c
index 4bdcd2b326bd..c254accb14de 100644
--- a/net/mptcp/options.c
+++ b/net/mptcp/options.c
@@ -1026,6 +1026,12 @@ u64 __mptcp_expand_seq(u64 old_seq, u64 cur_seq)
 	return cur_seq;
 }
 
+static void __mptcp_snd_una_update(struct mptcp_sock *msk, u64 new_snd_una)
+{
+	msk->bytes_acked += new_snd_una - msk->snd_una;
+	msk->snd_una = new_snd_una;
+}
+
 static void ack_update_msk(struct mptcp_sock *msk,
 			   struct sock *ssk,
 			   struct mptcp_options_received *mp_opt)
@@ -1057,7 +1063,7 @@ static void ack_update_msk(struct mptcp_sock *msk,
 		__mptcp_check_push(sk, ssk);
 
 	if (after64(new_snd_una, old_snd_una)) {
-		msk->snd_una = new_snd_una;
+		__mptcp_snd_una_update(msk, new_snd_una);
 		__mptcp_data_acked(sk);
 	}
 	mptcp_data_unlock(sk);
@@ -1123,7 +1129,7 @@ bool mptcp_incoming_options(struct sock *sk, struct sk_buff *skb)
 		/* on fallback we just need to ignore the msk-level snd_una, as
 		 * this is really plain TCP
 		 */
-		msk->snd_una = READ_ONCE(msk->snd_nxt);
+		__mptcp_snd_una_update(msk, READ_ONCE(msk->snd_nxt));
 
 		__mptcp_data_acked(subflow->conn);
 		mptcp_data_unlock(subflow->conn);
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 7ecd2f250909..79b3059ca3ef 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -377,6 +377,7 @@ static bool __mptcp_move_skb(struct mptcp_sock *msk, struct sock *ssk,
 
 	if (MPTCP_SKB_CB(skb)->map_seq == msk->ack_seq) {
 		/* in sequence */
+		msk->bytes_received += copy_len;
 		WRITE_ONCE(msk->ack_seq, msk->ack_seq + copy_len);
 		tail = skb_peek_tail(&sk->sk_receive_queue);
 		if (tail && mptcp_try_coalesce(sk, tail, skb))
@@ -760,6 +761,7 @@ static bool __mptcp_ofo_queue(struct mptcp_sock *msk)
 			MPTCP_SKB_CB(skb)->map_seq += delta;
 			__skb_queue_tail(&sk->sk_receive_queue, skb);
 		}
+		msk->bytes_received += end_seq - msk->ack_seq;
 		msk->ack_seq = end_seq;
 		moved = true;
 	}
@@ -1511,8 +1513,10 @@ static void mptcp_update_post_push(struct mptcp_sock *msk,
 	 * that has been handed to the subflow for transmission
 	 * and skip update in case it was old dfrag.
 	 */
-	if (likely(after64(snd_nxt_new, msk->snd_nxt)))
+	if (likely(after64(snd_nxt_new, msk->snd_nxt))) {
+		msk->bytes_sent += snd_nxt_new - msk->snd_nxt;
 		msk->snd_nxt = snd_nxt_new;
+	}
 }
 
 void mptcp_check_and_set_pending(struct sock *sk)
@@ -2637,6 +2641,8 @@ static void __mptcp_retrans(struct sock *sk)
 			msk->last_snd = ssk;
 		}
 	}
+
+	msk->bytes_retrans += len;
 	dfrag->already_sent = max(dfrag->already_sent, len);
 
 reset_timer:
@@ -3150,6 +3156,10 @@ static int mptcp_disconnect(struct sock *sk, int flags)
 	WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
 	mptcp_pm_data_reset(msk);
 	mptcp_ca_reset(sk);
+	msk->bytes_acked = 0;
+	msk->bytes_received = 0;
+	msk->bytes_sent = 0;
+	msk->bytes_retrans = 0;
 
 	WRITE_ONCE(sk->sk_shutdown, 0);
 	sk_error_report(sk);
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index bd3771c7d79d..ec733955f18e 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -261,10 +261,13 @@ struct mptcp_sock {
 	u64		local_key;
 	u64		remote_key;
 	u64		write_seq;
+	u64		bytes_sent;
 	u64		snd_nxt;
+	u64		bytes_received;
 	u64		ack_seq;
 	atomic64_t	rcv_wnd_sent;
 	u64		rcv_data_fin_seq;
+	u64		bytes_retrans;
 	int		rmem_fwd_alloc;
 	struct sock	*last_snd;
 	int		snd_burst;
@@ -273,6 +276,7 @@ struct mptcp_sock {
 						 * recovery related fields are under data_lock
 						 * protection
 						 */
+	u64		bytes_acked;
 	u64		snd_una;
 	u64		wnd_end;
 	unsigned long	timer_ival;
diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index d4258869ac48..770279e0a598 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -888,7 +888,9 @@ static int mptcp_getsockopt_first_sf_only(struct mptcp_sock *msk, int level, int
 
 void mptcp_diag_fill_info(struct mptcp_sock *msk, struct mptcp_info *info)
 {
+	struct sock *sk = (struct sock *)msk;
 	u32 flags = 0;
+	bool slow;
 
 	memset(info, 0, sizeof(*info));
 
@@ -914,11 +916,21 @@ void mptcp_diag_fill_info(struct mptcp_sock *msk, struct mptcp_info *info)
 	if (READ_ONCE(msk->can_ack))
 		flags |= MPTCP_INFO_FLAG_REMOTE_KEY_RECEIVED;
 	info->mptcpi_flags = flags;
-	info->mptcpi_token = READ_ONCE(msk->token);
-	info->mptcpi_write_seq = READ_ONCE(msk->write_seq);
-	info->mptcpi_snd_una = READ_ONCE(msk->snd_una);
-	info->mptcpi_rcv_nxt = READ_ONCE(msk->ack_seq);
-	info->mptcpi_csum_enabled = READ_ONCE(msk->csum_enabled);
+	mptcp_data_lock(sk);
+	info->mptcpi_snd_una = msk->snd_una;
+	info->mptcpi_rcv_nxt = msk->ack_seq;
+	info->mptcpi_bytes_acked = msk->bytes_acked;
+	mptcp_data_unlock(sk);
+
+	slow = lock_sock_fast(sk);
+	info->mptcpi_csum_enabled = msk->csum_enabled;
+	info->mptcpi_token = msk->token;
+	info->mptcpi_write_seq = msk->write_seq;
+	info->mptcpi_retransmits = inet_csk(sk)->icsk_retransmits;
+	info->mptcpi_bytes_sent = msk->bytes_sent;
+	info->mptcpi_bytes_received = msk->bytes_received;
+	info->mptcpi_bytes_retrans = msk->bytes_retrans;
+	unlock_sock_fast(sk, slow);
 }
 EXPORT_SYMBOL_GPL(mptcp_diag_fill_info);
 
-- 
2.40.1

[PATCH v5 mptcp-next 3/6] selftests: mptcp: explicitly tests aggregate counters

[Paolo Abeni]

Update the existing sockopt test-case to do some basic checks
on the newly added counters.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
v2 -> v3:
 - be more kind with older kernel (Matttbe)
---
 .../selftests/net/mptcp/mptcp_sockopt.c       | 27 ++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_sockopt.c b/tools/testing/selftests/net/mptcp/mptcp_sockopt.c
index ae61f39556ca..ff8fcdfccf76 100644
--- a/tools/testing/selftests/net/mptcp/mptcp_sockopt.c
+++ b/tools/testing/selftests/net/mptcp/mptcp_sockopt.c
@@ -51,6 +51,11 @@ struct mptcp_info {
 	__u8	mptcpi_local_addr_used;
 	__u8	mptcpi_local_addr_max;
 	__u8	mptcpi_csum_enabled;
+	__u32	mptcpi_retransmits;
+	__u64	mptcpi_bytes_retrans;
+	__u64	mptcpi_bytes_sent;
+	__u64	mptcpi_bytes_received;
+	__u64	mptcpi_bytes_acked;
 };
 
 struct mptcp_subflow_data {
@@ -83,8 +88,10 @@ struct mptcp_subflow_addrs {
 
 struct so_state {
 	struct mptcp_info mi;
+	struct mptcp_info last_sample;
 	uint64_t mptcpi_rcv_delta;
 	uint64_t tcpi_rcv_delta;
+	bool pkt_stats_avail;
 };
 
 static void die_perror(const char *msg)
@@ -318,8 +325,9 @@ static void do_getsockopt_mptcp_info(struct so_state *s, int fd, size_t w)
 	if (ret < 0)
 		die_perror("getsockopt MPTCP_INFO");
 
-	assert(olen == sizeof(i));
+	s->pkt_stats_avail = olen >= sizeof(i);
 
+	s->last_sample = i;
 	if (s->mi.mptcpi_write_seq == 0)
 		s->mi = i;
 
@@ -556,6 +564,23 @@ static void process_one_client(int fd, int pipefd)
 	do_getsockopts(&s, fd, ret, ret2);
 	if (s.mptcpi_rcv_delta != (uint64_t)ret + 1)
 		xerror("mptcpi_rcv_delta %" PRIu64 ", expect %" PRIu64, s.mptcpi_rcv_delta, ret + 1, s.mptcpi_rcv_delta - ret);
+
+	/* be nice when running on top of older kernel */
+	if (s.pkt_stats_avail) {
+		if (s.last_sample.mptcpi_bytes_sent != ret2)
+			xerror("mptcpi_bytes_sent %" PRIu64 ", expect %" PRIu64,
+			       s.last_sample.mptcpi_bytes_sent, ret2,
+			       s.last_sample.mptcpi_bytes_sent - ret2);
+		if (s.last_sample.mptcpi_bytes_received != ret)
+			xerror("mptcpi_bytes_received %" PRIu64 ", expect %" PRIu64,
+			       s.last_sample.mptcpi_bytes_received, ret,
+			       s.last_sample.mptcpi_bytes_received - ret);
+		if (s.last_sample.mptcpi_bytes_acked != ret)
+			xerror("mptcpi_bytes_acked %" PRIu64 ", expect %" PRIu64,
+			       s.last_sample.mptcpi_bytes_acked, ret2,
+			       s.last_sample.mptcpi_bytes_acked - ret2);
+	}
+
 	close(fd);
 }
 
-- 
2.40.1

[PATCH v5 mptcp-next 4/6] mptcp: add subflow unique id

[Paolo Abeni]

The user-space need to preperly account the data received/sent by
individual subflows. When additional subflows are created and/or
closed during the MPTCP socket lifetime, the information currently
exposed via MPTCP_TCPINFO are not enough: subflows are identifed only
by the sequential position inside the info dumps, and that will change
with the above mentioned events.

To solve the above problem, this patch introduces a new subflow identifier
that is unique inside the given mptcp socket scope.

The initial subflow get the id 1 and the other subflows get incremental
values at join time.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
v2 -> v3:
 - fix msk subflow_id init (Matttbe)

v1 -> v2:
 - properly set subflow_id for the first passive subflow and active subflows, too
 - drop the tcpi_fackets overload
---
 net/mptcp/protocol.c | 6 ++++++
 net/mptcp/protocol.h | 5 ++++-
 net/mptcp/subflow.c  | 2 ++
 3 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 79b3059ca3ef..895eda3d79e8 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -96,6 +96,7 @@ static int __mptcp_socket_create(struct mptcp_sock *msk)
 	list_add(&subflow->node, &msk->conn_list);
 	sock_hold(ssock->sk);
 	subflow->request_mptcp = 1;
+	subflow->subflow_id = msk->subflow_id++;
 
 	/* This is the first subflow, always with id 0 */
 	subflow->local_id_valid = 1;
@@ -847,6 +848,7 @@ static bool __mptcp_finish_join(struct mptcp_sock *msk, struct sock *ssk)
 	if (sk->sk_socket && !ssk->sk_socket)
 		mptcp_sock_graft(ssk, sk->sk_socket);
 
+	mptcp_subflow_ctx(ssk)->subflow_id = msk->subflow_id++;
 	mptcp_sockopt_sync_locked(msk, ssk);
 	mptcp_subflow_joined(msk, ssk);
 	return true;
@@ -2775,6 +2777,7 @@ static int __mptcp_init_sock(struct sock *sk)
 	WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
 	WRITE_ONCE(msk->allow_infinite_fallback, true);
 	msk->recovery = false;
+	msk->subflow_id = 1;
 
 	mptcp_pm_data_init(msk);
 
@@ -3210,6 +3213,9 @@ struct sock *mptcp_sk_clone_init(const struct sock *sk,
 	msk->setsockopt_seq = mptcp_sk(sk)->setsockopt_seq;
 	mptcp_init_sched(msk, mptcp_sk(sk)->sched);
 
+	/* passive msk is created after the first/MPC subflow */
+	msk->subflow_id = 2;
+
 	sock_reset_flag(nsk, SOCK_RCU_FREE);
 	security_inet_csk_clone(nsk, req);
 
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index ec733955f18e..b4b5c5004f34 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -323,7 +323,8 @@ struct mptcp_sock {
 		u64	rtt_us; /* last maximum rtt of subflows */
 	} rcvq_space;
 
-	u32 setsockopt_seq;
+	u32		subflow_id;
+	u32		setsockopt_seq;
 	char		ca_name[TCP_CA_NAME_MAX];
 	struct mptcp_sock	*dl_next;
 };
@@ -505,6 +506,8 @@ struct mptcp_subflow_context {
 	u8	reset_reason:4;
 	u8	stale_count;
 
+	u32	subflow_id;
+
 	long	delegated_status;
 	unsigned long	fail_tout;
 
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 63ac4dc621d4..c7001a23550a 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -819,6 +819,7 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 			if (!ctx->conn)
 				goto fallback;
 
+			ctx->subflow_id = 1;
 			owner = mptcp_sk(ctx->conn);
 			mptcp_pm_new_connection(owner, child, 1);
 
@@ -1574,6 +1575,7 @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc,
 	subflow->remote_id = remote_id;
 	subflow->request_join = 1;
 	subflow->request_bkup = !!(flags & MPTCP_PM_ADDR_FLAG_BACKUP);
+	subflow->subflow_id = msk->subflow_id++;
 	mptcp_info2sockaddr(remote, &addr, ssk->sk_family);
 
 	sock_hold(ssk);
-- 
2.40.1

[PATCH v5 mptcp-next 5/6] mptcp: introduce MPTCP_FULL_INFO getsockopt

[Paolo Abeni]

Some user-space applications want to monitor the subflows utilization.

Dumping the per subflow tcp_info is not enough, as the PM could close
and re-create the subflows under-the-hood, fooling the accounting.
Even checking the src/dst addresses used by each subflow could not
be enough, because new subflows could re-use the same address/port of
the just closed one.

This patch introduces a new socket option, allow dumping all the relevant
information all-at-once (everything, everywhere...), in a consistent manner.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
v4 -> v5:
 - full_info struct re-design (Florian)
 - fix build issue on 32 bit hosts

v3 -> v4:
 - full_info struct re-design (Florian)

v2 -> v3:
 - added missing changelog (oops)
---
 include/uapi/linux/mptcp.h |  24 +++++++
 net/mptcp/sockopt.c        | 133 ++++++++++++++++++++++++++++++++++++-
 2 files changed, 155 insertions(+), 2 deletions(-)

diff --git a/include/uapi/linux/mptcp.h b/include/uapi/linux/mptcp.h
index a124be6ebbba..ee9c49f949a2 100644
--- a/include/uapi/linux/mptcp.h
+++ b/include/uapi/linux/mptcp.h
@@ -249,9 +249,33 @@ struct mptcp_subflow_addrs {
 	};
 };
 
+struct mptcp_subflow_info {
+	__u32				id;
+	struct mptcp_subflow_addrs	addrs;
+};
+
+struct mptcp_full_info {
+	__u32		size_tcpinfo_kernel;	/* must be 0, set by kernel */
+	__u32		size_tcpinfo_user;
+	__u32		size_sfinfo_kernel;	/* must be 0, set by kernel */
+	__u32		size_sfinfo_user;
+	__u32		num_subflows;		/* must be 0, set by kernel (real subflow count) */
+	__u32		size_arrays_user;	/* max subflows that userspace is interested in;
+						 * the buffers at subflow_info/tcp_info
+						 * are respectively at least:
+						 *  size_arrays * size_sfinfo_user
+						 *  size_arrays * size_tcpinfo_user
+						 * bytes wide
+						 */
+	__aligned_u64		subflow_info;
+	__aligned_u64		tcp_info;
+	struct mptcp_info	mptcp_info;
+};
+
 /* MPTCP socket options */
 #define MPTCP_INFO		1
 #define MPTCP_TCPINFO		2
 #define MPTCP_SUBFLOW_ADDRS	3
+#define MPTCP_FULL_INFO		4
 
 #endif /* _UAPI_MPTCP_H */
diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 770279e0a598..b6f6fa1eb53b 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -14,7 +14,8 @@
 #include <net/mptcp.h>
 #include "protocol.h"
 
-#define MIN_INFO_OPTLEN_SIZE	16
+#define MIN_INFO_OPTLEN_SIZE		16
+#define MIN_FULL_INFO_OPTLEN_SIZE	40
 
 static struct sock *__mptcp_tcp_fallback(struct mptcp_sock *msk)
 {
@@ -977,7 +978,8 @@ static int mptcp_put_subflow_data(struct mptcp_subflow_data *sfd,
 }
 
 static int mptcp_get_subflow_data(struct mptcp_subflow_data *sfd,
-				  char __user *optval, int __user *optlen)
+				  char __user *optval,
+				  int __user *optlen)
 {
 	int len, copylen;
 
@@ -1158,6 +1160,131 @@ static int mptcp_getsockopt_subflow_addrs(struct mptcp_sock *msk, char __user *o
 	return 0;
 }
 
+static int mptcp_get_full_info(struct mptcp_full_info *mfi,
+			       char __user *optval,
+			       int __user *optlen)
+{
+	int len;
+
+	BUILD_BUG_ON(offsetof(struct mptcp_full_info, mptcp_info) !=
+		     MIN_FULL_INFO_OPTLEN_SIZE);
+
+	if (get_user(len, optlen))
+		return -EFAULT;
+
+	if (len < MIN_FULL_INFO_OPTLEN_SIZE)
+		return -EINVAL;
+
+	memset(mfi, 0, sizeof(*mfi));
+	if (copy_from_user(mfi, optval, MIN_FULL_INFO_OPTLEN_SIZE))
+		return -EFAULT;
+
+	if (mfi->size_tcpinfo_kernel ||
+	    mfi->size_sfinfo_kernel ||
+	    mfi->num_subflows)
+		return -EINVAL;
+
+	if (mfi->size_sfinfo_user > INT_MAX ||
+	    mfi->size_tcpinfo_user > INT_MAX)
+		return -EINVAL;
+
+	return len - MIN_FULL_INFO_OPTLEN_SIZE;
+}
+
+static int mptcp_put_full_info(struct mptcp_full_info *mfi,
+			       char __user *optval,
+			       u32 copied,
+			       int __user *optlen)
+{
+	copied += MIN_FULL_INFO_OPTLEN_SIZE;
+	if (put_user(copied, optlen))
+		return -EFAULT;
+
+	if (copy_to_user(optval, mfi, copied))
+		return -EFAULT;
+	return 0;
+}
+
+static int mptcp_getsockopt_full_info(struct mptcp_sock *msk, char __user *optval,
+				      int __user *optlen)
+{
+	struct mptcp_subflow_context *subflow;
+	struct sock *sk = (struct sock *)msk;
+	unsigned int sfcount = 0, copied = 0;
+	void __user *tcpinfoptr, *sfinfoptr;
+	struct mptcp_full_info mfi;
+	int len;
+
+	len = mptcp_get_full_info(&mfi, optval, optlen);
+	if (len < 0)
+		return len;
+
+	/* don't bother filling the mptcp info if there is not enough
+	 * user-space-provided storage
+	 */
+	if (len > 0) {
+		char __user *infoptr;
+		int mptcp_info_len;
+
+		infoptr = optval + MIN_FULL_INFO_OPTLEN_SIZE;
+		mptcp_info_len = min_t(unsigned int, len, sizeof(struct mptcp_info));
+		mptcp_diag_fill_info(msk, &mfi.mptcp_info);
+
+		copied += mptcp_info_len;
+	}
+
+	mfi.size_tcpinfo_kernel = sizeof(struct tcp_info);
+	mfi.size_tcpinfo_user = min_t(unsigned int, mfi.size_tcpinfo_user,
+				      sizeof(struct tcp_info));
+	sfinfoptr = u64_to_user_ptr(mfi.subflow_info);
+	mfi.size_sfinfo_kernel = sizeof(struct mptcp_subflow_info);
+	mfi.size_sfinfo_user = min_t(unsigned int, mfi.size_sfinfo_user,
+				     sizeof(struct mptcp_subflow_info));
+	tcpinfoptr = u64_to_user_ptr(mfi.tcp_info);
+
+	lock_sock(sk);
+	mptcp_for_each_subflow(msk, subflow) {
+		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
+		struct mptcp_subflow_info sfinfo;
+		struct tcp_info tcp_info;
+
+		if (sfcount++ >= mfi.size_arrays_user)
+			continue;
+
+		/* fetch addr/tcp_info only if the user space buffers
+		 * are wide enough
+		 */
+		memset(&sfinfo, 0, sizeof(sfinfo));
+		sfinfo.id = subflow->subflow_id;
+		if (mfi.size_sfinfo_user >
+		    offsetof(struct mptcp_subflow_info, addrs))
+			mptcp_get_sub_addrs(ssk, &sfinfo.addrs);
+		if (copy_to_user(sfinfoptr, &sfinfo, mfi.size_sfinfo_user))
+			goto fail_release;
+
+		if (mfi.size_tcpinfo_user) {
+			tcp_get_info(ssk, &tcp_info);
+			if (copy_to_user(tcpinfoptr, &tcp_info,
+					 mfi.size_tcpinfo_user))
+				goto fail_release;
+		}
+
+		tcpinfoptr += mfi.size_tcpinfo_user;
+		sfinfoptr += mfi.size_sfinfo_user;
+	}
+	release_sock(sk);
+
+	mfi.num_subflows = sfcount;
+	if (mptcp_put_full_info(&mfi, optval, copied, optlen))
+		return -EFAULT;
+
+	return 0;
+
+fail_release:
+	release_sock(sk);
+	return -EFAULT;
+}
+
 static int mptcp_put_int_option(struct mptcp_sock *msk, char __user *optval,
 				int __user *optlen, int val)
 {
@@ -1231,6 +1358,8 @@ static int mptcp_getsockopt_sol_mptcp(struct mptcp_sock *msk, int optname,
 	switch (optname) {
 	case MPTCP_INFO:
 		return mptcp_getsockopt_info(msk, optval, optlen);
+	case MPTCP_FULL_INFO:
+		return mptcp_getsockopt_full_info(msk, optval, optlen);
 	case MPTCP_TCPINFO:
 		return mptcp_getsockopt_tcpinfo(msk, optval, optlen);
 	case MPTCP_SUBFLOW_ADDRS:
-- 
2.40.1

[PATCH v5 mptcp-next 6/6] selftests: mptcp: add MPTCP_FULL_INFO testcase

[Paolo Abeni]

Add a testcase explicitly triggering the newly introduce
MPTCP_FULL_INFO getsockopt.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
 .../selftests/net/mptcp/mptcp_sockopt.c       | 91 ++++++++++++++++++-
 1 file changed, 89 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_sockopt.c b/tools/testing/selftests/net/mptcp/mptcp_sockopt.c
index ff8fcdfccf76..00f3659dd1eb 100644
--- a/tools/testing/selftests/net/mptcp/mptcp_sockopt.c
+++ b/tools/testing/selftests/net/mptcp/mptcp_sockopt.c
@@ -86,9 +86,38 @@ struct mptcp_subflow_addrs {
 #define MPTCP_SUBFLOW_ADDRS	3
 #endif
 
+#ifndef MPTCP_FULL_INFO
+struct mptcp_subflow_info {
+	__u32				id;
+	struct mptcp_subflow_addrs	addrs;
+};
+
+struct mptcp_full_info {
+	__u32		size_tcpinfo_kernel;	/* must be 0, set by kernel */
+	__u32		size_tcpinfo_user;
+	__u32		size_sfinfo_kernel;	/* must be 0, set by kernel */
+	__u32		size_sfinfo_user;
+	__u32		num_subflows;		/* must be 0, set by kernel (real subflow count) */
+	__u32		size_arrays_user;	/* max subflows that userspace is interested in;
+						 * the buffers at subflow_info/tcp_info
+						 * are respectively at least:
+						 *  size_arrays * size_sfinfo_user
+						 *  size_arrays * size_tcpinfo_user
+						 * bytes wide
+						 */
+	__aligned_u64		subflow_info;
+	__aligned_u64		tcp_info;
+	struct mptcp_info	mptcp_info;
+};
+
+#define MPTCP_FULL_INFO		4
+#endif
+
 struct so_state {
 	struct mptcp_info mi;
 	struct mptcp_info last_sample;
+	struct tcp_info tcp_info;
+	struct mptcp_subflow_addrs addrs;
 	uint64_t mptcpi_rcv_delta;
 	uint64_t tcpi_rcv_delta;
 	bool pkt_stats_avail;
@@ -365,6 +394,8 @@ static void do_getsockopt_tcp_info(struct so_state *s, int fd, size_t r, size_t
 		olen -= sizeof(struct mptcp_subflow_data);
 		assert(olen == sizeof(struct tcp_info));
 
+		s->tcp_info = ti.ti[0];
+
 		if (ti.ti[0].tcpi_bytes_sent == w &&
 		    ti.ti[0].tcpi_bytes_received == r)
 			goto done;
@@ -386,7 +417,7 @@ static void do_getsockopt_tcp_info(struct so_state *s, int fd, size_t r, size_t
 	do_getsockopt_bogus_sf_data(fd, MPTCP_TCPINFO);
 }
 
-static void do_getsockopt_subflow_addrs(int fd)
+static void do_getsockopt_subflow_addrs(struct so_state *s, int fd)
 {
 	struct sockaddr_storage remote, local;
 	socklen_t olen, rlen, llen;
@@ -433,6 +464,7 @@ static void do_getsockopt_subflow_addrs(int fd)
 
 	assert(memcmp(&local, &addrs.addr[0].ss_local, sizeof(local)) == 0);
 	assert(memcmp(&remote, &addrs.addr[0].ss_remote, sizeof(remote)) == 0);
+	s->addrs = addrs.addr[0];
 
 	memset(&addrs, 0, sizeof(addrs));
 
@@ -453,13 +485,68 @@ static void do_getsockopt_subflow_addrs(int fd)
 	do_getsockopt_bogus_sf_data(fd, MPTCP_SUBFLOW_ADDRS);
 }
 
+static void do_getsockopt_mptcp_full_info(struct so_state *s, int fd)
+{
+	size_t data_size = sizeof(struct mptcp_full_info);
+	struct mptcp_subflow_info sfinfo[2];
+	struct tcp_info tcp_info[2];
+	struct mptcp_full_info mfi;
+	socklen_t olen;
+	int ret;
+
+	memset(&mfi, 0, data_size);
+	memset(tcp_info, 0, sizeof(tcp_info));
+	memset(sfinfo, 0, sizeof(sfinfo));
+
+	mfi.size_tcpinfo_user = sizeof(struct tcp_info);
+	mfi.size_sfinfo_user = sizeof(struct mptcp_subflow_info);
+	mfi.size_arrays_user = 2;
+	mfi.subflow_info = (unsigned long) &sfinfo[0];
+	mfi.tcp_info = (unsigned long) &tcp_info[0];
+	olen = data_size;
+
+	ret = getsockopt(fd, SOL_MPTCP, MPTCP_FULL_INFO, &mfi, &olen);
+	if (ret < 0) {
+		if (errno == EOPNOTSUPP) {
+			fprintf(stderr, "\tMPTCP_FULL_INFO test skipped due to lack of kernel support\n");
+			return;
+		}
+		xerror("getsockopt MPTCP_FULL_INFO");
+	}
+
+	assert(olen <= data_size);
+	assert(mfi.size_tcpinfo_user == mfi.size_tcpinfo_kernel);
+	assert(mfi.size_tcpinfo_user == sizeof(struct tcp_info));
+	assert(mfi.size_sfinfo_user == mfi.size_sfinfo_kernel);
+	assert(mfi.size_sfinfo_user == sizeof(struct mptcp_subflow_info));
+	assert(mfi.num_subflows == 1);
+
+	/* Tolerate future extension to mptcp_info struct and running newer
+	 * test on top of older kernel.
+	 * Anyway any kernel supporting MPTCP_FULL_INFO must at least include
+	 * the following in mptcp_info.
+	 */
+	assert(olen > (socklen_t)__builtin_offsetof(struct mptcp_full_info, tcp_info));
+	assert(mfi.mptcp_info.mptcpi_subflows == 0);
+	assert(mfi.mptcp_info.mptcpi_bytes_sent == s->last_sample.mptcpi_bytes_sent);
+	assert(mfi.mptcp_info.mptcpi_bytes_received == s->last_sample.mptcpi_bytes_received);
+
+	assert(sfinfo[0].id == 1);
+	assert(tcp_info[0].tcpi_bytes_sent == s->tcp_info.tcpi_bytes_sent);
+	assert(tcp_info[0].tcpi_bytes_received == s->tcp_info.tcpi_bytes_received);
+	assert(!memcmp(&sfinfo->addrs, &s->addrs, sizeof(struct mptcp_subflow_addrs)));
+}
+
 static void do_getsockopts(struct so_state *s, int fd, size_t r, size_t w)
 {
 	do_getsockopt_mptcp_info(s, fd, w);
 
 	do_getsockopt_tcp_info(s, fd, r, w);
 
-	do_getsockopt_subflow_addrs(fd);
+	do_getsockopt_subflow_addrs(s, fd);
+
+	if (r)
+		do_getsockopt_mptcp_full_info(s, fd);
 }
 
 static void connect_one_server(int fd, int pipefd)
-- 
2.40.1

Re: selftests: mptcp: add MPTCP_FULL_INFO testcase: Build Failure

[MPTCP CI]

Hi Paolo,

Thank you for your modifications, that's great!

But sadly, our CI spotted some issues with it when trying to build it.

You can find more details there:

  https://patchwork.kernel.org/project/mptcp/patch/803034fc4f174c8180befe1e859675f99784874e.1684962909.git.pabeni@redhat.com/
  https://github.com/multipath-tcp/mptcp_net-next/actions/runs/5073733505

Status: failure
Initiator: MPTCPimporter
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/d241aa733adf

Feel free to reply to this email if you cannot access logs, if you need
some support to fix the error, if this doesn't seem to be caused by your
modifications or if the error is a false positive one.

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (Tessares)

Re: selftests: mptcp: add MPTCP_FULL_INFO testcase: Tests Results

[MPTCP CI]

Hi Paolo,

Thank you for your modifications, that's great!

Our CI did some validations and here is its report:

- KVM Validation: normal (except selftest_mptcp_join):
  - Unstable: 1 failed test(s): packetdrill_fastopen 🔴:
  - Task: https://cirrus-ci.com/task/5986916420550656
  - Summary: https://api.cirrus-ci.com/v1/artifact/task/5986916420550656/summary/summary.txt

- KVM Validation: normal (only selftest_mptcp_join):
  - Success! ✅:
  - Task: https://cirrus-ci.com/task/5423966467129344
  - Summary: https://api.cirrus-ci.com/v1/artifact/task/5423966467129344/summary/summary.txt

- KVM Validation: debug (only selftest_mptcp_join):
  - Success! ✅:
  - Task: https://cirrus-ci.com/task/4720279025352704
  - Summary: https://api.cirrus-ci.com/v1/artifact/task/4720279025352704/summary/summary.txt

- KVM Validation: debug (except selftest_mptcp_join):
  - Unstable: 4 failed test(s): packetdrill_fastopen packetdrill_mp_capable packetdrill_syscalls selftest_diag 🔴:
  - Task: https://cirrus-ci.com/task/6549866373971968
  - Summary: https://api.cirrus-ci.com/v1/artifact/task/6549866373971968/summary/summary.txt

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/d241aa733adf


If there are some issues, you can reproduce them using the same environment as
the one used by the CI thanks to a docker image, e.g.:

    $ cd [kernel source code]
    $ docker run -v "${PWD}:${PWD}:rw" -w "${PWD}" --privileged --rm -it \
        --pull always mptcp/mptcp-upstream-virtme-docker:latest \
        auto-debug

For more details:

    https://github.com/multipath-tcp/mptcp-upstream-virtme-docker


Please note that despite all the efforts that have been already done to have a
stable tests suite when executed on a public CI like here, it is possible some
reported issues are not due to your modifications. Still, do not hesitate to
help us improve that ;-)

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (Tessares)

Re: [PATCH v5 mptcp-next 5/6] mptcp: introduce MPTCP_FULL_INFO getsockopt

[kernel test robot]

Hi Paolo,

kernel test robot noticed the following build warnings:

[auto build test WARNING on mptcp/export]
[cannot apply to mptcp/export-net linus/master v6.4-rc3 next-20230524]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Paolo-Abeni/mptcp-move-snd_una-update-earlier-for-fallback-socket/20230525-053110
base:   https://github.com/multipath-tcp/mptcp_net-next.git export
patch link:    https://lore.kernel.org/r/400326ed4b59e6c9878fc4a406c2bf406876ee72.1684962909.git.pabeni%40redhat.com
patch subject: [PATCH v5 mptcp-next 5/6] mptcp: introduce MPTCP_FULL_INFO getsockopt
config: ia64-allyesconfig (https://download.01.org/0day-ci/archive/20230525/202305250816.Ie1tjg5a-lkp@intel.com/config)
compiler: ia64-linux-gcc (GCC) 12.1.0
reproduce (this is a W=1 build):
        mkdir -p ~/bin
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/intel-lab-lkp/linux/commit/8317ccb72cec68bc55db2258c67c166180f07712
        git remote add linux-review https://github.com/intel-lab-lkp/linux
        git fetch --no-tags linux-review Paolo-Abeni/mptcp-move-snd_una-update-earlier-for-fallback-socket/20230525-053110
        git checkout 8317ccb72cec68bc55db2258c67c166180f07712
        # save the config file
        mkdir build_dir && cp config build_dir/.config
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 ~/bin/make.cross W=1 O=build_dir ARCH=ia64 olddefconfig
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 ~/bin/make.cross W=1 O=build_dir ARCH=ia64 SHELL=/bin/bash net/mptcp/

If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202305250816.Ie1tjg5a-lkp@intel.com/

All warnings (new ones prefixed by >>):

   net/mptcp/sockopt.c: In function 'mptcp_getsockopt_full_info':
>> net/mptcp/sockopt.c:1226:30: warning: variable 'infoptr' set but not used [-Wunused-but-set-variable]
    1226 |                 char __user *infoptr;
         |                              ^~~~~~~


vim +/infoptr +1226 net/mptcp/sockopt.c

  1207	
  1208	static int mptcp_getsockopt_full_info(struct mptcp_sock *msk, char __user *optval,
  1209					      int __user *optlen)
  1210	{
  1211		struct mptcp_subflow_context *subflow;
  1212		struct sock *sk = (struct sock *)msk;
  1213		unsigned int sfcount = 0, copied = 0;
  1214		void __user *tcpinfoptr, *sfinfoptr;
  1215		struct mptcp_full_info mfi;
  1216		int len;
  1217	
  1218		len = mptcp_get_full_info(&mfi, optval, optlen);
  1219		if (len < 0)
  1220			return len;
  1221	
  1222		/* don't bother filling the mptcp info if there is not enough
  1223		 * user-space-provided storage
  1224		 */
  1225		if (len > 0) {
> 1226			char __user *infoptr;
  1227			int mptcp_info_len;
  1228	
  1229			infoptr = optval + MIN_FULL_INFO_OPTLEN_SIZE;
  1230			mptcp_info_len = min_t(unsigned int, len, sizeof(struct mptcp_info));
  1231			mptcp_diag_fill_info(msk, &mfi.mptcp_info);
  1232	
  1233			copied += mptcp_info_len;
  1234		}
  1235	
  1236		mfi.size_tcpinfo_kernel = sizeof(struct tcp_info);
  1237		mfi.size_tcpinfo_user = min_t(unsigned int, mfi.size_tcpinfo_user,
  1238					      sizeof(struct tcp_info));
  1239		sfinfoptr = u64_to_user_ptr(mfi.subflow_info);
  1240		mfi.size_sfinfo_kernel = sizeof(struct mptcp_subflow_info);
  1241		mfi.size_sfinfo_user = min_t(unsigned int, mfi.size_sfinfo_user,
  1242					     sizeof(struct mptcp_subflow_info));
  1243		tcpinfoptr = u64_to_user_ptr(mfi.tcp_info);
  1244	
  1245		lock_sock(sk);
  1246		mptcp_for_each_subflow(msk, subflow) {
  1247			struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
  1248			struct mptcp_subflow_info sfinfo;
  1249			struct tcp_info tcp_info;
  1250	
  1251			if (sfcount++ >= mfi.size_arrays_user)
  1252				continue;
  1253	
  1254			/* fetch addr/tcp_info only if the user space buffers
  1255			 * are wide enough
  1256			 */
  1257			memset(&sfinfo, 0, sizeof(sfinfo));
  1258			sfinfo.id = subflow->subflow_id;
  1259			if (mfi.size_sfinfo_user >
  1260			    offsetof(struct mptcp_subflow_info, addrs))
  1261				mptcp_get_sub_addrs(ssk, &sfinfo.addrs);
  1262			if (copy_to_user(sfinfoptr, &sfinfo, mfi.size_sfinfo_user))
  1263				goto fail_release;
  1264	
  1265			if (mfi.size_tcpinfo_user) {
  1266				tcp_get_info(ssk, &tcp_info);
  1267				if (copy_to_user(tcpinfoptr, &tcp_info,
  1268						 mfi.size_tcpinfo_user))
  1269					goto fail_release;
  1270			}
  1271	
  1272			tcpinfoptr += mfi.size_tcpinfo_user;
  1273			sfinfoptr += mfi.size_sfinfo_user;
  1274		}
  1275		release_sock(sk);
  1276	
  1277		mfi.num_subflows = sfcount;
  1278		if (mptcp_put_full_info(&mfi, optval, copied, optlen))
  1279			return -EFAULT;
  1280	
  1281		return 0;
  1282	
  1283	fail_release:
  1284		release_sock(sk);
  1285		return -EFAULT;
  1286	}
  1287	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki