From: Christoph Paasch cpaasch@openai.com

When skb_ext_add(skb, SKB_EXT_MPTCP) fails in mptcp_incoming_options(), we used to return true, letting the segment proceed through the TCP receive path without a DSS mapping. Such segments can leave inconsistent mapping state and trigger a mid-stream fallback to TCP, which in testing collapsed (by artificially forcing failures in skb_ext_add) throughput to zero.

Return false instead so the TCP input path drops the skb (see tcp_data_queue() and step-7 processing). This is the safer choice under memory pressure: it preserves MPTCP correctness and provides backpressure to the sender.

Control packets remain unaffected: ACK updates and DATA_FIN handling happen before attempting the extension allocation, and tcp_reset() continues to ignore the return value.

With this change, MPTCP continues to work at high throughput if we artificially inject failures into skb_ext_add.

Fixes: 6787b7e350d3 ("mptcp: avoid processing packet if a subflow reset") Cc: stable@vger.kernel.org Signed-off-by: Christoph Paasch cpaasch@openai.com Reviewed-by: Matthieu Baerts (NGI0) matttbe@kernel.org Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- net/mptcp/options.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/mptcp/options.c b/net/mptcp/options.c index 70c0ab0ecf905d282e5dc6c1b21ffc6476c8d71b..2a8ea28442b271675411d190604073f779bee7fa 100644 --- a/net/mptcp/options.c +++ b/net/mptcp/options.c @@ -1118,7 +1118,9 @@ static bool add_addr_hmac_valid(struct mptcp_sock *msk, return hmac == mp_opt->ahmac; }

-/* Return false if a subflow has been reset, else return true */ +/* Return false in case of error (or subflow has been reset), + * else return true. + */ bool mptcp_incoming_options(struct sock *sk, struct sk_buff *skb) { struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk); @@ -1222,7 +1224,7 @@ bool mptcp_incoming_options(struct sock *sk, struct sk_buff *skb)

mpext = skb_ext_add(skb, SKB_EXT_MPTCP); if (!mpext) - return true; + return false;

memset(mpext, 0, sizeof(*mpext));

This modification is linked to the parent commit where the received ADD_ADDR limit was accidentally reset when the endpoints were flushed.

To validate that, the test is now flushing endpoints after having set new limits, and before checking them.

The 'Fixes' tag here below is the same as the one from the previous commit: this patch here is not fixing anything wrong in the selftests, but it validates the previous fix for an issue introduced by this commit ID.

Fixes: 01cacb00b35c ("mptcp: add netlink-based PM") Cc: stable@vger.kernel.org Reviewed-by: Mat Martineau martineau@kernel.org Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- tools/testing/selftests/net/mptcp/pm_netlink.sh | 1 + 1 file changed, 1 insertion(+)

diff --git a/tools/testing/selftests/net/mptcp/pm_netlink.sh b/tools/testing/selftests/net/mptcp/pm_netlink.sh index 2e6648a2b2c0c6ea0e9f030c942077a0f483092f..ac7ec6f9402376a34602ef1ca6c4822e8dde0ded 100755 --- a/tools/testing/selftests/net/mptcp/pm_netlink.sh +++ b/tools/testing/selftests/net/mptcp/pm_netlink.sh @@ -198,6 +198,7 @@ set_limits 1 9 2>/dev/null check "get_limits" "${default_limits}" "subflows above hard limit"

set_limits 8 8 +flush_endpoint ## to make sure it doesn't affect the limits check "get_limits" "$(format_limits 8 8)" "set limits"

flush_endpoint

From: Geliang Tang tanggeliang@kylinos.cn

When add_addr_timeout was set to 0, this caused the ADD_ADDR to be retransmitted immediately, which looks like a buggy behaviour. Instead, interpret 0 as "no retransmissions needed".

The documentation is updated to explicitly state that setting the timeout to 0 disables retransmission.

Fixes: 93f323b9cccc ("mptcp: add a new sysctl add_addr_timeout") Cc: stable@vger.kernel.org Suggested-by: Matthieu Baerts matttbe@kernel.org Signed-off-by: Geliang Tang tanggeliang@kylinos.cn Reviewed-by: Matthieu Baerts (NGI0) matttbe@kernel.org Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- Documentation/networking/mptcp-sysctl.rst | 2 ++ net/mptcp/pm.c | 13 ++++++++++--- 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/Documentation/networking/mptcp-sysctl.rst b/Documentation/networking/mptcp-sysctl.rst index 5bfab01eff5a9db89e1484787953241c16e147cf..1683c139821e3ba6d9eaa3c59330a523d29f1164 100644 --- a/Documentation/networking/mptcp-sysctl.rst +++ b/Documentation/networking/mptcp-sysctl.rst @@ -12,6 +12,8 @@ add_addr_timeout - INTEGER (seconds) resent to an MPTCP peer that has not acknowledged a previous ADD_ADDR message.

+ Do not retransmit if set to 0. + The default value matches TCP_RTO_MAX. This is a per-namespace sysctl.

diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c index c5f6a53ce5f1e10dada52865e41b685d888a71fa..136a380602cae872b76560649c924330e5f42533 100644 --- a/net/mptcp/pm.c +++ b/net/mptcp/pm.c @@ -274,6 +274,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer) add_timer); struct mptcp_sock *msk = entry->sock; struct sock *sk = (struct sock *)msk; + unsigned int timeout;

pr_debug("msk=%p\n", msk);

@@ -291,6 +292,10 @@ static void mptcp_pm_add_timer(struct timer_list *timer) goto out; }

+ timeout = mptcp_get_add_addr_timeout(sock_net(sk)); + if (!timeout) + goto out; + spin_lock_bh(&msk->pm.lock);

if (!mptcp_pm_should_add_signal_addr(msk)) { @@ -302,7 +307,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer)

if (entry->retrans_times < ADD_ADDR_RETRANS_MAX) sk_reset_timer(sk, timer, - jiffies + mptcp_get_add_addr_timeout(sock_net(sk))); + jiffies + timeout);

spin_unlock_bh(&msk->pm.lock);

@@ -344,6 +349,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, struct mptcp_pm_add_entry *add_entry = NULL; struct sock *sk = (struct sock *)msk; struct net *net = sock_net(sk); + unsigned int timeout;

lockdep_assert_held(&msk->pm.lock);

@@ -368,8 +374,9 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,

timer_setup(&add_entry->add_timer, mptcp_pm_add_timer, 0); reset_timer: - sk_reset_timer(sk, &add_entry->add_timer, - jiffies + mptcp_get_add_addr_timeout(net)); + timeout = mptcp_get_add_addr_timeout(net); + if (timeout) + sk_reset_timer(sk, &add_entry->add_timer, jiffies + timeout);

return true; }

GCC was complaining about the new label:

mptcp_connect.c:187:2: warning: label followed by a declaration is a C23 extension [-Wc23-extensions] 187 | int err = getaddrinfo(node, service, hints, res); | ^

Simply declare 'err' before the label to avoid this warning.

Fixes: a862771d1aa4 ("selftests: mptcp: use IPPROTO_MPTCP for getaddrinfo") Cc: stable@vger.kernel.org Reviewed-by: Geliang Tang geliang@kernel.org Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- tools/testing/selftests/net/mptcp/mptcp_connect.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_connect.c b/tools/testing/selftests/net/mptcp/mptcp_connect.c index ac1349c4b9e5404c95935eb38b08a15d774eb1d9..4f07ac9fa207cb08a934582b98d688d0b9512f97 100644 --- a/tools/testing/selftests/net/mptcp/mptcp_connect.c +++ b/tools/testing/selftests/net/mptcp/mptcp_connect.c @@ -183,9 +183,10 @@ static void xgetaddrinfo(const char *node, const char *service, struct addrinfo *hints, struct addrinfo **res) { -again: - int err = getaddrinfo(node, service, hints, res); + int err;

+again: + err = getaddrinfo(node, service, hints, res); if (err) { const char *errstr;