From dea3be40464a1af2f7eca83ee2408b7d3fd21c23 Mon Sep 17 00:00:00 2001
From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Date: Mon, 16 Jun 2025 09:24:08 +0200
Subject: [PATCH 01/16] net: phy: qca807x: use new GPIO line value setter
 callbacks

struct gpio_chip now has callbacks for setting line values that return
an integer, allowing to indicate failures. Convert the driver to using
them.

Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20250616-gpiochip-set-rv-net-v2-5-cae0b182a552@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 drivers/net/phy/qcom/qca807x.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/drivers/net/phy/qcom/qca807x.c b/drivers/net/phy/qcom/qca807x.c
index 1af6b5ead74bf..6d10ef7e9a8a4 100644
--- a/drivers/net/phy/qcom/qca807x.c
+++ b/drivers/net/phy/qcom/qca807x.c
@@ -377,7 +377,7 @@ static int qca807x_gpio_get(struct gpio_chip *gc, unsigned int offset)
 	return FIELD_GET(QCA807X_GPIO_FORCE_MODE_MASK, val);
 }
 
-static void qca807x_gpio_set(struct gpio_chip *gc, unsigned int offset, int value)
+static int qca807x_gpio_set(struct gpio_chip *gc, unsigned int offset, int value)
 {
 	struct qca807x_gpio_priv *priv = gpiochip_get_data(gc);
 	u16 reg;
@@ -386,18 +386,19 @@ static void qca807x_gpio_set(struct gpio_chip *gc, unsigned int offset, int valu
 	reg = QCA807X_MMD7_LED_FORCE_CTRL(offset);
 
 	val = phy_read_mmd(priv->phy, MDIO_MMD_AN, reg);
+	if (val < 0)
+		return val;
+
 	val &= ~QCA807X_GPIO_FORCE_MODE_MASK;
 	val |= QCA807X_GPIO_FORCE_EN;
 	val |= FIELD_PREP(QCA807X_GPIO_FORCE_MODE_MASK, value);
 
-	phy_write_mmd(priv->phy, MDIO_MMD_AN, reg, val);
+	return phy_write_mmd(priv->phy, MDIO_MMD_AN, reg, val);
 }
 
 static int qca807x_gpio_dir_out(struct gpio_chip *gc, unsigned int offset, int value)
 {
-	qca807x_gpio_set(gc, offset, value);
-
-	return 0;
+	return qca807x_gpio_set(gc, offset, value);
 }
 
 static int qca807x_gpio(struct phy_device *phydev)
@@ -425,7 +426,7 @@ static int qca807x_gpio(struct phy_device *phydev)
 	gc->get_direction = qca807x_gpio_get_direction;
 	gc->direction_output = qca807x_gpio_dir_out;
 	gc->get = qca807x_gpio_get;
-	gc->set = qca807x_gpio_set;
+	gc->set_rv = qca807x_gpio_set;
 
 	return devm_gpiochip_add_data(dev, gc, priv);
 }
-- 
2.51.0


From 2de1ba0887e5d3bf02d7c212f380039b34e10aa3 Mon Sep 17 00:00:00 2001
From: Gal Pressman <gal@nvidia.com>
Date: Mon, 16 Jun 2025 16:26:24 +0300
Subject: [PATCH 02/16] net: vlan: Make is_vlan_dev() a stub when VLAN is not
 configured

Add a stub implementation of is_vlan_dev() that returns false when
VLAN support is not compiled in (CONFIG_VLAN_8021Q=n).

This allows us to compile-out VLAN-dependent dead code when it is not
needed.

This also resolves the following compilation error when:
* CONFIG_VLAN_8021Q=n
* CONFIG_OBJTOOL=y
* CONFIG_OBJTOOL_WERROR=y

drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.o: error: objtool: parse_mirred.isra.0+0x370: mlx5e_tc_act_vlan_add_push_action() missing __noreturn in .c/.h or NORETURN() in noreturns.h

The error occurs because objtool cannot determine that unreachable BUG()
(which doesn't return) calls in VLAN code paths are actually dead code
when VLAN support is disabled.

Signed-off-by: Gal Pressman <gal@nvidia.com>
Link: https://patch.msgid.link/20250616132626.1749331-2-gal@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/linux/if_vlan.h | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h
index 38456b42cdb55..618a973ff8ee1 100644
--- a/include/linux/if_vlan.h
+++ b/include/linux/if_vlan.h
@@ -79,11 +79,6 @@ static inline struct vlan_ethhdr *skb_vlan_eth_hdr(const struct sk_buff *skb)
 /* found in socket.c */
 extern void vlan_ioctl_set(int (*hook)(struct net *, void __user *));
 
-static inline bool is_vlan_dev(const struct net_device *dev)
-{
-        return dev->priv_flags & IFF_802_1Q_VLAN;
-}
-
 #define skb_vlan_tag_present(__skb)	(!!(__skb)->vlan_all)
 #define skb_vlan_tag_get(__skb)		((__skb)->vlan_tci)
 #define skb_vlan_tag_get_id(__skb)	((__skb)->vlan_tci & VLAN_VID_MASK)
@@ -200,6 +195,11 @@ struct vlan_dev_priv {
 #endif
 };
 
+static inline bool is_vlan_dev(const struct net_device *dev)
+{
+	return dev->priv_flags & IFF_802_1Q_VLAN;
+}
+
 static inline struct vlan_dev_priv *vlan_dev_priv(const struct net_device *dev)
 {
 	return netdev_priv(dev);
@@ -237,6 +237,11 @@ extern void vlan_vids_del_by_dev(struct net_device *dev,
 extern bool vlan_uses_dev(const struct net_device *dev);
 
 #else
+static inline bool is_vlan_dev(const struct net_device *dev)
+{
+	return false;
+}
+
 static inline struct net_device *
 __vlan_find_dev_deep_rcu(struct net_device *real_dev,
 		     __be16 vlan_proto, u16 vlan_id)
-- 
2.51.0


From 60a8b1a5d0824afda869f18dc0ecfe72f8dfda42 Mon Sep 17 00:00:00 2001
From: Gal Pressman <gal@nvidia.com>
Date: Mon, 16 Jun 2025 16:26:25 +0300
Subject: [PATCH 03/16] net: vlan: Replace BUG() with WARN_ON_ONCE() in
 vlan_dev_* stubs

When CONFIG_VLAN_8021Q=n, a set of stub helpers are used, three of these
helpers use BUG() unconditionally.

This code should not be reached, as callers of these functions should
always check for is_vlan_dev() first, but the usage of BUG() is not
recommended, replace it with WARN_ON() instead.

Reviewed-by: Alex Lazar <alazar@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Link: https://patch.msgid.link/20250616132626.1749331-3-gal@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/linux/if_vlan.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h
index 618a973ff8ee1..b9f699799cf6e 100644
--- a/include/linux/if_vlan.h
+++ b/include/linux/if_vlan.h
@@ -259,19 +259,19 @@ vlan_for_each(struct net_device *dev,
 
 static inline struct net_device *vlan_dev_real_dev(const struct net_device *dev)
 {
-	BUG();
+	WARN_ON_ONCE(1);
 	return NULL;
 }
 
 static inline u16 vlan_dev_vlan_id(const struct net_device *dev)
 {
-	BUG();
+	WARN_ON_ONCE(1);
 	return 0;
 }
 
 static inline __be16 vlan_dev_vlan_proto(const struct net_device *dev)
 {
-	BUG();
+	WARN_ON_ONCE(1);
 	return 0;
 }
 
-- 
2.51.0


From 9c5f5a5bf0da5cee2044b93907ac6d8d9af0492b Mon Sep 17 00:00:00 2001
From: Gal Pressman <gal@nvidia.com>
Date: Mon, 16 Jun 2025 16:26:26 +0300
Subject: [PATCH 04/16] net: vlan: Use IS_ENABLED() helper for
 CONFIG_VLAN_8021Q guard
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

The header currently tests the VLAN core with an explicit pair of 'if
defined' checks:
    #if defined(CONFIG_VLAN_8021Q) || defined(CONFIG_VLAN_8021Q_MODULE)

Instead, use IS_ENABLED() which is the kernel way to test whether an
option is configured as builtin/module.

This is purely cosmetic – no functional changes.

Reviewed-by: Alex Lazar <alazar@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Link: https://patch.msgid.link/20250616132626.1749331-4-gal@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/linux/if_vlan.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h
index b9f699799cf6e..15e01935d3fad 100644
--- a/include/linux/if_vlan.h
+++ b/include/linux/if_vlan.h
@@ -131,7 +131,7 @@ struct vlan_pcpu_stats {
 	u32			tx_dropped;
 };
 
-#if defined(CONFIG_VLAN_8021Q) || defined(CONFIG_VLAN_8021Q_MODULE)
+#if IS_ENABLED(CONFIG_VLAN_8021Q)
 
 extern struct net_device *__vlan_find_dev_deep_rcu(struct net_device *real_dev,
 					       __be16 vlan_proto, u16 vlan_id);
-- 
2.51.0


From e3411e326fa48c9be09ba449330352ba698db698 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:09 +0200
Subject: [PATCH 05/16] net: ipv4: Add a flags argument to iptunnel_xmit(),
 udp_tunnel_xmit_skb()

iptunnel_xmit() erases the contents of the SKB control block. In order to
be able to set particular IPCB flags on the SKB, add a corresponding
parameter, and propagate it to udp_tunnel_xmit_skb() as well.

In one of the following patches, VXLAN driver will use this facility to
mark packets as subject to IP multicast routing.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Acked-by: Antonio Quartulli <antonio@openvpn.net>
Link: https://patch.msgid.link/89c9daf9f2dc088b6b92ccebcc929f51742de91f.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 drivers/net/amt.c              |  9 ++++++---
 drivers/net/bareudp.c          |  4 ++--
 drivers/net/geneve.c           |  4 ++--
 drivers/net/gtp.c              | 10 ++++++----
 drivers/net/ovpn/udp.c         |  2 +-
 drivers/net/vxlan/vxlan_core.c |  2 +-
 drivers/net/wireguard/socket.c |  2 +-
 include/net/ip_tunnels.h       |  2 +-
 include/net/udp_tunnel.h       |  2 +-
 net/ipv4/ip_tunnel.c           |  4 ++--
 net/ipv4/ip_tunnel_core.c      |  4 +++-
 net/ipv4/udp_tunnel_core.c     |  5 +++--
 net/ipv6/sit.c                 |  2 +-
 net/sctp/protocol.c            |  3 ++-
 net/tipc/udp_media.c           |  2 +-
 15 files changed, 33 insertions(+), 24 deletions(-)

diff --git a/drivers/net/amt.c b/drivers/net/amt.c
index fb130fde68c07..ed86537b2f61c 100644
--- a/drivers/net/amt.c
+++ b/drivers/net/amt.c
@@ -1046,7 +1046,8 @@ static bool amt_send_membership_update(struct amt_dev *amt,
 			    amt->gw_port,
 			    amt->relay_port,
 			    false,
-			    false);
+			    false,
+			    0);
 	amt_update_gw_status(amt, AMT_STATUS_SENT_UPDATE, true);
 	return false;
 }
@@ -1103,7 +1104,8 @@ static void amt_send_multicast_data(struct amt_dev *amt,
 			    amt->relay_port,
 			    tunnel->source_port,
 			    false,
-			    false);
+			    false,
+			    0);
 }
 
 static bool amt_send_membership_query(struct amt_dev *amt,
@@ -1161,7 +1163,8 @@ static bool amt_send_membership_query(struct amt_dev *amt,
 			    amt->relay_port,
 			    tunnel->source_port,
 			    false,
-			    false);
+			    false,
+			    0);
 	amt_update_relay_status(tunnel, AMT_STATUS_SENT_QUERY, true);
 	return false;
 }
diff --git a/drivers/net/bareudp.c b/drivers/net/bareudp.c
index a9dffdcac8057..5e613080d3f80 100644
--- a/drivers/net/bareudp.c
+++ b/drivers/net/bareudp.c
@@ -362,8 +362,8 @@ static int bareudp_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	udp_tunnel_xmit_skb(rt, sock->sk, skb, saddr, info->key.u.ipv4.dst,
 			    tos, ttl, df, sport, bareudp->port,
 			    !net_eq(bareudp->net, dev_net(bareudp->dev)),
-			    !test_bit(IP_TUNNEL_CSUM_BIT,
-				      info->key.tun_flags));
+			    !test_bit(IP_TUNNEL_CSUM_BIT, info->key.tun_flags),
+			    0);
 	return 0;
 
 free_dst:
diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index ffc15a4326899..c668e8b00ed26 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -921,8 +921,8 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	udp_tunnel_xmit_skb(rt, gs4->sock->sk, skb, saddr, info->key.u.ipv4.dst,
 			    tos, ttl, df, sport, geneve->cfg.info.key.tp_dst,
 			    !net_eq(geneve->net, dev_net(geneve->dev)),
-			    !test_bit(IP_TUNNEL_CSUM_BIT,
-				      info->key.tun_flags));
+			    !test_bit(IP_TUNNEL_CSUM_BIT, info->key.tun_flags),
+			    0);
 	return 0;
 }
 
diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c
index d4dec741c7f44..14584793fe4e8 100644
--- a/drivers/net/gtp.c
+++ b/drivers/net/gtp.c
@@ -446,7 +446,8 @@ static int gtp0_send_echo_resp_ip(struct gtp_dev *gtp, struct sk_buff *skb)
 			    htons(GTP0_PORT), htons(GTP0_PORT),
 			    !net_eq(sock_net(gtp->sk1u),
 				    dev_net(gtp->dev)),
-			    false);
+			    false,
+			    0);
 
 	return 0;
 }
@@ -704,7 +705,8 @@ static int gtp1u_send_echo_resp(struct gtp_dev *gtp, struct sk_buff *skb)
 			    htons(GTP1U_PORT), htons(GTP1U_PORT),
 			    !net_eq(sock_net(gtp->sk1u),
 				    dev_net(gtp->dev)),
-			    false);
+			    false,
+			    0);
 	return 0;
 }
 
@@ -1304,7 +1306,7 @@ static netdev_tx_t gtp_dev_xmit(struct sk_buff *skb, struct net_device *dev)
 				    pktinfo.gtph_port, pktinfo.gtph_port,
 				    !net_eq(sock_net(pktinfo.pctx->sk),
 					    dev_net(dev)),
-				    false);
+				    false, 0);
 		break;
 	case AF_INET6:
 #if IS_ENABLED(CONFIG_IPV6)
@@ -2405,7 +2407,7 @@ static int gtp_genl_send_echo_req(struct sk_buff *skb, struct genl_info *info)
 			    port, port,
 			    !net_eq(sock_net(sk),
 				    dev_net(gtp->dev)),
-			    false);
+			    false, 0);
 	return 0;
 }
 
diff --git a/drivers/net/ovpn/udp.c b/drivers/net/ovpn/udp.c
index bff00946eae2d..d866e6bfda704 100644
--- a/drivers/net/ovpn/udp.c
+++ b/drivers/net/ovpn/udp.c
@@ -199,7 +199,7 @@ static int ovpn_udp4_output(struct ovpn_peer *peer, struct ovpn_bind *bind,
 transmit:
 	udp_tunnel_xmit_skb(rt, sk, skb, fl.saddr, fl.daddr, 0,
 			    ip4_dst_hoplimit(&rt->dst), 0, fl.fl4_sport,
-			    fl.fl4_dport, false, sk->sk_no_check_tx);
+			    fl.fl4_dport, false, sk->sk_no_check_tx, 0);
 	ret = 0;
 err:
 	local_bh_enable();
diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 97792de896b72..1cc18acd242dc 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2522,7 +2522,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 
 		udp_tunnel_xmit_skb(rt, sock4->sock->sk, skb, saddr,
 				    pkey->u.ipv4.dst, tos, ttl, df,
-				    src_port, dst_port, xnet, !udp_sum);
+				    src_port, dst_port, xnet, !udp_sum, 0);
 #if IS_ENABLED(CONFIG_IPV6)
 	} else {
 		struct vxlan_sock *sock6 = rcu_dereference(vxlan->vn6_sock);
diff --git a/drivers/net/wireguard/socket.c b/drivers/net/wireguard/socket.c
index 0414d7a6ce741..88e685667bc0d 100644
--- a/drivers/net/wireguard/socket.c
+++ b/drivers/net/wireguard/socket.c
@@ -84,7 +84,7 @@ static int send4(struct wg_device *wg, struct sk_buff *skb,
 	skb->ignore_df = 1;
 	udp_tunnel_xmit_skb(rt, sock, skb, fl.saddr, fl.daddr, ds,
 			    ip4_dst_hoplimit(&rt->dst), 0, fl.fl4_sport,
-			    fl.fl4_dport, false, false);
+			    fl.fl4_dport, false, false, 0);
 	goto out;
 
 err:
diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index 0c3d571a04a16..8cf1380f36562 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -603,7 +603,7 @@ static inline int iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
 
 void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
 		   __be32 src, __be32 dst, u8 proto,
-		   u8 tos, u8 ttl, __be16 df, bool xnet);
+		   u8 tos, u8 ttl, __be16 df, bool xnet, u16 ipcb_flags);
 struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md,
 					     gfp_t flags);
 int skb_tunnel_check_pmtu(struct sk_buff *skb, struct dst_entry *encap_dst,
diff --git a/include/net/udp_tunnel.h b/include/net/udp_tunnel.h
index 2df3b8344eb52..28102c8fd8a85 100644
--- a/include/net/udp_tunnel.h
+++ b/include/net/udp_tunnel.h
@@ -150,7 +150,7 @@ static inline void udp_tunnel_drop_rx_info(struct net_device *dev)
 void udp_tunnel_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *skb,
 			 __be32 src, __be32 dst, __u8 tos, __u8 ttl,
 			 __be16 df, __be16 src_port, __be16 dst_port,
-			 bool xnet, bool nocheck);
+			 bool xnet, bool nocheck, u16 ipcb_flags);
 
 int udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
 			 struct sk_buff *skb,
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 678b8f96e3e9c..aaeb5d16f0c9a 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -668,7 +668,7 @@ void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 	ip_tunnel_adj_headroom(dev, headroom);
 
 	iptunnel_xmit(NULL, rt, skb, fl4.saddr, fl4.daddr, proto, tos, ttl,
-		      df, !net_eq(tunnel->net, dev_net(dev)));
+		      df, !net_eq(tunnel->net, dev_net(dev)), 0);
 	return;
 tx_error:
 	DEV_STATS_INC(dev, tx_errors);
@@ -857,7 +857,7 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 	ip_tunnel_adj_headroom(dev, max_headroom);
 
 	iptunnel_xmit(NULL, rt, skb, fl4.saddr, fl4.daddr, protocol, tos, ttl,
-		      df, !net_eq(tunnel->net, dev_net(dev)));
+		      df, !net_eq(tunnel->net, dev_net(dev)), 0);
 	return;
 
 #if IS_ENABLED(CONFIG_IPV6)
diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index f65d2f7273813..cc9915543637d 100644
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -49,7 +49,8 @@ EXPORT_SYMBOL(ip6tun_encaps);
 
 void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
 		   __be32 src, __be32 dst, __u8 proto,
-		   __u8 tos, __u8 ttl, __be16 df, bool xnet)
+		   __u8 tos, __u8 ttl, __be16 df, bool xnet,
+		   u16 ipcb_flags)
 {
 	int pkt_len = skb->len - skb_inner_network_offset(skb);
 	struct net *net = dev_net(rt->dst.dev);
@@ -62,6 +63,7 @@ void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
 	skb_clear_hash_if_not_l4(skb);
 	skb_dst_set(skb, &rt->dst);
 	memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
+	IPCB(skb)->flags = ipcb_flags;
 
 	/* Push down and install the IP header. */
 	skb_push(skb, sizeof(struct iphdr));
diff --git a/net/ipv4/udp_tunnel_core.c b/net/ipv4/udp_tunnel_core.c
index 2326548997d3f..9efd625059163 100644
--- a/net/ipv4/udp_tunnel_core.c
+++ b/net/ipv4/udp_tunnel_core.c
@@ -169,7 +169,7 @@ EXPORT_SYMBOL_GPL(udp_tunnel_notify_del_rx_port);
 void udp_tunnel_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *skb,
 			 __be32 src, __be32 dst, __u8 tos, __u8 ttl,
 			 __be16 df, __be16 src_port, __be16 dst_port,
-			 bool xnet, bool nocheck)
+			 bool xnet, bool nocheck, u16 ipcb_flags)
 {
 	struct udphdr *uh;
 
@@ -185,7 +185,8 @@ void udp_tunnel_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *skb
 
 	udp_set_csum(nocheck, skb, src, dst, skb->len);
 
-	iptunnel_xmit(sk, rt, skb, src, dst, IPPROTO_UDP, tos, ttl, df, xnet);
+	iptunnel_xmit(sk, rt, skb, src, dst, IPPROTO_UDP, tos, ttl, df, xnet,
+		      ipcb_flags);
 }
 EXPORT_SYMBOL_GPL(udp_tunnel_xmit_skb);
 
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index a72dbca9e8fca..12496ba1b7d4d 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -1035,7 +1035,7 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
 	skb_set_inner_ipproto(skb, IPPROTO_IPV6);
 
 	iptunnel_xmit(NULL, rt, skb, fl4.saddr, fl4.daddr, protocol, tos, ttl,
-		      df, !net_eq(tunnel->net, dev_net(dev)));
+		      df, !net_eq(tunnel->net, dev_net(dev)), 0);
 	return NETDEV_TX_OK;
 
 tx_error_icmp:
diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c
index f402f90eb6b67..a5ccada55f2b2 100644
--- a/net/sctp/protocol.c
+++ b/net/sctp/protocol.c
@@ -1103,7 +1103,8 @@ static inline int sctp_v4_xmit(struct sk_buff *skb, struct sctp_transport *t)
 	skb_set_inner_ipproto(skb, IPPROTO_SCTP);
 	udp_tunnel_xmit_skb(dst_rtable(dst), sk, skb, fl4->saddr,
 			    fl4->daddr, dscp, ip4_dst_hoplimit(dst), df,
-			    sctp_sk(sk)->udp_port, t->encap_port, false, false);
+			    sctp_sk(sk)->udp_port, t->encap_port, false, false,
+			    0);
 	return 0;
 }
 
diff --git a/net/tipc/udp_media.c b/net/tipc/udp_media.c
index 108a4cc2e0010..87e8c1e6d5504 100644
--- a/net/tipc/udp_media.c
+++ b/net/tipc/udp_media.c
@@ -197,7 +197,7 @@ static int tipc_udp_xmit(struct net *net, struct sk_buff *skb,
 		ttl = ip4_dst_hoplimit(&rt->dst);
 		udp_tunnel_xmit_skb(rt, ub->ubsock->sk, skb, src->ipv4.s_addr,
 				    dst->ipv4.s_addr, 0, ttl, 0, src->port,
-				    dst->port, false, true);
+				    dst->port, false, true, 0);
 #if IS_ENABLED(CONFIG_IPV6)
 	} else {
 		if (!ndst) {
-- 
2.51.0


From 3b7bc938e0ada6a791103faae261dd2a770df86d Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:10 +0200
Subject: [PATCH 06/16] net: ipv4: ipmr: ipmr_queue_xmit(): Drop local variable
 `dev'

The variable is used for caching of rt->dst.dev. The netdevice referenced
therein does not change during the scope of validity of that local. At the
same time, the local is only used twice, and each of these uses will end up
in a different function in the following patches, further eliminating any
use the local could have had.

Drop the local altogether and inline the uses.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/c80600a4b51679fe78f429ccb6d60892c2f9e4de.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ipv4/ipmr.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index a7d09ae9d7616..d2ac630bea3a8 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -1859,7 +1859,6 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 	const struct iphdr *iph = ip_hdr(skb);
 	struct vif_device *vif = &mrt->vif_table[vifi];
 	struct net_device *vif_dev;
-	struct net_device *dev;
 	struct rtable *rt;
 	struct flowi4 fl4;
 	int    encap = 0;
@@ -1898,8 +1897,6 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 			goto out_free;
 	}
 
-	dev = rt->dst.dev;
-
 	if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) {
 		/* Do not fragment multicasts. Alas, IPv4 does not
 		 * allow to send ICMP, so that packets will disappear
@@ -1910,7 +1907,7 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 		goto out_free;
 	}
 
-	encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len;
+	encap += LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len;
 
 	if (skb_cow(skb, encap)) {
 		ip_rt_put(rt);
@@ -1947,7 +1944,7 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 	 * result in receiving multiple packets.
 	 */
 	NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD,
-		net, NULL, skb, skb->dev, dev,
+		net, NULL, skb, skb->dev, rt->dst.dev,
 		ipmr_forward_finish);
 	return;
 
-- 
2.51.0


From b2e653bcff0f3eb43183393548f7142c176faa6d Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:11 +0200
Subject: [PATCH 07/16] net: ipv4: ipmr: Split ipmr_queue_xmit() in two

Some of the work of ipmr_queue_xmit() is specific to IPMR forwarding, and
should not take place on the output path. In order to allow reuse of the
common parts, split the function into two: the ipmr_prepare_xmit() helper
that takes care of the common bits, and the ipmr_queue_fwd_xmit(), which
invokes the former and encapsulates the whole forwarding algorithm.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/4e8db165572a4f8bd29a723a801e854e9d20df4d.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ipv4/ipmr.c | 45 +++++++++++++++++++++++++++++----------------
 1 file changed, 29 insertions(+), 16 deletions(-)

diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index d2ac630bea3a8..74d45fd5d11ea 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -1853,8 +1853,8 @@ static bool ipmr_forward_offloaded(struct sk_buff *skb, struct mr_table *mrt,
 
 /* Processing handlers for ipmr_forward, under rcu_read_lock() */
 
-static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
-			    int in_vifi, struct sk_buff *skb, int vifi)
+static int ipmr_prepare_xmit(struct net *net, struct mr_table *mrt,
+			     struct sk_buff *skb, int vifi)
 {
 	const struct iphdr *iph = ip_hdr(skb);
 	struct vif_device *vif = &mrt->vif_table[vifi];
@@ -1865,7 +1865,7 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 
 	vif_dev = vif_dev_read(vif);
 	if (!vif_dev)
-		goto out_free;
+		return -1;
 
 	if (vif->flags & VIFF_REGISTER) {
 		WRITE_ONCE(vif->pkt_out, vif->pkt_out + 1);
@@ -1873,12 +1873,9 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 		DEV_STATS_ADD(vif_dev, tx_bytes, skb->len);
 		DEV_STATS_INC(vif_dev, tx_packets);
 		ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT);
-		goto out_free;
+		return -1;
 	}
 
-	if (ipmr_forward_offloaded(skb, mrt, in_vifi, vifi))
-		goto out_free;
-
 	if (vif->flags & VIFF_TUNNEL) {
 		rt = ip_route_output_ports(net, &fl4, NULL,
 					   vif->remote, vif->local,
@@ -1886,7 +1883,7 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 					   IPPROTO_IPIP,
 					   iph->tos & INET_DSCP_MASK, vif->link);
 		if (IS_ERR(rt))
-			goto out_free;
+			return -1;
 		encap = sizeof(struct iphdr);
 	} else {
 		rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0,
@@ -1894,7 +1891,7 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 					   IPPROTO_IPIP,
 					   iph->tos & INET_DSCP_MASK, vif->link);
 		if (IS_ERR(rt))
-			goto out_free;
+			return -1;
 	}
 
 	if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) {
@@ -1904,14 +1901,14 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 		 */
 		IP_INC_STATS(net, IPSTATS_MIB_FRAGFAILS);
 		ip_rt_put(rt);
-		goto out_free;
+		return -1;
 	}
 
 	encap += LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len;
 
 	if (skb_cow(skb, encap)) {
 		ip_rt_put(rt);
-		goto out_free;
+		return -1;
 	}
 
 	WRITE_ONCE(vif->pkt_out, vif->pkt_out + 1);
@@ -1931,6 +1928,22 @@ static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
 		DEV_STATS_ADD(vif_dev, tx_bytes, skb->len);
 	}
 
+	return 0;
+}
+
+static void ipmr_queue_fwd_xmit(struct net *net, struct mr_table *mrt,
+				int in_vifi, struct sk_buff *skb, int vifi)
+{
+	struct rtable *rt;
+
+	if (ipmr_forward_offloaded(skb, mrt, in_vifi, vifi))
+		goto out_free;
+
+	if (ipmr_prepare_xmit(net, mrt, skb, vifi))
+		goto out_free;
+
+	rt = skb_rtable(skb);
+
 	IPCB(skb)->flags |= IPSKB_FORWARDED;
 
 	/* RFC1584 teaches, that DVMRP/PIM router must deliver packets locally
@@ -2062,8 +2075,8 @@ forward:
 				struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
 
 				if (skb2)
-					ipmr_queue_xmit(net, mrt, true_vifi,
-							skb2, psend);
+					ipmr_queue_fwd_xmit(net, mrt, true_vifi,
+							    skb2, psend);
 			}
 			psend = ct;
 		}
@@ -2074,10 +2087,10 @@ last_forward:
 			struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
 
 			if (skb2)
-				ipmr_queue_xmit(net, mrt, true_vifi, skb2,
-						psend);
+				ipmr_queue_fwd_xmit(net, mrt, true_vifi, skb2,
+						    psend);
 		} else {
-			ipmr_queue_xmit(net, mrt, true_vifi, skb, psend);
+			ipmr_queue_fwd_xmit(net, mrt, true_vifi, skb, psend);
 			return;
 		}
 	}
-- 
2.51.0


From 35bec72a24ace52a7f57642ff2813f22733b08fd Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:12 +0200
Subject: [PATCH 08/16] net: ipv4: Add ip_mr_output()

Multicast routing is today handled in the input path. Locally generated MC
packets don't hit the IPMR code today. Thus if a VXLAN remote address is
multicast, the driver needs to set an OIF during route lookup. Thus MC
routing configuration needs to be kept in sync with the VXLAN FDB and MDB.
Ideally, the VXLAN packets would be routed by the MC routing code instead.

To that end, this patch adds support to route locally generated multicast
packets. The newly-added routines do largely what ip_mr_input() and
ip_mr_forward() do: make an MR cache lookup to find where to send the
packets, and use ip_mc_output() to send each of them. When no cache entry
is found, the packet is punted to the daemon for resolution.

However, an installation that uses a VXLAN underlay netdevice for which it
also has matching MC routes, would get a different routing with this patch.
Previously, the MC packets would be delivered directly to the underlay
port, whereas now they would be MC-routed. In order to avoid this change in
behavior, introduce an IPCB flag. Only if the flag is set will
ip_mr_output() actually engage, otherwise it reverts to ip_mc_output().

This code is based on work by Roopa Prabhu and Nikolay Aleksandrov.

Signed-off-by: Roopa Prabhu <roopa@nvidia.com>
Signed-off-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/0aadbd49330471c0f758d54afb05eb3b6e3a6b65.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/net/ip.h |   2 +
 net/ipv4/ipmr.c  | 117 +++++++++++++++++++++++++++++++++++++++++++++++
 net/ipv4/route.c |   2 +-
 3 files changed, 120 insertions(+), 1 deletion(-)

diff --git a/include/net/ip.h b/include/net/ip.h
index 47ed6d23853d7..375304bb99f69 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -59,6 +59,7 @@ struct inet_skb_parm {
 #define IPSKB_L3SLAVE		BIT(7)
 #define IPSKB_NOPOLICY		BIT(8)
 #define IPSKB_MULTIPATH		BIT(9)
+#define IPSKB_MCROUTE		BIT(10)
 
 	u16			frag_max_size;
 };
@@ -167,6 +168,7 @@ void ip_list_rcv(struct list_head *head, struct packet_type *pt,
 int ip_local_deliver(struct sk_buff *skb);
 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
 int ip_mr_input(struct sk_buff *skb);
+int ip_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb);
 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index 74d45fd5d11ea..f78c4e53dc8c1 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -1965,6 +1965,19 @@ out_free:
 	kfree_skb(skb);
 }
 
+static void ipmr_queue_output_xmit(struct net *net, struct mr_table *mrt,
+				   struct sk_buff *skb, int vifi)
+{
+	if (ipmr_prepare_xmit(net, mrt, skb, vifi))
+		goto out_free;
+
+	ip_mc_output(net, NULL, skb);
+	return;
+
+out_free:
+	kfree_skb(skb);
+}
+
 /* Called with mrt_lock or rcu_read_lock() */
 static int ipmr_find_vif(const struct mr_table *mrt, struct net_device *dev)
 {
@@ -2224,6 +2237,110 @@ dont_forward:
 	return 0;
 }
 
+static void ip_mr_output_finish(struct net *net, struct mr_table *mrt,
+				struct net_device *dev, struct sk_buff *skb,
+				struct mfc_cache *c)
+{
+	int psend = -1;
+	int ct;
+
+	atomic_long_inc(&c->_c.mfc_un.res.pkt);
+	atomic_long_add(skb->len, &c->_c.mfc_un.res.bytes);
+	WRITE_ONCE(c->_c.mfc_un.res.lastuse, jiffies);
+
+	/* Forward the frame */
+	if (c->mfc_origin == htonl(INADDR_ANY) &&
+	    c->mfc_mcastgrp == htonl(INADDR_ANY)) {
+		if (ip_hdr(skb)->ttl >
+		    c->_c.mfc_un.res.ttls[c->_c.mfc_parent]) {
+			/* It's an (*,*) entry and the packet is not coming from
+			 * the upstream: forward the packet to the upstream
+			 * only.
+			 */
+			psend = c->_c.mfc_parent;
+			goto last_xmit;
+		}
+		goto dont_xmit;
+	}
+
+	for (ct = c->_c.mfc_un.res.maxvif - 1;
+	     ct >= c->_c.mfc_un.res.minvif; ct--) {
+		if (ip_hdr(skb)->ttl > c->_c.mfc_un.res.ttls[ct]) {
+			if (psend != -1) {
+				struct sk_buff *skb2;
+
+				skb2 = skb_clone(skb, GFP_ATOMIC);
+				if (skb2)
+					ipmr_queue_output_xmit(net, mrt,
+							       skb2, psend);
+			}
+			psend = ct;
+		}
+	}
+
+last_xmit:
+	if (psend != -1) {
+		ipmr_queue_output_xmit(net, mrt, skb, psend);
+		return;
+	}
+
+dont_xmit:
+	kfree_skb(skb);
+}
+
+/* Multicast packets for forwarding arrive here
+ * Called with rcu_read_lock();
+ */
+int ip_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb)
+{
+	struct rtable *rt = skb_rtable(skb);
+	struct mfc_cache *cache;
+	struct net_device *dev;
+	struct mr_table *mrt;
+	int vif;
+
+	WARN_ON_ONCE(!rcu_read_lock_held());
+	dev = rt->dst.dev;
+
+	if (IPCB(skb)->flags & IPSKB_FORWARDED)
+		goto mc_output;
+	if (!(IPCB(skb)->flags & IPSKB_MCROUTE))
+		goto mc_output;
+
+	skb->dev = dev;
+
+	mrt = ipmr_rt_fib_lookup(net, skb);
+	if (IS_ERR(mrt))
+		goto mc_output;
+
+	/* already under rcu_read_lock() */
+	cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr);
+	if (!cache) {
+		vif = ipmr_find_vif(mrt, dev);
+		if (vif >= 0)
+			cache = ipmr_cache_find_any(mrt, ip_hdr(skb)->daddr,
+						    vif);
+	}
+
+	/* No usable cache entry */
+	if (!cache) {
+		vif = ipmr_find_vif(mrt, dev);
+		if (vif >= 0)
+			return ipmr_cache_unresolved(mrt, vif, skb, dev);
+		goto mc_output;
+	}
+
+	vif = cache->_c.mfc_parent;
+	if (rcu_access_pointer(mrt->vif_table[vif].dev) != dev)
+		goto mc_output;
+
+	ip_mr_output_finish(net, mrt, dev, skb, cache);
+	return 0;
+
+mc_output:
+	return ip_mc_output(net, sk, skb);
+}
+
 #ifdef CONFIG_IP_PIMSM_V1
 /* Handle IGMP messages of PIMv1 */
 int pim_rcv_v1(struct sk_buff *skb)
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index fccb05fb3a794..3ddf6bf403579 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -2660,7 +2660,7 @@ add:
 			if (IN_DEV_MFORWARD(in_dev) &&
 			    !ipv4_is_local_multicast(fl4->daddr)) {
 				rth->dst.input = ip_mr_input;
-				rth->dst.output = ip_mc_output;
+				rth->dst.output = ip_mr_output;
 			}
 		}
 #endif
-- 
2.51.0


From 6a7d88ca15f73c5c570c372238f71d63da1fda55 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:13 +0200
Subject: [PATCH 09/16] net: ipv6: Make udp_tunnel6_xmit_skb() void

The function always returns zero, thus the return value does not carry any
signal. Just make it void.

Most callers already ignore the return value. However:

- Refold arguments of the call from sctp_v6_xmit() so that they fit into
  the 80-column limit.

- tipc_udp_xmit() initializes err from the return value, but that should
  already be always zero at that point. So there's no practical change, but
  elision of the assignment prompts a couple more tweaks to clean up the
  function.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/7facacf9d8ca3ca9391a4aee88160913671b868d.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/net/udp_tunnel.h  | 14 +++++++-------
 net/ipv6/ip6_udp_tunnel.c | 15 +++++++--------
 net/sctp/ipv6.c           |  7 ++++---
 net/tipc/udp_media.c      | 10 +++++-----
 4 files changed, 23 insertions(+), 23 deletions(-)

diff --git a/include/net/udp_tunnel.h b/include/net/udp_tunnel.h
index 28102c8fd8a85..0b01f6ade20da 100644
--- a/include/net/udp_tunnel.h
+++ b/include/net/udp_tunnel.h
@@ -152,13 +152,13 @@ void udp_tunnel_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *skb
 			 __be16 df, __be16 src_port, __be16 dst_port,
 			 bool xnet, bool nocheck, u16 ipcb_flags);
 
-int udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
-			 struct sk_buff *skb,
-			 struct net_device *dev,
-			 const struct in6_addr *saddr,
-			 const struct in6_addr *daddr,
-			 __u8 prio, __u8 ttl, __be32 label,
-			 __be16 src_port, __be16 dst_port, bool nocheck);
+void udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
+			  struct sk_buff *skb,
+			  struct net_device *dev,
+			  const struct in6_addr *saddr,
+			  const struct in6_addr *daddr,
+			  __u8 prio, __u8 ttl, __be32 label,
+			  __be16 src_port, __be16 dst_port, bool nocheck);
 
 void udp_tunnel_sock_release(struct socket *sock);
 
diff --git a/net/ipv6/ip6_udp_tunnel.c b/net/ipv6/ip6_udp_tunnel.c
index c99053189ea8a..21681718b7bb0 100644
--- a/net/ipv6/ip6_udp_tunnel.c
+++ b/net/ipv6/ip6_udp_tunnel.c
@@ -74,13 +74,13 @@ error:
 }
 EXPORT_SYMBOL_GPL(udp_sock_create6);
 
-int udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
-			 struct sk_buff *skb,
-			 struct net_device *dev,
-			 const struct in6_addr *saddr,
-			 const struct in6_addr *daddr,
-			 __u8 prio, __u8 ttl, __be32 label,
-			 __be16 src_port, __be16 dst_port, bool nocheck)
+void udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
+			  struct sk_buff *skb,
+			  struct net_device *dev,
+			  const struct in6_addr *saddr,
+			  const struct in6_addr *daddr,
+			  __u8 prio, __u8 ttl, __be32 label,
+			  __be16 src_port, __be16 dst_port, bool nocheck)
 {
 	struct udphdr *uh;
 	struct ipv6hdr *ip6h;
@@ -109,7 +109,6 @@ int udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
 	ip6h->saddr	  = *saddr;
 
 	ip6tunnel_xmit(sk, skb, dev);
-	return 0;
 }
 EXPORT_SYMBOL_GPL(udp_tunnel6_xmit_skb);
 
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index a9ed2ccab1bdb..d1ecf7454827b 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -261,9 +261,10 @@ static int sctp_v6_xmit(struct sk_buff *skb, struct sctp_transport *t)
 	skb_set_inner_ipproto(skb, IPPROTO_SCTP);
 	label = ip6_make_flowlabel(sock_net(sk), skb, fl6->flowlabel, true, fl6);
 
-	return udp_tunnel6_xmit_skb(dst, sk, skb, NULL, &fl6->saddr,
-				    &fl6->daddr, tclass, ip6_dst_hoplimit(dst),
-				    label, sctp_sk(sk)->udp_port, t->encap_port, false);
+	udp_tunnel6_xmit_skb(dst, sk, skb, NULL, &fl6->saddr, &fl6->daddr,
+			     tclass, ip6_dst_hoplimit(dst), label,
+			     sctp_sk(sk)->udp_port, t->encap_port, false);
+	return 0;
 }
 
 /* Returns the dst cache entry for the given source and destination ip
diff --git a/net/tipc/udp_media.c b/net/tipc/udp_media.c
index 87e8c1e6d5504..414713fcd8c5f 100644
--- a/net/tipc/udp_media.c
+++ b/net/tipc/udp_media.c
@@ -172,7 +172,7 @@ static int tipc_udp_xmit(struct net *net, struct sk_buff *skb,
 			 struct udp_media_addr *dst, struct dst_cache *cache)
 {
 	struct dst_entry *ndst;
-	int ttl, err = 0;
+	int ttl, err;
 
 	local_bh_disable();
 	ndst = dst_cache_get(cache);
@@ -217,13 +217,13 @@ static int tipc_udp_xmit(struct net *net, struct sk_buff *skb,
 			dst_cache_set_ip6(cache, ndst, &fl6.saddr);
 		}
 		ttl = ip6_dst_hoplimit(ndst);
-		err = udp_tunnel6_xmit_skb(ndst, ub->ubsock->sk, skb, NULL,
-					   &src->ipv6, &dst->ipv6, 0, ttl, 0,
-					   src->port, dst->port, false);
+		udp_tunnel6_xmit_skb(ndst, ub->ubsock->sk, skb, NULL,
+				     &src->ipv6, &dst->ipv6, 0, ttl, 0,
+				     src->port, dst->port, false);
 #endif
 	}
 	local_bh_enable();
-	return err;
+	return 0;
 
 tx_error:
 	local_bh_enable();
-- 
2.51.0


From f78c75d84fe83898f0a00658f593d4f17b38cbc6 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:14 +0200
Subject: [PATCH 10/16] net: ipv6: Add a flags argument to ip6tunnel_xmit(),
 udp_tunnel6_xmit_skb()

ip6tunnel_xmit() erases the contents of the SKB control block. In order to
be able to set particular IP6CB flags on the SKB, add a corresponding
parameter, and propagate it to udp_tunnel6_xmit_skb() as well.

In one of the following patches, VXLAN driver will use this facility to
mark packets as subject to IPv6 multicast routing.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/acb4f9f3e40c3a931236c3af08a720b017fbfbfb.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 drivers/net/bareudp.c          | 3 ++-
 drivers/net/geneve.c           | 3 ++-
 drivers/net/gtp.c              | 2 +-
 drivers/net/ovpn/udp.c         | 2 +-
 drivers/net/vxlan/vxlan_core.c | 3 ++-
 drivers/net/wireguard/socket.c | 2 +-
 include/net/ip6_tunnel.h       | 3 ++-
 include/net/udp_tunnel.h       | 3 ++-
 net/ipv6/ip6_tunnel.c          | 2 +-
 net/ipv6/ip6_udp_tunnel.c      | 5 +++--
 net/sctp/ipv6.c                | 2 +-
 net/tipc/udp_media.c           | 2 +-
 12 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/drivers/net/bareudp.c b/drivers/net/bareudp.c
index 5e613080d3f80..0df3208783ad9 100644
--- a/drivers/net/bareudp.c
+++ b/drivers/net/bareudp.c
@@ -431,7 +431,8 @@ static int bareudp6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 			     &saddr, &daddr, prio, ttl,
 			     info->key.label, sport, bareudp->port,
 			     !test_bit(IP_TUNNEL_CSUM_BIT,
-				       info->key.tun_flags));
+				       info->key.tun_flags),
+			     0);
 	return 0;
 
 free_dst:
diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index c668e8b00ed26..f6bd155aae7fe 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -1014,7 +1014,8 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 			     &saddr, &key->u.ipv6.dst, prio, ttl,
 			     info->key.label, sport, geneve->cfg.info.key.tp_dst,
 			     !test_bit(IP_TUNNEL_CSUM_BIT,
-				       info->key.tun_flags));
+				       info->key.tun_flags),
+			     0);
 	return 0;
 }
 #endif
diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c
index 14584793fe4e8..4b668ebaa0f71 100644
--- a/drivers/net/gtp.c
+++ b/drivers/net/gtp.c
@@ -1316,7 +1316,7 @@ static netdev_tx_t gtp_dev_xmit(struct sk_buff *skb, struct net_device *dev)
 				     ip6_dst_hoplimit(&pktinfo.rt->dst),
 				     0,
 				     pktinfo.gtph_port, pktinfo.gtph_port,
-				     false);
+				     false, 0);
 #else
 		goto tx_err;
 #endif
diff --git a/drivers/net/ovpn/udp.c b/drivers/net/ovpn/udp.c
index d866e6bfda704..254cc94c46173 100644
--- a/drivers/net/ovpn/udp.c
+++ b/drivers/net/ovpn/udp.c
@@ -274,7 +274,7 @@ transmit:
 	skb->ignore_df = 1;
 	udp_tunnel6_xmit_skb(dst, sk, skb, skb->dev, &fl.saddr, &fl.daddr, 0,
 			     ip6_dst_hoplimit(dst), 0, fl.fl6_sport,
-			     fl.fl6_dport, udp_get_no_check6_tx(sk));
+			     fl.fl6_dport, udp_get_no_check6_tx(sk), 0);
 	ret = 0;
 err:
 	local_bh_enable();
diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 1cc18acd242dc..b22f9866be8eb 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2586,7 +2586,8 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 
 		udp_tunnel6_xmit_skb(ndst, sock6->sock->sk, skb, dev,
 				     &saddr, &pkey->u.ipv6.dst, tos, ttl,
-				     pkey->label, src_port, dst_port, !udp_sum);
+				     pkey->label, src_port, dst_port, !udp_sum,
+				     0);
 #endif
 	}
 	vxlan_vnifilter_count(vxlan, vni, NULL, VXLAN_VNI_STATS_TX, pkt_len);
diff --git a/drivers/net/wireguard/socket.c b/drivers/net/wireguard/socket.c
index 88e685667bc0d..253488f8c00f8 100644
--- a/drivers/net/wireguard/socket.c
+++ b/drivers/net/wireguard/socket.c
@@ -151,7 +151,7 @@ static int send6(struct wg_device *wg, struct sk_buff *skb,
 	skb->ignore_df = 1;
 	udp_tunnel6_xmit_skb(dst, sock, skb, skb->dev, &fl.saddr, &fl.daddr, ds,
 			     ip6_dst_hoplimit(dst), 0, fl.fl6_sport,
-			     fl.fl6_dport, false);
+			     fl.fl6_dport, false, 0);
 	goto out;
 
 err:
diff --git a/include/net/ip6_tunnel.h b/include/net/ip6_tunnel.h
index 399592405c72a..dd163495f3539 100644
--- a/include/net/ip6_tunnel.h
+++ b/include/net/ip6_tunnel.h
@@ -152,11 +152,12 @@ int ip6_tnl_get_iflink(const struct net_device *dev);
 int ip6_tnl_change_mtu(struct net_device *dev, int new_mtu);
 
 static inline void ip6tunnel_xmit(struct sock *sk, struct sk_buff *skb,
-				  struct net_device *dev)
+				  struct net_device *dev, u16 ip6cb_flags)
 {
 	int pkt_len, err;
 
 	memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
+	IP6CB(skb)->flags = ip6cb_flags;
 	pkt_len = skb->len - skb_inner_network_offset(skb);
 	err = ip6_local_out(dev_net(skb_dst(skb)->dev), sk, skb);
 
diff --git a/include/net/udp_tunnel.h b/include/net/udp_tunnel.h
index 0b01f6ade20da..e3c70b5790951 100644
--- a/include/net/udp_tunnel.h
+++ b/include/net/udp_tunnel.h
@@ -158,7 +158,8 @@ void udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
 			  const struct in6_addr *saddr,
 			  const struct in6_addr *daddr,
 			  __u8 prio, __u8 ttl, __be32 label,
-			  __be16 src_port, __be16 dst_port, bool nocheck);
+			  __be16 src_port, __be16 dst_port, bool nocheck,
+			  u16 ip6cb_flags);
 
 void udp_tunnel_sock_release(struct socket *sock);
 
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index 894d3158a6f05..a885bb5c98ea8 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -1278,7 +1278,7 @@ route_lookup:
 	ipv6h->nexthdr = proto;
 	ipv6h->saddr = fl6->saddr;
 	ipv6h->daddr = fl6->daddr;
-	ip6tunnel_xmit(NULL, skb, dev);
+	ip6tunnel_xmit(NULL, skb, dev, 0);
 	return 0;
 tx_err_link_failure:
 	DEV_STATS_INC(dev, tx_carrier_errors);
diff --git a/net/ipv6/ip6_udp_tunnel.c b/net/ipv6/ip6_udp_tunnel.c
index 21681718b7bb0..8ebe17a6058ad 100644
--- a/net/ipv6/ip6_udp_tunnel.c
+++ b/net/ipv6/ip6_udp_tunnel.c
@@ -80,7 +80,8 @@ void udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
 			  const struct in6_addr *saddr,
 			  const struct in6_addr *daddr,
 			  __u8 prio, __u8 ttl, __be32 label,
-			  __be16 src_port, __be16 dst_port, bool nocheck)
+			  __be16 src_port, __be16 dst_port, bool nocheck,
+			  u16 ip6cb_flags)
 {
 	struct udphdr *uh;
 	struct ipv6hdr *ip6h;
@@ -108,7 +109,7 @@ void udp_tunnel6_xmit_skb(struct dst_entry *dst, struct sock *sk,
 	ip6h->daddr	  = *daddr;
 	ip6h->saddr	  = *saddr;
 
-	ip6tunnel_xmit(sk, skb, dev);
+	ip6tunnel_xmit(sk, skb, dev, ip6cb_flags);
 }
 EXPORT_SYMBOL_GPL(udp_tunnel6_xmit_skb);
 
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index d1ecf7454827b..3336dcfb45150 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -263,7 +263,7 @@ static int sctp_v6_xmit(struct sk_buff *skb, struct sctp_transport *t)
 
 	udp_tunnel6_xmit_skb(dst, sk, skb, NULL, &fl6->saddr, &fl6->daddr,
 			     tclass, ip6_dst_hoplimit(dst), label,
-			     sctp_sk(sk)->udp_port, t->encap_port, false);
+			     sctp_sk(sk)->udp_port, t->encap_port, false, 0);
 	return 0;
 }
 
diff --git a/net/tipc/udp_media.c b/net/tipc/udp_media.c
index 414713fcd8c5f..a024fcc8c0cb7 100644
--- a/net/tipc/udp_media.c
+++ b/net/tipc/udp_media.c
@@ -219,7 +219,7 @@ static int tipc_udp_xmit(struct net *net, struct sk_buff *skb,
 		ttl = ip6_dst_hoplimit(ndst);
 		udp_tunnel6_xmit_skb(ndst, ub->ubsock->sk, skb, NULL,
 				     &src->ipv6, &dst->ipv6, 0, ttl, 0,
-				     src->port, dst->port, false);
+				     src->port, dst->port, false, 0);
 #endif
 	}
 	local_bh_enable();
-- 
2.51.0


From 3365afd3abda5f6a54f4a822dad5c9314e94c3fc Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:15 +0200
Subject: [PATCH 11/16] net: ipv6: ip6mr: Fix in/out netdev to pass to the
 FORWARD chain

The netfilter hook is invoked with skb->dev for input netdevice, and
vif_dev for output netdevice. However at the point of invocation, skb->dev
is already set to vif_dev, and MR-forwarded packets are reported with
in=out:

 # ip6tables -A FORWARD -j LOG --log-prefix '[forw]'
 # cd tools/testing/selftests/net/forwarding
 # ./router_multicast.sh
 # dmesg | fgrep '[forw]'
 [ 1670.248245] [forw]IN=v5 OUT=v5 [...]

For reference, IPv4 MR code shows in and out as appropriate.
Fix by caching skb->dev and using the updated value for output netdev.

Fixes: 7bc570c8b4f7 ("[IPV6] MROUTE: Support multicast forwarding.")
Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/3141ae8386fbe13fef4b793faa75e6bae58d798a.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ipv6/ip6mr.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 9db31e5b998c1..426859cd34092 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -2039,6 +2039,7 @@ static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
 			  struct sk_buff *skb, int vifi)
 {
 	struct vif_device *vif = &mrt->vif_table[vifi];
+	struct net_device *indev = skb->dev;
 	struct net_device *vif_dev;
 	struct ipv6hdr *ipv6h;
 	struct dst_entry *dst;
@@ -2101,7 +2102,7 @@ static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
 	IP6CB(skb)->flags |= IP6SKB_FORWARDED;
 
 	return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
-		       net, NULL, skb, skb->dev, vif_dev,
+		       net, NULL, skb, indev, skb->dev,
 		       ip6mr_forward2_finish);
 
 out_free:
-- 
2.51.0


From 094f39d5e84d6c48bb38ded6b30f68b12cb8145b Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:16 +0200
Subject: [PATCH 12/16] net: ipv6: ip6mr: Make ip6mr_forward2() void

Nobody uses the return value, so convert the function to void.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Link: https://patch.msgid.link/e0bee259da0da58da96647ea9e21e6360c8f7e11.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ipv6/ip6mr.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 426859cd34092..41c348209e1b4 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -2035,8 +2035,8 @@ static inline int ip6mr_forward2_finish(struct net *net, struct sock *sk, struct
  *	Processing handlers for ip6mr_forward
  */
 
-static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
-			  struct sk_buff *skb, int vifi)
+static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
+			   struct sk_buff *skb, int vifi)
 {
 	struct vif_device *vif = &mrt->vif_table[vifi];
 	struct net_device *indev = skb->dev;
@@ -2101,13 +2101,13 @@ static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
 
 	IP6CB(skb)->flags |= IP6SKB_FORWARDED;
 
-	return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
-		       net, NULL, skb, indev, skb->dev,
-		       ip6mr_forward2_finish);
+	NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
+		net, NULL, skb, indev, skb->dev,
+		ip6mr_forward2_finish);
+	return;
 
 out_free:
 	kfree_skb(skb);
-	return 0;
 }
 
 /* Called with rcu_read_lock() */
-- 
2.51.0


From 1b02f4475d29c6e2c4b7f1bae74149b9c1369791 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:17 +0200
Subject: [PATCH 13/16] net: ipv6: ip6mr: Split ip6mr_forward2() in two

Some of the work of ip6mr_forward2() is specific to IPMR forwarding, and
should not take place on the output path. In order to allow reuse of the
common parts, extract out of the function a helper,
ip6mr_prepare_forward().

Signed-off-by: Petr Machata <petrm@nvidia.com>
Link: https://patch.msgid.link/8932bd5c0fbe3f662b158803b8509604fa7bc113.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ipv6/ip6mr.c | 23 ++++++++++++++++-------
 1 file changed, 16 insertions(+), 7 deletions(-)

diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 41c348209e1b4..bd964564160d0 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -2035,11 +2035,10 @@ static inline int ip6mr_forward2_finish(struct net *net, struct sock *sk, struct
  *	Processing handlers for ip6mr_forward
  */
 
-static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
-			   struct sk_buff *skb, int vifi)
+static int ip6mr_prepare_xmit(struct net *net, struct mr_table *mrt,
+			      struct sk_buff *skb, int vifi)
 {
 	struct vif_device *vif = &mrt->vif_table[vifi];
-	struct net_device *indev = skb->dev;
 	struct net_device *vif_dev;
 	struct ipv6hdr *ipv6h;
 	struct dst_entry *dst;
@@ -2047,7 +2046,7 @@ static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
 
 	vif_dev = vif_dev_read(vif);
 	if (!vif_dev)
-		goto out_free;
+		return -1;
 
 #ifdef CONFIG_IPV6_PIMSM_V2
 	if (vif->flags & MIFF_REGISTER) {
@@ -2056,7 +2055,7 @@ static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
 		DEV_STATS_ADD(vif_dev, tx_bytes, skb->len);
 		DEV_STATS_INC(vif_dev, tx_packets);
 		ip6mr_cache_report(mrt, skb, vifi, MRT6MSG_WHOLEPKT);
-		goto out_free;
+		return -1;
 	}
 #endif
 
@@ -2070,7 +2069,7 @@ static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
 	dst = ip6_route_output(net, NULL, &fl6);
 	if (dst->error) {
 		dst_release(dst);
-		goto out_free;
+		return -1;
 	}
 
 	skb_dst_drop(skb);
@@ -2094,10 +2093,20 @@ static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
 	/* We are about to write */
 	/* XXX: extension headers? */
 	if (skb_cow(skb, sizeof(*ipv6h) + LL_RESERVED_SPACE(vif_dev)))
-		goto out_free;
+		return -1;
 
 	ipv6h = ipv6_hdr(skb);
 	ipv6h->hop_limit--;
+	return 0;
+}
+
+static void ip6mr_forward2(struct net *net, struct mr_table *mrt,
+			   struct sk_buff *skb, int vifi)
+{
+	struct net_device *indev = skb->dev;
+
+	if (ip6mr_prepare_xmit(net, mrt, skb, vifi))
+		goto out_free;
 
 	IP6CB(skb)->flags |= IP6SKB_FORWARDED;
 
-- 
2.51.0


From 96e8f5a9fe2d91b9f9eb8b45cc13ce1ca6a8af82 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:18 +0200
Subject: [PATCH 14/16] net: ipv6: Add ip6_mr_output()

Multicast routing is today handled in the input path. Locally generated MC
packets don't hit the IPMR code today. Thus if a VXLAN remote address is
multicast, the driver needs to set an OIF during route lookup. Thus MC
routing configuration needs to be kept in sync with the VXLAN FDB and MDB.
Ideally, the VXLAN packets would be routed by the MC routing code instead.

To that end, this patch adds support to route locally generated multicast
packets. The newly-added routines do largely what ip6_mr_input() and
ip6_mr_forward() do: make an MR cache lookup to find where to send the
packets, and use ip6_output() to send each of them. When no cache entry is
found, the packet is punted to the daemon for resolution.

Similarly to the IPv4 case in a previous patch, the new logic is contingent
on a newly-added IP6CB flag being set.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Link: https://patch.msgid.link/3bcc034a3ab4d3c291072fff38f78d7fbbeef4e6.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/linux/ipv6.h    |   1 +
 include/linux/mroute6.h |   7 +++
 net/ipv6/ip6mr.c        | 118 ++++++++++++++++++++++++++++++++++++++++
 net/ipv6/route.c        |   1 +
 4 files changed, 127 insertions(+)

diff --git a/include/linux/ipv6.h b/include/linux/ipv6.h
index 5aeeed22f35bf..db0eb0d86b641 100644
--- a/include/linux/ipv6.h
+++ b/include/linux/ipv6.h
@@ -156,6 +156,7 @@ struct inet6_skb_parm {
 #define IP6SKB_SEG6	      256
 #define IP6SKB_FAKEJUMBO      512
 #define IP6SKB_MULTIPATH      1024
+#define IP6SKB_MCROUTE        2048
 };
 
 #if defined(CONFIG_NET_L3_MASTER_DEV)
diff --git a/include/linux/mroute6.h b/include/linux/mroute6.h
index 63ef5191cc579..fddafdc168f73 100644
--- a/include/linux/mroute6.h
+++ b/include/linux/mroute6.h
@@ -31,6 +31,7 @@ extern int ip6_mroute_getsockopt(struct sock *, int, sockptr_t, sockptr_t);
 extern int ip6_mr_input(struct sk_buff *skb);
 extern int ip6mr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg);
 extern int ip6_mr_init(void);
+extern int ip6_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb);
 extern void ip6_mr_cleanup(void);
 int ip6mr_ioctl(struct sock *sk, int cmd, void *arg);
 #else
@@ -58,6 +59,12 @@ static inline int ip6_mr_init(void)
 	return 0;
 }
 
+static inline int
+ip6_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb)
+{
+	return ip6_output(net, sk, skb);
+}
+
 static inline void ip6_mr_cleanup(void)
 {
 	return;
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index bd964564160d0..a35f4f1c65896 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -2119,6 +2119,19 @@ out_free:
 	kfree_skb(skb);
 }
 
+static void ip6mr_output2(struct net *net, struct mr_table *mrt,
+			  struct sk_buff *skb, int vifi)
+{
+	if (ip6mr_prepare_xmit(net, mrt, skb, vifi))
+		goto out_free;
+
+	ip6_output(net, NULL, skb);
+	return;
+
+out_free:
+	kfree_skb(skb);
+}
+
 /* Called with rcu_read_lock() */
 static int ip6mr_find_vif(struct mr_table *mrt, struct net_device *dev)
 {
@@ -2231,6 +2244,56 @@ dont_forward:
 	kfree_skb(skb);
 }
 
+/* Called under rcu_read_lock() */
+static void ip6_mr_output_finish(struct net *net, struct mr_table *mrt,
+				 struct net_device *dev, struct sk_buff *skb,
+				 struct mfc6_cache *c)
+{
+	int psend = -1;
+	int ct;
+
+	WARN_ON_ONCE(!rcu_read_lock_held());
+
+	atomic_long_inc(&c->_c.mfc_un.res.pkt);
+	atomic_long_add(skb->len, &c->_c.mfc_un.res.bytes);
+	WRITE_ONCE(c->_c.mfc_un.res.lastuse, jiffies);
+
+	/* Forward the frame */
+	if (ipv6_addr_any(&c->mf6c_origin) &&
+	    ipv6_addr_any(&c->mf6c_mcastgrp)) {
+		if (ipv6_hdr(skb)->hop_limit >
+		    c->_c.mfc_un.res.ttls[c->_c.mfc_parent]) {
+			/* It's an (*,*) entry and the packet is not coming from
+			 * the upstream: forward the packet to the upstream
+			 * only.
+			 */
+			psend = c->_c.mfc_parent;
+			goto last_forward;
+		}
+		goto dont_forward;
+	}
+	for (ct = c->_c.mfc_un.res.maxvif - 1;
+	     ct >= c->_c.mfc_un.res.minvif; ct--) {
+		if (ipv6_hdr(skb)->hop_limit > c->_c.mfc_un.res.ttls[ct]) {
+			if (psend != -1) {
+				struct sk_buff *skb2;
+
+				skb2 = skb_clone(skb, GFP_ATOMIC);
+				if (skb2)
+					ip6mr_output2(net, mrt, skb2, psend);
+			}
+			psend = ct;
+		}
+	}
+last_forward:
+	if (psend != -1) {
+		ip6mr_output2(net, mrt, skb, psend);
+		return;
+	}
+
+dont_forward:
+	kfree_skb(skb);
+}
 
 /*
  *	Multicast packets for forwarding arrive here
@@ -2298,6 +2361,61 @@ int ip6_mr_input(struct sk_buff *skb)
 	return 0;
 }
 
+int ip6_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb)
+{
+	struct net_device *dev = skb_dst(skb)->dev;
+	struct flowi6 fl6 = (struct flowi6) {
+		.flowi6_iif = LOOPBACK_IFINDEX,
+		.flowi6_mark = skb->mark,
+	};
+	struct mfc6_cache *cache;
+	struct mr_table *mrt;
+	int err;
+	int vif;
+
+	WARN_ON_ONCE(!rcu_read_lock_held());
+
+	if (IP6CB(skb)->flags & IP6SKB_FORWARDED)
+		goto ip6_output;
+	if (!(IP6CB(skb)->flags & IP6SKB_MCROUTE))
+		goto ip6_output;
+
+	err = ip6mr_fib_lookup(net, &fl6, &mrt);
+	if (err < 0) {
+		kfree_skb(skb);
+		return err;
+	}
+
+	cache = ip6mr_cache_find(mrt,
+				 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr);
+	if (!cache) {
+		vif = ip6mr_find_vif(mrt, dev);
+		if (vif >= 0)
+			cache = ip6mr_cache_find_any(mrt,
+						     &ipv6_hdr(skb)->daddr,
+						     vif);
+	}
+
+	/* No usable cache entry */
+	if (!cache) {
+		vif = ip6mr_find_vif(mrt, dev);
+		if (vif >= 0)
+			return ip6mr_cache_unresolved(mrt, vif, skb, dev);
+		goto ip6_output;
+	}
+
+	/* Wrong interface */
+	vif = cache->_c.mfc_parent;
+	if (rcu_access_pointer(mrt->vif_table[vif].dev) != dev)
+		goto ip6_output;
+
+	ip6_mr_output_finish(net, mrt, dev, skb, cache);
+	return 0;
+
+ip6_output:
+	return ip6_output(net, sk, skb);
+}
+
 int ip6mr_get_route(struct net *net, struct sk_buff *skb, struct rtmsg *rtm,
 		    u32 portid)
 {
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 79c8f1acf8a35..df0caffefb382 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -1145,6 +1145,7 @@ static void ip6_rt_init_dst(struct rt6_info *rt, const struct fib6_result *res)
 		rt->dst.input = ip6_input;
 	} else if (ipv6_addr_type(&f6i->fib6_dst.addr) & IPV6_ADDR_MULTICAST) {
 		rt->dst.input = ip6_mc_input;
+		rt->dst.output = ip6_mr_output;
 	} else {
 		rt->dst.input = ip6_forward;
 	}
-- 
2.51.0


From f8337efa4ff5a27e6c1d4e384166413eecd21a65 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:19 +0200
Subject: [PATCH 15/16] vxlan: Support MC routing in the underlay

Locally-generated MC packets have so far not been subject to MC routing.
Instead an MC-enabled installation would maintain the MC routing tables,
and separately from that the list of interfaces to send packets to as part
of the VXLAN FDB and MDB.

In a previous patch, a ip_mr_output() and ip6_mr_output() routines were
added for IPv4 and IPv6. All locally generated MC traffic is now passed
through these functions. For reasons of backward compatibility, an SKB
(IPCB / IP6CB) flag guards the actual MC routing.

This patch adds logic to set the flag, and the UAPI to enable the behavior.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/d899655bb7e9b2521ee8c793e67056b9fd02ba12.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 drivers/net/vxlan/vxlan_core.c | 22 ++++++++++++++++++++--
 include/net/vxlan.h            |  5 ++++-
 include/uapi/linux/if_link.h   |  1 +
 3 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index b22f9866be8eb..a6cc1de4d8b85 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2451,6 +2451,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 	rcu_read_lock();
 	if (addr_family == AF_INET) {
 		struct vxlan_sock *sock4 = rcu_dereference(vxlan->vn4_sock);
+		u16 ipcb_flags = 0;
 		struct rtable *rt;
 		__be16 df = 0;
 		__be32 saddr;
@@ -2467,6 +2468,9 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 			goto tx_error;
 		}
 
+		if (flags & VXLAN_F_MC_ROUTE)
+			ipcb_flags |= IPSKB_MCROUTE;
+
 		if (!info) {
 			/* Bypass encapsulation if the destination is local */
 			err = encap_bypass_if_local(skb, dev, vxlan, AF_INET,
@@ -2522,11 +2526,13 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 
 		udp_tunnel_xmit_skb(rt, sock4->sock->sk, skb, saddr,
 				    pkey->u.ipv4.dst, tos, ttl, df,
-				    src_port, dst_port, xnet, !udp_sum, 0);
+				    src_port, dst_port, xnet, !udp_sum,
+				    ipcb_flags);
 #if IS_ENABLED(CONFIG_IPV6)
 	} else {
 		struct vxlan_sock *sock6 = rcu_dereference(vxlan->vn6_sock);
 		struct in6_addr saddr;
+		u16 ip6cb_flags = 0;
 
 		if (!ifindex)
 			ifindex = sock6->sock->sk->sk_bound_dev_if;
@@ -2542,6 +2548,9 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 			goto tx_error;
 		}
 
+		if (flags & VXLAN_F_MC_ROUTE)
+			ip6cb_flags |= IP6SKB_MCROUTE;
+
 		if (!info) {
 			u32 rt6i_flags = dst_rt6_info(ndst)->rt6i_flags;
 
@@ -2587,7 +2596,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 		udp_tunnel6_xmit_skb(ndst, sock6->sock->sk, skb, dev,
 				     &saddr, &pkey->u.ipv6.dst, tos, ttl,
 				     pkey->label, src_port, dst_port, !udp_sum,
-				     0);
+				     ip6cb_flags);
 #endif
 	}
 	vxlan_vnifilter_count(vxlan, vni, NULL, VXLAN_VNI_STATS_TX, pkt_len);
@@ -3402,6 +3411,7 @@ static const struct nla_policy vxlan_policy[IFLA_VXLAN_MAX + 1] = {
 	[IFLA_VXLAN_LOCALBYPASS]	= NLA_POLICY_MAX(NLA_U8, 1),
 	[IFLA_VXLAN_LABEL_POLICY]       = NLA_POLICY_MAX(NLA_U32, VXLAN_LABEL_MAX),
 	[IFLA_VXLAN_RESERVED_BITS] = NLA_POLICY_EXACT_LEN(sizeof(struct vxlanhdr)),
+	[IFLA_VXLAN_MC_ROUTE]		= NLA_POLICY_MAX(NLA_U8, 1),
 };
 
 static int vxlan_validate(struct nlattr *tb[], struct nlattr *data[],
@@ -4315,6 +4325,14 @@ static int vxlan_nl2conf(struct nlattr *tb[], struct nlattr *data[],
 			return err;
 	}
 
+	if (data[IFLA_VXLAN_MC_ROUTE]) {
+		err = vxlan_nl2flag(conf, data, IFLA_VXLAN_MC_ROUTE,
+				    VXLAN_F_MC_ROUTE, changelink,
+				    true, extack);
+		if (err)
+			return err;
+	}
+
 	if (tb[IFLA_MTU]) {
 		if (changelink) {
 			NL_SET_ERR_MSG_ATTR(extack, tb[IFLA_MTU],
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index e2f7ca045d3e5..0ee50785f4f1c 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -332,6 +332,7 @@ struct vxlan_dev {
 #define VXLAN_F_VNIFILTER               0x20000
 #define VXLAN_F_MDB			0x40000
 #define VXLAN_F_LOCALBYPASS		0x80000
+#define VXLAN_F_MC_ROUTE		0x100000
 
 /* Flags that are used in the receive path. These flags must match in
  * order for a socket to be shareable
@@ -353,7 +354,9 @@ struct vxlan_dev {
 					 VXLAN_F_UDP_ZERO_CSUM6_RX |	\
 					 VXLAN_F_COLLECT_METADATA  |	\
 					 VXLAN_F_VNIFILTER         |    \
-					 VXLAN_F_LOCALBYPASS)
+					 VXLAN_F_LOCALBYPASS       |	\
+					 VXLAN_F_MC_ROUTE          |	\
+					 0)
 
 struct net_device *vxlan_dev_create(struct net *net, const char *name,
 				    u8 name_assign_type, struct vxlan_config *conf);
diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h
index 3ad2d5d980347..873c285996feb 100644
--- a/include/uapi/linux/if_link.h
+++ b/include/uapi/linux/if_link.h
@@ -1398,6 +1398,7 @@ enum {
 	IFLA_VXLAN_LOCALBYPASS,
 	IFLA_VXLAN_LABEL_POLICY, /* IPv6 flow label policy; ifla_vxlan_label_policy */
 	IFLA_VXLAN_RESERVED_BITS,
+	IFLA_VXLAN_MC_ROUTE,
 	__IFLA_VXLAN_MAX
 };
 #define IFLA_VXLAN_MAX	(__IFLA_VXLAN_MAX - 1)
-- 
2.51.0


From 2a719b7bacc74be9f04147be01262b6289ad8dc5 Mon Sep 17 00:00:00 2001
From: Petr Machata <petrm@nvidia.com>
Date: Tue, 17 Jun 2025 00:44:20 +0200
Subject: [PATCH 16/16] selftests: forwarding: lib: Move smcrouted helpers here

router_multicast.sh has several helpers for work with smcrouted. Extract
them to lib.sh so that other selftests can use them as well. Convert the
helpers to defer in the process, because that simplifies the interface
quite a bit. Therefore have router_multicast.sh invoke
defer_scopes_cleanup() in its cleanup() function.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Link: https://patch.msgid.link/410411c1a81225ce6e44542289b9c3ec21e5786c.1750113335.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 tools/testing/selftests/net/forwarding/lib.sh | 33 +++++++++++++++++
 .../net/forwarding/router_multicast.sh        | 35 ++++---------------
 2 files changed, 39 insertions(+), 29 deletions(-)

diff --git a/tools/testing/selftests/net/forwarding/lib.sh b/tools/testing/selftests/net/forwarding/lib.sh
index 508f3c700d719..253847372062e 100644
--- a/tools/testing/selftests/net/forwarding/lib.sh
+++ b/tools/testing/selftests/net/forwarding/lib.sh
@@ -37,6 +37,7 @@ declare -A NETIFS=(
 : "${TEAMD:=teamd}"
 : "${MCD:=smcrouted}"
 : "${MC_CLI:=smcroutectl}"
+: "${MCD_TABLE_NAME:=selftests}"
 
 # Constants for netdevice bring-up:
 # Default time in seconds to wait for an interface to come up before giving up
@@ -1757,6 +1758,38 @@ mc_send()
 		msend -g $groups -I $if_name -c 1 > /dev/null 2>&1
 }
 
+adf_mcd_start()
+{
+	local table_name="$MCD_TABLE_NAME"
+	local smcroutedir
+	local pid
+	local i
+
+	check_command "$MCD" || return 1
+	check_command "$MC_CLI" || return 1
+
+	smcroutedir=$(mktemp -d)
+	defer rm -rf "$smcroutedir"
+
+	for ((i = 1; i <= NUM_NETIFS; ++i)); do
+		echo "phyint ${NETIFS[p$i]} enable" >> \
+			"$smcroutedir/$table_name.conf"
+	done
+
+	"$MCD" -N -I "$table_name" -f "$smcroutedir/$table_name.conf" \
+		-P "$smcroutedir/$table_name.pid"
+	busywait "$BUSYWAIT_TIMEOUT" test -e "$smcroutedir/$table_name.pid"
+	pid=$(cat "$smcroutedir/$table_name.pid")
+	defer kill_process "$pid"
+}
+
+mc_cli()
+{
+	local table_name="$MCD_TABLE_NAME"
+
+        "$MC_CLI" -I "$table_name" "$@"
+}
+
 start_ip_monitor()
 {
 	local mtype=$1; shift
diff --git a/tools/testing/selftests/net/forwarding/router_multicast.sh b/tools/testing/selftests/net/forwarding/router_multicast.sh
index 5a58b1ec8aefb..83e52abdbc2ed 100755
--- a/tools/testing/selftests/net/forwarding/router_multicast.sh
+++ b/tools/testing/selftests/net/forwarding/router_multicast.sh
@@ -33,10 +33,6 @@ NUM_NETIFS=6
 source lib.sh
 source tc_common.sh
 
-require_command $MCD
-require_command $MC_CLI
-table_name=selftests
-
 h1_create()
 {
 	simple_if_init $h1 198.51.100.2/28 2001:db8:1::2/64
@@ -149,25 +145,6 @@ router_destroy()
 	ip link set dev $rp1 down
 }
 
-start_mcd()
-{
-	SMCROUTEDIR="$(mktemp -d)"
-
-	for ((i = 1; i <= $NUM_NETIFS; ++i)); do
-		echo "phyint ${NETIFS[p$i]} enable" >> \
-			$SMCROUTEDIR/$table_name.conf
-	done
-
-	$MCD -N -I $table_name -f $SMCROUTEDIR/$table_name.conf \
-		-P $SMCROUTEDIR/$table_name.pid
-}
-
-kill_mcd()
-{
-	pkill $MCD
-	rm -rf $SMCROUTEDIR
-}
-
 setup_prepare()
 {
 	h1=${NETIFS[p1]}
@@ -179,7 +156,7 @@ setup_prepare()
 	rp3=${NETIFS[p5]}
 	h3=${NETIFS[p6]}
 
-	start_mcd
+	adf_mcd_start || exit "$EXIT_STATUS"
 
 	vrf_prepare
 
@@ -206,7 +183,7 @@ cleanup()
 
 	vrf_cleanup
 
-	kill_mcd
+	defer_scopes_cleanup
 }
 
 create_mcast_sg()
@@ -214,9 +191,9 @@ create_mcast_sg()
 	local if_name=$1; shift
 	local s_addr=$1; shift
 	local mcast=$1; shift
-	local dest_ifs=${@}
+	local dest_ifs=("${@}")
 
-	$MC_CLI -I $table_name add $if_name $s_addr $mcast $dest_ifs
+	mc_cli add "$if_name" "$s_addr" "$mcast" "${dest_ifs[@]}"
 }
 
 delete_mcast_sg()
@@ -224,9 +201,9 @@ delete_mcast_sg()
 	local if_name=$1; shift
 	local s_addr=$1; shift
 	local mcast=$1; shift
-	local dest_ifs=${@}
+	local dest_ifs=("${@}")
 
-        $MC_CLI -I $table_name remove $if_name $s_addr $mcast $dest_ifs
+        mc_cli remove "$if_name" "$s_addr" "$mcast" "${dest_ifs[@]}"
 }
 
 mcast_v4()
-- 
2.51.0