ipmr, ip6mr: Allow MC-routing locally-generated MC packets

PATCH Superseded Petr Machata 5 participants 36 messages Jun 9, 2025 – Jun 12, 2025

Subsystems (10)

Networking Network drivers Kselftest IPv4/IPv6 +6 more subsystems

lore

Share & tools

lore

Petr Machata

author

2025-06-09 20:51 UTC #0 lore raw

cover ipmr, ip6mr: Allow MC-routing locally-generated MC packets

Multicast routing is today handled in the input path. Locally generated MC packets don't hit the IPMR code. Thus if a VXLAN remote address is multicast, the driver needs to set an OIF during route lookup. In practice that means that 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 patchset adds support to route locally generated multicast packets.

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/IP6CB flag. Unless the flag is set, the new MC-routing code is skipped.

All this is keyed to a new VXLAN attribute, IFLA_VXLAN_MC_ROUTE. Only when it is set does any of the above engage.

In addition to that, and as is the case today with MC forwarding, IPV4_DEVCONF_MC_FORWARDING must be enabled for the netdevice that acts as a source of MC traffic (i.e. the VXLAN PHYS_DEV), so an MC daemon must be attached to the netdevice.

When a VXLAN netdevice with a MC remote is brought up, the physical netdevice joins the indicated MC group. This is important for local delivery of MC packets, so it is still necessary to configure a physical netdevice -- the parameter cannot go away. The netdevice would however typically not be a front panel port, but a dummy. An MC daemon would then sit on top of that netdevice as well as any front panel ports that it needs to service, and have routes set up between the two.

A way to configure the VXLAN netdevice to take advantage of the new MC routing would be:

 # ip link add name d up type dummy
 # ip link add name vx10 up type vxlan id 1000 dstport 4789 \
	local 192.0.2.1 group 225.0.0.1 ttl 16 dev d mrcoute
 # ip link set dev vx10 master br # plus vlans etc.

With the following MC routes:

(192.0.2.1, 225.0.0.1) iif=d oil=swp1,swp2 # TX route
(*, 225.0.0.1) iif=swp1 oil=d,swp2 # RX route
(*, 225.0.0.1) iif=swp2 oil=d,swp1 # RX route

The RX path has not changed, with the exception of an extra MC hop. Packets are delivered to the front panel port and MC-forwarded to the VXLAN physical port, here "d". Since the port has joined the multicast group, the packets are locally delivered, and end up being processed by the VXLAN netdevice.

This patchset is based on earlier patches from Nikolay Aleksandrov and Roopa Prabhu, though it underwent significant changes. Roopa broadly presented the topic on LPC 2019 [0].

Patchset progression:

- Patches #1 to #4 add ip_mr_output()
- Patches #5 to #9 add ip6_mr_output()
- Patch #10 adds the VXLAN bits to enable MR engagement
- Patches #11 to #13 prepare selftest libraries
- Patch #14 includes a new test suite

[0] https://www.youtube.com/watch?v=xlReECfi-uo

Petr Machata (14):
net: ipv4: Add a flags argument to iptunnel_xmit(),
udp_tunnel_xmit_skb()
net: ipv4: ipmr: ipmr_queue_xmit(): Drop local variable `dev'
net: ipv4: ipmr: Split ipmr_queue_xmit() in two
net: ipv4: Add ip_mr_output()
net: ipv6: Make udp_tunnel6_xmit_skb() void
net: ipv6: Add a flags argument to ip6tunnel_xmit(),

    udp_tunnel6_xmit_skb()

net: ipv6: ip6mr: Fix in/out netdev to pass to the FORWARD chain
net: ipv6: ip6mr: Extract a helper out of ip6mr_forward2()
net: ipv6: Add ip6_mr_output()
vxlan: Support MC routing in the underlay
selftests: forwarding: lib: Move smcrouted helpers here
selftests: net: lib: Add ip_link_has_flag()
selftests: forwarding: adf_mcd_start(): Allow configuring custom
interfaces
selftests: forwarding: Add a test for verifying VXLAN MC underlay

 drivers/net/amt.c                             |   9 +-
 drivers/net/bareudp.c                         |   7 +-
 drivers/net/geneve.c                          |   7 +-
 drivers/net/gtp.c                             |  12 +-
 drivers/net/ovpn/udp.c                        |   4 +-
 drivers/net/vxlan/vxlan_core.c                |  23 +-
 drivers/net/wireguard/socket.c                |   4 +-
 include/linux/ipv6.h                          |   1 +
 include/linux/mroute6.h                       |   7 +
 include/net/ip.h                              |   2 +
 include/net/ip6_tunnel.h                      |   3 +-
 include/net/ip_tunnels.h                      |   2 +-
 include/net/udp_tunnel.h                      |  17 +-
 include/net/vxlan.h                           |   5 +-
 include/uapi/linux/if_link.h                  |   1 +
 net/ipv4/ip_tunnel.c                          |   4 +-
 net/ipv4/ip_tunnel_core.c                     |   4 +-
 net/ipv4/ipmr.c                               | 169 +++-
 net/ipv4/route.c                              |   2 +-
 net/ipv4/udp_tunnel_core.c                    |   5 +-
 net/ipv6/ip6_tunnel.c                         |   2 +-
 net/ipv6/ip6_udp_tunnel.c                     |  18 +-
 net/ipv6/ip6mr.c                              | 137 +++-
 net/ipv6/route.c                              |   1 +
 net/ipv6/sit.c                                |   2 +-
 net/sctp/ipv6.c                               |   7 +-
 net/sctp/protocol.c                           |   3 +-
 net/tipc/udp_media.c                          |  12 +-
 .../testing/selftests/net/forwarding/Makefile |   1 +
 tools/testing/selftests/net/forwarding/lib.sh |  43 +
 .../net/forwarding/router_multicast.sh        |  31 +-
 .../net/forwarding/vxlan_bridge_1q_mc_ul.sh   | 757 ++++++++++++++++++
 tools/testing/selftests/net/lib.sh            |  12 +-
 33 files changed, 1199 insertions(+), 115 deletions(-)
 create mode 100755 tools/testing/selftests/net/forwarding/vxlan_bridge_1q_mc_ul.sh

Signature · 1 line

2.49.0

Petr Machata

author

2025-06-09 20:51 UTC #1 lore raw

PATCH 1/14 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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---

Notes:
CC: Pablo Neira Ayuso <[email protected]>
CC: [email protected]
CC: Andrew Lunn <[email protected]>
CC: Taehee Yoo <[email protected]>
CC: Antonio Quartulli <[email protected]>
CC: "Jason A. Donenfeld" <[email protected]>
CC: [email protected]
CC: Marcelo Ricardo Leitner <[email protected]>
CC: [email protected]
CC: Jon Maloy <[email protected]>
CC: [email protected]

 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(-)

drivers/net/amt.c +6 -3

index 734a0b3242a9..d0f719531499 100644
@@ -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;
 }

drivers/net/bareudp.c +2 -2

index a9dffdcac805..5e613080d3f8 100644
@@ -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:

drivers/net/geneve.c +2 -2

index ffc15a432689..c668e8b00ed2 100644
@@ -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;
 }

drivers/net/gtp.c +6 -4

index d4dec741c7f4..14584793fe4e 100644
@@ -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;
 }

drivers/net/ovpn/udp.c +1 -1

index bff00946eae2..d866e6bfda70 100644
@@ -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();

drivers/net/vxlan/vxlan_core.c +1 -1

index a56d7239b127..d7a5d8873a1b 100644
@@ -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);

drivers/net/wireguard/socket.c +1 -1

index 0414d7a6ce74..88e685667bc0 100644
@@ -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:

include/net/ip_tunnels.h +1 -1

index 0c3d571a04a1..8cf1380f3656 100644
@@ -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,

include/net/udp_tunnel.h +1 -1

index 2df3b8344eb5..28102c8fd8a8 100644
@@ -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,

net/ipv4/ip_tunnel.c +2 -2

index 678b8f96e3e9..aaeb5d16f0c9 100644
@@ -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)

net/ipv4/ip_tunnel_core.c +3 -1

index f65d2f727381..cc9915543637 100644
@@ -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));

net/ipv4/udp_tunnel_core.c +3 -2

index 2326548997d3..9efd62505916 100644
@@ -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);

net/ipv6/sit.c +1 -1

index a72dbca9e8fc..12496ba1b7d4 100644
@@ -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:

net/sctp/protocol.c +2 -1

index 8c3b80c4d40b..bfbb73e359f5 100644
@@ -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;
 }

net/tipc/udp_media.c +1 -1

index 108a4cc2e001..87e8c1e6d550 100644
@@ -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) {

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 1/14

2025-06-12 10:28 UTC #2 lore raw

Re: [PATCH net-next 01/14] net: ipv4: Add a flags argument to iptunnel_xmit(), udp_tunnel_xmit_skb()

Quoted 24 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---
Notes:
CC: Pablo Neira Ayuso <[email protected]>
CC: [email protected]
CC: Andrew Lunn <[email protected]>
CC: Taehee Yoo <[email protected]>
CC: Antonio Quartulli <[email protected]>
CC: "Jason A. Donenfeld" <[email protected]>
CC: [email protected]
CC: Marcelo Ricardo Leitner <[email protected]>
CC: [email protected]
CC: Jon Maloy <[email protected]>
CC: [email protected]

Quoted 17 lines

  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(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Antonio Quartulli

maintainer Acked-by on v1 1/14

2025-06-12 11:21 UTC #3 lore raw

Re: [PATCH net-next 01/14] net: ipv4: Add a flags argument to iptunnel_xmit(), udp_tunnel_xmit_skb()

In reply to Petr Machata

Quoted 11 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---

Quoted 12 lines

Notes:
CC: Pablo Neira Ayuso <[email protected]>
CC: [email protected]
CC: Andrew Lunn <[email protected]>
CC: Taehee Yoo <[email protected]>
CC: Antonio Quartulli <[email protected]>
CC: "Jason A. Donenfeld" <[email protected]>
CC: [email protected]
CC: Marcelo Ricardo Leitner <[email protected]>
CC: [email protected]
CC: Jon Maloy <[email protected]>
CC: [email protected]

Acked-by Antonio Quartulli <[email protected]>

Signature · 2 lines

Antonio Quartulli
OpenVPN Inc.

Petr Machata

author

2025-06-09 20:51 UTC #4 lore raw

PATCH 2/14 net: ipv4: ipmr: ipmr_queue_xmit(): Drop local variable `dev'

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---
 net/ipv4/ipmr.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

net/ipv4/ipmr.c +2 -5

index 2ff2f79c7351..1c5e6167cd76 100644
@@ -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;

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 2/14

2025-06-12 10:28 UTC #5 lore raw

Re: [PATCH net-next 02/14] net: ipv4: ipmr: ipmr_queue_xmit(): Drop local variable `dev'

Quoted 6 lines · Petr Machata

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.

Quoted 8 lines

Drop the local altogether and inline the uses.
Signed-off-by Petr Machata <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---
  net/ipv4/ipmr.c | 7 ++-----
  1 file changed, 2 insertions(+), 5 deletions(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:51 UTC #6 lore raw

PATCH 3/14 net: ipv4: ipmr: Split ipmr_queue_xmit() in two

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---
 net/ipv4/ipmr.c | 45 +++++++++++++++++++++++++++++----------------
 1 file changed, 29 insertions(+), 16 deletions(-)

net/ipv4/ipmr.c +29 ~~-16~~

index 1c5e6167cd76..7c1045d67ea8 100644
@@ -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 @@ static void ip_mr_forward(struct net *net, struct mr_table *mrt,
 				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 @@ static void ip_mr_forward(struct net *net, struct mr_table *mrt,
 			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;
 		}
 	}

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 3/14

2025-06-12 10:29 UTC #7 lore raw

Re: [PATCH net-next 03/14] net: ipv4: ipmr: Split ipmr_queue_xmit() in two

Quoted 6 lines · Petr Machata

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.

Quoted 6 lines

Signed-off-by Petr Machata <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---
  net/ipv4/ipmr.c | 45 +++++++++++++++++++++++++++++----------------
  1 file changed, 29 insertions(+), 16 deletions(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:51 UTC #8 lore raw

PATCH 4/14 net: ipv4: Add ip_mr_output()

In reply to Petr Machata

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 <[email protected]>

Signed-off-by Nikolay Aleksandrov <[email protected]>

Signed-off-by Benjamin Poirier <[email protected]>

Signed-off-by Petr Machata <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---
 include/net/ip.h |   2 +
 net/ipv4/ipmr.c  | 117 +++++++++++++++++++++++++++++++++++++++++++++++
 net/ipv4/route.c |   2 +-
 3 files changed, 120 insertions(+), 1 deletion(-)

include/net/ip.h +2 -0

index 47ed6d23853d..375304bb99f6 100644
@@ -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,

net/ipv4/ipmr.c +117 -0

129 lines not on this page · raw message

net/ipv4/route.c +1 -1

index fccb05fb3a79..3ddf6bf40357 100644
@@ -2660,7 +2660,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
 			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

Signature · 1 line

2.49.0

Petr Machata

author

2025-06-09 20:52 UTC #9 lore raw

PATCH 5/14 net: ipv6: Make udp_tunnel6_xmit_skb() void

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---

Notes:
CC:Marcelo Ricardo Leitner <[email protected]> CC:[email protected]
CC:Jon Maloy <[email protected]> CC:[email protected]

 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(-)

include/net/udp_tunnel.h +7 -7

index 28102c8fd8a8..0b01f6ade20d 100644
@@ -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);

net/ipv6/ip6_udp_tunnel.c +7 -8

index c99053189ea8..21681718b7bb 100644
@@ -74,13 +74,13 @@ int udp_sock_create6(struct net *net, struct udp_port_cfg *cfg,
 }
 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);

net/sctp/ipv6.c +4 -3

index a9ed2ccab1bd..d1ecf7454827 100644
@@ -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

net/tipc/udp_media.c +5 -5

index 87e8c1e6d550..414713fcd8c5 100644
@@ -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();

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 5/14

2025-06-12 10:29 UTC #10 lore raw

Re: [PATCH net-next 05/14] net: ipv6: Make udp_tunnel6_xmit_skb() void

Quoted 23 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---
Notes:
CC:Marcelo Ricardo Leitner <[email protected]> CC:[email protected]
CC:Jon Maloy <[email protected]> CC:[email protected]

Quoted 6 lines

  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(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #11 lore raw

PATCH 6/14 net: ipv6: Add a flags argument to ip6tunnel_xmit(), udp_tunnel6_xmit_skb()

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---

Notes:
CC:Pablo Neira Ayuso <[email protected]> CC:[email protected] CC:Andrew Lunn <[email protected]> CC:Antonio Quartulli <[email protected]> CC:"Jason A. Donenfeld" <[email protected]> CC:[email protected]
CC:Marcelo Ricardo Leitner <[email protected]> CC:[email protected]
CC:Jon Maloy <[email protected]> CC:[email protected]

 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(-)

drivers/net/bareudp.c +2 -1

index 5e613080d3f8..0df3208783ad 100644
@@ -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:

drivers/net/geneve.c +2 -1

index c668e8b00ed2..f6bd155aae7f 100644
@@ -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

drivers/net/gtp.c +1 -1

index 14584793fe4e..4b668ebaa0f7 100644
@@ -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

drivers/net/ovpn/udp.c +1 -1

index d866e6bfda70..254cc94c4617 100644
@@ -274,7 +274,7 @@ static int ovpn_udp6_output(struct ovpn_peer *peer, struct ovpn_bind *bind,
 	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();

drivers/net/vxlan/vxlan_core.c +2 -1

index d7a5d8873a1b..c4af6c652560 100644
@@ -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);

drivers/net/wireguard/socket.c +1 -1

index 88e685667bc0..253488f8c00f 100644
@@ -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:

include/net/ip6_tunnel.h +2 -1

index 399592405c72..dd163495f353 100644
@@ -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);

include/net/udp_tunnel.h +2 -1

index 0b01f6ade20d..e3c70b579095 100644
@@ -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);

net/ipv6/ip6_tunnel.c +1 -1

index 894d3158a6f0..a885bb5c98ea 100644
@@ -1278,7 +1278,7 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 	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);

net/ipv6/ip6_udp_tunnel.c +3 -2

index 21681718b7bb..8ebe17a6058a 100644
@@ -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);

net/sctp/ipv6.c +1 -1

index d1ecf7454827..3336dcfb4515 100644
@@ -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;
 }

net/tipc/udp_media.c +1 -1

index 414713fcd8c5..a024fcc8c0cb 100644
@@ -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();

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 6/14

2025-06-12 10:30 UTC #12 lore raw

Re: [PATCH net-next 06/14] net: ipv6: Add a flags argument to ip6tunnel_xmit(), udp_tunnel6_xmit_skb()

Quoted 23 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---
Notes:
CC:Pablo Neira Ayuso <[email protected]> CC:[email protected] CC:Andrew Lunn <[email protected]> CC:Antonio Quartulli <[email protected]> CC:"Jason A. Donenfeld" <[email protected]> CC:[email protected]
CC:Marcelo Ricardo Leitner <[email protected]> CC:[email protected]
CC:Jon Maloy <[email protected]> CC:[email protected]

Quoted 14 lines

  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(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #13 lore raw

PATCH 7/14 net: ipv6: ip6mr: Fix in/out netdev to pass to the FORWARD chain

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---

Notes:
This never worked correctly, hence going through net-next.
---
CC: YOSHIFUJI Hideaki <[email protected]>

 net/ipv6/ip6mr.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

net/ipv6/ip6mr.c +2 -1

index 3276cde5ebd7..63c90dae6cbf 100644
@@ -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:

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 7/14

2025-06-12 10:31 UTC #14 lore raw

Re: [PATCH net-next 07/14] net: ipv6: ip6mr: Fix in/out netdev to pass to the FORWARD chain

Quoted 19 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---

Quoted 8 lines

Notes:
This never worked correctly, hence going through net-next.
---
CC: YOSHIFUJI Hideaki <[email protected]>
  net/ipv6/ip6mr.c | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #15 lore raw

PATCH 8/14 net: ipv6: ip6mr: Extract a helper out of ip6mr_forward2()

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---
 net/ipv6/ip6mr.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

net/ipv6/ip6mr.c +16 -6

index 63c90dae6cbf..03bfc0b65175 100644
@@ -2035,11 +2035,10 @@ 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 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;

@@ -2098,6 +2097,20 @@ static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
 
 	ipv6h = ipv6_hdr(skb);
 	ipv6h->hop_limit--;
+	return 0;
+
+out_free:
+	kfree_skb(skb);
+	return -1;
+}
+
+static int 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))
+		return 0;
 
 	IP6CB(skb)->flags |= IP6SKB_FORWARDED;

@@ -2105,9 +2118,6 @@ static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
 		       net, NULL, skb, indev, skb->dev,
 		       ip6mr_forward2_finish);
 
-out_free:
-	kfree_skb(skb);
-	return 0;
 }
 
 /* Called with rcu_read_lock() */

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 8/14

2025-06-12 10:31 UTC #16 lore raw

Re: [PATCH net-next 08/14] net: ipv6: ip6mr: Extract a helper out of ip6mr_forward2()

Quoted 11 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---
  net/ipv6/ip6mr.c | 22 ++++++++++++++++------
  1 file changed, 16 insertions(+), 6 deletions(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #17 lore raw

PATCH 9/14 net: ipv6: Add ip6_mr_output()

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---
 include/linux/ipv6.h    |   1 +
 include/linux/mroute6.h |   7 +++
 net/ipv6/ip6mr.c        | 114 ++++++++++++++++++++++++++++++++++++++++
 net/ipv6/route.c        |   1 +
 4 files changed, 123 insertions(+)

include/linux/ipv6.h +1 -0

index 5aeeed22f35b..db0eb0d86b64 100644
@@ -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)

include/linux/mroute6.h +7 -0

index 63ef5191cc57..6526787de67e 100644
@@ -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 0;
+}
+
 static inline void ip6_mr_cleanup(void)
 {
 	return;

net/ipv6/ip6mr.c +114 -0

132 lines not on this page · raw message

net/ipv6/route.c +1 -0

index 0143262094b0..86f88fd8b385 100644
@@ -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;
 	}

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 9/14

2025-06-12 10:35 UTC #18 lore raw

Re: [PATCH net-next 09/14] net: ipv6: Add ip6_mr_output()

Quoted 15 lines · Petr Machata

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.

Quoted 9 lines

Signed-off-by Petr Machata <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---
  include/linux/ipv6.h    |   1 +
  include/linux/mroute6.h |   7 +++
  net/ipv6/ip6mr.c        | 114 ++++++++++++++++++++++++++++++++++++++++
  net/ipv6/route.c        |   1 +
  4 files changed, 123 insertions(+)

Looks good to me,

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #19 lore raw

PATCH 10/14 vxlan: Support MC routing in the underlay

In reply to Petr Machata

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 <[email protected]>

Reviewed-by Ido Schimmel <[email protected]>

---

Notes:
CC:Andrew Lunn <[email protected]>
CC:Menglong Dong <[email protected]>

 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(-)

drivers/net/vxlan/vxlan_core.c +20 -2

index c4af6c652560..02eba9235406 100644
@@ -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],

include/net/vxlan.h +4 -1

index e2f7ca045d3e..0ee50785f4f1 100644
@@ -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);

include/uapi/linux/if_link.h +1 -0

index 3ad2d5d98034..873c285996fe 100644
@@ -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)

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 10/14

2025-06-12 10:36 UTC #20 lore raw

Re: [PATCH net-next 10/14] vxlan: Support MC routing in the underlay

Quoted 16 lines · Petr Machata

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 <[email protected]>
Reviewed-by Ido Schimmel <[email protected]>
---

Quoted 9 lines

Notes:
CC:Andrew Lunn <[email protected]>
CC:Menglong Dong <[email protected]>

  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(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #21 lore raw

PATCH 11/14 selftests: forwarding: lib: Move smcrouted helpers here

In reply to Petr Machata

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 <[email protected]>

---

Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]

 tools/testing/selftests/net/forwarding/lib.sh | 31 +++++++++++++++++++
 .../net/forwarding/router_multicast.sh        | 31 +++----------------
 2 files changed, 35 insertions(+), 27 deletions(-)

tools/testing/selftests/net/forwarding/lib.sh +31 -0

index 508f3c700d71..88e63562f5c5 100644
@@ -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 +1758,36 @@ mc_send()
 		msend -g $groups -I $if_name -c 1 > /dev/null 2>&1
 }
 
+adf_mcd_start()
+{
+	local i
+
+	check_command $MCD || return 1
+	check_command $MC_CLI || return 1
+
+	local table_name="$MCD_TABLE_NAME"
+	local 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
+	defer kill_process $(cat $smcroutedir/$table_name.pid)
+}
+
+mc_cli()
+{
+	local table_name="$MCD_TABLE_NAME"
+
+        $MC_CLI -I $table_name "$@"
+}
+
 start_ip_monitor()
 {
 	local mtype=$1; shift

tools/testing/selftests/net/forwarding/router_multicast.sh +4 ~~-27~~

index 5a58b1ec8aef..1e2378777b48 100755
@@ -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()

@@ -216,7 +193,7 @@ create_mcast_sg()
 	local mcast=$1; shift
 	local dest_ifs=${@}
 
-	$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()

@@ -226,7 +203,7 @@ delete_mcast_sg()
 	local mcast=$1; shift
 	local dest_ifs=${@}
 
-        $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()

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 11/14

2025-06-12 10:38 UTC #22 lore raw

Re: [PATCH net-next 11/14] selftests: forwarding: lib: Move smcrouted helpers here

Quoted 9 lines · Petr Machata

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 <[email protected]>
---

Quoted 8 lines

Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]

  tools/testing/selftests/net/forwarding/lib.sh | 31 +++++++++++++++++++
  .../net/forwarding/router_multicast.sh        | 31 +++----------------
  2 files changed, 35 insertions(+), 27 deletions(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #23 lore raw

PATCH 12/14 selftests: net: lib: Add ip_link_has_flag()

In reply to Petr Machata

Add a helper to determine whether a given netdevice has a given flag.

Rewrite ip_link_is_up() in terms of the new helper.

Signed-off-by Petr Machata <[email protected]>

---

Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]

 tools/testing/selftests/net/lib.sh | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

tools/testing/selftests/net/lib.sh +9 -3

index 006fdadcc4b9..ff0dbe23e8e0 100644
@@ -547,13 +547,19 @@ ip_link_set_addr()
 	defer ip link set dev "$name" address "$old_addr"
 }
 
-ip_link_is_up()
+ip_link_has_flag()
 {
 	local name=$1; shift
+	local flag=$1; shift
 
 	local state=$(ip -j link show "$name" |
-		      jq -r '(.[].flags[] | select(. == "UP")) // "DOWN"')
-	[[ $state == "UP" ]]
+		      jq --arg flag "$flag" 'any(.[].flags.[]; . == $flag)')
+	[[ $state == true ]]
+}
+
+ip_link_is_up()
+{
+	ip_link_has_flag "$1" UP
 }
 
 ip_link_set_up()

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 12/14

2025-06-12 10:38 UTC #24 lore raw

Re: [PATCH net-next 12/14] selftests: net: lib: Add ip_link_has_flag()

Quoted 5 lines · Petr Machata

Add a helper to determine whether a given netdevice has a given flag.
Rewrite ip_link_is_up() in terms of the new helper.
Signed-off-by Petr Machata <[email protected]>

Quoted 9 lines

---
Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]
  tools/testing/selftests/net/lib.sh | 12 +++++++++---
  1 file changed, 9 insertions(+), 3 deletions(-)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:52 UTC #25 lore raw

PATCH 13/14 selftests: forwarding: adf_mcd_start(): Allow configuring custom interfaces

In reply to Petr Machata

Tests may wish to add other interfaces to listen on. Notably locally generated traffic uses dummy interfaces. The multicast daemon needs to know about these so that it allows forming rules that involve these interfaces, and so that net.ipv4.conf.X.mc_forwarding is set for the interfaces.

To that end, allow passing in a list of interfaces to configure in addition to all the physical ones.

Signed-off-by Petr Machata <[email protected]>

---

Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]

 tools/testing/selftests/net/forwarding/lib.sh | 12 ++++++++++++
 1 file changed, 12 insertions(+)

tools/testing/selftests/net/forwarding/lib.sh +12 -0

index 88e63562f5c5..5f144d75167a 100644
@@ -1760,6 +1760,8 @@ mc_send()
 
 adf_mcd_start()
 {
+	local ifs=("$@")
+	local if
 	local i
 
 	check_command $MCD || return 1

@@ -1775,6 +1777,16 @@ adf_mcd_start()
 			$smcroutedir/$table_name.conf
 	done
 
+	for if in ${ifs[@]}; do
+		if ! ip_link_has_flag "$if" MULTICAST; then
+			ip link set dev "$if" multicast on
+			defer ip link set dev "$if" multicast off
+		fi
+
+		echo "phyint $if 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

Signature · 1 line

2.49.0

Nikolay Aleksandrov

Reviewed-by on v1 13/14

2025-06-12 10:38 UTC #26 lore raw

Re: [PATCH net-next 13/14] selftests: forwarding: adf_mcd_start(): Allow configuring custom interfaces

Quoted 9 lines · Petr Machata

Tests may wish to add other interfaces to listen on. Notably locally generated traffic uses dummy interfaces. The multicast daemon needs to know about these so that it allows forming rules that involve these interfaces, and so that net.ipv4.conf.X.mc_forwarding is set for the interfaces.
To that end, allow passing in a list of interfaces to configure in addition to all the physical ones.
Signed-off-by Petr Machata <[email protected]>

Quoted 9 lines

---
Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]
  tools/testing/selftests/net/forwarding/lib.sh | 12 ++++++++++++
  1 file changed, 12 insertions(+)

Reviewed-by Nikolay Aleksandrov <[email protected]>

Petr Machata

author

2025-06-09 20:53 UTC #27 lore raw

PATCH 14/14 selftests: forwarding: Add a test for verifying VXLAN MC underlay

In reply to Petr Machata

Add tests for MC-routing underlay VXLAN traffic.

Signed-off-by Petr Machata <[email protected]>

---

Notes:
CC: Shuah Khan <[email protected]>
CC: [email protected]

 .../testing/selftests/net/forwarding/Makefile |   1 +
 .../net/forwarding/vxlan_bridge_1q_mc_ul.sh   | 757 ++++++++++++++++++
 2 files changed, 758 insertions(+)
 create mode 100755 tools/testing/selftests/net/forwarding/vxlan_bridge_1q_mc_ul.sh

tools/testing/selftests/net/forwarding/Makefile +1 -0

index 00bde7b6f39e..d7bb2e80e88c 100644
@@ -102,6 +102,7 @@ TEST_PROGS = bridge_fdb_learning_limit.sh \
 	vxlan_bridge_1d_port_8472.sh \
 	vxlan_bridge_1d.sh \
 	vxlan_bridge_1q_ipv6.sh \
+	vxlan_bridge_1q_mc_ul.sh \
 	vxlan_bridge_1q_port_8472_ipv6.sh \
 	vxlan_bridge_1q_port_8472.sh \
 	vxlan_bridge_1q.sh \

tools/testing/selftests/net/forwarding/vxlan_bridge_1q_mc_ul.sh new file +757 -0

757 lines not on this page · raw message

Signature · 1 line

2.49.0

Jakub Kicinski

maintainer

2025-06-10 13:03 UTC #28 lore raw

Re: [PATCH net-next 14/14] selftests: forwarding: Add a test for verifying VXLAN MC underlay

Quoted 1 line · On Mon, 9 Jun 2025 22:50:30 +0200 Petr Machata wrote:

Add tests for MC-routing underlay VXLAN traffic.

nit: we started using shellcheck, may be worth addressing the warning-level complaints?

Petr Machata

author

2025-06-10 16:23 UTC #29 lore raw

Re: [PATCH net-next 14/14] selftests: forwarding: Add a test for verifying VXLAN MC underlay

Quoted 5 lines · Jakub Kicinski

Quoted 1 line · On Mon, 9 Jun 2025 22:50:30 +0200 Petr Machata wrote:
Add tests for MC-routing underlay VXLAN traffic.
nit: we started using shellcheck, may be worth addressing the warning-level complaints?

Sure. Sorry, I noticed and meant to do that, but it slipped my mind :-|

Jakub Kicinski

maintainer

2025-06-10 12:58 UTC #30 lore raw

In reply to Petr Machata

Quoted 6 lines · On Mon, 9 Jun 2025 22:50:16 +0200 Petr Machata wrote:

Multicast routing is today handled in the input path. Locally generated MC packets don't hit the IPMR code. Thus if a VXLAN remote address is multicast, the driver needs to set an OIF during route lookup. In practice that means that 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.

I think this leads to kmemleaks:

unreferenced object 0xffff88800aabe740 (size 232):
comm "kworker/0:2", pid 471, jiffies 4295215616
hex dump (first 32 bytes):
00 40 df 08 80 88 ff ff 00 f7 5d 98 ff ff ff ff .@........].....
a1 55 19 95 ff ff ff ff 00 00 00 00 00 00 00 00 .U..............
backtrace (crc b1fabddb):
kmem_cache_alloc_noprof (./include/linux/kmemleak.h:43 mm/slub.c:4152 mm/slub.c:4197 mm/slub.c:4204) dst_alloc (net/core/dst.c:89)

ip6_rt_pcpu_alloc (net/ipv6/route.c:342 net/ipv6/route.c:1419) 
ip6_pol_route (net/ipv6/route.c:1468 net/ipv6/route.c:2305) 
fib6_rule_lookup (./include/net/ip6_fib.h:617 net/ipv6/ip6_fib.c:326) 
ip6_route_output_flags (net/ipv6/route.c:2699) 
ip6_dst_lookup_tail.constprop.0 (net/ipv6/ip6_output.c:1128) 
ip6_dst_lookup_flow (net/ipv6/ip6_output.c:1260) 
udp_tunnel6_dst_lookup (net/ipv6/ip6_udp_tunnel.c:165 net/ipv6/ip6_udp_tunnel.c:135) ip6_udp_tunnel 
vxlan_xmit_one (drivers/net/vxlan/vxlan_core.c:2540 (discriminator 4)) vxlan 
vxlan_xmit (drivers/net/vxlan/vxlan_core.c:2809) vxlan 
dev_hard_start_xmit (./include/linux/netdevice.h:5215 ./include/linux/netdevice.h:5224 net/core/dev.c:3830 net/core/dev.c:3846) 
__dev_queue_xmit (net/core/dev.h:356 net/core/dev.c:4714) 
ip6_finish_output2 (./include/net/neighbour.h:539 net/ipv6/ip6_output.c:141) 
ip6_finish_output.constprop.0 (net/ipv6/ip6_output.c:215 net/ipv6/ip6_output.c:226)

mld_sendpack (net/ipv6/mcast.c:1872)

hit by netdevsim udp_tunnel_nic.sh

Also, do you have a branch with the iproute2 patches we could pull in the CI?

Petr Machata

author

2025-06-10 13:14 UTC #31 lore raw

Quoted 11 lines · Jakub Kicinski

Quoted 6 lines · On Mon, 9 Jun 2025 22:50:16 +0200 Petr Machata wrote:
Multicast routing is today handled in the input path. Locally generated MC packets don't hit the IPMR code. Thus if a VXLAN remote address is multicast, the driver needs to set an OIF during route lookup. In practice that means that 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.
I think this leads to kmemleaks:
[...]
hit by netdevsim udp_tunnel_nic.sh

Thanks, I'll take a look.

Quoted 2 lines

Also, do you have a branch with the iproute2 patches we could pull in the CI?

My bad, didn't think of it:
https://github.com/pmachata/iproute2/commits/vxlan_mc_ul/

Petr Machata

author

2025-06-11 16:34 UTC #32 lore raw

Quoted 10 lines · Petr Machata

Quoted 8 lines · Jakub Kicinski
Quoted 6 lines · On Mon, 9 Jun 2025 22:50:16 +0200 Petr Machata wrote:
Multicast routing is today handled in the input path. Locally generated MC packets don't hit the IPMR code. Thus if a VXLAN remote address is multicast, the driver needs to set an OIF during route lookup. In practice that means that 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.

Quoted 5 lines

Quoted 3 lines
I think this leads to kmemleaks:
[...]
hit by netdevsim udp_tunnel_nic.sh
Thanks, I'll take a look.

Hmm, I can't reproduce this :-| I'm using the following incantation to build the kernel:

    vng --build --config tools/testing/selftests/net/forwarding/config \
                --config tools/testing/selftests/drivers/net/config \
                --config tools/testing/selftests/drivers/net/netdevsim/config \
                --config kernel/configs/debug.config

And run the test like so:

    vng -v --run . --user root --cpus 4 -- \
        make -C tools/testing/selftests TARGETS=drivers/net/netdevsim \
                TEST_PROGS=udp_tunnel_nic.sh TEST_GEN_PROGS="" run_tests

vng -v --run . --user root --cpus 4 -- \
        bash -c 'make -C tools/testing/selftests TARGETS=drivers/net/netdevsim \
                        TEST_PROGS=udp_tunnel_nic.sh TEST_GEN_PROGS="" run_tests; \
                 echo scan > /sys/kernel/debug/kmemleak; \
                 cat /sys/kernel/debug/kmemleak'

Anything I'm missing, or is this what the CI is doing, more or less?

Could it actually have been caused by another test? The howto page mentions that the CI is running the tests one at a time, so I don't suppose that's a possibility.

I'll try to run a more fuller suite tomorrow and star at the code a bit to see if I might be missing an error branch or something.

Jakub Kicinski

maintainer

2025-06-11 20:23 UTC #33 lore raw

Quoted 6 lines · On Wed, 11 Jun 2025 17:30:15 +0200 Petr Machata wrote:

Could it actually have been caused by another test? The howto page mentions that the CI is running the tests one at a time, so I don't suppose that's a possibility.
I'll try to run a more fuller suite tomorrow and star at the code a bit to see if I might be missing an error branch or something.

We also hit a crash in ipv6 fcnal.sh, too. Looks like this is either a kmemleak false positive or possibly related to the rtnl changes in ipv6. Either way I it's not related to you changes, sorry about that! :(

[ 2900.792890] BUG: kernel NULL pointer dereference, address: 0000000000000108
[ 2900.792961] #PF: supervisor read access in kernel mode
[ 2900.793017] #PF: error_code(0x0000) - not-present page
[ 2900.793053] PGD 8fd6067 P4D 8fd6067 PUD 6402067 PMD 0 
[ 2900.793097] Oops: Oops: 0000 [#1] SMP NOPTI
[ 2900.793127] CPU: 0 UID: 0 PID: 15652 Comm: nettest Not tainted 6.15.0-virtme #1 PREEMPT(voluntary) 
[ 2900.793200] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 2900.793245] RIP: 0010:ip6_pol_route+0x286/0x4a0
[ 2900.793290] Code: 0c 24 0f 85 fb 01 00 00 09 ca 0f 88 2f 01 00 00 e8 cf 11 43 ff 83 cb 08 48 8d 7c 24 18 e8 32 7b ff ff 0f b7 cb ba ff ff ff ff <4c> 8b 80 08 01 00 00 48 89 c6 49 89 c7 49 8d b8 80 06 00 00 4c 89
[ 2900.793422] RSP: 0018:ffffc08a0932f480 EFLAGS: 00010246
[ 2900.793460] RAX: 0000000000000000 RBX: 0000000000000008 RCX: 0000000000000008
[ 2900.793521] RDX: 00000000ffffffff RSI: ffffc08a0932f740 RDI: ffff9adac8c8f1a8
[ 2900.793580] RBP: ffff9adac87458c0 R08: 0000000000000000 R09: 0000000000000000
[ 2900.793635] R10: 0000000000000000 R11: 0000000000000040 R12: ffff9adac82e362c
[ 2900.793692] R13: ffff9adac82e3600 R14: 0000000000000080 R15: 0000000000000000
[ 2900.793752] FS:  00007f3418913740(0000) GS:ffff9adb7373a000(0000) knlGS:0000000000000000
[ 2900.793816] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2900.793864] CR2: 0000000000000108 CR3: 0000000008007004 CR4: 0000000000772ef0
[ 2900.793920] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 2900.793977] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 2900.794031] PKRU: 55555554
[ 2900.794050] Call Trace:
[ 2900.794070]  <TASK>
[ 2900.794090]  ? __pfx_ip6_pol_route_output+0x10/0x10
[ 2900.794131]  fib6_rule_action+0xe3/0x310
[ 2900.794166]  fib_rules_lookup+0x1b2/0x2b0
[ 2900.794200]  ? __pfx_ip6_pol_route_output+0x10/0x10
[ 2900.794241]  fib6_rule_lookup+0xa9/0x270
[ 2900.794271]  ? __pfx_ip6_pol_route_output+0x10/0x10
[ 2900.794310]  ip6_route_output_flags+0xab/0x180
[ 2900.794353]  ip6_dst_lookup_tail.constprop.0+0x282/0x340
[ 2900.794394]  ip6_dst_lookup_flow+0x46/0xc0
[ 2900.794422]  vrf_xmit+0x100/0x4a0
[ 2900.794459]  dev_hard_start_xmit+0x8d/0x1c0

https://netdev-3.bots.linux.dev/vmksft-net/results/160541/vm-crash-thr0-0

Kuniyuki Iwashima

2025-06-11 21:06 UTC #34 lore raw

From: Jakub Kicinski <[email protected]>
Date: Wed, 11 Jun 2025 13:23:20 -0700

Quoted 12 lines · On Wed, 11 Jun 2025 17:30:15 +0200 Petr Machata wrote:

Quoted 6 lines
Could it actually have been caused by another test? The howto page mentions that the CI is running the tests one at a time, so I don't suppose that's a possibility.
I'll try to run a more fuller suite tomorrow and star at the code a bit to see if I might be missing an error branch or something.
We also hit a crash in ipv6 fcnal.sh, too. Looks like this is either a kmemleak false positive or possibly related to the rtnl changes in ipv6. Either way I it's not related to you changes, sorry about that! :(

Quoted 8 lines

[ 2900.792890] BUG: kernel NULL pointer dereference, address: 0000000000000108
[ 2900.792961] #PF: supervisor read access in kernel mode
[ 2900.793017] #PF: error_code(0x0000) - not-present page
[ 2900.793053] PGD 8fd6067 P4D 8fd6067 PUD 6402067 PMD 0 
[ 2900.793097] Oops: Oops: 0000 [#1] SMP NOPTI
[ 2900.793127] CPU: 0 UID: 0 PID: 15652 Comm: nettest Not tainted 6.15.0-virtme #1 PREEMPT(voluntary) 
[ 2900.793200] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 2900.793245] RIP: 0010:ip6_pol_route+0x286/0x4a0

fwiw, my local syzkaller had the same splat on 6a325aed130b, where my IPv6 series hadn't landed, and syzbot had a similar one on f1b785f4c787, where my RTNL work was on the prep stage, so I think this is an old? bug :)

I'll see if the .syz repro in the report still works on the latest.

syzbot:
https://lore.kernel.org/netdev/[email protected]/

syzkaller:
---8<---

BUG: unable to handle page fault for address: 00007f93e4d51a23
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 12cd4c067 P4D 125130067 PUD 0 
Oops: Oops: 0000 [#1] SMP
CPU: 1 UID: 0 PID: 2259 Comm: syz.2.555 Not tainted 6.15.0-rc1-00220-g6a325aed130b #2 PREEMPT(voluntary)  fb2957dd255dfc2983199dfa2ccdedc09370f316
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014
RIP: 0010:arch_atomic_read arch/x86/include/asm/atomic.h:23 [inline]
RIP: 0010:raw_atomic_read include/linux/atomic/atomic-arch-fallback.h:457 [inline]
RIP: 0010:atomic_read include/linux/atomic/atomic-instrumented.h:33 [inline]
RIP: 0010:rt_genid_ipv6 include/net/net_namespace.h:537 [inline]
RIP: 0010:rt6_is_valid net/ipv6/route.c:1435 [inline]
RIP: 0010:rt6_get_pcpu_route net/ipv6/route.c:1445 [inline]
RIP: 0010:ip6_pol_route+0x301/0x9d0 net/ipv6/route.c:2298
Code: a9 fe 4d 85 ff 4c 8b 74 24 10 74 3b 41 8b 9f 98 00 00 00 31 ff 89 de e8 3d 37 a9 fe 85 db 74 33 49 8b 07 48 8b 80 08 01 00 00 <8b> a8 24 0a 00 00 89 df 89 ee e8 f0 34 a9 fe 39 eb 75 20 e8 67 33
RSP: 0018:ffa0000000c3ba20 EFLAGS: 00010202
RAX: 00007f93e4d50fff RBX: 000000001ac62540 RCX: 0000000000000002
RDX: ff11000005e99700 RSI: 000000001ac62540 RDI: 0000000000000000
RBP: 0000000000000003 R08: 0000000000000000 R09: 0000000000000000
R10: 00000000ffff0000 R11: 0000000000000002 R12: ffffffff829af127
R13: 0000000000000000 R14: ff110001053e2580 R15: ff1100012505f400
FS:  00007f190e8276c0(0000) GS:ff110001b79d2000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f93e4d51a23 CR3: 000000011d7b5004 CR4: 0000000000771ef0
DR0: 0000000080000001 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe07f0 DR7: 0000000000000600
PKRU: 80000000
Call Trace:
 <TASK>
 pol_lookup_func include/net/ip6_fib.h:616 [inline]
 fib6_rule_lookup+0xe8/0x2a0 net/ipv6/fib6_rules.c:120
 ip6_route_output_flags_noref net/ipv6/route.c:2673 [inline]
 ip6_route_output_flags+0x188/0x260 net/ipv6/route.c:2685
 ip6_route_output include/net/ip6_route.h:93 [inline]
 ip6_dst_lookup_tail+0x9a/0x7d0 net/ipv6/ip6_output.c:1128
 ip6_dst_lookup_flow+0x47/0xe0 net/ipv6/ip6_output.c:1259
 tcp_v6_connect+0x50c/0x8e0 net/ipv6/tcp_ipv6.c:277
 mptcp_connect+0x389/0x680 net/mptcp/protocol.c:3683
 __inet_stream_connect+0x11f/0x5c0 net/ipv4/af_inet.c:677
 inet_stream_connect+0x36/0x50 net/ipv4/af_inet.c:748
 __sys_connect_file net/socket.c:2038 [inline]
 __sys_connect+0x17b/0x220 net/socket.c:2057
 __do_sys_connect net/socket.c:2063 [inline]
 __se_sys_connect net/socket.c:2060 [inline]
 __x64_sys_connect+0x1c/0x20 net/socket.c:2060
 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
 do_syscall_64+0xc8/0x1a0 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x4b/0x53
RIP: 0033:0x7f19101bd169
Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f190e827038 EFLAGS: 00000246 ORIG_RAX: 000000000000002a
RAX: ffffffffffffffda RBX: 00007f19103d5fa0 RCX: 00007f19101bd169
RDX: 000000000000001c RSI: 0000200000000000 RDI: 0000000000000004
RBP: 00007f191023e730 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 0000000000000000 R14: 00007f19103d5fa0 R15: 00007f19104ffa28
 </TASK>
Modules linked in:
CR2: 00007f93e4d51a23
---[ end trace 0000000000000000 ]---
RIP: 0010:arch_atomic_read arch/x86/include/asm/atomic.h:23 [inline]
RIP: 0010:raw_atomic_read include/linux/atomic/atomic-arch-fallback.h:457 [inline]
RIP: 0010:atomic_read include/linux/atomic/atomic-instrumented.h:33 [inline]
RIP: 0010:rt_genid_ipv6 include/net/net_namespace.h:537 [inline]
RIP: 0010:rt6_is_valid net/ipv6/route.c:1435 [inline]
RIP: 0010:rt6_get_pcpu_route net/ipv6/route.c:1445 [inline]
RIP: 0010:ip6_pol_route+0x301/0x9d0 net/ipv6/route.c:2298
Code: a9 fe 4d 85 ff 4c 8b 74 24 10 74 3b 41 8b 9f 98 00 00 00 31 ff 89 de e8 3d 37 a9 fe 85 db 74 33 49 8b 07 48 8b 80 08 01 00 00 <8b> a8 24 0a 00 00 89 df 89 ee e8 f0 34 a9 fe 39 eb 75 20 e8 67 33
RSP: 0018:ffa0000000c3ba20 EFLAGS: 00010202
RAX: 00007f93e4d50fff RBX: 000000001ac62540 RCX: 0000000000000002
RDX: ff11000005e99700 RSI: 000000001ac62540 RDI: 0000000000000000
RBP: 0000000000000003 R08: 0000000000000000 R09: 0000000000000000
R10: 00000000ffff0000 R11: 0000000000000002 R12: ffffffff829af127
R13: 0000000000000000 R14: ff110001053e2580 R15: ff1100012505f400
FS:  00007f190e8276c0(0000) GS:ff110001b79d2000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f93e4d51a23 CR3: 000000011d7b5004 CR4: 0000000000771ef0
DR0: 0000000080000001 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe07f0 DR7: 0000000000000600
PKRU: 80000000
----------------
Code disassembly (best guess):
   0:	a9 fe 4d 85 ff       	test   $0xff854dfe,%eax
   5:	4c 8b 74 24 10       	mov    0x10(%rsp),%r14
   a:	74 3b                	je     0x47
   c:	41 8b 9f 98 00 00 00 	mov    0x98(%r15),%ebx
  13:	31 ff                	xor    %edi,%edi
  15:	89 de                	mov    %ebx,%esi
  17:	e8 3d 37 a9 fe       	call   0xfea93759
  1c:	85 db                	test   %ebx,%ebx
  1e:	74 33                	je     0x53
  20:	49 8b 07             	mov    (%r15),%rax
  23:	48 8b 80 08 01 00 00 	mov    0x108(%rax),%rax
* 2a:	8b a8 24 0a 00 00    	mov    0xa24(%rax),%ebp <-- trapping instruction
  30:	89 df                	mov    %ebx,%edi
  32:	89 ee                	mov    %ebp,%esi
  34:	e8 f0 34 a9 fe       	call   0xfea93529
  39:	39 eb                	cmp    %ebp,%ebx
  3b:	75 20                	jne    0x5d
  3d:	e8                   	.byte 0xe8
  3e:	67                   	addr32
  3f:	33                   	.byte 0x33

---8<---

Petr Machata

author

2025-06-12 16:25 UTC #35 lore raw

In reply to Jakub Kicinski

Quoted 11 lines · Jakub Kicinski

Quoted 6 lines · On Wed, 11 Jun 2025 17:30:15 +0200 Petr Machata wrote:
Could it actually have been caused by another test? The howto page mentions that the CI is running the tests one at a time, so I don't suppose that's a possibility.
I'll try to run a more fuller suite tomorrow and star at the code a bit to see if I might be missing an error branch or something.
We also hit a crash in ipv6 fcnal.sh, too. Looks like this is either a kmemleak false positive or possibly related to the rtnl changes in ipv6. Either way I it's not related to you changes, sorry about that! :(

No problem.

Quoted 35 lines

[ 2900.792890] BUG: kernel NULL pointer dereference, address: 0000000000000108
[ 2900.792961] #PF: supervisor read access in kernel mode
[ 2900.793017] #PF: error_code(0x0000) - not-present page
[ 2900.793053] PGD 8fd6067 P4D 8fd6067 PUD 6402067 PMD 0 
[ 2900.793097] Oops: Oops: 0000 [#1] SMP NOPTI
[ 2900.793127] CPU: 0 UID: 0 PID: 15652 Comm: nettest Not tainted 6.15.0-virtme #1 PREEMPT(voluntary) 
[ 2900.793200] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 2900.793245] RIP: 0010:ip6_pol_route+0x286/0x4a0
[ 2900.793290] Code: 0c 24 0f 85 fb 01 00 00 09 ca 0f 88 2f 01 00 00 e8 cf 11 43 ff 83 cb 08 48 8d 7c 24 18 e8 32 7b ff ff 0f b7 cb ba ff ff ff ff <4c> 8b 80 08 01 00 00 48 89 c6 49 89 c7 49 8d b8 80 06 00 00 4c 89
[ 2900.793422] RSP: 0018:ffffc08a0932f480 EFLAGS: 00010246
[ 2900.793460] RAX: 0000000000000000 RBX: 0000000000000008 RCX: 0000000000000008
[ 2900.793521] RDX: 00000000ffffffff RSI: ffffc08a0932f740 RDI: ffff9adac8c8f1a8
[ 2900.793580] RBP: ffff9adac87458c0 R08: 0000000000000000 R09: 0000000000000000
[ 2900.793635] R10: 0000000000000000 R11: 0000000000000040 R12: ffff9adac82e362c
[ 2900.793692] R13: ffff9adac82e3600 R14: 0000000000000080 R15: 0000000000000000
[ 2900.793752] FS:  00007f3418913740(0000) GS:ffff9adb7373a000(0000) knlGS:0000000000000000
[ 2900.793816] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2900.793864] CR2: 0000000000000108 CR3: 0000000008007004 CR4: 0000000000772ef0
[ 2900.793920] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 2900.793977] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 2900.794031] PKRU: 55555554
[ 2900.794050] Call Trace:
[ 2900.794070]  <TASK>
[ 2900.794090]  ? __pfx_ip6_pol_route_output+0x10/0x10
[ 2900.794131]  fib6_rule_action+0xe3/0x310
[ 2900.794166]  fib_rules_lookup+0x1b2/0x2b0
[ 2900.794200]  ? __pfx_ip6_pol_route_output+0x10/0x10
[ 2900.794241]  fib6_rule_lookup+0xa9/0x270
[ 2900.794271]  ? __pfx_ip6_pol_route_output+0x10/0x10
[ 2900.794310]  ip6_route_output_flags+0xab/0x180
[ 2900.794353]  ip6_dst_lookup_tail.constprop.0+0x282/0x340
[ 2900.794394]  ip6_dst_lookup_flow+0x46/0xc0
[ 2900.794422]  vrf_xmit+0x100/0x4a0
[ 2900.794459]  dev_hard_start_xmit+0x8d/0x1c0

Quoted 1 line

https://netdev-3.bots.linux.dev/vmksft-net/results/160541/vm-crash-thr0-0

Didn't see this issue either FWIW.