Merge branch 'mptcp-various-small-and-unrelated-improvements'

Matthieu Baerts says:

====================
mptcp: various small and unrelated improvements

Here are various unrelated patches:

- Patch 1: sched: remove unused structure.

- Patch 2: sched: split the validation part, a preparation for later.

- Patch 3: pm: clarify code, not to think there is a possible UaF.
  Note: a previous version has already been sent individually to Netdev.

- Patch 4: subflow: simplify subflow_hmac_valid by passing subflow_req.

- Patch 5: mib: add counter for MPJoin rejected by the PM.

- Patch 6: selftests: validate this new MPJoinRejected counter.

- Patch 7: selftests: define nlh variable only where needed.

- Patch 8: selftests: show how to use IPPROTO_MPTCP with getaddrinfo.
  Note: a previous version has already been sent individually to Netdev.

v1: https://lore.kernel.org/20250411-net-next-mptcp-sched-mib-sft-misc-v1-0-85ac8c6654c3@kernel.org
====================

Link: https://patch.msgid.link/20250413-net-next-mptcp-sched-mib-sft-misc-v2-0-0f83a4350150@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

11 files changed, 80 insertions, 46 deletions

diff --git a/include/net/mptcp.h b/include/net/mptcp.h
index bfbad695951c..f7263fe2a2e4 100644
--- a/include/net/mptcp.h
+++ b/include/net/mptcp.h
@@ -101,18 +101,9 @@ struct mptcp_out_options {
 #define MPTCP_SCHED_MAX		128
 #define MPTCP_SCHED_BUF_MAX	(MPTCP_SCHED_NAME_MAX * MPTCP_SCHED_MAX)
 
-#define MPTCP_SUBFLOWS_MAX	8
-
-struct mptcp_sched_data {
-	u8	subflows;
-	struct mptcp_subflow_context *contexts[MPTCP_SUBFLOWS_MAX];
-};
-
 struct mptcp_sched_ops {
-	int (*get_send)(struct mptcp_sock *msk,
-			struct mptcp_sched_data *data);
-	int (*get_retrans)(struct mptcp_sock *msk,
-			   struct mptcp_sched_data *data);
+	int (*get_send)(struct mptcp_sock *msk);
+	int (*get_retrans)(struct mptcp_sock *msk);
 
 	char			name[MPTCP_SCHED_NAME_MAX];
 	struct module		*owner;
diff --git a/net/mptcp/mib.c b/net/mptcp/mib.c
index 19eb9292bd60..0c24545f0e8d 100644
--- a/net/mptcp/mib.c
+++ b/net/mptcp/mib.c
@@ -28,6 +28,7 @@ static const struct snmp_mib mptcp_snmp_list[] = {
 	SNMP_MIB_ITEM("MPJoinSynAckHMacFailure", MPTCP_MIB_JOINSYNACKMAC),
 	SNMP_MIB_ITEM("MPJoinAckRx", MPTCP_MIB_JOINACKRX),
 	SNMP_MIB_ITEM("MPJoinAckHMacFailure", MPTCP_MIB_JOINACKMAC),
+	SNMP_MIB_ITEM("MPJoinRejected", MPTCP_MIB_JOINREJECTED),
 	SNMP_MIB_ITEM("MPJoinSynTx", MPTCP_MIB_JOINSYNTX),
 	SNMP_MIB_ITEM("MPJoinSynTxCreatSkErr", MPTCP_MIB_JOINSYNTXCREATSKERR),
 	SNMP_MIB_ITEM("MPJoinSynTxBindErr", MPTCP_MIB_JOINSYNTXBINDERR),
diff --git a/net/mptcp/mib.h b/net/mptcp/mib.h
index 128282982843..250c6b77977e 100644
--- a/net/mptcp/mib.h
+++ b/net/mptcp/mib.h
@@ -23,6 +23,7 @@ enum linux_mptcp_mib_field {
 	MPTCP_MIB_JOINSYNACKMAC,	/* HMAC was wrong on SYN/ACK + MP_JOIN */
 	MPTCP_MIB_JOINACKRX,		/* Received an ACK + MP_JOIN */
 	MPTCP_MIB_JOINACKMAC,		/* HMAC was wrong on ACK + MP_JOIN */
+	MPTCP_MIB_JOINREJECTED,		/* The PM rejected the JOIN request */
 	MPTCP_MIB_JOINSYNTX,		/* Sending a SYN + MP_JOIN */
 	MPTCP_MIB_JOINSYNTXCREATSKERR,	/* Not able to create a socket when sending a SYN + MP_JOIN */
 	MPTCP_MIB_JOINSYNTXBINDERR,	/* Not able to bind() the address when sending a SYN + MP_JOIN */
diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c
index 31747f974941..1306d4dc287b 100644
--- a/net/mptcp/pm.c
+++ b/net/mptcp/pm.c
@@ -151,10 +151,13 @@ bool mptcp_remove_anno_list_by_saddr(struct mptcp_sock *msk,
 				     const struct mptcp_addr_info *addr)
 {
 	struct mptcp_pm_add_entry *entry;
+	bool ret;
 
 	entry = mptcp_pm_del_add_timer(msk, addr, false);
+	ret = entry;
 	kfree(entry);
-	return entry;
+
+	return ret;
 }
 
 bool mptcp_pm_sport_in_anno_list(struct mptcp_sock *msk, const struct sock *sk)
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 44f7ab463d75..26ffa06c21e8 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -3527,8 +3527,10 @@ bool mptcp_finish_join(struct sock *ssk)
 		return true;
 	}
 
-	if (!mptcp_pm_allow_new_subflow(msk))
+	if (!mptcp_pm_allow_new_subflow(msk)) {
+		MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_JOINREJECTED);
 		goto err_prohibited;
+	}
 
 	/* If we can't acquire msk socket lock here, let the release callback
 	 * handle it
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index d409586b5977..7aa38d74fef6 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -744,6 +744,7 @@ void mptcp_info2sockaddr(const struct mptcp_addr_info *info,
 			 struct sockaddr_storage *addr,
 			 unsigned short family);
 struct mptcp_sched_ops *mptcp_sched_find(const char *name);
+int mptcp_validate_scheduler(struct mptcp_sched_ops *sched);
 int mptcp_register_scheduler(struct mptcp_sched_ops *sched);
 void mptcp_unregister_scheduler(struct mptcp_sched_ops *sched);
 void mptcp_sched_init(void);
diff --git a/net/mptcp/sched.c b/net/mptcp/sched.c
index c16c6fbd4ba2..1e59072d478c 100644
--- a/net/mptcp/sched.c
+++ b/net/mptcp/sched.c
@@ -16,8 +16,7 @@
 static DEFINE_SPINLOCK(mptcp_sched_list_lock);
 static LIST_HEAD(mptcp_sched_list);
 
-static int mptcp_sched_default_get_send(struct mptcp_sock *msk,
-					struct mptcp_sched_data *data)
+static int mptcp_sched_default_get_send(struct mptcp_sock *msk)
 {
 	struct sock *ssk;
 
@@ -29,8 +28,7 @@ static int mptcp_sched_default_get_send(struct mptcp_sock *msk,
 	return 0;
 }
 
-static int mptcp_sched_default_get_retrans(struct mptcp_sock *msk,
-					   struct mptcp_sched_data *data)
+static int mptcp_sched_default_get_retrans(struct mptcp_sock *msk)
 {
 	struct sock *ssk;
 
@@ -84,10 +82,23 @@ void mptcp_get_available_schedulers(char *buf, size_t maxlen)
 	rcu_read_unlock();
 }
 
-int mptcp_register_scheduler(struct mptcp_sched_ops *sched)
+int mptcp_validate_scheduler(struct mptcp_sched_ops *sched)
 {
-	if (!sched->get_send)
+	if (!sched->get_send) {
+		pr_err("%s does not implement required ops\n", sched->name);
 		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int mptcp_register_scheduler(struct mptcp_sched_ops *sched)
+{
+	int ret;
+
+	ret = mptcp_validate_scheduler(sched);
+	if (ret)
+		return ret;
 
 	spin_lock(&mptcp_sched_list_lock);
 	if (mptcp_sched_find(sched->name)) {
@@ -157,7 +168,6 @@ void mptcp_subflow_set_scheduled(struct mptcp_subflow_context *subflow,
 int mptcp_sched_get_send(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow;
-	struct mptcp_sched_data *data = NULL;
 
 	msk_owned_by_me(msk);
 
@@ -178,14 +188,13 @@ int mptcp_sched_get_send(struct mptcp_sock *msk)
 	}
 
 	if (msk->sched == &mptcp_sched_default || !msk->sched)
-		return mptcp_sched_default_get_send(msk, data);
-	return msk->sched->get_send(msk, data);
+		return mptcp_sched_default_get_send(msk);
+	return msk->sched->get_send(msk);
 }
 
 int mptcp_sched_get_retrans(struct mptcp_sock *msk)
 {
 	struct mptcp_subflow_context *subflow;
-	struct mptcp_sched_data *data = NULL;
 
 	msk_owned_by_me(msk);
 
@@ -199,8 +208,8 @@ int mptcp_sched_get_retrans(struct mptcp_sock *msk)
 	}
 
 	if (msk->sched == &mptcp_sched_default || !msk->sched)
-		return mptcp_sched_default_get_retrans(msk, data);
+		return mptcp_sched_default_get_retrans(msk);
 	if (msk->sched->get_retrans)
-		return msk->sched->get_retrans(msk, data);
-	return msk->sched->get_send(msk, data);
+		return msk->sched->get_retrans(msk);
+	return msk->sched->get_send(msk);
 }
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 24c2de1891bd..15613d691bfe 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -247,6 +247,7 @@ again:
 
 		if (unlikely(req->syncookie)) {
 			if (!mptcp_can_accept_new_subflow(subflow_req->msk)) {
+				SUBFLOW_REQ_INC_STATS(req, MPTCP_MIB_JOINREJECTED);
 				subflow_add_reset_reason(skb, MPTCP_RST_EPROHIBIT);
 				return -EPERM;
 			}
@@ -745,15 +746,11 @@ struct request_sock *mptcp_subflow_reqsk_alloc(const struct request_sock_ops *op
 EXPORT_SYMBOL(mptcp_subflow_reqsk_alloc);
 
 /* validate hmac received in third ACK */
-static bool subflow_hmac_valid(const struct request_sock *req,
+static bool subflow_hmac_valid(const struct mptcp_subflow_request_sock *subflow_req,
 			       const struct mptcp_options_received *mp_opt)
 {
-	const struct mptcp_subflow_request_sock *subflow_req;
+	struct mptcp_sock *msk = subflow_req->msk;
 	u8 hmac[SHA256_DIGEST_SIZE];
-	struct mptcp_sock *msk;
-
-	subflow_req = mptcp_subflow_rsk(req);
-	msk = subflow_req->msk;
 
 	subflow_generate_hmac(READ_ONCE(msk->remote_key),
 			      READ_ONCE(msk->local_key),
@@ -899,13 +896,14 @@ create_child:
 				goto dispose_child;
 			}
 
-			if (!subflow_hmac_valid(req, &mp_opt)) {
+			if (!subflow_hmac_valid(subflow_req, &mp_opt)) {
 				SUBFLOW_REQ_INC_STATS(req, MPTCP_MIB_JOINACKMAC);
 				subflow_add_reset_reason(skb, MPTCP_RST_EPROHIBIT);
 				goto dispose_child;
 			}
 
 			if (!mptcp_can_accept_new_subflow(owner)) {
+				SUBFLOW_REQ_INC_STATS(req, MPTCP_MIB_JOINREJECTED);
 				subflow_add_reset_reason(skb, MPTCP_RST_EPROHIBIT);
 				goto dispose_child;
 			}
diff --git a/tools/testing/selftests/net/mptcp/mptcp_connect.c b/tools/testing/selftests/net/mptcp/mptcp_connect.c
index c83a8b47bbdf..ac1349c4b9e5 100644
--- a/tools/testing/selftests/net/mptcp/mptcp_connect.c
+++ b/tools/testing/selftests/net/mptcp/mptcp_connect.c
@@ -180,13 +180,26 @@ static void xgetnameinfo(const struct sockaddr *addr, socklen_t addrlen,
 }
 
 static void xgetaddrinfo(const char *node, const char *service,
-			 const struct addrinfo *hints,
+			 struct addrinfo *hints,
 			 struct addrinfo **res)
 {
+again:
 	int err = getaddrinfo(node, service, hints, res);
 
 	if (err) {
-		const char *errstr = getxinfo_strerr(err);
+		const char *errstr;
+
+		/* glibc starts to support MPTCP since v2.42.
+		 * For older versions, use IPPROTO_TCP to resolve,
+		 * and use TCP/MPTCP to create socket.
+		 * Link: https://sourceware.org/git/?p=glibc.git;a=commit;h=a8e9022e0f82
+		 */
+		if (err == EAI_SOCKTYPE) {
+			hints->ai_protocol = IPPROTO_TCP;
+			goto again;
+		}
+
+		errstr = getxinfo_strerr(err);
 
 		fprintf(stderr, "Fatal: getaddrinfo(%s:%s): %s\n",
 			node ? node : "", service ? service : "", errstr);
@@ -292,7 +305,7 @@ static int sock_listen_mptcp(const char * const listenaddr,
 {
 	int sock = -1;
 	struct addrinfo hints = {
-		.ai_protocol = IPPROTO_TCP,
+		.ai_protocol = IPPROTO_MPTCP,
 		.ai_socktype = SOCK_STREAM,
 		.ai_flags = AI_PASSIVE | AI_NUMERICHOST
 	};
@@ -356,7 +369,7 @@ static int sock_connect_mptcp(const char * const remoteaddr,
 			      int infd, struct wstate *winfo)
 {
 	struct addrinfo hints = {
-		.ai_protocol = IPPROTO_TCP,
+		.ai_protocol = IPPROTO_MPTCP,
 		.ai_socktype = SOCK_STREAM,
 	};
 	struct addrinfo *a, *addr;
diff --git a/tools/testing/selftests/net/mptcp/mptcp_diag.c b/tools/testing/selftests/net/mptcp/mptcp_diag.c
index 284286c524cf..37d5015ad08c 100644
--- a/tools/testing/selftests/net/mptcp/mptcp_diag.c
+++ b/tools/testing/selftests/net/mptcp/mptcp_diag.c
@@ -185,9 +185,10 @@ static void parse_nlmsg(struct nlmsghdr *nlh)
 	}
 }
 
-static void recv_nlmsg(int fd, struct nlmsghdr *nlh)
+static void recv_nlmsg(int fd)
 {
 	char rcv_buff[8192];
+	struct nlmsghdr *nlh = (struct nlmsghdr *)rcv_buff;
 	struct sockaddr_nl rcv_nladdr = {
 		.nl_family = AF_NETLINK
 	};
@@ -204,7 +205,6 @@ static void recv_nlmsg(int fd, struct nlmsghdr *nlh)
 	int len;
 
 	len = recvmsg(fd, &rcv_msg, 0);
-	nlh = (struct nlmsghdr *)rcv_buff;
 
 	while (NLMSG_OK(nlh, len)) {
 		if (nlh->nlmsg_type == NLMSG_DONE) {
@@ -225,7 +225,6 @@ static void recv_nlmsg(int fd, struct nlmsghdr *nlh)
 
 static void get_mptcpinfo(__u32 token)
 {
-	struct nlmsghdr *nlh = NULL;
 	int fd;
 
 	fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
@@ -233,7 +232,7 @@ static void get_mptcpinfo(__u32 token)
 		die_perror("Netlink socket");
 
 	send_query(fd, token);
-	recv_nlmsg(fd, nlh);
+	recv_nlmsg(fd);
 
 	close(fd);
 }
diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index befa66f5a366..b8af65373b3a 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -62,6 +62,7 @@ unset sflags
 unset fastclose
 unset fullmesh
 unset speed
+unset join_syn_rej
 unset join_csum_ns1
 unset join_csum_ns2
 unset join_fail_nr
@@ -1403,6 +1404,7 @@ chk_join_nr()
 	local syn_nr=$1
 	local syn_ack_nr=$2
 	local ack_nr=$3
+	local syn_rej=${join_syn_rej:-0}
 	local csum_ns1=${join_csum_ns1:-0}
 	local csum_ns2=${join_csum_ns2:-0}
 	local fail_nr=${join_fail_nr:-0}
@@ -1468,6 +1470,15 @@ chk_join_nr()
 		fail_test "got $count JOIN[s] ack HMAC failure expected 0"
 	fi
 
+	count=$(mptcp_lib_get_counter ${ns1} "MPTcpExtMPJoinRejected")
+	if [ -z "$count" ]; then
+		rc=${KSFT_SKIP}
+	elif [ "$count" != "$syn_rej" ]; then
+		rc=${KSFT_FAIL}
+		print_check "syn rejected"
+		fail_test "got $count JOIN[s] syn rejected expected $syn_rej"
+	fi
+
 	print_results "join Rx" ${rc}
 
 	join_syn_tx="${join_syn_tx:-${syn_nr}}" \
@@ -1963,7 +1974,8 @@ subflows_tests()
 		pm_nl_set_limits $ns2 0 1
 		pm_nl_add_endpoint $ns2 10.0.3.2 flags subflow
 		run_tests $ns1 $ns2 10.0.1.1
-		chk_join_nr 1 1 0
+		join_syn_rej=1 \
+			chk_join_nr 1 1 0
 	fi
 
 	# subflow
@@ -1992,7 +2004,8 @@ subflows_tests()
 		pm_nl_add_endpoint $ns2 10.0.3.2 flags subflow
 		pm_nl_add_endpoint $ns2 10.0.2.2 flags subflow
 		run_tests $ns1 $ns2 10.0.1.1
-		chk_join_nr 2 2 1
+		join_syn_rej=1 \
+			chk_join_nr 2 2 1
 	fi
 
 	# single subflow, dev
@@ -3061,7 +3074,8 @@ syncookies_tests()
 		pm_nl_add_endpoint $ns2 10.0.3.2 flags subflow
 		pm_nl_add_endpoint $ns2 10.0.2.2 flags subflow
 		run_tests $ns1 $ns2 10.0.1.1
-		chk_join_nr 2 1 1
+		join_syn_rej=1 \
+			chk_join_nr 2 1 1
 	fi
 
 	# test signal address with cookies
@@ -3545,7 +3559,8 @@ userspace_tests()
 		pm_nl_set_limits $ns2 1 1
 		pm_nl_add_endpoint $ns2 10.0.3.2 flags subflow
 		run_tests $ns1 $ns2 10.0.1.1
-		chk_join_nr 1 1 0
+		join_syn_rej=1 \
+			chk_join_nr 1 1 0
 	fi
 
 	# userspace pm type does not send join
@@ -3568,7 +3583,8 @@ userspace_tests()
 		pm_nl_add_endpoint $ns2 10.0.3.2 flags subflow
 		sflags=backup speed=slow \
 			run_tests $ns1 $ns2 10.0.1.1
-		chk_join_nr 1 1 0
+		join_syn_rej=1 \
+			chk_join_nr 1 1 0
 		chk_prio_nr 0 0 0 0
 	fi