[PATCH mptcp-next v5 00/11] Fix socket options used by NVMe over MPTCP

[PATCH mptcp-next v5 00/11] Fix socket options used by NVMe over MPTCP

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

NVMe over MPTCP relies on SO_LINGER, SO_PRIORITY, SO_REUSEADDR, TCP_SYNCNT,
TCP_NODELAY, IP_TOS, and SO_BINDTODEVICE. This series contains fixes to
make all of them work correctly, and adds tclass support for it.

v5:
 - Include Gang Yan's TCP_MAXSEG cleanup.
 - Include two v4-mapped addr fixes.
 - Patch 9, set rcv_flowinfo to 0 for v4-mapped addr.
 - Patch 11, use rcv_flowinfo instead of np->tclass.
 - All comments from Sashiko on v4 regarding "CONFIG_IPV6=m" are false
   positives.

v4:
 - patch 1: skip ssk->sk_bound_dev_if = local->ifindex when local->ifindex
   is 0, so __mptcp_subflow_connect() doesn't overwrite the inherited
   SO_BINDTODEVICE binding.
 - patch 4: assign tcp_sock_set_syncnt()'s return to ret so an invalid
   TCP_SYNCNT is propagated, not silently dropped.
 - patch 6: skip IPV6_TCLASS setsockopt/getsockopt paths on AF_INET
   sockets/subflows (inet6_sk is NULL there).
 - patch 7: treat IPv4-mapped IPv6 as AF_INET for the TOS/tclass path.
 - patch 9: gate ip6_sock_set_tclass() behind sk_family == AF_INET6 and
   IS_ENABLED(CONFIG_IPV6) to avoid IPv4 NULL-deref and link failure when
   CONFIG_IPV6 is off.
 - patch 10: gate tclass on sk_family, skip IPv6 branch for IPv4-mapped
   connections, wrap IPv6 in CONFIG_IPV6.
 - https://patchwork.kernel.org/project/mptcp/cover/cover.1785378180.git.tanggeliang@kylinos.cn/

v3:
 - include tclass patches.
 - I also included three NVMe patches here because they have dependencies.
 - https://patchwork.kernel.org/project/mptcp/cover/cover.1785238723.git.tanggeliang@kylinos.cn/

v2:
 - Drop "mptcp: don't reset dst when setting default 0 tos"
   and "selftests: mptcp: sockopt: cover LINGER, REUSEADDR,
   PRIORITY, NODELAY, SYNCNT": the 'if (val > 0)' guard
   blocked the legitimate "reset to 0" path, and the test
   only ran val_in=1, missing the SK_CAN_REUSE "any non-zero
   -> 1" normalization.
 - mptcp: bump setsockopt_seq for subflow-only socket options,
   so secondary subflows created via MP_JOIN re-sync
   sk_reuse / sk_reuseport / sk_bound_dev_if from msk.
 - mptcp: copy the subflow's TOS to the msk on accept
   (alongside the existing ssk->rcv_tos copy), so MP_JOIN'd
   subflows inherit the reflected outgoing TOS.
 - mptcp: propagate sk_reuseport to subflows via
   sync_socket_options, so secondary subflows inherit
   SO_REUSEPORT, not just SO_REUSEADDR.
 - mptcp: tighten TCP_SYNCNT bounds check
   ('val < 1 || val > MAX_TCP_SYNCNT'), so out-of-bounds
   values are rejected even when msk has no subflows yet
   (where __mptcp_setsockopt_set_val would otherwise return 0
   without invoking the set_val callback).
 - https://patchwork.kernel.org/project/mptcp/cover/cover.1785054808.git.tanggeliang@kylinos.cn/

v1:
 - https://patchwork.kernel.org/project/mptcp/cover/cover.1784985085.git.tanggeliang@kylinos.cn/

David 'equinox' Lamparter (1):
  mptcp: sockopt: implement IPV6_TCLASS

Gang Yan (1):
  mptcp: take TCP_MAXSEG handling into setsockopt_set_val

Geliang Tang (9):
  mptcp: handle TCP_MAXSEG getsockopt in common case
  mptcp: inherit sk_reuse/sk_reuseport and fix dev bind override
  mptcp: add TCP_SYNCNT setsockopt/getsockopt
  mptcp: skip ip6 options on v4-mapped clone
  mptcp: report v4-mapped subflow addrs as IPv4
  ipv6: extract and export ip6_sock_set_tclass helpers
  mptcp: copy the subflow's tos/tclass to the msk on accept
  nvme-tcp: support IPv6 traffic class
  nvmet-tcp: support IPv6 traffic class

 drivers/nvme/host/fabrics.c | 18 ++++++++
 drivers/nvme/host/fabrics.h |  3 ++
 drivers/nvme/host/tcp.c     |  8 +++-
 drivers/nvme/target/tcp.c   |  8 ++++
 include/net/ipv6.h          |  3 ++
 net/ipv6/ipv6_sockglue.c    | 31 +++++++++----
 net/mptcp/protocol.c        | 20 ++++++++-
 net/mptcp/sockopt.c         | 89 ++++++++++++++++++++++++++++++++++---
 net/mptcp/subflow.c         |  4 +-
 9 files changed, 166 insertions(+), 18 deletions(-)

-- 
2.53.0

[PATCH mptcp-next v5 01/11] mptcp: handle TCP_MAXSEG getsockopt in common case

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

Move TCP_MAXSEG from a tail-clause into the common case alongside other
"first subflow only" options (TCP_FASTOPEN_*), making the dispatch logic
more uniform.

No behavioural change.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/sockopt.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 922f6ae5c80c..2c1eaf66202a 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -1499,6 +1499,7 @@ static int mptcp_getsockopt_sol_tcp(struct mptcp_sock *msk, int optname,
 	case TCP_FASTOPEN_CONNECT:
 	case TCP_FASTOPEN_KEY:
 	case TCP_FASTOPEN_NO_COOKIE:
+	case TCP_MAXSEG:
 		return mptcp_getsockopt_first_sf_only(msk, SOL_TCP, optname,
 						      optval, optlen);
 	case TCP_INQ:
@@ -1523,9 +1524,6 @@ static int mptcp_getsockopt_sol_tcp(struct mptcp_sock *msk, int optname,
 		return mptcp_put_int_option(msk, optval, optlen, msk->notsent_lowat);
 	case TCP_IS_MPTCP:
 		return mptcp_put_int_option(msk, optval, optlen, 1);
-	case TCP_MAXSEG:
-		return mptcp_getsockopt_first_sf_only(msk, SOL_TCP, optname,
-						      optval, optlen);
 	}
 	return -EOPNOTSUPP;
 }
-- 
2.53.0

[PATCH mptcp-next v5 02/11] mptcp: take TCP_MAXSEG handling into setsockopt_set_val

[Geliang Tang]

From: Gang Yan <yangang@kylinos.cn>

Replace mptcp_setsockopt_all_sf() with __mptcp_setsockopt_set_val()
for TCP_MAXSEG option handling. This aligns the implementation with
other socket options that use the same helper and simplifies the code.

Co-developed-by: Geliang Tang <geliang@kernel.org>
Signed-off-by: Geliang Tang <geliang@kernel.org>
Signed-off-by: Gang Yan <yangang@kylinos.cn>
---
 net/mptcp/sockopt.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 2c1eaf66202a..5de371e29721 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -968,9 +968,9 @@ static int mptcp_setsockopt_sol_tcp(struct mptcp_sock *msk, int optname,
 						 val);
 		break;
 	case TCP_MAXSEG:
-		msk->maxseg = val;
-		ret = mptcp_setsockopt_all_sf(msk, SOL_TCP, optname, optval,
-					      optlen);
+		ret = __mptcp_setsockopt_set_val(msk, MAX_TCP_WINDOW,
+						 &tcp_sock_set_maxseg,
+						 &msk->maxseg, val);
 		break;
 	default:
 		ret = -ENOPROTOOPT;
-- 
2.53.0

[PATCH mptcp-next v5 03/11] mptcp: inherit sk_reuse/sk_reuseport and fix dev bind override

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

Propagate sk_reuse and sk_reuseport from master socket to subflows, and
bump setsockopt_seq so later subflows pick up the changes.

This patch also fixes unconditional sk_bound_dev_if override in
__mptcp_subflow_connect(): only set it when local->ifindex is non-zero,
otherwise preserve the value inherited from the master socket via
sync_socket_options().

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/sockopt.c | 5 +++++
 net/mptcp/subflow.c | 4 +++-
 2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 5de371e29721..d79139eb5b8a 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -335,6 +335,8 @@ static int mptcp_setsockopt_sol_socket(struct mptcp_sock *msk, int optname,
 				sk->sk_bound_dev_if = ssk->sk_bound_dev_if;
 			else if (optname == SO_BINDTOIFINDEX)
 				sk->sk_bound_dev_if = ssk->sk_bound_dev_if;
+
+			sockopt_seq_inc(msk);
 		}
 		release_sock(sk);
 		return ret;
@@ -1704,6 +1706,9 @@ static void sync_socket_options(struct mptcp_sock *msk, struct sock *ssk)
 		assign_bit(b, &inet_sk(ssk)->inet_flags, src & BIT(b));
 
 	WRITE_ONCE(inet_sk(ssk)->local_port_range, READ_ONCE(inet_sk(sk)->local_port_range));
+
+	ssk->sk_reuse = sk->sk_reuse;
+	ssk->sk_reuseport = sk->sk_reuseport;
 }
 
 void mptcp_sockopt_sync_locked(struct mptcp_sock *msk, struct sock *ssk)
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index af81ad5e699d..47ff736c4d66 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1672,7 +1672,9 @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_pm_local *local,
 	if (addr.ss_family == AF_INET6)
 		addrlen = sizeof(struct sockaddr_in6);
 #endif
-	ssk->sk_bound_dev_if = local->ifindex;
+	/* Only override the bound device if the path manager picked one. */
+	if (local->ifindex)
+		ssk->sk_bound_dev_if = local->ifindex;
 	err = kernel_bind(sf, (struct sockaddr_unsized *)&addr, addrlen);
 	if (err) {
 		MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_JOINSYNTXBINDERR);
-- 
2.53.0

[PATCH mptcp-next v5 04/11] mptcp: add TCP_SYNCNT setsockopt/getsockopt

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

This patch adds TCP_SYNCNT support on the MPTCP socket.

The setsockopt() path stores the value in inet_csk(sk)->icsk_syn_retries
on the msk and all subflow sockets. The getsockopt() path reads the same
field back from the master socket.

A newly created subflow picks up the master socket's syn_retries through
sync_socket_options(), so the setting takes effect for outgoing SYNs
without further plumbing.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/sockopt.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index d79139eb5b8a..71a4a2aaa838 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -974,6 +974,13 @@ static int mptcp_setsockopt_sol_tcp(struct mptcp_sock *msk, int optname,
 						 &tcp_sock_set_maxseg,
 						 &msk->maxseg, val);
 		break;
+	case TCP_SYNCNT:
+		ret = __mptcp_setsockopt_set_val(msk, MAX_TCP_SYNCNT,
+						 &tcp_sock_set_syncnt,
+						 &val, val);
+		if (ret == 0)
+			ret = tcp_sock_set_syncnt(sk, val);
+		break;
 	default:
 		ret = -ENOPROTOOPT;
 	}
@@ -1526,6 +1533,10 @@ static int mptcp_getsockopt_sol_tcp(struct mptcp_sock *msk, int optname,
 		return mptcp_put_int_option(msk, optval, optlen, msk->notsent_lowat);
 	case TCP_IS_MPTCP:
 		return mptcp_put_int_option(msk, optval, optlen, 1);
+	case TCP_SYNCNT:
+		return mptcp_put_int_option(msk, optval, optlen,
+					    READ_ONCE(inet_csk(sk)->icsk_syn_retries) ? :
+					    READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_syn_retries));
 	}
 	return -EOPNOTSUPP;
 }
@@ -1646,6 +1657,7 @@ static void sync_socket_options(struct mptcp_sock *msk, struct sock *ssk)
 	struct sock *sk = (struct sock *)msk;
 	unsigned long src;
 	bool keep_open;
+	int syncnt;
 	int b;
 
 	keep_open = sock_flag(sk, SOCK_KEEPOPEN);
@@ -1709,6 +1721,9 @@ static void sync_socket_options(struct mptcp_sock *msk, struct sock *ssk)
 
 	ssk->sk_reuse = sk->sk_reuse;
 	ssk->sk_reuseport = sk->sk_reuseport;
+	syncnt = READ_ONCE(inet_csk(sk)->icsk_syn_retries);
+	if (syncnt > 0 && tcp_sock_set_syncnt(ssk, syncnt))
+		pr_warn("Failed to sync TCP_SYNCNT=%u to subflow\n", syncnt);
 }
 
 void mptcp_sockopt_sync_locked(struct mptcp_sock *msk, struct sock *ssk)
-- 
2.53.0

[PATCH mptcp-next v5 05/11] mptcp: skip ip6 options on v4-mapped clone

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

v4-mapped IPv6 destinations carry no IPv6-specific options, so skip
mptcp_copy_ip6_options() when sk_v6_daddr is in ::ffff:0:0/96. Native
IPv6 subflows are unchanged.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/protocol.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index c2762d74f29d..f19777c04d32 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -3750,7 +3750,8 @@ struct sock *mptcp_sk_clone_init(const struct sock *sk,
 	__mptcp_init_sock(nsk);
 
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
-	if (nsk->sk_family == AF_INET6)
+	if (nsk->sk_family == AF_INET6 &&
+	    !ipv6_addr_v4mapped(&nsk->sk_v6_daddr))
 		mptcp_copy_ip6_options(nsk, sk);
 	else
 #endif
-- 
2.53.0

[PATCH mptcp-next v5 06/11] mptcp: report v4-mapped subflow addrs as IPv4

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

mptcp_get_sub_addrs() currently fills sin6_* for any AF_INET6 subflow.
Detect v4-mapped destinations (guarded by CONFIG_IPV6) and reuse the
IPv4 branch so MPTCP_SUBFLOW_ADDRS reports match userspace's view.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/sockopt.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 71a4a2aaa838..43a8d1c37137 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -1256,10 +1256,15 @@ static int mptcp_getsockopt_tcpinfo(struct mptcp_sock *msk, char __user *optval,
 static void mptcp_get_sub_addrs(const struct sock *sk, struct mptcp_subflow_addrs *a)
 {
 	const struct inet_sock *inet = inet_sk(sk);
+	bool is_v4 = sk->sk_family == AF_INET;
 
 	memset(a, 0, sizeof(*a));
 
-	if (sk->sk_family == AF_INET) {
+#if IS_ENABLED(CONFIG_IPV6)
+	if (!is_v4 && sk->sk_family == AF_INET6)
+		is_v4 = ipv6_addr_v4mapped(&sk->sk_v6_daddr);
+#endif
+	if (is_v4) {
 		a->sin_local.sin_family = AF_INET;
 		a->sin_local.sin_port = inet->inet_sport;
 		a->sin_local.sin_addr.s_addr = inet->inet_rcv_saddr;
-- 
2.53.0

[PATCH mptcp-next v5 07/11] ipv6: extract and export ip6_sock_set_tclass helpers

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

The IPV6_TCLASS case in do_ipv6_setsockopt() inlined the same ECN mask
handling and inet6_sk(sk)->tclass write that callers need to do when
propagating a tclass value onto a freshly created socket.

Pull this into small helpers exported via <net/ipv6.h>, so external
modules (e.g. nvme-tcp, nvmet-tcp) can apply IPV6_TCLASS without
duplicating the ECN handling.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 include/net/ipv6.h       |  3 +++
 net/ipv6/ipv6_sockglue.c | 31 +++++++++++++++++++++++--------
 2 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index 3de07e738538..82e57b34ba57 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -1252,6 +1252,9 @@ static inline void ip6_sock_set_recverr(struct sock *sk)
 	inet6_set_bit(RECVERR6, sk);
 }
 
+void __ip6_sock_set_tclass(struct sock *sk, int val);
+void ip6_sock_set_tclass(struct sock *sk, int val);
+
 #define IPV6_PREFER_SRC_MASK (IPV6_PREFER_SRC_TMP | IPV6_PREFER_SRC_PUBLIC | \
 			      IPV6_PREFER_SRC_COA)
 
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index b4c977434c2e..c59f25fbb336 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -373,6 +373,28 @@ static int ipv6_set_opt_hdr(struct sock *sk, int optname, sockptr_t optval,
 	return err;
 }
 
+void __ip6_sock_set_tclass(struct sock *sk, int val)
+{
+	u8 old_tclass = inet6_sk(sk)->tclass;
+
+	if (sk->sk_type == SOCK_STREAM) {
+		val &= ~INET_ECN_MASK;
+		val |= old_tclass & INET_ECN_MASK;
+	}
+	if (old_tclass != val) {
+		WRITE_ONCE(inet6_sk(sk)->tclass, val);
+		sk_dst_reset(sk);
+	}
+}
+
+void ip6_sock_set_tclass(struct sock *sk, int val)
+{
+	sockopt_lock_sock(sk);
+	__ip6_sock_set_tclass(sk, val);
+	sockopt_release_sock(sk);
+}
+EXPORT_SYMBOL(ip6_sock_set_tclass);
+
 int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
 		       sockptr_t optval, unsigned int optlen)
 {
@@ -713,14 +735,7 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
 		/* RFC 3542, 6.5: default traffic class of 0x0 */
 		if (val == -1)
 			val = 0;
-		if (sk->sk_type == SOCK_STREAM) {
-			val &= ~INET_ECN_MASK;
-			val |= np->tclass & INET_ECN_MASK;
-		}
-		if (np->tclass != val) {
-			np->tclass = val;
-			sk_dst_reset(sk);
-		}
+		__ip6_sock_set_tclass(sk, val);
 		retv = 0;
 		break;
 
-- 
2.53.0

[PATCH mptcp-next v5 08/11] mptcp: sockopt: implement IPV6_TCLASS

[Geliang Tang]

From: David 'equinox' Lamparter <equinox@diac24.net>

The IPV6_TCLASS setsockopt just needs to be forwarded to the individual
TCP sockets, like IP_TOS is already handled for IPv4.

Coincidentally, ssh uses this sockopt and prints an error in the middle
of your ongoing SSH session when it doesn't work (very annoying when
doing SCP/SFTP on a multiplexed session.)

For setsockopt, the value is first applied to the MPTCP socket itself via
ipv6_setsockopt(), then propagated to all existing subflows using
__ip6_sock_set_tclass(). The setsockopt_seq is bumped to ensure that
subflows created later will pick up the setting through
sync_socket_options().

For getsockopt, the value is read directly from the MPTCP socket.

Also add sync_socket_options() support for IPV6_TCLASS, so that new
subflows created after the option is set inherit the correct tclass
value.

Cc: Mat Martineau <martineau@kernel.org>
Cc: Matthieu Baerts <matttbe@kernel.org>
Co-developed-by: Geliang Tang <geliang@kernel.org>
Signed-off-by: Geliang Tang <geliang@kernel.org>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/568
Signed-off-by: David 'equinox' Lamparter <equinox@diac24.net>
---
 net/mptcp/sockopt.c | 52 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 52 insertions(+)

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 43a8d1c37137..ac326e3f289e 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -485,6 +485,42 @@ static int mptcp_setsockopt_recverr(struct mptcp_sock *msk, int level,
 	return ret;
 }
 
+#if IS_ENABLED(CONFIG_IPV6)
+static int mptcp_setsockopt_v6_set_tclass(struct mptcp_sock *msk, int optname,
+					  sockptr_t optval, unsigned int optlen)
+{
+	struct mptcp_subflow_context *subflow;
+	struct sock *sk = (struct sock *)msk;
+	int err, val;
+
+	if (sk->sk_family != AF_INET6)
+		return -EOPNOTSUPP;
+
+	err = ipv6_setsockopt(sk, SOL_IPV6, optname, optval, optlen);
+
+	if (err != 0)
+		return err;
+
+	lock_sock(sk);
+	sockopt_seq_inc(msk);
+	val = READ_ONCE(inet6_sk(sk)->tclass);
+	mptcp_for_each_subflow(msk, subflow) {
+		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
+		bool slow;
+
+		if (ssk->sk_family != AF_INET6)
+			continue;
+
+		slow = lock_sock_fast(ssk);
+		__ip6_sock_set_tclass(ssk, val);
+		unlock_sock_fast(ssk, slow);
+	}
+	release_sock(sk);
+
+	return 0;
+}
+#endif
+
 static int mptcp_setsockopt_v6(struct mptcp_sock *msk, int optname,
 			       sockptr_t optval, unsigned int optlen)
 {
@@ -532,6 +568,11 @@ static int mptcp_setsockopt_v6(struct mptcp_sock *msk, int optname,
 		ret = mptcp_setsockopt_recverr(msk, SOL_IPV6, optname, optval,
 					       optlen);
 		break;
+#if IS_ENABLED(CONFIG_IPV6)
+	case IPV6_TCLASS:
+		return mptcp_setsockopt_v6_set_tclass(msk, optname, optval,
+						      optlen);
+#endif
 	}
 
 	return ret;
@@ -1602,6 +1643,13 @@ static int mptcp_getsockopt_v6(struct mptcp_sock *msk, int optname,
 			return -ENOPROTOOPT;
 		return mptcp_put_int_option(msk, optval, optlen,
 					    inet6_test_bit(RECVERR6_RFC4884, sk));
+#if IS_ENABLED(CONFIG_IPV6)
+	case IPV6_TCLASS:
+		if (sk->sk_family != AF_INET6)
+			return -EOPNOTSUPP;
+		return mptcp_put_int_option(msk, optval, optlen,
+					    READ_ONCE(inet6_sk(sk)->tclass));
+#endif
 	}
 
 	return -EOPNOTSUPP;
@@ -1729,6 +1777,10 @@ static void sync_socket_options(struct mptcp_sock *msk, struct sock *ssk)
 	syncnt = READ_ONCE(inet_csk(sk)->icsk_syn_retries);
 	if (syncnt > 0 && tcp_sock_set_syncnt(ssk, syncnt))
 		pr_warn("Failed to sync TCP_SYNCNT=%u to subflow\n", syncnt);
+#if IS_ENABLED(CONFIG_IPV6)
+	if (sk->sk_family == AF_INET6 && ssk->sk_family == AF_INET6)
+		__ip6_sock_set_tclass(ssk, READ_ONCE(inet6_sk(sk)->tclass));
+#endif
 }
 
 void mptcp_sockopt_sync_locked(struct mptcp_sock *msk, struct sock *ssk)
-- 
2.53.0

[PATCH mptcp-next v5 09/11] mptcp: copy the subflow's tos/tclass to the msk on accept

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

When a new MPTCP socket is accepted, copy the tos (and rcv_tos for IPv4)
or tclass (and rcv_flowinfo for IPv6) from the first subflow to the
newly-created msk.

This mirrors what plain TCP does in tcp_v4_syn_recv_sock() and
tcp_v6_syn_recv_sock(), where these values are copied from the incoming
skb to the accepted socket.

Without this, MP_JOIN subflows created later would inherit tos/tclass=0
from the msk and lose the peer's reflected value, causing inconsistent
traffic class behavior.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/protocol.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index f19777c04d32..e4fc929abe55 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -3757,6 +3757,23 @@ struct sock *mptcp_sk_clone_init(const struct sock *sk,
 #endif
 		mptcp_copy_ip_options(nsk, sk);
 
+	if (ssk->sk_family == AF_INET) {
+		inet_sk(nsk)->rcv_tos = inet_sk(ssk)->rcv_tos;
+		__ip_sock_set_tos(nsk, inet_sk(ssk)->tos);
+	}
+#if IS_ENABLED(CONFIG_MPTCP_IPV6)
+	else if (ssk->sk_family == AF_INET6 &&
+		 ipv6_addr_v4mapped(&ssk->sk_v6_daddr)) {
+		inet_sk(nsk)->rcv_tos = inet_sk(ssk)->rcv_tos;
+		__ip_sock_set_tos(nsk, inet_sk(ssk)->tos);
+		inet6_sk(nsk)->rcv_flowinfo = 0;
+	}
+	else if (ssk->sk_family == AF_INET6) {
+		inet6_sk(nsk)->rcv_flowinfo = inet6_sk(ssk)->rcv_flowinfo;
+		__ip6_sock_set_tclass(nsk, inet6_sk(ssk)->tclass);
+	}
+#endif
+
 	msk = mptcp_sk(nsk);
 	WRITE_ONCE(msk->local_key, subflow_req->local_key);
 	WRITE_ONCE(msk->token, subflow_req->token);
-- 
2.53.0

[PATCH mptcp-next v5 10/11] nvme-tcp: support IPv6 traffic class

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

The NVMe/TCP transport needs to configure the IPv6 traffic class on
its queue sockets, but the fabrics option parser currently has no
corresponding option.

Add the tclass option to the fabrics parser and store the validated
value in struct nvmf_ctrl_options. Negative values are rejected, while
values greater than 255 are clamped to 255, matching the valid range
of the IPv6 traffic class field.

Keep -1 as the default value to let the transport use its default
traffic class.

The fabrics layer now supports parsing the tclass option, but the
NVMe/TCP transport does not currently advertise or apply it.

Allow NVMe/TCP to use NVMF_OPT_TCLASS and set IPV6_TCLASS on each queue
socket when a traffic class is specified. Keep the existing IPv4 TOS
handling unchanged.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 drivers/nvme/host/fabrics.c | 18 ++++++++++++++++++
 drivers/nvme/host/fabrics.h |  3 +++
 drivers/nvme/host/tcp.c     |  8 +++++++-
 3 files changed, 28 insertions(+), 1 deletion(-)

diff --git a/drivers/nvme/host/fabrics.c b/drivers/nvme/host/fabrics.c
index ac3d4f400601..643c03dc7bcb 100644
--- a/drivers/nvme/host/fabrics.c
+++ b/drivers/nvme/host/fabrics.c
@@ -695,6 +695,7 @@ static const match_table_t opt_tokens = {
 	{ NVMF_OPT_NR_WRITE_QUEUES,	"nr_write_queues=%d"	},
 	{ NVMF_OPT_NR_POLL_QUEUES,	"nr_poll_queues=%d"	},
 	{ NVMF_OPT_TOS,			"tos=%d"		},
+	{ NVMF_OPT_TCLASS,		"tclass=%d"		},
 #ifdef CONFIG_NVME_TCP_TLS
 	{ NVMF_OPT_KEYRING,		"keyring=%d"		},
 	{ NVMF_OPT_TLS_KEY,		"tls_key=%d"		},
@@ -734,6 +735,7 @@ static int nvmf_parse_options(struct nvmf_ctrl_options *opts,
 	opts->hdr_digest = false;
 	opts->data_digest = false;
 	opts->tos = -1; /* < 0 == use transport default */
+	opts->tclass = -1; /* < 0 == use transport default */
 	opts->tls = false;
 	opts->tls_key = NULL;
 	opts->keyring = NULL;
@@ -991,6 +993,22 @@ static int nvmf_parse_options(struct nvmf_ctrl_options *opts,
 			}
 			opts->tos = token;
 			break;
+		case NVMF_OPT_TCLASS:
+			if (match_int(args, &token)) {
+				ret = -EINVAL;
+				goto out;
+			}
+			if (token < 0) {
+				pr_err("Invalid traffic class %d\n", token);
+				ret = -EINVAL;
+				goto out;
+			}
+			if (token > 255) {
+				pr_warn("Clamping traffic class to 255\n");
+				token = 255;
+			}
+			opts->tclass = token;
+			break;
 		case NVMF_OPT_KEYRING:
 			if (match_int(args, &key_id) || key_id <= 0) {
 				ret = -EINVAL;
diff --git a/drivers/nvme/host/fabrics.h b/drivers/nvme/host/fabrics.h
index caf5503d0833..3dfd40bd8960 100644
--- a/drivers/nvme/host/fabrics.h
+++ b/drivers/nvme/host/fabrics.h
@@ -67,6 +67,7 @@ enum {
 	NVMF_OPT_KEYRING	= 1 << 26,
 	NVMF_OPT_TLS_KEY	= 1 << 27,
 	NVMF_OPT_CONCAT		= 1 << 28,
+	NVMF_OPT_TCLASS		= 1 << 29,
 };
 
 /**
@@ -109,6 +110,7 @@ enum {
  * @nr_write_queues: number of queues for write I/O
  * @nr_poll_queues: number of queues for polling I/O
  * @tos: type of service
+ * @tclass: IPv6 traffic class
  * @fast_io_fail_tmo: Fast I/O fail timeout in seconds
  */
 struct nvmf_ctrl_options {
@@ -139,6 +141,7 @@ struct nvmf_ctrl_options {
 	unsigned int		nr_write_queues;
 	unsigned int		nr_poll_queues;
 	int			tos;
+	int			tclass;
 	int			fast_io_fail_tmo;
 };
 
diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
index ba5c7b3e2a7c..cabbf67d24f8 100644
--- a/drivers/nvme/host/tcp.c
+++ b/drivers/nvme/host/tcp.c
@@ -1837,6 +1837,11 @@ static int nvme_tcp_alloc_queue(struct nvme_ctrl *nctrl, int qid,
 	/* Set socket type of service */
 	if (nctrl->opts->tos >= 0)
 		ip_sock_set_tos(queue->sock->sk, nctrl->opts->tos);
+#if IS_ENABLED(CONFIG_IPV6)
+	if (nctrl->opts->tclass >= 0 &&
+	    queue->sock->sk->sk_family == AF_INET6)
+		ip6_sock_set_tclass(queue->sock->sk, nctrl->opts->tclass);
+#endif
 
 	/* Set 10 seconds timeout for icresp recvmsg */
 	queue->sock->sk->sk_rcvtimeo = 10 * HZ;
@@ -3041,7 +3046,8 @@ static struct nvmf_transport_ops nvme_tcp_transport = {
 			  NVMF_OPT_HOST_TRADDR | NVMF_OPT_CTRL_LOSS_TMO |
 			  NVMF_OPT_HDR_DIGEST | NVMF_OPT_DATA_DIGEST |
 			  NVMF_OPT_NR_WRITE_QUEUES | NVMF_OPT_NR_POLL_QUEUES |
-			  NVMF_OPT_TOS | NVMF_OPT_HOST_IFACE | NVMF_OPT_TLS |
+			  NVMF_OPT_TOS | NVMF_OPT_TCLASS |
+			  NVMF_OPT_HOST_IFACE | NVMF_OPT_TLS |
 			  NVMF_OPT_KEYRING | NVMF_OPT_TLS_KEY | NVMF_OPT_CONCAT,
 	.create_ctrl	= nvme_tcp_create_ctrl,
 };
-- 
2.53.0

[PATCH mptcp-next v5 11/11] nvmet-tcp: support IPv6 traffic class

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

nvmet-tcp currently applies the received IPv4 TOS value when setting
up a queue socket, but does not handle the IPv6 traffic class.

Extend the queue socket setup to handle AF_INET6 sockets. Obtain the
traffic class from the IPv6 socket and apply it through IPV6_TCLASS.
Keep the existing IPv4 TOS handling unchanged.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 drivers/nvme/target/tcp.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/nvme/target/tcp.c b/drivers/nvme/target/tcp.c
index 75a276d73be3..1d65519ebc0f 100644
--- a/drivers/nvme/target/tcp.c
+++ b/drivers/nvme/target/tcp.c
@@ -1725,6 +1725,14 @@ static int nvmet_tcp_set_queue_sock(struct nvmet_tcp_queue *queue)
 	/* Set socket type of service */
 	if (inet->rcv_tos > 0)
 		ip_sock_set_tos(sock->sk, inet->rcv_tos);
+#if IS_ENABLED(CONFIG_IPV6)
+	if (sock->sk->sk_family == AF_INET6) {
+		u8 rcv_tclass = ip6_tclass(inet6_sk(sock->sk)->rcv_flowinfo);
+
+		if (rcv_tclass > 0)
+			ip6_sock_set_tclass(sock->sk, rcv_tclass);
+	}
+#endif
 
 	ret = 0;
 	write_lock_bh(&sock->sk->sk_callback_lock);
-- 
2.53.0

Re: [PATCH mptcp-next v5 00/11] Fix socket options used by NVMe over MPTCP

[MPTCP CI]

Hi Geliang,

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): Success! ✅
- KVM Validation: normal (only selftest_mptcp_join): Success! ✅
- KVM Validation: debug (except selftest_mptcp_join): Success! ✅
- KVM Validation: debug (only selftest_mptcp_join): Success! ✅
- KVM Validation: btf-normal (only bpftest_all): Success! ✅
- KVM Validation: btf-debug (only bpftest_all): Success! ✅
- Task: https://github.com/multipath-tcp/mptcp_net-next/actions/runs/31238427573

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/b9ff91daca52
Patchwork: https://patchwork.kernel.org/project/mptcp/list/?series=1142524


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-normal

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 (NGI0 Core)