1 // SPDX-License-Identifier: GPL-2.0-or-later
3 * IPv6 virtual tunneling interface
5 * Copyright (C) 2013 secunet Security Networks AG
8 * Steffen Klassert <steffen.klassert@secunet.com>
11 * net/ipv6/ip6_tunnel.c
14 #include <linux/module.h>
15 #include <linux/capability.h>
16 #include <linux/errno.h>
17 #include <linux/types.h>
18 #include <linux/sockios.h>
19 #include <linux/icmp.h>
23 #include <linux/net.h>
24 #include <linux/in6.h>
25 #include <linux/netdevice.h>
26 #include <linux/if_arp.h>
27 #include <linux/icmpv6.h>
28 #include <linux/init.h>
29 #include <linux/route.h>
30 #include <linux/rtnetlink.h>
31 #include <linux/netfilter_ipv6.h>
32 #include <linux/slab.h>
33 #include <linux/hash.h>
35 #include <linux/uaccess.h>
36 #include <linux/atomic.h>
40 #include <net/ip_tunnels.h>
42 #include <net/ip6_route.h>
43 #include <net/addrconf.h>
44 #include <net/ip6_tunnel.h>
46 #include <net/net_namespace.h>
47 #include <net/netns/generic.h>
48 #include <linux/etherdevice.h>
50 #define IP6_VTI_HASH_SIZE_SHIFT 5
51 #define IP6_VTI_HASH_SIZE (1 << IP6_VTI_HASH_SIZE_SHIFT)
53 static u32
HASH(const struct in6_addr
*addr1
, const struct in6_addr
*addr2
)
55 u32 hash
= ipv6_addr_hash(addr1
) ^ ipv6_addr_hash(addr2
);
57 return hash_32(hash
, IP6_VTI_HASH_SIZE_SHIFT
);
60 static int vti6_dev_init(struct net_device
*dev
);
61 static void vti6_dev_setup(struct net_device
*dev
);
62 static struct rtnl_link_ops vti6_link_ops __read_mostly
;
64 static unsigned int vti6_net_id __read_mostly
;
66 /* the vti6 tunnel fallback device */
67 struct net_device
*fb_tnl_dev
;
68 /* lists for storing tunnels in use */
69 struct ip6_tnl __rcu
*tnls_r_l
[IP6_VTI_HASH_SIZE
];
70 struct ip6_tnl __rcu
*tnls_wc
[1];
71 struct ip6_tnl __rcu
**tnls
[2];
74 #define for_each_vti6_tunnel_rcu(start) \
75 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
78 * vti6_tnl_lookup - fetch tunnel matching the end-point addresses
79 * @net: network namespace
80 * @remote: the address of the tunnel exit-point
81 * @local: the address of the tunnel entry-point
84 * tunnel matching given end-points if found,
85 * else fallback tunnel if its device is up,
88 static struct ip6_tnl
*
89 vti6_tnl_lookup(struct net
*net
, const struct in6_addr
*remote
,
90 const struct in6_addr
*local
)
92 unsigned int hash
= HASH(remote
, local
);
94 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
97 for_each_vti6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
98 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
99 ipv6_addr_equal(remote
, &t
->parms
.raddr
) &&
100 (t
->dev
->flags
& IFF_UP
))
104 memset(&any
, 0, sizeof(any
));
105 hash
= HASH(&any
, local
);
106 for_each_vti6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
107 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
108 (t
->dev
->flags
& IFF_UP
))
112 hash
= HASH(remote
, &any
);
113 for_each_vti6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
114 if (ipv6_addr_equal(remote
, &t
->parms
.raddr
) &&
115 (t
->dev
->flags
& IFF_UP
))
119 t
= rcu_dereference(ip6n
->tnls_wc
[0]);
120 if (t
&& (t
->dev
->flags
& IFF_UP
))
127 * vti6_tnl_bucket - get head of list matching given tunnel parameters
128 * @p: parameters containing tunnel end-points
131 * vti6_tnl_bucket() returns the head of the list matching the
132 * &struct in6_addr entries laddr and raddr in @p.
134 * Return: head of IPv6 tunnel list
136 static struct ip6_tnl __rcu
**
137 vti6_tnl_bucket(struct vti6_net
*ip6n
, const struct __ip6_tnl_parm
*p
)
139 const struct in6_addr
*remote
= &p
->raddr
;
140 const struct in6_addr
*local
= &p
->laddr
;
144 if (!ipv6_addr_any(remote
) || !ipv6_addr_any(local
)) {
146 h
= HASH(remote
, local
);
148 return &ip6n
->tnls
[prio
][h
];
152 vti6_tnl_link(struct vti6_net
*ip6n
, struct ip6_tnl
*t
)
154 struct ip6_tnl __rcu
**tp
= vti6_tnl_bucket(ip6n
, &t
->parms
);
156 rcu_assign_pointer(t
->next
, rtnl_dereference(*tp
));
157 rcu_assign_pointer(*tp
, t
);
161 vti6_tnl_unlink(struct vti6_net
*ip6n
, struct ip6_tnl
*t
)
163 struct ip6_tnl __rcu
**tp
;
164 struct ip6_tnl
*iter
;
166 for (tp
= vti6_tnl_bucket(ip6n
, &t
->parms
);
167 (iter
= rtnl_dereference(*tp
)) != NULL
;
170 rcu_assign_pointer(*tp
, t
->next
);
176 static void vti6_dev_free(struct net_device
*dev
)
178 free_percpu(dev
->tstats
);
181 static int vti6_tnl_create2(struct net_device
*dev
)
183 struct ip6_tnl
*t
= netdev_priv(dev
);
184 struct net
*net
= dev_net(dev
);
185 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
188 dev
->rtnl_link_ops
= &vti6_link_ops
;
189 err
= register_netdevice(dev
);
193 strcpy(t
->parms
.name
, dev
->name
);
196 vti6_tnl_link(ip6n
, t
);
204 static struct ip6_tnl
*vti6_tnl_create(struct net
*net
, struct __ip6_tnl_parm
*p
)
206 struct net_device
*dev
;
212 if (!dev_valid_name(p
->name
))
214 strlcpy(name
, p
->name
, IFNAMSIZ
);
216 sprintf(name
, "ip6_vti%%d");
219 dev
= alloc_netdev(sizeof(*t
), name
, NET_NAME_UNKNOWN
, vti6_dev_setup
);
223 dev_net_set(dev
, net
);
225 t
= netdev_priv(dev
);
227 t
->net
= dev_net(dev
);
229 err
= vti6_tnl_create2(dev
);
242 * vti6_locate - find or create tunnel matching given parameters
243 * @net: network namespace
244 * @p: tunnel parameters
245 * @create: != 0 if allowed to create new tunnel if no match found
248 * vti6_locate() first tries to locate an existing tunnel
249 * based on @parms. If this is unsuccessful, but @create is set a new
250 * tunnel device is created and registered for use.
253 * matching tunnel or NULL
255 static struct ip6_tnl
*vti6_locate(struct net
*net
, struct __ip6_tnl_parm
*p
,
258 const struct in6_addr
*remote
= &p
->raddr
;
259 const struct in6_addr
*local
= &p
->laddr
;
260 struct ip6_tnl __rcu
**tp
;
262 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
264 for (tp
= vti6_tnl_bucket(ip6n
, p
);
265 (t
= rtnl_dereference(*tp
)) != NULL
;
267 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
268 ipv6_addr_equal(remote
, &t
->parms
.raddr
)) {
277 return vti6_tnl_create(net
, p
);
281 * vti6_dev_uninit - tunnel device uninitializer
282 * @dev: the device to be destroyed
285 * vti6_dev_uninit() removes tunnel from its list
287 static void vti6_dev_uninit(struct net_device
*dev
)
289 struct ip6_tnl
*t
= netdev_priv(dev
);
290 struct vti6_net
*ip6n
= net_generic(t
->net
, vti6_net_id
);
292 if (dev
== ip6n
->fb_tnl_dev
)
293 RCU_INIT_POINTER(ip6n
->tnls_wc
[0], NULL
);
295 vti6_tnl_unlink(ip6n
, t
);
299 static int vti6_rcv(struct sk_buff
*skb
)
302 const struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
305 t
= vti6_tnl_lookup(dev_net(skb
->dev
), &ipv6h
->saddr
, &ipv6h
->daddr
);
307 if (t
->parms
.proto
!= IPPROTO_IPV6
&& t
->parms
.proto
!= 0) {
312 if (!xfrm6_policy_check(NULL
, XFRM_POLICY_IN
, skb
)) {
317 ipv6h
= ipv6_hdr(skb
);
318 if (!ip6_tnl_rcv_ctl(t
, &ipv6h
->daddr
, &ipv6h
->saddr
)) {
319 t
->dev
->stats
.rx_dropped
++;
326 return xfrm6_rcv_tnl(skb
, t
);
335 static int vti6_rcv_cb(struct sk_buff
*skb
, int err
)
337 unsigned short family
;
338 struct net_device
*dev
;
339 struct pcpu_sw_netstats
*tstats
;
340 struct xfrm_state
*x
;
341 const struct xfrm_mode
*inner_mode
;
342 struct ip6_tnl
*t
= XFRM_TUNNEL_SKB_CB(skb
)->tunnel
.ip6
;
343 u32 orig_mark
= skb
->mark
;
352 dev
->stats
.rx_errors
++;
353 dev
->stats
.rx_dropped
++;
358 x
= xfrm_input_state(skb
);
360 inner_mode
= &x
->inner_mode
;
362 if (x
->sel
.family
== AF_UNSPEC
) {
363 inner_mode
= xfrm_ip2inner_mode(x
, XFRM_MODE_SKB_CB(skb
)->protocol
);
364 if (inner_mode
== NULL
) {
365 XFRM_INC_STATS(dev_net(skb
->dev
),
366 LINUX_MIB_XFRMINSTATEMODEERROR
);
371 family
= inner_mode
->family
;
373 skb
->mark
= be32_to_cpu(t
->parms
.i_key
);
374 ret
= xfrm_policy_check(NULL
, XFRM_POLICY_IN
, skb
, family
);
375 skb
->mark
= orig_mark
;
380 skb_scrub_packet(skb
, !net_eq(t
->net
, dev_net(skb
->dev
)));
383 tstats
= this_cpu_ptr(dev
->tstats
);
384 u64_stats_update_begin(&tstats
->syncp
);
385 tstats
->rx_packets
++;
386 tstats
->rx_bytes
+= skb
->len
;
387 u64_stats_update_end(&tstats
->syncp
);
393 * vti6_addr_conflict - compare packet addresses to tunnel's own
394 * @t: the outgoing tunnel device
395 * @hdr: IPv6 header from the incoming packet
398 * Avoid trivial tunneling loop by checking that tunnel exit-point
399 * doesn't match source of incoming packet.
406 vti6_addr_conflict(const struct ip6_tnl
*t
, const struct ipv6hdr
*hdr
)
408 return ipv6_addr_equal(&t
->parms
.raddr
, &hdr
->saddr
);
411 static bool vti6_state_check(const struct xfrm_state
*x
,
412 const struct in6_addr
*dst
,
413 const struct in6_addr
*src
)
415 xfrm_address_t
*daddr
= (xfrm_address_t
*)dst
;
416 xfrm_address_t
*saddr
= (xfrm_address_t
*)src
;
418 /* if there is no transform then this tunnel is not functional.
419 * Or if the xfrm is not mode tunnel.
421 if (!x
|| x
->props
.mode
!= XFRM_MODE_TUNNEL
||
422 x
->props
.family
!= AF_INET6
)
425 if (ipv6_addr_any(dst
))
426 return xfrm_addr_equal(saddr
, &x
->props
.saddr
, AF_INET6
);
428 if (!xfrm_state_addr_check(x
, daddr
, saddr
, AF_INET6
))
435 * vti6_xmit - send a packet
436 * @skb: the outgoing socket buffer
437 * @dev: the outgoing tunnel device
438 * @fl: the flow informations for the xfrm_lookup
441 vti6_xmit(struct sk_buff
*skb
, struct net_device
*dev
, struct flowi
*fl
)
443 struct ip6_tnl
*t
= netdev_priv(dev
);
444 struct net_device_stats
*stats
= &t
->dev
->stats
;
445 struct dst_entry
*dst
= skb_dst(skb
);
446 struct net_device
*tdev
;
447 struct xfrm_state
*x
;
448 int pkt_len
= skb
->len
;
453 goto tx_err_link_failure
;
456 dst
= xfrm_lookup(t
->net
, dst
, fl
, NULL
, 0);
460 goto tx_err_link_failure
;
464 if (!vti6_state_check(x
, &t
->parms
.raddr
, &t
->parms
.laddr
))
465 goto tx_err_link_failure
;
467 if (!ip6_tnl_xmit_ctl(t
, (const struct in6_addr
*)&x
->props
.saddr
,
468 (const struct in6_addr
*)&x
->id
.daddr
))
469 goto tx_err_link_failure
;
475 net_warn_ratelimited("%s: Local routing loop detected!\n",
477 goto tx_err_dst_release
;
481 if (skb
->len
> mtu
) {
482 skb_dst_update_pmtu(skb
, mtu
);
484 if (skb
->protocol
== htons(ETH_P_IPV6
)) {
485 if (mtu
< IPV6_MIN_MTU
)
488 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
490 icmp_send(skb
, ICMP_DEST_UNREACH
, ICMP_FRAG_NEEDED
,
495 goto tx_err_dst_release
;
498 skb_scrub_packet(skb
, !net_eq(t
->net
, dev_net(dev
)));
499 skb_dst_set(skb
, dst
);
500 skb
->dev
= skb_dst(skb
)->dev
;
502 err
= dst_output(t
->net
, skb
->sk
, skb
);
503 if (net_xmit_eval(err
) == 0)
505 iptunnel_xmit_stats(dev
, err
);
509 stats
->tx_carrier_errors
++;
510 dst_link_failure(skb
);
517 vti6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
519 struct ip6_tnl
*t
= netdev_priv(dev
);
520 struct net_device_stats
*stats
= &t
->dev
->stats
;
524 if (!pskb_inet_may_pull(skb
))
527 memset(&fl
, 0, sizeof(fl
));
529 switch (skb
->protocol
) {
530 case htons(ETH_P_IPV6
):
531 if ((t
->parms
.proto
!= IPPROTO_IPV6
&& t
->parms
.proto
!= 0) ||
532 vti6_addr_conflict(t
, ipv6_hdr(skb
)))
535 xfrm_decode_session(skb
, &fl
, AF_INET6
);
536 memset(IP6CB(skb
), 0, sizeof(*IP6CB(skb
)));
538 case htons(ETH_P_IP
):
539 xfrm_decode_session(skb
, &fl
, AF_INET
);
540 memset(IPCB(skb
), 0, sizeof(*IPCB(skb
)));
546 /* override mark with tunnel output key */
547 fl
.flowi_mark
= be32_to_cpu(t
->parms
.o_key
);
549 ret
= vti6_xmit(skb
, dev
, &fl
);
562 static int vti6_err(struct sk_buff
*skb
, struct inet6_skb_parm
*opt
,
563 u8 type
, u8 code
, int offset
, __be32 info
)
567 struct xfrm_state
*x
;
569 struct ip_esp_hdr
*esph
;
570 struct ip_auth_hdr
*ah
;
571 struct ip_comp_hdr
*ipch
;
572 struct net
*net
= dev_net(skb
->dev
);
573 const struct ipv6hdr
*iph
= (const struct ipv6hdr
*)skb
->data
;
574 int protocol
= iph
->nexthdr
;
576 t
= vti6_tnl_lookup(dev_net(skb
->dev
), &iph
->daddr
, &iph
->saddr
);
580 mark
= be32_to_cpu(t
->parms
.o_key
);
584 esph
= (struct ip_esp_hdr
*)(skb
->data
+ offset
);
588 ah
= (struct ip_auth_hdr
*)(skb
->data
+ offset
);
592 ipch
= (struct ip_comp_hdr
*)(skb
->data
+ offset
);
593 spi
= htonl(ntohs(ipch
->cpi
));
599 if (type
!= ICMPV6_PKT_TOOBIG
&&
600 type
!= NDISC_REDIRECT
)
603 x
= xfrm_state_lookup(net
, mark
, (const xfrm_address_t
*)&iph
->daddr
,
604 spi
, protocol
, AF_INET6
);
608 if (type
== NDISC_REDIRECT
)
609 ip6_redirect(skb
, net
, skb
->dev
->ifindex
, 0,
610 sock_net_uid(net
, NULL
));
612 ip6_update_pmtu(skb
, net
, info
, 0, 0, sock_net_uid(net
, NULL
));
618 static void vti6_link_config(struct ip6_tnl
*t
, bool keep_mtu
)
620 struct net_device
*dev
= t
->dev
;
621 struct __ip6_tnl_parm
*p
= &t
->parms
;
622 struct net_device
*tdev
= NULL
;
625 memcpy(dev
->dev_addr
, &p
->laddr
, sizeof(struct in6_addr
));
626 memcpy(dev
->broadcast
, &p
->raddr
, sizeof(struct in6_addr
));
628 p
->flags
&= ~(IP6_TNL_F_CAP_XMIT
| IP6_TNL_F_CAP_RCV
|
629 IP6_TNL_F_CAP_PER_PACKET
);
630 p
->flags
|= ip6_tnl_get_cap(t
, &p
->laddr
, &p
->raddr
);
632 if (p
->flags
& IP6_TNL_F_CAP_XMIT
&& p
->flags
& IP6_TNL_F_CAP_RCV
)
633 dev
->flags
|= IFF_POINTOPOINT
;
635 dev
->flags
&= ~IFF_POINTOPOINT
;
637 if (keep_mtu
&& dev
->mtu
) {
638 dev
->mtu
= clamp(dev
->mtu
, dev
->min_mtu
, dev
->max_mtu
);
642 if (p
->flags
& IP6_TNL_F_CAP_XMIT
) {
643 int strict
= (ipv6_addr_type(&p
->raddr
) &
644 (IPV6_ADDR_MULTICAST
| IPV6_ADDR_LINKLOCAL
));
645 struct rt6_info
*rt
= rt6_lookup(t
->net
,
646 &p
->raddr
, &p
->laddr
,
647 p
->link
, NULL
, strict
);
654 if (!tdev
&& p
->link
)
655 tdev
= __dev_get_by_index(t
->net
, p
->link
);
658 mtu
= tdev
->mtu
- sizeof(struct ipv6hdr
);
660 mtu
= ETH_DATA_LEN
- LL_MAX_HEADER
- sizeof(struct ipv6hdr
);
662 dev
->mtu
= max_t(int, mtu
, IPV4_MIN_MTU
);
666 * vti6_tnl_change - update the tunnel parameters
667 * @t: tunnel to be changed
668 * @p: tunnel configuration parameters
669 * @keep_mtu: MTU was set from userspace, don't re-compute it
672 * vti6_tnl_change() updates the tunnel parameters
675 vti6_tnl_change(struct ip6_tnl
*t
, const struct __ip6_tnl_parm
*p
,
678 t
->parms
.laddr
= p
->laddr
;
679 t
->parms
.raddr
= p
->raddr
;
680 t
->parms
.link
= p
->link
;
681 t
->parms
.i_key
= p
->i_key
;
682 t
->parms
.o_key
= p
->o_key
;
683 t
->parms
.proto
= p
->proto
;
684 t
->parms
.fwmark
= p
->fwmark
;
685 dst_cache_reset(&t
->dst_cache
);
686 vti6_link_config(t
, keep_mtu
);
690 static int vti6_update(struct ip6_tnl
*t
, struct __ip6_tnl_parm
*p
,
693 struct net
*net
= dev_net(t
->dev
);
694 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
697 vti6_tnl_unlink(ip6n
, t
);
699 err
= vti6_tnl_change(t
, p
, keep_mtu
);
700 vti6_tnl_link(ip6n
, t
);
701 netdev_state_change(t
->dev
);
706 vti6_parm_from_user(struct __ip6_tnl_parm
*p
, const struct ip6_tnl_parm2
*u
)
715 memcpy(p
->name
, u
->name
, sizeof(u
->name
));
719 vti6_parm_to_user(struct ip6_tnl_parm2
*u
, const struct __ip6_tnl_parm
*p
)
727 u
->i_flags
|= GRE_KEY
;
729 u
->o_flags
|= GRE_KEY
;
732 memcpy(u
->name
, p
->name
, sizeof(u
->name
));
736 * vti6_ioctl - configure vti6 tunnels from userspace
737 * @dev: virtual device associated with tunnel
738 * @ifr: parameters passed from userspace
739 * @cmd: command to be performed
742 * vti6_ioctl() is used for managing vti6 tunnels
745 * The possible commands are the following:
746 * %SIOCGETTUNNEL: get tunnel parameters for device
747 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
748 * %SIOCCHGTUNNEL: change tunnel parameters to those given
749 * %SIOCDELTUNNEL: delete tunnel
751 * The fallback device "ip6_vti0", created during module
752 * initialization, can be used for creating other tunnel devices.
756 * %-EFAULT if unable to copy data to or from userspace,
757 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
758 * %-EINVAL if passed tunnel parameters are invalid,
759 * %-EEXIST if changing a tunnel's parameters would cause a conflict
760 * %-ENODEV if attempting to change or delete a nonexisting device
763 vti6_ioctl(struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
766 struct ip6_tnl_parm2 p
;
767 struct __ip6_tnl_parm p1
;
768 struct ip6_tnl
*t
= NULL
;
769 struct net
*net
= dev_net(dev
);
770 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
774 if (dev
== ip6n
->fb_tnl_dev
) {
775 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
))) {
779 vti6_parm_from_user(&p1
, &p
);
780 t
= vti6_locate(net
, &p1
, 0);
782 memset(&p
, 0, sizeof(p
));
785 t
= netdev_priv(dev
);
786 vti6_parm_to_user(&p
, &t
->parms
);
787 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof(p
)))
793 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
796 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
799 if (p
.proto
!= IPPROTO_IPV6
&& p
.proto
!= 0)
801 vti6_parm_from_user(&p1
, &p
);
802 t
= vti6_locate(net
, &p1
, cmd
== SIOCADDTUNNEL
);
803 if (dev
!= ip6n
->fb_tnl_dev
&& cmd
== SIOCCHGTUNNEL
) {
810 t
= netdev_priv(dev
);
812 err
= vti6_update(t
, &p1
, false);
816 vti6_parm_to_user(&p
, &t
->parms
);
817 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof(p
)))
821 err
= (cmd
== SIOCADDTUNNEL
? -ENOBUFS
: -ENOENT
);
825 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
828 if (dev
== ip6n
->fb_tnl_dev
) {
830 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
833 vti6_parm_from_user(&p1
, &p
);
834 t
= vti6_locate(net
, &p1
, 0);
838 if (t
->dev
== ip6n
->fb_tnl_dev
)
843 unregister_netdevice(dev
);
851 static const struct net_device_ops vti6_netdev_ops
= {
852 .ndo_init
= vti6_dev_init
,
853 .ndo_uninit
= vti6_dev_uninit
,
854 .ndo_start_xmit
= vti6_tnl_xmit
,
855 .ndo_do_ioctl
= vti6_ioctl
,
856 .ndo_get_stats64
= ip_tunnel_get_stats64
,
857 .ndo_get_iflink
= ip6_tnl_get_iflink
,
861 * vti6_dev_setup - setup virtual tunnel device
862 * @dev: virtual device associated with tunnel
865 * Initialize function pointers and device parameters
867 static void vti6_dev_setup(struct net_device
*dev
)
869 dev
->netdev_ops
= &vti6_netdev_ops
;
870 dev
->needs_free_netdev
= true;
871 dev
->priv_destructor
= vti6_dev_free
;
873 dev
->type
= ARPHRD_TUNNEL6
;
874 dev
->min_mtu
= IPV4_MIN_MTU
;
875 dev
->max_mtu
= IP_MAX_MTU
- sizeof(struct ipv6hdr
);
876 dev
->flags
|= IFF_NOARP
;
877 dev
->addr_len
= sizeof(struct in6_addr
);
879 /* This perm addr will be used as interface identifier by IPv6 */
880 dev
->addr_assign_type
= NET_ADDR_RANDOM
;
881 eth_random_addr(dev
->perm_addr
);
885 * vti6_dev_init_gen - general initializer for all tunnel devices
886 * @dev: virtual device associated with tunnel
888 static inline int vti6_dev_init_gen(struct net_device
*dev
)
890 struct ip6_tnl
*t
= netdev_priv(dev
);
893 t
->net
= dev_net(dev
);
894 dev
->tstats
= netdev_alloc_pcpu_stats(struct pcpu_sw_netstats
);
901 * vti6_dev_init - initializer for all non fallback tunnel devices
902 * @dev: virtual device associated with tunnel
904 static int vti6_dev_init(struct net_device
*dev
)
906 struct ip6_tnl
*t
= netdev_priv(dev
);
907 int err
= vti6_dev_init_gen(dev
);
911 vti6_link_config(t
, true);
916 * vti6_fb_tnl_dev_init - initializer for fallback tunnel device
917 * @dev: fallback device
921 static int __net_init
vti6_fb_tnl_dev_init(struct net_device
*dev
)
923 struct ip6_tnl
*t
= netdev_priv(dev
);
924 struct net
*net
= dev_net(dev
);
925 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
927 t
->parms
.proto
= IPPROTO_IPV6
;
930 rcu_assign_pointer(ip6n
->tnls_wc
[0], t
);
934 static int vti6_validate(struct nlattr
*tb
[], struct nlattr
*data
[],
935 struct netlink_ext_ack
*extack
)
940 static void vti6_netlink_parms(struct nlattr
*data
[],
941 struct __ip6_tnl_parm
*parms
)
943 memset(parms
, 0, sizeof(*parms
));
948 if (data
[IFLA_VTI_LINK
])
949 parms
->link
= nla_get_u32(data
[IFLA_VTI_LINK
]);
951 if (data
[IFLA_VTI_LOCAL
])
952 parms
->laddr
= nla_get_in6_addr(data
[IFLA_VTI_LOCAL
]);
954 if (data
[IFLA_VTI_REMOTE
])
955 parms
->raddr
= nla_get_in6_addr(data
[IFLA_VTI_REMOTE
]);
957 if (data
[IFLA_VTI_IKEY
])
958 parms
->i_key
= nla_get_be32(data
[IFLA_VTI_IKEY
]);
960 if (data
[IFLA_VTI_OKEY
])
961 parms
->o_key
= nla_get_be32(data
[IFLA_VTI_OKEY
]);
963 if (data
[IFLA_VTI_FWMARK
])
964 parms
->fwmark
= nla_get_u32(data
[IFLA_VTI_FWMARK
]);
967 static int vti6_newlink(struct net
*src_net
, struct net_device
*dev
,
968 struct nlattr
*tb
[], struct nlattr
*data
[],
969 struct netlink_ext_ack
*extack
)
971 struct net
*net
= dev_net(dev
);
974 nt
= netdev_priv(dev
);
975 vti6_netlink_parms(data
, &nt
->parms
);
977 nt
->parms
.proto
= IPPROTO_IPV6
;
979 if (vti6_locate(net
, &nt
->parms
, 0))
982 return vti6_tnl_create2(dev
);
985 static void vti6_dellink(struct net_device
*dev
, struct list_head
*head
)
987 struct net
*net
= dev_net(dev
);
988 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
990 if (dev
!= ip6n
->fb_tnl_dev
)
991 unregister_netdevice_queue(dev
, head
);
994 static int vti6_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
995 struct nlattr
*data
[],
996 struct netlink_ext_ack
*extack
)
999 struct __ip6_tnl_parm p
;
1000 struct net
*net
= dev_net(dev
);
1001 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
1003 if (dev
== ip6n
->fb_tnl_dev
)
1006 vti6_netlink_parms(data
, &p
);
1008 t
= vti6_locate(net
, &p
, 0);
1014 t
= netdev_priv(dev
);
1016 return vti6_update(t
, &p
, tb
&& tb
[IFLA_MTU
]);
1019 static size_t vti6_get_size(const struct net_device
*dev
)
1024 /* IFLA_VTI_LOCAL */
1025 nla_total_size(sizeof(struct in6_addr
)) +
1026 /* IFLA_VTI_REMOTE */
1027 nla_total_size(sizeof(struct in6_addr
)) +
1032 /* IFLA_VTI_FWMARK */
1037 static int vti6_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
1039 struct ip6_tnl
*tunnel
= netdev_priv(dev
);
1040 struct __ip6_tnl_parm
*parm
= &tunnel
->parms
;
1042 if (nla_put_u32(skb
, IFLA_VTI_LINK
, parm
->link
) ||
1043 nla_put_in6_addr(skb
, IFLA_VTI_LOCAL
, &parm
->laddr
) ||
1044 nla_put_in6_addr(skb
, IFLA_VTI_REMOTE
, &parm
->raddr
) ||
1045 nla_put_be32(skb
, IFLA_VTI_IKEY
, parm
->i_key
) ||
1046 nla_put_be32(skb
, IFLA_VTI_OKEY
, parm
->o_key
) ||
1047 nla_put_u32(skb
, IFLA_VTI_FWMARK
, parm
->fwmark
))
1048 goto nla_put_failure
;
1055 static const struct nla_policy vti6_policy
[IFLA_VTI_MAX
+ 1] = {
1056 [IFLA_VTI_LINK
] = { .type
= NLA_U32
},
1057 [IFLA_VTI_LOCAL
] = { .len
= sizeof(struct in6_addr
) },
1058 [IFLA_VTI_REMOTE
] = { .len
= sizeof(struct in6_addr
) },
1059 [IFLA_VTI_IKEY
] = { .type
= NLA_U32
},
1060 [IFLA_VTI_OKEY
] = { .type
= NLA_U32
},
1061 [IFLA_VTI_FWMARK
] = { .type
= NLA_U32
},
1064 static struct rtnl_link_ops vti6_link_ops __read_mostly
= {
1066 .maxtype
= IFLA_VTI_MAX
,
1067 .policy
= vti6_policy
,
1068 .priv_size
= sizeof(struct ip6_tnl
),
1069 .setup
= vti6_dev_setup
,
1070 .validate
= vti6_validate
,
1071 .newlink
= vti6_newlink
,
1072 .dellink
= vti6_dellink
,
1073 .changelink
= vti6_changelink
,
1074 .get_size
= vti6_get_size
,
1075 .fill_info
= vti6_fill_info
,
1076 .get_link_net
= ip6_tnl_get_link_net
,
1079 static void __net_exit
vti6_destroy_tunnels(struct vti6_net
*ip6n
,
1080 struct list_head
*list
)
1085 for (h
= 0; h
< IP6_VTI_HASH_SIZE
; h
++) {
1086 t
= rtnl_dereference(ip6n
->tnls_r_l
[h
]);
1088 unregister_netdevice_queue(t
->dev
, list
);
1089 t
= rtnl_dereference(t
->next
);
1093 t
= rtnl_dereference(ip6n
->tnls_wc
[0]);
1095 unregister_netdevice_queue(t
->dev
, list
);
1098 static int __net_init
vti6_init_net(struct net
*net
)
1100 struct vti6_net
*ip6n
= net_generic(net
, vti6_net_id
);
1101 struct ip6_tnl
*t
= NULL
;
1104 ip6n
->tnls
[0] = ip6n
->tnls_wc
;
1105 ip6n
->tnls
[1] = ip6n
->tnls_r_l
;
1107 if (!net_has_fallback_tunnels(net
))
1110 ip6n
->fb_tnl_dev
= alloc_netdev(sizeof(struct ip6_tnl
), "ip6_vti0",
1111 NET_NAME_UNKNOWN
, vti6_dev_setup
);
1113 if (!ip6n
->fb_tnl_dev
)
1115 dev_net_set(ip6n
->fb_tnl_dev
, net
);
1116 ip6n
->fb_tnl_dev
->rtnl_link_ops
= &vti6_link_ops
;
1118 err
= vti6_fb_tnl_dev_init(ip6n
->fb_tnl_dev
);
1122 err
= register_netdev(ip6n
->fb_tnl_dev
);
1126 t
= netdev_priv(ip6n
->fb_tnl_dev
);
1128 strcpy(t
->parms
.name
, ip6n
->fb_tnl_dev
->name
);
1132 free_netdev(ip6n
->fb_tnl_dev
);
1137 static void __net_exit
vti6_exit_batch_net(struct list_head
*net_list
)
1139 struct vti6_net
*ip6n
;
1144 list_for_each_entry(net
, net_list
, exit_list
) {
1145 ip6n
= net_generic(net
, vti6_net_id
);
1146 vti6_destroy_tunnels(ip6n
, &list
);
1148 unregister_netdevice_many(&list
);
1152 static struct pernet_operations vti6_net_ops
= {
1153 .init
= vti6_init_net
,
1154 .exit_batch
= vti6_exit_batch_net
,
1156 .size
= sizeof(struct vti6_net
),
1159 static struct xfrm6_protocol vti_esp6_protocol __read_mostly
= {
1160 .handler
= vti6_rcv
,
1161 .cb_handler
= vti6_rcv_cb
,
1162 .err_handler
= vti6_err
,
1166 static struct xfrm6_protocol vti_ah6_protocol __read_mostly
= {
1167 .handler
= vti6_rcv
,
1168 .cb_handler
= vti6_rcv_cb
,
1169 .err_handler
= vti6_err
,
1173 static struct xfrm6_protocol vti_ipcomp6_protocol __read_mostly
= {
1174 .handler
= vti6_rcv
,
1175 .cb_handler
= vti6_rcv_cb
,
1176 .err_handler
= vti6_err
,
1181 * vti6_tunnel_init - register protocol and reserve needed resources
1183 * Return: 0 on success
1185 static int __init
vti6_tunnel_init(void)
1190 msg
= "tunnel device";
1191 err
= register_pernet_device(&vti6_net_ops
);
1193 goto pernet_dev_failed
;
1195 msg
= "tunnel protocols";
1196 err
= xfrm6_protocol_register(&vti_esp6_protocol
, IPPROTO_ESP
);
1198 goto xfrm_proto_esp_failed
;
1199 err
= xfrm6_protocol_register(&vti_ah6_protocol
, IPPROTO_AH
);
1201 goto xfrm_proto_ah_failed
;
1202 err
= xfrm6_protocol_register(&vti_ipcomp6_protocol
, IPPROTO_COMP
);
1204 goto xfrm_proto_comp_failed
;
1206 msg
= "netlink interface";
1207 err
= rtnl_link_register(&vti6_link_ops
);
1209 goto rtnl_link_failed
;
1214 xfrm6_protocol_deregister(&vti_ipcomp6_protocol
, IPPROTO_COMP
);
1215 xfrm_proto_comp_failed
:
1216 xfrm6_protocol_deregister(&vti_ah6_protocol
, IPPROTO_AH
);
1217 xfrm_proto_ah_failed
:
1218 xfrm6_protocol_deregister(&vti_esp6_protocol
, IPPROTO_ESP
);
1219 xfrm_proto_esp_failed
:
1220 unregister_pernet_device(&vti6_net_ops
);
1222 pr_err("vti6 init: failed to register %s\n", msg
);
1227 * vti6_tunnel_cleanup - free resources and unregister protocol
1229 static void __exit
vti6_tunnel_cleanup(void)
1231 rtnl_link_unregister(&vti6_link_ops
);
1232 xfrm6_protocol_deregister(&vti_ipcomp6_protocol
, IPPROTO_COMP
);
1233 xfrm6_protocol_deregister(&vti_ah6_protocol
, IPPROTO_AH
);
1234 xfrm6_protocol_deregister(&vti_esp6_protocol
, IPPROTO_ESP
);
1235 unregister_pernet_device(&vti6_net_ops
);
1238 module_init(vti6_tunnel_init
);
1239 module_exit(vti6_tunnel_cleanup
);
1240 MODULE_LICENSE("GPL");
1241 MODULE_ALIAS_RTNL_LINK("vti6");
1242 MODULE_ALIAS_NETDEV("ip6_vti0");
1243 MODULE_AUTHOR("Steffen Klassert");
1244 MODULE_DESCRIPTION("IPv6 virtual tunnel interface");