1 // SPDX-License-Identifier: GPL-2.0-or-later
3 * Linux NET3: IP/IP protocol decoder modified to support
4 * virtual tunnel interface
7 * Saurabh Mohan (saurabh.mohan@vyatta.com) 05/07/2012
11 This version of net/ipv4/ip_vti.c is cloned of net/ipv4/ipip.c
13 For comments look at net/ipv4/ip_gre.c --ANK
17 #include <linux/capability.h>
18 #include <linux/module.h>
19 #include <linux/types.h>
20 #include <linux/kernel.h>
21 #include <linux/uaccess.h>
22 #include <linux/skbuff.h>
23 #include <linux/netdevice.h>
25 #include <linux/tcp.h>
26 #include <linux/udp.h>
27 #include <linux/if_arp.h>
28 #include <linux/init.h>
29 #include <linux/netfilter_ipv4.h>
30 #include <linux/if_ether.h>
31 #include <linux/icmpv6.h>
36 #include <net/ip_tunnels.h>
37 #include <net/inet_ecn.h>
39 #include <net/net_namespace.h>
40 #include <net/netns/generic.h>
42 static struct rtnl_link_ops vti_link_ops __read_mostly
;
44 static unsigned int vti_net_id __read_mostly
;
45 static int vti_tunnel_init(struct net_device
*dev
);
47 static int vti_input(struct sk_buff
*skb
, int nexthdr
, __be32 spi
,
48 int encap_type
, bool update_skb_dev
)
50 struct ip_tunnel
*tunnel
;
51 const struct iphdr
*iph
= ip_hdr(skb
);
52 struct net
*net
= dev_net(skb
->dev
);
53 struct ip_tunnel_net
*itn
= net_generic(net
, vti_net_id
);
55 tunnel
= ip_tunnel_lookup(itn
, skb
->dev
->ifindex
, TUNNEL_NO_KEY
,
56 iph
->saddr
, iph
->daddr
, 0);
58 if (!xfrm4_policy_check(NULL
, XFRM_POLICY_IN
, skb
))
61 XFRM_TUNNEL_SKB_CB(skb
)->tunnel
.ip4
= tunnel
;
64 skb
->dev
= tunnel
->dev
;
66 return xfrm_input(skb
, nexthdr
, spi
, encap_type
);
75 static int vti_input_proto(struct sk_buff
*skb
, int nexthdr
, __be32 spi
,
78 return vti_input(skb
, nexthdr
, spi
, encap_type
, false);
81 static int vti_rcv(struct sk_buff
*skb
, __be32 spi
, bool update_skb_dev
)
83 XFRM_SPI_SKB_CB(skb
)->family
= AF_INET
;
84 XFRM_SPI_SKB_CB(skb
)->daddroff
= offsetof(struct iphdr
, daddr
);
86 return vti_input(skb
, ip_hdr(skb
)->protocol
, spi
, 0, update_skb_dev
);
89 static int vti_rcv_proto(struct sk_buff
*skb
)
91 return vti_rcv(skb
, 0, false);
94 static int vti_rcv_tunnel(struct sk_buff
*skb
)
96 return vti_rcv(skb
, ip_hdr(skb
)->saddr
, true);
99 static int vti_rcv_cb(struct sk_buff
*skb
, int err
)
101 unsigned short family
;
102 struct net_device
*dev
;
103 struct pcpu_sw_netstats
*tstats
;
104 struct xfrm_state
*x
;
105 const struct xfrm_mode
*inner_mode
;
106 struct ip_tunnel
*tunnel
= XFRM_TUNNEL_SKB_CB(skb
)->tunnel
.ip4
;
107 u32 orig_mark
= skb
->mark
;
116 dev
->stats
.rx_errors
++;
117 dev
->stats
.rx_dropped
++;
122 x
= xfrm_input_state(skb
);
124 inner_mode
= &x
->inner_mode
;
126 if (x
->sel
.family
== AF_UNSPEC
) {
127 inner_mode
= xfrm_ip2inner_mode(x
, XFRM_MODE_SKB_CB(skb
)->protocol
);
128 if (inner_mode
== NULL
) {
129 XFRM_INC_STATS(dev_net(skb
->dev
),
130 LINUX_MIB_XFRMINSTATEMODEERROR
);
135 family
= inner_mode
->family
;
137 skb
->mark
= be32_to_cpu(tunnel
->parms
.i_key
);
138 ret
= xfrm_policy_check(NULL
, XFRM_POLICY_IN
, skb
, family
);
139 skb
->mark
= orig_mark
;
144 skb_scrub_packet(skb
, !net_eq(tunnel
->net
, dev_net(skb
->dev
)));
147 tstats
= this_cpu_ptr(dev
->tstats
);
149 u64_stats_update_begin(&tstats
->syncp
);
150 tstats
->rx_packets
++;
151 tstats
->rx_bytes
+= skb
->len
;
152 u64_stats_update_end(&tstats
->syncp
);
157 static bool vti_state_check(const struct xfrm_state
*x
, __be32 dst
, __be32 src
)
159 xfrm_address_t
*daddr
= (xfrm_address_t
*)&dst
;
160 xfrm_address_t
*saddr
= (xfrm_address_t
*)&src
;
162 /* if there is no transform then this tunnel is not functional.
163 * Or if the xfrm is not mode tunnel.
165 if (!x
|| x
->props
.mode
!= XFRM_MODE_TUNNEL
||
166 x
->props
.family
!= AF_INET
)
170 return xfrm_addr_equal(saddr
, &x
->props
.saddr
, AF_INET
);
172 if (!xfrm_state_addr_check(x
, daddr
, saddr
, AF_INET
))
178 static netdev_tx_t
vti_xmit(struct sk_buff
*skb
, struct net_device
*dev
,
181 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
182 struct ip_tunnel_parm
*parms
= &tunnel
->parms
;
183 struct dst_entry
*dst
= skb_dst(skb
);
184 struct net_device
*tdev
; /* Device to other host */
185 int pkt_len
= skb
->len
;
190 dev
->stats
.tx_carrier_errors
++;
195 dst
= xfrm_lookup(tunnel
->net
, dst
, fl
, NULL
, 0);
197 dev
->stats
.tx_carrier_errors
++;
201 if (!vti_state_check(dst
->xfrm
, parms
->iph
.daddr
, parms
->iph
.saddr
)) {
202 dev
->stats
.tx_carrier_errors
++;
211 dev
->stats
.collisions
++;
216 if (skb
->len
> mtu
) {
217 skb_dst_update_pmtu(skb
, mtu
);
218 if (skb
->protocol
== htons(ETH_P_IP
)) {
219 icmp_send(skb
, ICMP_DEST_UNREACH
, ICMP_FRAG_NEEDED
,
222 if (mtu
< IPV6_MIN_MTU
)
225 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
232 skb_scrub_packet(skb
, !net_eq(tunnel
->net
, dev_net(dev
)));
233 skb_dst_set(skb
, dst
);
234 skb
->dev
= skb_dst(skb
)->dev
;
236 err
= dst_output(tunnel
->net
, skb
->sk
, skb
);
237 if (net_xmit_eval(err
) == 0)
239 iptunnel_xmit_stats(dev
, err
);
243 dst_link_failure(skb
);
245 dev
->stats
.tx_errors
++;
250 /* This function assumes it is being called from dev_queue_xmit()
251 * and that skb is filled properly by that function.
253 static netdev_tx_t
vti_tunnel_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
255 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
258 if (!pskb_inet_may_pull(skb
))
261 memset(&fl
, 0, sizeof(fl
));
263 switch (skb
->protocol
) {
264 case htons(ETH_P_IP
):
265 xfrm_decode_session(skb
, &fl
, AF_INET
);
266 memset(IPCB(skb
), 0, sizeof(*IPCB(skb
)));
268 case htons(ETH_P_IPV6
):
269 xfrm_decode_session(skb
, &fl
, AF_INET6
);
270 memset(IP6CB(skb
), 0, sizeof(*IP6CB(skb
)));
276 /* override mark with tunnel output key */
277 fl
.flowi_mark
= be32_to_cpu(tunnel
->parms
.o_key
);
279 return vti_xmit(skb
, dev
, &fl
);
282 dev
->stats
.tx_errors
++;
287 static int vti4_err(struct sk_buff
*skb
, u32 info
)
291 struct xfrm_state
*x
;
292 struct ip_tunnel
*tunnel
;
293 struct ip_esp_hdr
*esph
;
294 struct ip_auth_hdr
*ah
;
295 struct ip_comp_hdr
*ipch
;
296 struct net
*net
= dev_net(skb
->dev
);
297 const struct iphdr
*iph
= (const struct iphdr
*)skb
->data
;
298 int protocol
= iph
->protocol
;
299 struct ip_tunnel_net
*itn
= net_generic(net
, vti_net_id
);
301 tunnel
= ip_tunnel_lookup(itn
, skb
->dev
->ifindex
, TUNNEL_NO_KEY
,
302 iph
->daddr
, iph
->saddr
, 0);
306 mark
= be32_to_cpu(tunnel
->parms
.o_key
);
310 esph
= (struct ip_esp_hdr
*)(skb
->data
+(iph
->ihl
<<2));
314 ah
= (struct ip_auth_hdr
*)(skb
->data
+(iph
->ihl
<<2));
318 ipch
= (struct ip_comp_hdr
*)(skb
->data
+(iph
->ihl
<<2));
319 spi
= htonl(ntohs(ipch
->cpi
));
325 switch (icmp_hdr(skb
)->type
) {
326 case ICMP_DEST_UNREACH
:
327 if (icmp_hdr(skb
)->code
!= ICMP_FRAG_NEEDED
)
335 x
= xfrm_state_lookup(net
, mark
, (const xfrm_address_t
*)&iph
->daddr
,
336 spi
, protocol
, AF_INET
);
340 if (icmp_hdr(skb
)->type
== ICMP_DEST_UNREACH
)
341 ipv4_update_pmtu(skb
, net
, info
, 0, protocol
);
343 ipv4_redirect(skb
, net
, 0, protocol
);
350 vti_tunnel_ioctl(struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
353 struct ip_tunnel_parm p
;
355 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
358 if (cmd
== SIOCADDTUNNEL
|| cmd
== SIOCCHGTUNNEL
) {
359 if (p
.iph
.version
!= 4 || p
.iph
.protocol
!= IPPROTO_IPIP
||
364 if (!(p
.i_flags
& GRE_KEY
))
366 if (!(p
.o_flags
& GRE_KEY
))
369 p
.i_flags
= VTI_ISVTI
;
371 err
= ip_tunnel_ioctl(dev
, &p
, cmd
);
375 if (cmd
!= SIOCDELTUNNEL
) {
376 p
.i_flags
|= GRE_KEY
;
377 p
.o_flags
|= GRE_KEY
;
380 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof(p
)))
385 static const struct net_device_ops vti_netdev_ops
= {
386 .ndo_init
= vti_tunnel_init
,
387 .ndo_uninit
= ip_tunnel_uninit
,
388 .ndo_start_xmit
= vti_tunnel_xmit
,
389 .ndo_do_ioctl
= vti_tunnel_ioctl
,
390 .ndo_change_mtu
= ip_tunnel_change_mtu
,
391 .ndo_get_stats64
= ip_tunnel_get_stats64
,
392 .ndo_get_iflink
= ip_tunnel_get_iflink
,
395 static void vti_tunnel_setup(struct net_device
*dev
)
397 dev
->netdev_ops
= &vti_netdev_ops
;
398 dev
->type
= ARPHRD_TUNNEL
;
399 ip_tunnel_setup(dev
, vti_net_id
);
402 static int vti_tunnel_init(struct net_device
*dev
)
404 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
405 struct iphdr
*iph
= &tunnel
->parms
.iph
;
407 memcpy(dev
->dev_addr
, &iph
->saddr
, 4);
408 memcpy(dev
->broadcast
, &iph
->daddr
, 4);
410 dev
->flags
= IFF_NOARP
;
412 dev
->features
|= NETIF_F_LLTX
;
415 return ip_tunnel_init(dev
);
418 static void __net_init
vti_fb_tunnel_init(struct net_device
*dev
)
420 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
421 struct iphdr
*iph
= &tunnel
->parms
.iph
;
424 iph
->protocol
= IPPROTO_IPIP
;
428 static struct xfrm4_protocol vti_esp4_protocol __read_mostly
= {
429 .handler
= vti_rcv_proto
,
430 .input_handler
= vti_input_proto
,
431 .cb_handler
= vti_rcv_cb
,
432 .err_handler
= vti4_err
,
436 static struct xfrm4_protocol vti_ah4_protocol __read_mostly
= {
437 .handler
= vti_rcv_proto
,
438 .input_handler
= vti_input_proto
,
439 .cb_handler
= vti_rcv_cb
,
440 .err_handler
= vti4_err
,
444 static struct xfrm4_protocol vti_ipcomp4_protocol __read_mostly
= {
445 .handler
= vti_rcv_proto
,
446 .input_handler
= vti_input_proto
,
447 .cb_handler
= vti_rcv_cb
,
448 .err_handler
= vti4_err
,
452 static struct xfrm_tunnel ipip_handler __read_mostly
= {
453 .handler
= vti_rcv_tunnel
,
454 .err_handler
= vti4_err
,
458 static int __net_init
vti_init_net(struct net
*net
)
461 struct ip_tunnel_net
*itn
;
463 err
= ip_tunnel_init_net(net
, vti_net_id
, &vti_link_ops
, "ip_vti0");
466 itn
= net_generic(net
, vti_net_id
);
467 if (itn
->fb_tunnel_dev
)
468 vti_fb_tunnel_init(itn
->fb_tunnel_dev
);
472 static void __net_exit
vti_exit_batch_net(struct list_head
*list_net
)
474 ip_tunnel_delete_nets(list_net
, vti_net_id
, &vti_link_ops
);
477 static struct pernet_operations vti_net_ops
= {
478 .init
= vti_init_net
,
479 .exit_batch
= vti_exit_batch_net
,
481 .size
= sizeof(struct ip_tunnel_net
),
484 static int vti_tunnel_validate(struct nlattr
*tb
[], struct nlattr
*data
[],
485 struct netlink_ext_ack
*extack
)
490 static void vti_netlink_parms(struct nlattr
*data
[],
491 struct ip_tunnel_parm
*parms
,
494 memset(parms
, 0, sizeof(*parms
));
496 parms
->iph
.protocol
= IPPROTO_IPIP
;
501 parms
->i_flags
= VTI_ISVTI
;
503 if (data
[IFLA_VTI_LINK
])
504 parms
->link
= nla_get_u32(data
[IFLA_VTI_LINK
]);
506 if (data
[IFLA_VTI_IKEY
])
507 parms
->i_key
= nla_get_be32(data
[IFLA_VTI_IKEY
]);
509 if (data
[IFLA_VTI_OKEY
])
510 parms
->o_key
= nla_get_be32(data
[IFLA_VTI_OKEY
]);
512 if (data
[IFLA_VTI_LOCAL
])
513 parms
->iph
.saddr
= nla_get_in_addr(data
[IFLA_VTI_LOCAL
]);
515 if (data
[IFLA_VTI_REMOTE
])
516 parms
->iph
.daddr
= nla_get_in_addr(data
[IFLA_VTI_REMOTE
]);
518 if (data
[IFLA_VTI_FWMARK
])
519 *fwmark
= nla_get_u32(data
[IFLA_VTI_FWMARK
]);
522 static int vti_newlink(struct net
*src_net
, struct net_device
*dev
,
523 struct nlattr
*tb
[], struct nlattr
*data
[],
524 struct netlink_ext_ack
*extack
)
526 struct ip_tunnel_parm parms
;
529 vti_netlink_parms(data
, &parms
, &fwmark
);
530 return ip_tunnel_newlink(dev
, tb
, &parms
, fwmark
);
533 static int vti_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
534 struct nlattr
*data
[],
535 struct netlink_ext_ack
*extack
)
537 struct ip_tunnel
*t
= netdev_priv(dev
);
538 __u32 fwmark
= t
->fwmark
;
539 struct ip_tunnel_parm p
;
541 vti_netlink_parms(data
, &p
, &fwmark
);
542 return ip_tunnel_changelink(dev
, tb
, &p
, fwmark
);
545 static size_t vti_get_size(const struct net_device
*dev
)
556 /* IFLA_VTI_REMOTE */
558 /* IFLA_VTI_FWMARK */
563 static int vti_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
565 struct ip_tunnel
*t
= netdev_priv(dev
);
566 struct ip_tunnel_parm
*p
= &t
->parms
;
568 if (nla_put_u32(skb
, IFLA_VTI_LINK
, p
->link
) ||
569 nla_put_be32(skb
, IFLA_VTI_IKEY
, p
->i_key
) ||
570 nla_put_be32(skb
, IFLA_VTI_OKEY
, p
->o_key
) ||
571 nla_put_in_addr(skb
, IFLA_VTI_LOCAL
, p
->iph
.saddr
) ||
572 nla_put_in_addr(skb
, IFLA_VTI_REMOTE
, p
->iph
.daddr
) ||
573 nla_put_u32(skb
, IFLA_VTI_FWMARK
, t
->fwmark
))
579 static const struct nla_policy vti_policy
[IFLA_VTI_MAX
+ 1] = {
580 [IFLA_VTI_LINK
] = { .type
= NLA_U32
},
581 [IFLA_VTI_IKEY
] = { .type
= NLA_U32
},
582 [IFLA_VTI_OKEY
] = { .type
= NLA_U32
},
583 [IFLA_VTI_LOCAL
] = { .len
= sizeof_field(struct iphdr
, saddr
) },
584 [IFLA_VTI_REMOTE
] = { .len
= sizeof_field(struct iphdr
, daddr
) },
585 [IFLA_VTI_FWMARK
] = { .type
= NLA_U32
},
588 static struct rtnl_link_ops vti_link_ops __read_mostly
= {
590 .maxtype
= IFLA_VTI_MAX
,
591 .policy
= vti_policy
,
592 .priv_size
= sizeof(struct ip_tunnel
),
593 .setup
= vti_tunnel_setup
,
594 .validate
= vti_tunnel_validate
,
595 .newlink
= vti_newlink
,
596 .changelink
= vti_changelink
,
597 .dellink
= ip_tunnel_dellink
,
598 .get_size
= vti_get_size
,
599 .fill_info
= vti_fill_info
,
600 .get_link_net
= ip_tunnel_get_link_net
,
603 static int __init
vti_init(void)
608 pr_info("IPv4 over IPsec tunneling driver\n");
610 msg
= "tunnel device";
611 err
= register_pernet_device(&vti_net_ops
);
613 goto pernet_dev_failed
;
615 msg
= "tunnel protocols";
616 err
= xfrm4_protocol_register(&vti_esp4_protocol
, IPPROTO_ESP
);
618 goto xfrm_proto_esp_failed
;
619 err
= xfrm4_protocol_register(&vti_ah4_protocol
, IPPROTO_AH
);
621 goto xfrm_proto_ah_failed
;
622 err
= xfrm4_protocol_register(&vti_ipcomp4_protocol
, IPPROTO_COMP
);
624 goto xfrm_proto_comp_failed
;
627 err
= xfrm4_tunnel_register(&ipip_handler
, AF_INET
);
629 goto xfrm_tunnel_failed
;
631 msg
= "netlink interface";
632 err
= rtnl_link_register(&vti_link_ops
);
634 goto rtnl_link_failed
;
639 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
641 xfrm4_protocol_deregister(&vti_ipcomp4_protocol
, IPPROTO_COMP
);
642 xfrm_proto_comp_failed
:
643 xfrm4_protocol_deregister(&vti_ah4_protocol
, IPPROTO_AH
);
644 xfrm_proto_ah_failed
:
645 xfrm4_protocol_deregister(&vti_esp4_protocol
, IPPROTO_ESP
);
646 xfrm_proto_esp_failed
:
647 unregister_pernet_device(&vti_net_ops
);
649 pr_err("vti init: failed to register %s\n", msg
);
653 static void __exit
vti_fini(void)
655 rtnl_link_unregister(&vti_link_ops
);
656 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
657 xfrm4_protocol_deregister(&vti_ipcomp4_protocol
, IPPROTO_COMP
);
658 xfrm4_protocol_deregister(&vti_ah4_protocol
, IPPROTO_AH
);
659 xfrm4_protocol_deregister(&vti_esp4_protocol
, IPPROTO_ESP
);
660 unregister_pernet_device(&vti_net_ops
);
663 module_init(vti_init
);
664 module_exit(vti_fini
);
665 MODULE_LICENSE("GPL");
666 MODULE_ALIAS_RTNL_LINK("vti");
667 MODULE_ALIAS_NETDEV("ip_vti0");