Series comparison

From: Gang Yan <yangang@kylinos.cn>

After an MPTCP connection is established, the sk_sndbuf of client's msk
can be updated through 'subflow_finish_connect'. However, the newly
accepted msk on the server side has a small sk_sndbuf than
msk->first->sk_sndbuf:

'''
MPTCP: msk:00000000e55b09db, msk->sndbuf:20480, msk->first->sndbuf:2626560
'''

This means that when the server immediately sends MSG_DONTWAIT data to
the client after the connection is established, it is more likely to
encounter EAGAIN.

This patch synchronizes the sk_sndbuf by triggering its update during accept.

Fixes: 8005184fd1ca ("mptcp: refactor sndbuf auto-tuning")
Link: https://github.com/multipath-tcp/mptcp_net-next/issues/602
Signed-off-by: Gang Yan <yangang@kylinos.cn>
---

Notes:
    Hi Paolo, Matt,
    
    Sorry for the late response for this patch. I've been analyzing this
    issue recently, and the basic picture is as follows:
    
    The root cause is a timing gap between msk creation and TCP sndbuf
    auto-tuning on the server side:
    
    1. When the server receives the SYN, mptcp_sk_clone_init() creates the
       msk and calls __mptcp_propagate_sndbuf(). At this point, the TCP
       subflow is still in SYN_RCVD state, so its sk_sndbuf has only the
       initial value (tcp_wmem[1], typically ~16KB).
    
    2. When the 3-way handshake completes (ACK received), the TCP stack
       calls tcp_init_buffer_space() -> tcp_sndbuf_expand(), which grows
       the subflow's sk_sndbuf based on MSS, congestion window, etc.
       (potentially up to tcp_wmem[2], ~4MB).
    
    3. However, this auto-tuning happens deep in the TCP stack without
       any callback to MPTCP, so msk->sk_sndbuf is never updated to
       reflect the new subflow sndbuf value.
    
    4. When accept() returns, msk->sk_sndbuf still holds the small initial
       value, while msk->first->sk_sndbuf has been auto-tuned to a much
       larger value.
    
    In contrast, the active (client) side doesn't have this issue because
    subflow_finish_connect() calls mptcp_propagate_state() after the TCP
    sndbuf auto-tuning has already occurred, ensuring proper synchronization.
    
    Thanks
    Gang

net/mptcp/protocol.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index XXXXXXX..XXXXXXX 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -XXX,XX +XXX,XX @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 
 		mptcp_graft_subflows(newsk);
 		mptcp_rps_record_subflows(msk);
+		__mptcp_propagate_sndbuf(newsk, mptcp_subflow_tcp_sock(subflow));
 
 		/* Do late cleanup for the first subflow as necessary. Also
 		 * deal with bad peers not doing a complete shutdown.
-- 
2.43.0

From: Gang Yan <yangang@kylinos.cn>

On passive MPTCP connections, the msk sndbuf is not updated correctly.

The root cause is a timing issue in the accept path:
- tcp_check_req() -> subflow_syn_recv_sock() -> mptcp_sk_clone_init()
  calls __mptcp_propagate_sndbuf() to copy the ssk sndbuf into msk
- Later, tcp_child_process() -> tcp_init_transfer() ->
  tcp_sndbuf_expand() grows the ssk sndbuf.

So __mptcp_propagate_sndbuf() runs before the ssk sndbuf has been
expanded and the msk ends up with a much smaller sndbuf than the
subflow:
  MPTCP: msk->sndbuf:20480, msk->first->sndbuf:2626560

Fix this by removing the __mptcp_propagate_sndbuf() call in
mptcp_sk_clone_init(), as the ssk sndbuf is not yet finalized there.
Instead, call __mptcp_propagate_sndbuf() at accept() time, when the
ssk sndbuf has been fully expanded by tcp_sndbuf_expand().

Fixes: 8005184fd1ca ("mptcp: refactor sndbuf auto-tuning")
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/602
Signed-off-by: Gang Yan <yangang@kylinos.cn>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
 net/mptcp/protocol.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index XXXXXXX..XXXXXXX 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -XXX,XX +XXX,XX @@ struct sock *mptcp_sk_clone_init(const struct sock *sk,
 	 * uses the correct data
 	 */
 	mptcp_copy_inaddrs(nsk, ssk);
-	__mptcp_propagate_sndbuf(nsk, ssk);
 
 	mptcp_rcv_space_init(msk, ssk);
 	msk->rcvq_space.time = mptcp_stamp();
@@ -XXX,XX +XXX,XX @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 
 		mptcp_graft_subflows(newsk);
 		mptcp_rps_record_subflows(msk);
+		__mptcp_propagate_sndbuf(newsk, mptcp_subflow_tcp_sock(subflow));
 
 		/* Do late cleanup for the first subflow as necessary. Also
 		 * deal with bad peers not doing a complete shutdown.
-- 
2.43.0

From: Gang Yan <yangang@kylinos.cn>

Add a new chk_sndbuf() helper to diag.sh that extracts the sndbuf
(the 'tb' field from 'ss -m' skmem output) for both server and
client MPTCP sockets, and verifies they are equal.

Without the previous patch, it will fail:

'''
07 ....chk sndbuf server/client    [FAIL] sndbuf S=20480 != C=2630656
'''

Signed-off-by: Gang Yan <yangang@kylinos.cn>
---
 tools/testing/selftests/net/mptcp/diag.sh | 28 +++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/tools/testing/selftests/net/mptcp/diag.sh b/tools/testing/selftests/net/mptcp/diag.sh
index XXXXXXX..XXXXXXX 100755
--- a/tools/testing/selftests/net/mptcp/diag.sh
+++ b/tools/testing/selftests/net/mptcp/diag.sh
@@ -XXX,XX +XXX,XX @@ wait_connected()
 	done
 }
 
+chk_sndbuf()
+{
+	local server_sndbuf client_sndbuf msg
+	local port=${1}
+
+	msg="....chk sndbuf server/client"
+	server_sndbuf=$(ss -N "${ns}" -inmHM "sport" "${port}" | \
+			sed -n 's/.*tb$[0-9]\+$.*/\1/p')
+	client_sndbuf=$(ss -N "${ns}" -inmHM "dport" "${port}" | \
+			sed -n 's/.*tb$[0-9]\+$.*/\1/p')
+
+	mptcp_lib_print_title "${msg}"
+	if [ -z "${server_sndbuf}" ] || [ -z "${client_sndbuf}" ]; then
+		mptcp_lib_pr_fail "sndbuf S=${server_sndbuf} C=${client_sndbuf}"
+		mptcp_lib_result_fail "${msg}"
+		ret=${KSFT_FAIL}
+	elif [ "${server_sndbuf}" != "${client_sndbuf}" ]; then
+		mptcp_lib_pr_fail "sndbuf S=${server_sndbuf} != C=${client_sndbuf}"
+		mptcp_lib_result_fail "${msg}"
+		ret=${KSFT_FAIL}
+	else
+		mptcp_lib_pr_ok
+		mptcp_lib_result_pass "${msg}"
+	fi
+}
+
+
 trap cleanup EXIT
 mptcp_lib_ns_init ns
 
@@ -XXX,XX +XXX,XX @@ echo "b" | \
 				127.0.0.1 >/dev/null &
 wait_connected $ns 10000
 chk_msk_nr 2 "after MPC handshake"
+chk_sndbuf 10000
 chk_last_time_info 10000
 chk_msk_remote_key_nr 2 "....chk remote_key"
 chk_msk_fallback_nr 0 "....chk no fallback"
-- 
2.43.0