[PATCH net-next v2 0/4] mptcp: misc. features for v6.18

[PATCH net-next v2 0/4] mptcp: misc. features for v6.18

[Matthieu Baerts (NGI0)]

This series contains 4 independent new features:

- Patch 1: use HMAC-SHA256 library instead of open-coded HMAC.

- Patch 2: selftests: check for unexpected fallback counter increments.

- Patches 3-4: record subflows in RPS table, for aRFS support.

Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
---
Changes in v2:
- Drop previous patches 2 ("mptcp: make ADD_ADDR retransmission timeout
  adaptive") + 3 ("selftests: mptcp: remove add_addr_timeout settings"):
  They were introducing instabilities in the selftests.
- Rebased. Other patches have not been modified.
- Link to v1: https://lore.kernel.org/r/20250901-net-next-mptcp-misc-feat-6-18-v1-0-80ae80d2b903@kernel.org

---
Christoph Paasch (2):
      net: Add rfs_needed() helper
      mptcp: record subflows in RPS table

Eric Biggers (1):
      mptcp: use HMAC-SHA256 library instead of open-coded HMAC

Gang Yan (1):
      selftests: mptcp: add checks for fallback counters

 include/net/rps.h                               |  85 ++++++++++------
 net/mptcp/crypto.c                              |  35 +------
 net/mptcp/protocol.c                            |  21 ++++
 tools/testing/selftests/net/mptcp/mptcp_join.sh | 123 ++++++++++++++++++++++++
 4 files changed, 202 insertions(+), 62 deletions(-)
---
base-commit: cd8a4cfa6bb43a441901e82f5c222dddc75a18a3
change-id: 20250829-net-next-mptcp-misc-feat-6-18-722fa87a60f1

Best regards,
-- 
Matthieu Baerts (NGI0) <matttbe@kernel.org>

[PATCH net-next v2 1/4] mptcp: use HMAC-SHA256 library instead of open-coded HMAC

[Matthieu Baerts (NGI0)]

From: Eric Biggers <ebiggers@kernel.org>

Now that there are easy-to-use HMAC-SHA256 library functions, use these
in net/mptcp/crypto.c instead of open-coding the HMAC algorithm.

Remove the WARN_ON_ONCE() for messages longer than SHA256_DIGEST_SIZE.
The new implementation handles all message lengths correctly.

The mptcp-crypto KUnit test still passes after this change.

Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
---
 net/mptcp/crypto.c | 35 ++---------------------------------
 1 file changed, 2 insertions(+), 33 deletions(-)

diff --git a/net/mptcp/crypto.c b/net/mptcp/crypto.c
index b08ba959ac4fd485bb833043ff58d4c8bac8a37a..31948e18d97da7ee0ee2ae9e4f7c9ca0e3b330a7 100644
--- a/net/mptcp/crypto.c
+++ b/net/mptcp/crypto.c
@@ -22,7 +22,6 @@
 
 #include <linux/kernel.h>
 #include <crypto/sha2.h>
-#include <linux/unaligned.h>
 
 #include "protocol.h"
 
@@ -43,39 +42,9 @@ void mptcp_crypto_key_sha(u64 key, u32 *token, u64 *idsn)
 
 void mptcp_crypto_hmac_sha(u64 key1, u64 key2, u8 *msg, int len, void *hmac)
 {
-	u8 input[SHA256_BLOCK_SIZE + SHA256_DIGEST_SIZE];
-	u8 key1be[8];
-	u8 key2be[8];
-	int i;
+	__be64 key[2] = { cpu_to_be64(key1), cpu_to_be64(key2) };
 
-	if (WARN_ON_ONCE(len > SHA256_DIGEST_SIZE))
-		len = SHA256_DIGEST_SIZE;
-
-	put_unaligned_be64(key1, key1be);
-	put_unaligned_be64(key2, key2be);
-
-	/* Generate key xored with ipad */
-	memset(input, 0x36, SHA256_BLOCK_SIZE);
-	for (i = 0; i < 8; i++)
-		input[i] ^= key1be[i];
-	for (i = 0; i < 8; i++)
-		input[i + 8] ^= key2be[i];
-
-	memcpy(&input[SHA256_BLOCK_SIZE], msg, len);
-
-	/* emit sha256(K1 || msg) on the second input block, so we can
-	 * reuse 'input' for the last hashing
-	 */
-	sha256(input, SHA256_BLOCK_SIZE + len, &input[SHA256_BLOCK_SIZE]);
-
-	/* Prepare second part of hmac */
-	memset(input, 0x5C, SHA256_BLOCK_SIZE);
-	for (i = 0; i < 8; i++)
-		input[i] ^= key1be[i];
-	for (i = 0; i < 8; i++)
-		input[i + 8] ^= key2be[i];
-
-	sha256(input, SHA256_BLOCK_SIZE + SHA256_DIGEST_SIZE, hmac);
+	hmac_sha256_usingrawkey((const u8 *)key, sizeof(key), msg, len, hmac);
 }
 
 #if IS_MODULE(CONFIG_MPTCP_KUNIT_TEST)

-- 
2.51.0

[PATCH net-next v2 2/4] selftests: mptcp: add checks for fallback counters

[Matthieu Baerts (NGI0)]

From: Gang Yan <yangang@kylinos.cn>

Recently, some mib counters about fallback has been added, this patch
provides a method to check the expected behavior of these mib counters
during the test execution.

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/571
Signed-off-by: Gang Yan <yangang@kylinos.cn>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
---
 tools/testing/selftests/net/mptcp/mptcp_join.sh | 123 ++++++++++++++++++++++++
 1 file changed, 123 insertions(+)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index 82cae37d9c2026cc55466636d53a76f929a03452..2f046167a0b6cc6fb5531a033d8d95c9ea399cf9 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -74,6 +74,17 @@ unset join_create_err
 unset join_bind_err
 unset join_connect_err
 
+unset fb_ns1
+unset fb_ns2
+unset fb_infinite_map_tx
+unset fb_dss_corruption
+unset fb_simult_conn
+unset fb_mpc_passive
+unset fb_mpc_active
+unset fb_mpc_data
+unset fb_md5_sig
+unset fb_dss
+
 # generated using "nfbpf_compile '(ip && (ip[54] & 0xf0) == 0x30) ||
 #				  (ip6 && (ip6[74] & 0xf0) == 0x30)'"
 CBPF_MPTCP_SUBOPTION_ADD_ADDR="14,
@@ -1399,6 +1410,115 @@ chk_join_tx_nr()
 	print_results "join Tx" ${rc}
 }
 
+chk_fallback_nr()
+{
+	local infinite_map_tx=${fb_infinite_map_tx:-0}
+	local dss_corruption=${fb_dss_corruption:-0}
+	local simult_conn=${fb_simult_conn:-0}
+	local mpc_passive=${fb_mpc_passive:-0}
+	local mpc_active=${fb_mpc_active:-0}
+	local mpc_data=${fb_mpc_data:-0}
+	local md5_sig=${fb_md5_sig:-0}
+	local dss=${fb_dss:-0}
+	local rc=${KSFT_PASS}
+	local ns=$1
+	local count
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtInfiniteMapTx")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$infinite_map_tx" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns infinite map tx fallback"
+		fail_test "got $count infinite map tx fallback[s] in $ns expected $infinite_map_tx"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtDSSCorruptionFallback")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$dss_corruption" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns dss corruption fallback"
+		fail_test "got $count dss corruption fallback[s] in $ns expected $dss_corruption"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtSimultConnectFallback")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$simult_conn" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns simult conn fallback"
+		fail_test "got $count simult conn fallback[s] in $ns expected $simult_conn"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtMPCapableFallbackACK")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$mpc_passive" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns mpc passive fallback"
+		fail_test "got $count mpc passive fallback[s] in $ns expected $mpc_passive"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtMPCapableFallbackSYNACK")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$mpc_active" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns mpc active fallback"
+		fail_test "got $count mpc active fallback[s] in $ns expected $mpc_active"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtMPCapableDataFallback")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$mpc_data" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns mpc data fallback"
+		fail_test "got $count mpc data fallback[s] in $ns expected $mpc_data"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtMD5SigFallback")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$md5_sig" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns MD5 Sig fallback"
+		fail_test "got $count MD5 Sig fallback[s] in $ns expected $md5_sig"
+	fi
+
+	count=$(mptcp_lib_get_counter ${!ns} "MPTcpExtDssFallback")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$dss" ]; then
+		rc=${KSFT_FAIL}
+		print_check "$ns dss fallback"
+		fail_test "got $count dss fallback[s] in $ns expected $dss"
+	fi
+
+	return $rc
+}
+
+chk_fallback_nr_all()
+{
+	local netns=("ns1" "ns2")
+	local fb_ns=("fb_ns1" "fb_ns2")
+	local rc=${KSFT_PASS}
+
+	for i in 0 1; do
+		if [ -n "${!fb_ns[i]}" ]; then
+			eval "${!fb_ns[i]}" \
+				chk_fallback_nr ${netns[i]} || rc=${?}
+		else
+			chk_fallback_nr ${netns[i]} || rc=${?}
+		fi
+	done
+
+	if [ "${rc}" != "${KSFT_PASS}" ]; then
+		print_results "fallback" ${rc}
+	fi
+}
+
 chk_join_nr()
 {
 	local syn_nr=$1
@@ -1484,6 +1604,8 @@ chk_join_nr()
 	join_syn_tx="${join_syn_tx:-${syn_nr}}" \
 		chk_join_tx_nr
 
+	chk_fallback_nr_all
+
 	if $validate_checksum; then
 		chk_csum_nr $csum_ns1 $csum_ns2
 		chk_fail_nr $fail_nr $fail_nr
@@ -3337,6 +3459,7 @@ fail_tests()
 		join_csum_ns1=+1 join_csum_ns2=+0 \
 			join_fail_nr=1 join_rst_nr=0 join_infi_nr=1 \
 			join_corrupted_pkts="$(pedit_action_pkts)" \
+			fb_ns1="fb_dss=1" fb_ns2="fb_infinite_map_tx=1" \
 			chk_join_nr 0 0 0
 		chk_fail_nr 1 -1 invert
 	fi

-- 
2.51.0

[PATCH net-next v2 3/4] net: Add rfs_needed() helper

[Matthieu Baerts (NGI0)]

From: Christoph Paasch <cpaasch@openai.com>

Add a helper to check if RFS is needed or not. Allows to make the code a
bit cleaner and the next patch to have MPTCP use this helper to decide
whether or not to iterate over the subflows.

tun_flow_update() was calling sock_rps_record_flow_hash() regardless of
the state of rfs_needed. This was not really a bug as sock_flow_table
simply ends up being NULL and thus everything will be fine.
This commit here thus also implicitly makes tun_flow_update() respect
the state of rfs_needed.

Suggested-by: Matthieu Baerts <matttbe@kernel.org>
Signed-off-by: Christoph Paasch <cpaasch@openai.com>
Acked-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
---
 include/net/rps.h | 85 ++++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 56 insertions(+), 29 deletions(-)

diff --git a/include/net/rps.h b/include/net/rps.h
index 9917dce42ca457e9c25d9e84ee450235f771d09b..f1794cd2e7fb32a36bde9959fab651663ab190fd 100644
--- a/include/net/rps.h
+++ b/include/net/rps.h
@@ -85,11 +85,8 @@ static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
 		WRITE_ONCE(table->ents[index], val);
 }
 
-#endif /* CONFIG_RPS */
-
-static inline void sock_rps_record_flow_hash(__u32 hash)
+static inline void _sock_rps_record_flow_hash(__u32 hash)
 {
-#ifdef CONFIG_RPS
 	struct rps_sock_flow_table *sock_flow_table;
 
 	if (!hash)
@@ -99,42 +96,33 @@ static inline void sock_rps_record_flow_hash(__u32 hash)
 	if (sock_flow_table)
 		rps_record_sock_flow(sock_flow_table, hash);
 	rcu_read_unlock();
-#endif
 }
 
-static inline void sock_rps_record_flow(const struct sock *sk)
+static inline void _sock_rps_record_flow(const struct sock *sk)
 {
-#ifdef CONFIG_RPS
-	if (static_branch_unlikely(&rfs_needed)) {
-		/* Reading sk->sk_rxhash might incur an expensive cache line
-		 * miss.
-		 *
-		 * TCP_ESTABLISHED does cover almost all states where RFS
-		 * might be useful, and is cheaper [1] than testing :
-		 *	IPv4: inet_sk(sk)->inet_daddr
-		 * 	IPv6: ipv6_addr_any(&sk->sk_v6_daddr)
-		 * OR	an additional socket flag
-		 * [1] : sk_state and sk_prot are in the same cache line.
+	/* Reading sk->sk_rxhash might incur an expensive cache line
+	 * miss.
+	 *
+	 * TCP_ESTABLISHED does cover almost all states where RFS
+	 * might be useful, and is cheaper [1] than testing :
+	 *	IPv4: inet_sk(sk)->inet_daddr
+	 *	IPv6: ipv6_addr_any(&sk->sk_v6_daddr)
+	 * OR	an additional socket flag
+	 * [1] : sk_state and sk_prot are in the same cache line.
+	 */
+	if (sk->sk_state == TCP_ESTABLISHED) {
+		/* This READ_ONCE() is paired with the WRITE_ONCE()
+		 * from sock_rps_save_rxhash() and sock_rps_reset_rxhash().
 		 */
-		if (sk->sk_state == TCP_ESTABLISHED) {
-			/* This READ_ONCE() is paired with the WRITE_ONCE()
-			 * from sock_rps_save_rxhash() and sock_rps_reset_rxhash().
-			 */
-			sock_rps_record_flow_hash(READ_ONCE(sk->sk_rxhash));
-		}
+		_sock_rps_record_flow_hash(READ_ONCE(sk->sk_rxhash));
 	}
-#endif
 }
 
-static inline void sock_rps_delete_flow(const struct sock *sk)
+static inline void _sock_rps_delete_flow(const struct sock *sk)
 {
-#ifdef CONFIG_RPS
 	struct rps_sock_flow_table *table;
 	u32 hash, index;
 
-	if (!static_branch_unlikely(&rfs_needed))
-		return;
-
 	hash = READ_ONCE(sk->sk_rxhash);
 	if (!hash)
 		return;
@@ -147,6 +135,45 @@ static inline void sock_rps_delete_flow(const struct sock *sk)
 			WRITE_ONCE(table->ents[index], RPS_NO_CPU);
 	}
 	rcu_read_unlock();
+}
+#endif /* CONFIG_RPS */
+
+static inline bool rfs_is_needed(void)
+{
+#ifdef CONFIG_RPS
+	return static_branch_unlikely(&rfs_needed);
+#else
+	return false;
+#endif
+}
+
+static inline void sock_rps_record_flow_hash(__u32 hash)
+{
+#ifdef CONFIG_RPS
+	if (!rfs_is_needed())
+		return;
+
+	_sock_rps_record_flow_hash(hash);
+#endif
+}
+
+static inline void sock_rps_record_flow(const struct sock *sk)
+{
+#ifdef CONFIG_RPS
+	if (!rfs_is_needed())
+		return;
+
+	_sock_rps_record_flow(sk);
+#endif
+}
+
+static inline void sock_rps_delete_flow(const struct sock *sk)
+{
+#ifdef CONFIG_RPS
+	if (!rfs_is_needed())
+		return;
+
+	_sock_rps_delete_flow(sk);
 #endif
 }
 

-- 
2.51.0

[PATCH net-next v2 4/4] mptcp: record subflows in RPS table

[Matthieu Baerts (NGI0)]

From: Christoph Paasch <cpaasch@openai.com>

Accelerated Receive Flow Steering (aRFS) relies on sockets recording
their RX flow hash into the rps_sock_flow_table so that incoming packets
are steered to the CPU where the application runs.

With MPTCP, the application interacts with the parent MPTCP socket while
data is carried over per-subflow TCP sockets. Without recording these
subflows, aRFS cannot steer interrupts and RX processing for the flows
to the desired CPU.

Record all subflows in the RPS table by calling sock_rps_record_flow()
for each subflow at the start of mptcp_sendmsg(), mptcp_recvmsg() and
mptcp_stream_accept(), by using the new helper
mptcp_rps_record_subflows().

It does not by itself improve throughput, but ensures that IRQ and RX
processing are directed to the right CPU, which is a
prerequisite for effective aRFS.

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/protocol.c | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index ad41c48126e44fda646f1ec1c81957db1407a6cc..a8d57b88578dfea807d3d55e430849aa8005c637 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -12,6 +12,7 @@
 #include <linux/sched/signal.h>
 #include <linux/atomic.h>
 #include <net/aligned_data.h>
+#include <net/rps.h>
 #include <net/sock.h>
 #include <net/inet_common.h>
 #include <net/inet_hashtables.h>
@@ -1740,6 +1741,20 @@ static u32 mptcp_send_limit(const struct sock *sk)
 	return limit - not_sent;
 }
 
+static void mptcp_rps_record_subflows(const struct mptcp_sock *msk)
+{
+	struct mptcp_subflow_context *subflow;
+
+	if (!rfs_is_needed())
+		return;
+
+	mptcp_for_each_subflow(msk, subflow) {
+		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
+
+		sock_rps_record_flow(ssk);
+	}
+}
+
 static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 {
 	struct mptcp_sock *msk = mptcp_sk(sk);
@@ -1753,6 +1768,8 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 
 	lock_sock(sk);
 
+	mptcp_rps_record_subflows(msk);
+
 	if (unlikely(inet_test_bit(DEFER_CONNECT, sk) ||
 		     msg->msg_flags & MSG_FASTOPEN)) {
 		int copied_syn = 0;
@@ -2131,6 +2148,8 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 		goto out_err;
 	}
 
+	mptcp_rps_record_subflows(msk);
+
 	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
 
 	len = min_t(size_t, len, INT_MAX);
@@ -3922,6 +3941,8 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock,
 				mptcp_sock_graft(ssk, newsock);
 		}
 
+		mptcp_rps_record_subflows(msk);
+
 		/* Do late cleanup for the first subflow as necessary. Also
 		 * deal with bad peers not doing a complete shutdown.
 		 */

-- 
2.51.0

Re: [PATCH net-next v2 0/4] mptcp: misc. features for v6.18

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Tue, 02 Sep 2025 23:11:32 +0200 you wrote:
> This series contains 4 independent new features:
> 
> - Patch 1: use HMAC-SHA256 library instead of open-coded HMAC.
> 
> - Patch 2: selftests: check for unexpected fallback counter increments.
> 
> - Patches 3-4: record subflows in RPS table, for aRFS support.
> 
> [...]

Here is the summary with links:
  - [net-next,v2,1/4] mptcp: use HMAC-SHA256 library instead of open-coded HMAC
    https://git.kernel.org/netdev/net-next/c/2d5be5629ce7
  - [net-next,v2,2/4] selftests: mptcp: add checks for fallback counters
    https://git.kernel.org/netdev/net-next/c/3fff72f827ad
  - [net-next,v2,3/4] net: Add rfs_needed() helper
    https://git.kernel.org/netdev/net-next/c/929324913e0c
  - [net-next,v2,4/4] mptcp: record subflows in RPS table
    https://git.kernel.org/netdev/net-next/c/3bd4f98a4e2c

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html