2 * IPv6 tunneling device
3 * Linux INET6 implementation
6 * Ville Nuorvala <vnuorval@tcs.hut.fi>
7 * Yasuyuki Kozakai <kozakai@linux-ipv6.org>
10 * linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c
14 * This program is free software; you can redistribute it and/or
15 * modify it under the terms of the GNU General Public License
16 * as published by the Free Software Foundation; either version
17 * 2 of the License, or (at your option) any later version.
21 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
23 #include <linux/module.h>
24 #include <linux/capability.h>
25 #include <linux/errno.h>
26 #include <linux/types.h>
27 #include <linux/sockios.h>
28 #include <linux/icmp.h>
32 #include <linux/if_tunnel.h>
33 #include <linux/net.h>
34 #include <linux/in6.h>
35 #include <linux/netdevice.h>
36 #include <linux/if_arp.h>
37 #include <linux/icmpv6.h>
38 #include <linux/init.h>
39 #include <linux/route.h>
40 #include <linux/rtnetlink.h>
41 #include <linux/netfilter_ipv6.h>
42 #include <linux/slab.h>
43 #include <linux/hash.h>
45 #include <asm/uaccess.h>
46 #include <linux/atomic.h>
50 #include <net/ip_tunnels.h>
52 #include <net/ip6_route.h>
53 #include <net/addrconf.h>
54 #include <net/ip6_tunnel.h>
56 #include <net/dsfield.h>
57 #include <net/inet_ecn.h>
58 #include <net/net_namespace.h>
59 #include <net/netns/generic.h>
61 MODULE_AUTHOR("Ville Nuorvala");
62 MODULE_DESCRIPTION("IPv6 tunneling device");
63 MODULE_LICENSE("GPL");
64 MODULE_ALIAS_RTNL_LINK("ip6tnl");
65 MODULE_ALIAS_NETDEV("ip6tnl0");
68 #define IP6_TNL_TRACE(x...) pr_debug("%s:" x "\n", __func__)
70 #define IP6_TNL_TRACE(x...) do {;} while(0)
73 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
74 #define IPV6_TCLASS_SHIFT 20
76 #define HASH_SIZE_SHIFT 5
77 #define HASH_SIZE (1 << HASH_SIZE_SHIFT)
79 static bool log_ecn_error
= true;
80 module_param(log_ecn_error
, bool, 0644);
81 MODULE_PARM_DESC(log_ecn_error
, "Log packets received with corrupted ECN");
83 static u32
HASH(const struct in6_addr
*addr1
, const struct in6_addr
*addr2
)
85 u32 hash
= ipv6_addr_hash(addr1
) ^ ipv6_addr_hash(addr2
);
87 return hash_32(hash
, HASH_SIZE_SHIFT
);
90 static int ip6_tnl_dev_init(struct net_device
*dev
);
91 static void ip6_tnl_dev_setup(struct net_device
*dev
);
92 static struct rtnl_link_ops ip6_link_ops __read_mostly
;
94 static int ip6_tnl_net_id __read_mostly
;
96 /* the IPv6 tunnel fallback device */
97 struct net_device
*fb_tnl_dev
;
98 /* lists for storing tunnels in use */
99 struct ip6_tnl __rcu
*tnls_r_l
[HASH_SIZE
];
100 struct ip6_tnl __rcu
*tnls_wc
[1];
101 struct ip6_tnl __rcu
**tnls
[2];
104 static struct net_device_stats
*ip6_get_stats(struct net_device
*dev
)
106 struct pcpu_tstats sum
= { 0 };
109 for_each_possible_cpu(i
) {
110 const struct pcpu_tstats
*tstats
= per_cpu_ptr(dev
->tstats
, i
);
112 sum
.rx_packets
+= tstats
->rx_packets
;
113 sum
.rx_bytes
+= tstats
->rx_bytes
;
114 sum
.tx_packets
+= tstats
->tx_packets
;
115 sum
.tx_bytes
+= tstats
->tx_bytes
;
117 dev
->stats
.rx_packets
= sum
.rx_packets
;
118 dev
->stats
.rx_bytes
= sum
.rx_bytes
;
119 dev
->stats
.tx_packets
= sum
.tx_packets
;
120 dev
->stats
.tx_bytes
= sum
.tx_bytes
;
125 * Locking : hash tables are protected by RCU and RTNL
128 struct dst_entry
*ip6_tnl_dst_check(struct ip6_tnl
*t
)
130 struct dst_entry
*dst
= t
->dst_cache
;
132 if (dst
&& dst
->obsolete
&&
133 dst
->ops
->check(dst
, t
->dst_cookie
) == NULL
) {
141 EXPORT_SYMBOL_GPL(ip6_tnl_dst_check
);
143 void ip6_tnl_dst_reset(struct ip6_tnl
*t
)
145 dst_release(t
->dst_cache
);
148 EXPORT_SYMBOL_GPL(ip6_tnl_dst_reset
);
150 void ip6_tnl_dst_store(struct ip6_tnl
*t
, struct dst_entry
*dst
)
152 struct rt6_info
*rt
= (struct rt6_info
*) dst
;
153 t
->dst_cookie
= rt
->rt6i_node
? rt
->rt6i_node
->fn_sernum
: 0;
154 dst_release(t
->dst_cache
);
157 EXPORT_SYMBOL_GPL(ip6_tnl_dst_store
);
160 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
161 * @remote: the address of the tunnel exit-point
162 * @local: the address of the tunnel entry-point
165 * tunnel matching given end-points if found,
166 * else fallback tunnel if its device is up,
170 #define for_each_ip6_tunnel_rcu(start) \
171 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
173 static struct ip6_tnl
*
174 ip6_tnl_lookup(struct net
*net
, const struct in6_addr
*remote
, const struct in6_addr
*local
)
176 unsigned int hash
= HASH(remote
, local
);
178 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
180 for_each_ip6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
181 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
182 ipv6_addr_equal(remote
, &t
->parms
.raddr
) &&
183 (t
->dev
->flags
& IFF_UP
))
186 t
= rcu_dereference(ip6n
->tnls_wc
[0]);
187 if (t
&& (t
->dev
->flags
& IFF_UP
))
194 * ip6_tnl_bucket - get head of list matching given tunnel parameters
195 * @p: parameters containing tunnel end-points
198 * ip6_tnl_bucket() returns the head of the list matching the
199 * &struct in6_addr entries laddr and raddr in @p.
201 * Return: head of IPv6 tunnel list
204 static struct ip6_tnl __rcu
**
205 ip6_tnl_bucket(struct ip6_tnl_net
*ip6n
, const struct __ip6_tnl_parm
*p
)
207 const struct in6_addr
*remote
= &p
->raddr
;
208 const struct in6_addr
*local
= &p
->laddr
;
212 if (!ipv6_addr_any(remote
) || !ipv6_addr_any(local
)) {
214 h
= HASH(remote
, local
);
216 return &ip6n
->tnls
[prio
][h
];
220 * ip6_tnl_link - add tunnel to hash table
221 * @t: tunnel to be added
225 ip6_tnl_link(struct ip6_tnl_net
*ip6n
, struct ip6_tnl
*t
)
227 struct ip6_tnl __rcu
**tp
= ip6_tnl_bucket(ip6n
, &t
->parms
);
229 rcu_assign_pointer(t
->next
, rtnl_dereference(*tp
));
230 rcu_assign_pointer(*tp
, t
);
234 * ip6_tnl_unlink - remove tunnel from hash table
235 * @t: tunnel to be removed
239 ip6_tnl_unlink(struct ip6_tnl_net
*ip6n
, struct ip6_tnl
*t
)
241 struct ip6_tnl __rcu
**tp
;
242 struct ip6_tnl
*iter
;
244 for (tp
= ip6_tnl_bucket(ip6n
, &t
->parms
);
245 (iter
= rtnl_dereference(*tp
)) != NULL
;
248 rcu_assign_pointer(*tp
, t
->next
);
254 static void ip6_dev_free(struct net_device
*dev
)
256 free_percpu(dev
->tstats
);
260 static int ip6_tnl_create2(struct net_device
*dev
)
262 struct ip6_tnl
*t
= netdev_priv(dev
);
263 struct net
*net
= dev_net(dev
);
264 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
267 t
= netdev_priv(dev
);
269 dev
->rtnl_link_ops
= &ip6_link_ops
;
270 err
= register_netdevice(dev
);
274 strcpy(t
->parms
.name
, dev
->name
);
277 ip6_tnl_link(ip6n
, t
);
285 * ip6_tnl_create - create a new tunnel
286 * @p: tunnel parameters
287 * @pt: pointer to new tunnel
290 * Create tunnel matching given parameters.
293 * created tunnel or NULL
296 static struct ip6_tnl
*ip6_tnl_create(struct net
*net
, struct __ip6_tnl_parm
*p
)
298 struct net_device
*dev
;
304 strlcpy(name
, p
->name
, IFNAMSIZ
);
306 sprintf(name
, "ip6tnl%%d");
308 dev
= alloc_netdev(sizeof (*t
), name
, ip6_tnl_dev_setup
);
312 dev_net_set(dev
, net
);
314 t
= netdev_priv(dev
);
316 err
= ip6_tnl_create2(dev
);
329 * ip6_tnl_locate - find or create tunnel matching given parameters
330 * @p: tunnel parameters
331 * @create: != 0 if allowed to create new tunnel if no match found
334 * ip6_tnl_locate() first tries to locate an existing tunnel
335 * based on @parms. If this is unsuccessful, but @create is set a new
336 * tunnel device is created and registered for use.
339 * matching tunnel or NULL
342 static struct ip6_tnl
*ip6_tnl_locate(struct net
*net
,
343 struct __ip6_tnl_parm
*p
, int create
)
345 const struct in6_addr
*remote
= &p
->raddr
;
346 const struct in6_addr
*local
= &p
->laddr
;
347 struct ip6_tnl __rcu
**tp
;
349 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
351 for (tp
= ip6_tnl_bucket(ip6n
, p
);
352 (t
= rtnl_dereference(*tp
)) != NULL
;
354 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
355 ipv6_addr_equal(remote
, &t
->parms
.raddr
))
360 return ip6_tnl_create(net
, p
);
364 * ip6_tnl_dev_uninit - tunnel device uninitializer
365 * @dev: the device to be destroyed
368 * ip6_tnl_dev_uninit() removes tunnel from its list
372 ip6_tnl_dev_uninit(struct net_device
*dev
)
374 struct ip6_tnl
*t
= netdev_priv(dev
);
375 struct net
*net
= dev_net(dev
);
376 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
378 if (dev
== ip6n
->fb_tnl_dev
)
379 RCU_INIT_POINTER(ip6n
->tnls_wc
[0], NULL
);
381 ip6_tnl_unlink(ip6n
, t
);
382 ip6_tnl_dst_reset(t
);
387 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
388 * @skb: received socket buffer
391 * 0 if none was found,
392 * else index to encapsulation limit
395 __u16
ip6_tnl_parse_tlv_enc_lim(struct sk_buff
*skb
, __u8
*raw
)
397 const struct ipv6hdr
*ipv6h
= (const struct ipv6hdr
*) raw
;
398 __u8 nexthdr
= ipv6h
->nexthdr
;
399 __u16 off
= sizeof (*ipv6h
);
401 while (ipv6_ext_hdr(nexthdr
) && nexthdr
!= NEXTHDR_NONE
) {
403 struct ipv6_opt_hdr
*hdr
;
404 if (raw
+ off
+ sizeof (*hdr
) > skb
->data
&&
405 !pskb_may_pull(skb
, raw
- skb
->data
+ off
+ sizeof (*hdr
)))
408 hdr
= (struct ipv6_opt_hdr
*) (raw
+ off
);
409 if (nexthdr
== NEXTHDR_FRAGMENT
) {
410 struct frag_hdr
*frag_hdr
= (struct frag_hdr
*) hdr
;
411 if (frag_hdr
->frag_off
)
414 } else if (nexthdr
== NEXTHDR_AUTH
) {
415 optlen
= (hdr
->hdrlen
+ 2) << 2;
417 optlen
= ipv6_optlen(hdr
);
419 if (nexthdr
== NEXTHDR_DEST
) {
422 struct ipv6_tlv_tnl_enc_lim
*tel
;
424 /* No more room for encapsulation limit */
425 if (i
+ sizeof (*tel
) > off
+ optlen
)
428 tel
= (struct ipv6_tlv_tnl_enc_lim
*) &raw
[i
];
429 /* return index of option if found and valid */
430 if (tel
->type
== IPV6_TLV_TNL_ENCAP_LIMIT
&&
433 /* else jump to next option */
435 i
+= tel
->length
+ 2;
440 nexthdr
= hdr
->nexthdr
;
445 EXPORT_SYMBOL(ip6_tnl_parse_tlv_enc_lim
);
448 * ip6_tnl_err - tunnel error handler
451 * ip6_tnl_err() should handle errors in the tunnel according
452 * to the specifications in RFC 2473.
456 ip6_tnl_err(struct sk_buff
*skb
, __u8 ipproto
, struct inet6_skb_parm
*opt
,
457 u8
*type
, u8
*code
, int *msg
, __u32
*info
, int offset
)
459 const struct ipv6hdr
*ipv6h
= (const struct ipv6hdr
*) skb
->data
;
462 u8 rel_type
= ICMPV6_DEST_UNREACH
;
463 u8 rel_code
= ICMPV6_ADDR_UNREACH
;
468 /* If the packet doesn't contain the original IPv6 header we are
469 in trouble since we might need the source address for further
470 processing of the error. */
473 if ((t
= ip6_tnl_lookup(dev_net(skb
->dev
), &ipv6h
->daddr
,
474 &ipv6h
->saddr
)) == NULL
)
477 if (t
->parms
.proto
!= ipproto
&& t
->parms
.proto
!= 0)
484 struct ipv6_tlv_tnl_enc_lim
*tel
;
486 case ICMPV6_DEST_UNREACH
:
487 net_warn_ratelimited("%s: Path to destination invalid or inactive!\n",
491 case ICMPV6_TIME_EXCEED
:
492 if ((*code
) == ICMPV6_EXC_HOPLIMIT
) {
493 net_warn_ratelimited("%s: Too small hop limit or routing loop in tunnel!\n",
498 case ICMPV6_PARAMPROB
:
500 if ((*code
) == ICMPV6_HDR_FIELD
)
501 teli
= ip6_tnl_parse_tlv_enc_lim(skb
, skb
->data
);
503 if (teli
&& teli
== *info
- 2) {
504 tel
= (struct ipv6_tlv_tnl_enc_lim
*) &skb
->data
[teli
];
505 if (tel
->encap_limit
== 0) {
506 net_warn_ratelimited("%s: Too small encapsulation limit or routing loop in tunnel!\n",
511 net_warn_ratelimited("%s: Recipient unable to parse tunneled packet!\n",
515 case ICMPV6_PKT_TOOBIG
:
516 mtu
= *info
- offset
;
517 if (mtu
< IPV6_MIN_MTU
)
521 if ((len
= sizeof (*ipv6h
) + ntohs(ipv6h
->payload_len
)) > mtu
) {
522 rel_type
= ICMPV6_PKT_TOOBIG
;
541 ip4ip6_err(struct sk_buff
*skb
, struct inet6_skb_parm
*opt
,
542 u8 type
, u8 code
, int offset
, __be32 info
)
547 __u32 rel_info
= ntohl(info
);
549 struct sk_buff
*skb2
;
550 const struct iphdr
*eiph
;
554 err
= ip6_tnl_err(skb
, IPPROTO_IPIP
, opt
, &rel_type
, &rel_code
,
555 &rel_msg
, &rel_info
, offset
);
563 case ICMPV6_DEST_UNREACH
:
564 if (rel_code
!= ICMPV6_ADDR_UNREACH
)
566 rel_type
= ICMP_DEST_UNREACH
;
567 rel_code
= ICMP_HOST_UNREACH
;
569 case ICMPV6_PKT_TOOBIG
:
572 rel_type
= ICMP_DEST_UNREACH
;
573 rel_code
= ICMP_FRAG_NEEDED
;
576 rel_type
= ICMP_REDIRECT
;
577 rel_code
= ICMP_REDIR_HOST
;
582 if (!pskb_may_pull(skb
, offset
+ sizeof(struct iphdr
)))
585 skb2
= skb_clone(skb
, GFP_ATOMIC
);
591 skb_pull(skb2
, offset
);
592 skb_reset_network_header(skb2
);
595 /* Try to guess incoming interface */
596 rt
= ip_route_output_ports(dev_net(skb
->dev
), &fl4
, NULL
,
599 IPPROTO_IPIP
, RT_TOS(eiph
->tos
), 0);
603 skb2
->dev
= rt
->dst
.dev
;
605 /* route "incoming" packet */
606 if (rt
->rt_flags
& RTCF_LOCAL
) {
609 rt
= ip_route_output_ports(dev_net(skb
->dev
), &fl4
, NULL
,
610 eiph
->daddr
, eiph
->saddr
,
613 RT_TOS(eiph
->tos
), 0);
615 rt
->dst
.dev
->type
!= ARPHRD_TUNNEL
) {
620 skb_dst_set(skb2
, &rt
->dst
);
623 if (ip_route_input(skb2
, eiph
->daddr
, eiph
->saddr
, eiph
->tos
,
625 skb_dst(skb2
)->dev
->type
!= ARPHRD_TUNNEL
)
629 /* change mtu on this route */
630 if (rel_type
== ICMP_DEST_UNREACH
&& rel_code
== ICMP_FRAG_NEEDED
) {
631 if (rel_info
> dst_mtu(skb_dst(skb2
)))
634 skb_dst(skb2
)->ops
->update_pmtu(skb_dst(skb2
), NULL
, skb2
, rel_info
);
636 if (rel_type
== ICMP_REDIRECT
)
637 skb_dst(skb2
)->ops
->redirect(skb_dst(skb2
), NULL
, skb2
);
639 icmp_send(skb2
, rel_type
, rel_code
, htonl(rel_info
));
647 ip6ip6_err(struct sk_buff
*skb
, struct inet6_skb_parm
*opt
,
648 u8 type
, u8 code
, int offset
, __be32 info
)
653 __u32 rel_info
= ntohl(info
);
656 err
= ip6_tnl_err(skb
, IPPROTO_IPV6
, opt
, &rel_type
, &rel_code
,
657 &rel_msg
, &rel_info
, offset
);
661 if (rel_msg
&& pskb_may_pull(skb
, offset
+ sizeof(struct ipv6hdr
))) {
663 struct sk_buff
*skb2
= skb_clone(skb
, GFP_ATOMIC
);
669 skb_pull(skb2
, offset
);
670 skb_reset_network_header(skb2
);
672 /* Try to guess incoming interface */
673 rt
= rt6_lookup(dev_net(skb
->dev
), &ipv6_hdr(skb2
)->saddr
,
676 if (rt
&& rt
->dst
.dev
)
677 skb2
->dev
= rt
->dst
.dev
;
679 icmpv6_send(skb2
, rel_type
, rel_code
, rel_info
);
689 static int ip4ip6_dscp_ecn_decapsulate(const struct ip6_tnl
*t
,
690 const struct ipv6hdr
*ipv6h
,
693 __u8 dsfield
= ipv6_get_dsfield(ipv6h
) & ~INET_ECN_MASK
;
695 if (t
->parms
.flags
& IP6_TNL_F_RCV_DSCP_COPY
)
696 ipv4_change_dsfield(ip_hdr(skb
), INET_ECN_MASK
, dsfield
);
698 return IP6_ECN_decapsulate(ipv6h
, skb
);
701 static int ip6ip6_dscp_ecn_decapsulate(const struct ip6_tnl
*t
,
702 const struct ipv6hdr
*ipv6h
,
705 if (t
->parms
.flags
& IP6_TNL_F_RCV_DSCP_COPY
)
706 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h
), ipv6_hdr(skb
));
708 return IP6_ECN_decapsulate(ipv6h
, skb
);
711 __u32
ip6_tnl_get_cap(struct ip6_tnl
*t
,
712 const struct in6_addr
*laddr
,
713 const struct in6_addr
*raddr
)
715 struct __ip6_tnl_parm
*p
= &t
->parms
;
716 int ltype
= ipv6_addr_type(laddr
);
717 int rtype
= ipv6_addr_type(raddr
);
720 if (ltype
== IPV6_ADDR_ANY
|| rtype
== IPV6_ADDR_ANY
) {
721 flags
= IP6_TNL_F_CAP_PER_PACKET
;
722 } else if (ltype
& (IPV6_ADDR_UNICAST
|IPV6_ADDR_MULTICAST
) &&
723 rtype
& (IPV6_ADDR_UNICAST
|IPV6_ADDR_MULTICAST
) &&
724 !((ltype
|rtype
) & IPV6_ADDR_LOOPBACK
) &&
725 (!((ltype
|rtype
) & IPV6_ADDR_LINKLOCAL
) || p
->link
)) {
726 if (ltype
&IPV6_ADDR_UNICAST
)
727 flags
|= IP6_TNL_F_CAP_XMIT
;
728 if (rtype
&IPV6_ADDR_UNICAST
)
729 flags
|= IP6_TNL_F_CAP_RCV
;
733 EXPORT_SYMBOL(ip6_tnl_get_cap
);
735 /* called with rcu_read_lock() */
736 int ip6_tnl_rcv_ctl(struct ip6_tnl
*t
,
737 const struct in6_addr
*laddr
,
738 const struct in6_addr
*raddr
)
740 struct __ip6_tnl_parm
*p
= &t
->parms
;
742 struct net
*net
= dev_net(t
->dev
);
744 if ((p
->flags
& IP6_TNL_F_CAP_RCV
) ||
745 ((p
->flags
& IP6_TNL_F_CAP_PER_PACKET
) &&
746 (ip6_tnl_get_cap(t
, laddr
, raddr
) & IP6_TNL_F_CAP_RCV
))) {
747 struct net_device
*ldev
= NULL
;
750 ldev
= dev_get_by_index_rcu(net
, p
->link
);
752 if ((ipv6_addr_is_multicast(laddr
) ||
753 likely(ipv6_chk_addr(net
, laddr
, ldev
, 0))) &&
754 likely(!ipv6_chk_addr(net
, raddr
, NULL
, 0)))
759 EXPORT_SYMBOL_GPL(ip6_tnl_rcv_ctl
);
762 * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
763 * @skb: received socket buffer
764 * @protocol: ethernet protocol ID
765 * @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
770 static int ip6_tnl_rcv(struct sk_buff
*skb
, __u16 protocol
,
772 int (*dscp_ecn_decapsulate
)(const struct ip6_tnl
*t
,
773 const struct ipv6hdr
*ipv6h
,
774 struct sk_buff
*skb
))
777 const struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
782 if ((t
= ip6_tnl_lookup(dev_net(skb
->dev
), &ipv6h
->saddr
,
783 &ipv6h
->daddr
)) != NULL
) {
784 struct pcpu_tstats
*tstats
;
786 if (t
->parms
.proto
!= ipproto
&& t
->parms
.proto
!= 0) {
791 if (!xfrm6_policy_check(NULL
, XFRM_POLICY_IN
, skb
)) {
796 if (!ip6_tnl_rcv_ctl(t
, &ipv6h
->daddr
, &ipv6h
->saddr
)) {
797 t
->dev
->stats
.rx_dropped
++;
802 skb
->mac_header
= skb
->network_header
;
803 skb_reset_network_header(skb
);
804 skb
->protocol
= htons(protocol
);
805 skb
->pkt_type
= PACKET_HOST
;
806 memset(skb
->cb
, 0, sizeof(struct inet6_skb_parm
));
808 __skb_tunnel_rx(skb
, t
->dev
);
810 err
= dscp_ecn_decapsulate(t
, ipv6h
, skb
);
813 net_info_ratelimited("non-ECT from %pI6 with dsfield=%#x\n",
815 ipv6_get_dsfield(ipv6h
));
817 ++t
->dev
->stats
.rx_frame_errors
;
818 ++t
->dev
->stats
.rx_errors
;
824 tstats
= this_cpu_ptr(t
->dev
->tstats
);
825 tstats
->rx_packets
++;
826 tstats
->rx_bytes
+= skb
->len
;
841 static int ip4ip6_rcv(struct sk_buff
*skb
)
843 return ip6_tnl_rcv(skb
, ETH_P_IP
, IPPROTO_IPIP
,
844 ip4ip6_dscp_ecn_decapsulate
);
847 static int ip6ip6_rcv(struct sk_buff
*skb
)
849 return ip6_tnl_rcv(skb
, ETH_P_IPV6
, IPPROTO_IPV6
,
850 ip6ip6_dscp_ecn_decapsulate
);
853 struct ipv6_tel_txoption
{
854 struct ipv6_txoptions ops
;
858 static void init_tel_txopt(struct ipv6_tel_txoption
*opt
, __u8 encap_limit
)
860 memset(opt
, 0, sizeof(struct ipv6_tel_txoption
));
862 opt
->dst_opt
[2] = IPV6_TLV_TNL_ENCAP_LIMIT
;
864 opt
->dst_opt
[4] = encap_limit
;
865 opt
->dst_opt
[5] = IPV6_TLV_PADN
;
868 opt
->ops
.dst0opt
= (struct ipv6_opt_hdr
*) opt
->dst_opt
;
869 opt
->ops
.opt_nflen
= 8;
873 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
874 * @t: the outgoing tunnel device
875 * @hdr: IPv6 header from the incoming packet
878 * Avoid trivial tunneling loop by checking that tunnel exit-point
879 * doesn't match source of incoming packet.
887 ip6_tnl_addr_conflict(const struct ip6_tnl
*t
, const struct ipv6hdr
*hdr
)
889 return ipv6_addr_equal(&t
->parms
.raddr
, &hdr
->saddr
);
892 int ip6_tnl_xmit_ctl(struct ip6_tnl
*t
)
894 struct __ip6_tnl_parm
*p
= &t
->parms
;
896 struct net
*net
= dev_net(t
->dev
);
898 if (p
->flags
& IP6_TNL_F_CAP_XMIT
) {
899 struct net_device
*ldev
= NULL
;
903 ldev
= dev_get_by_index_rcu(net
, p
->link
);
905 if (unlikely(!ipv6_chk_addr(net
, &p
->laddr
, ldev
, 0)))
906 pr_warn("%s xmit: Local address not yet configured!\n",
908 else if (!ipv6_addr_is_multicast(&p
->raddr
) &&
909 unlikely(ipv6_chk_addr(net
, &p
->raddr
, NULL
, 0)))
910 pr_warn("%s xmit: Routing loop! Remote address found on this node!\n",
918 EXPORT_SYMBOL_GPL(ip6_tnl_xmit_ctl
);
921 * ip6_tnl_xmit2 - encapsulate packet and send
922 * @skb: the outgoing socket buffer
923 * @dev: the outgoing tunnel device
924 * @dsfield: dscp code for outer header
925 * @fl: flow of tunneled packet
926 * @encap_limit: encapsulation limit
927 * @pmtu: Path MTU is stored if packet is too big
930 * Build new header and do some sanity checks on the packet before sending
936 * %-EMSGSIZE message too big. return mtu in this case.
939 static int ip6_tnl_xmit2(struct sk_buff
*skb
,
940 struct net_device
*dev
,
946 struct net
*net
= dev_net(dev
);
947 struct ip6_tnl
*t
= netdev_priv(dev
);
948 struct net_device_stats
*stats
= &t
->dev
->stats
;
949 struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
950 struct ipv6_tel_txoption opt
;
951 struct dst_entry
*dst
= NULL
, *ndst
= NULL
;
952 struct net_device
*tdev
;
953 bool use_cache
= false;
955 unsigned int max_headroom
= sizeof(struct ipv6hdr
);
959 if (!(t
->parms
.flags
&
960 (IP6_TNL_F_USE_ORIG_TCLASS
| IP6_TNL_F_USE_ORIG_FWMARK
))) {
961 /* enable the cache only only if the routing decision does
962 * not depend on the current inner header value
968 dst
= ip6_tnl_dst_check(t
);
970 ndst
= ip6_route_output(net
, NULL
, fl6
);
973 goto tx_err_link_failure
;
974 ndst
= xfrm_lookup(net
, ndst
, flowi6_to_flowi(fl6
), NULL
, 0);
978 goto tx_err_link_failure
;
987 net_warn_ratelimited("%s: Local routing loop detected!\n",
989 goto tx_err_dst_release
;
991 mtu
= dst_mtu(dst
) - sizeof (*ipv6h
);
992 if (encap_limit
>= 0) {
996 if (mtu
< IPV6_MIN_MTU
)
999 skb_dst(skb
)->ops
->update_pmtu(skb_dst(skb
), NULL
, skb
, mtu
);
1000 if (skb
->len
> mtu
) {
1003 goto tx_err_dst_release
;
1007 * Okay, now see if we can stuff it in the buffer as-is.
1009 max_headroom
+= LL_RESERVED_SPACE(tdev
);
1011 if (skb_headroom(skb
) < max_headroom
|| skb_shared(skb
) ||
1012 (skb_cloned(skb
) && !skb_clone_writable(skb
, 0))) {
1013 struct sk_buff
*new_skb
;
1015 if (!(new_skb
= skb_realloc_headroom(skb
, max_headroom
)))
1016 goto tx_err_dst_release
;
1019 skb_set_owner_w(new_skb
, skb
->sk
);
1025 skb_dst_set(skb
, dst
);
1028 skb_dst_set_noref(skb
, dst
);
1030 skb
->transport_header
= skb
->network_header
;
1032 proto
= fl6
->flowi6_proto
;
1033 if (encap_limit
>= 0) {
1034 init_tel_txopt(&opt
, encap_limit
);
1035 ipv6_push_nfrag_opts(skb
, &opt
.ops
, &proto
, NULL
);
1037 skb_push(skb
, sizeof(struct ipv6hdr
));
1038 skb_reset_network_header(skb
);
1039 ipv6h
= ipv6_hdr(skb
);
1040 ip6_flow_hdr(ipv6h
, INET_ECN_encapsulate(0, dsfield
), fl6
->flowlabel
);
1041 ipv6h
->hop_limit
= t
->parms
.hop_limit
;
1042 ipv6h
->nexthdr
= proto
;
1043 ipv6h
->saddr
= fl6
->saddr
;
1044 ipv6h
->daddr
= fl6
->daddr
;
1045 ip6tunnel_xmit(skb
, dev
);
1047 ip6_tnl_dst_store(t
, ndst
);
1049 tx_err_link_failure
:
1050 stats
->tx_carrier_errors
++;
1051 dst_link_failure(skb
);
1058 ip4ip6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1060 struct ip6_tnl
*t
= netdev_priv(dev
);
1061 const struct iphdr
*iph
= ip_hdr(skb
);
1062 int encap_limit
= -1;
1068 if ((t
->parms
.proto
!= IPPROTO_IPIP
&& t
->parms
.proto
!= 0) ||
1069 !ip6_tnl_xmit_ctl(t
))
1072 if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1073 encap_limit
= t
->parms
.encap_limit
;
1075 memcpy(&fl6
, &t
->fl
.u
.ip6
, sizeof (fl6
));
1076 fl6
.flowi6_proto
= IPPROTO_IPIP
;
1078 dsfield
= ipv4_get_dsfield(iph
);
1080 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_TCLASS
)
1081 fl6
.flowlabel
|= htonl((__u32
)iph
->tos
<< IPV6_TCLASS_SHIFT
)
1083 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_FWMARK
)
1084 fl6
.flowi6_mark
= skb
->mark
;
1086 err
= ip6_tnl_xmit2(skb
, dev
, dsfield
, &fl6
, encap_limit
, &mtu
);
1088 /* XXX: send ICMP error even if DF is not set. */
1089 if (err
== -EMSGSIZE
)
1090 icmp_send(skb
, ICMP_DEST_UNREACH
, ICMP_FRAG_NEEDED
,
1099 ip6ip6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1101 struct ip6_tnl
*t
= netdev_priv(dev
);
1102 struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
1103 int encap_limit
= -1;
1110 if ((t
->parms
.proto
!= IPPROTO_IPV6
&& t
->parms
.proto
!= 0) ||
1111 !ip6_tnl_xmit_ctl(t
) || ip6_tnl_addr_conflict(t
, ipv6h
))
1114 offset
= ip6_tnl_parse_tlv_enc_lim(skb
, skb_network_header(skb
));
1116 struct ipv6_tlv_tnl_enc_lim
*tel
;
1117 tel
= (struct ipv6_tlv_tnl_enc_lim
*)&skb_network_header(skb
)[offset
];
1118 if (tel
->encap_limit
== 0) {
1119 icmpv6_send(skb
, ICMPV6_PARAMPROB
,
1120 ICMPV6_HDR_FIELD
, offset
+ 2);
1123 encap_limit
= tel
->encap_limit
- 1;
1124 } else if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1125 encap_limit
= t
->parms
.encap_limit
;
1127 memcpy(&fl6
, &t
->fl
.u
.ip6
, sizeof (fl6
));
1128 fl6
.flowi6_proto
= IPPROTO_IPV6
;
1130 dsfield
= ipv6_get_dsfield(ipv6h
);
1131 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_TCLASS
)
1132 fl6
.flowlabel
|= (*(__be32
*) ipv6h
& IPV6_TCLASS_MASK
);
1133 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_FLOWLABEL
)
1134 fl6
.flowlabel
|= (*(__be32
*) ipv6h
& IPV6_FLOWLABEL_MASK
);
1135 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_FWMARK
)
1136 fl6
.flowi6_mark
= skb
->mark
;
1138 err
= ip6_tnl_xmit2(skb
, dev
, dsfield
, &fl6
, encap_limit
, &mtu
);
1140 if (err
== -EMSGSIZE
)
1141 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
1149 ip6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1151 struct ip6_tnl
*t
= netdev_priv(dev
);
1152 struct net_device_stats
*stats
= &t
->dev
->stats
;
1155 switch (skb
->protocol
) {
1156 case htons(ETH_P_IP
):
1157 ret
= ip4ip6_tnl_xmit(skb
, dev
);
1159 case htons(ETH_P_IPV6
):
1160 ret
= ip6ip6_tnl_xmit(skb
, dev
);
1169 return NETDEV_TX_OK
;
1173 stats
->tx_dropped
++;
1175 return NETDEV_TX_OK
;
1178 static void ip6_tnl_link_config(struct ip6_tnl
*t
)
1180 struct net_device
*dev
= t
->dev
;
1181 struct __ip6_tnl_parm
*p
= &t
->parms
;
1182 struct flowi6
*fl6
= &t
->fl
.u
.ip6
;
1184 memcpy(dev
->dev_addr
, &p
->laddr
, sizeof(struct in6_addr
));
1185 memcpy(dev
->broadcast
, &p
->raddr
, sizeof(struct in6_addr
));
1187 /* Set up flowi template */
1188 fl6
->saddr
= p
->laddr
;
1189 fl6
->daddr
= p
->raddr
;
1190 fl6
->flowi6_oif
= p
->link
;
1193 if (!(p
->flags
&IP6_TNL_F_USE_ORIG_TCLASS
))
1194 fl6
->flowlabel
|= IPV6_TCLASS_MASK
& p
->flowinfo
;
1195 if (!(p
->flags
&IP6_TNL_F_USE_ORIG_FLOWLABEL
))
1196 fl6
->flowlabel
|= IPV6_FLOWLABEL_MASK
& p
->flowinfo
;
1198 p
->flags
&= ~(IP6_TNL_F_CAP_XMIT
|IP6_TNL_F_CAP_RCV
|IP6_TNL_F_CAP_PER_PACKET
);
1199 p
->flags
|= ip6_tnl_get_cap(t
, &p
->laddr
, &p
->raddr
);
1201 if (p
->flags
&IP6_TNL_F_CAP_XMIT
&& p
->flags
&IP6_TNL_F_CAP_RCV
)
1202 dev
->flags
|= IFF_POINTOPOINT
;
1204 dev
->flags
&= ~IFF_POINTOPOINT
;
1206 dev
->iflink
= p
->link
;
1208 if (p
->flags
& IP6_TNL_F_CAP_XMIT
) {
1209 int strict
= (ipv6_addr_type(&p
->raddr
) &
1210 (IPV6_ADDR_MULTICAST
|IPV6_ADDR_LINKLOCAL
));
1212 struct rt6_info
*rt
= rt6_lookup(dev_net(dev
),
1213 &p
->raddr
, &p
->laddr
,
1220 dev
->hard_header_len
= rt
->dst
.dev
->hard_header_len
+
1221 sizeof (struct ipv6hdr
);
1223 dev
->mtu
= rt
->dst
.dev
->mtu
- sizeof (struct ipv6hdr
);
1224 if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1227 if (dev
->mtu
< IPV6_MIN_MTU
)
1228 dev
->mtu
= IPV6_MIN_MTU
;
1235 * ip6_tnl_change - update the tunnel parameters
1236 * @t: tunnel to be changed
1237 * @p: tunnel configuration parameters
1240 * ip6_tnl_change() updates the tunnel parameters
1244 ip6_tnl_change(struct ip6_tnl
*t
, const struct __ip6_tnl_parm
*p
)
1246 t
->parms
.laddr
= p
->laddr
;
1247 t
->parms
.raddr
= p
->raddr
;
1248 t
->parms
.flags
= p
->flags
;
1249 t
->parms
.hop_limit
= p
->hop_limit
;
1250 t
->parms
.encap_limit
= p
->encap_limit
;
1251 t
->parms
.flowinfo
= p
->flowinfo
;
1252 t
->parms
.link
= p
->link
;
1253 t
->parms
.proto
= p
->proto
;
1254 ip6_tnl_dst_reset(t
);
1255 ip6_tnl_link_config(t
);
1259 static int ip6_tnl_update(struct ip6_tnl
*t
, struct __ip6_tnl_parm
*p
)
1261 struct net
*net
= dev_net(t
->dev
);
1262 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1265 ip6_tnl_unlink(ip6n
, t
);
1267 err
= ip6_tnl_change(t
, p
);
1268 ip6_tnl_link(ip6n
, t
);
1269 netdev_state_change(t
->dev
);
1274 ip6_tnl_parm_from_user(struct __ip6_tnl_parm
*p
, const struct ip6_tnl_parm
*u
)
1276 p
->laddr
= u
->laddr
;
1277 p
->raddr
= u
->raddr
;
1278 p
->flags
= u
->flags
;
1279 p
->hop_limit
= u
->hop_limit
;
1280 p
->encap_limit
= u
->encap_limit
;
1281 p
->flowinfo
= u
->flowinfo
;
1283 p
->proto
= u
->proto
;
1284 memcpy(p
->name
, u
->name
, sizeof(u
->name
));
1288 ip6_tnl_parm_to_user(struct ip6_tnl_parm
*u
, const struct __ip6_tnl_parm
*p
)
1290 u
->laddr
= p
->laddr
;
1291 u
->raddr
= p
->raddr
;
1292 u
->flags
= p
->flags
;
1293 u
->hop_limit
= p
->hop_limit
;
1294 u
->encap_limit
= p
->encap_limit
;
1295 u
->flowinfo
= p
->flowinfo
;
1297 u
->proto
= p
->proto
;
1298 memcpy(u
->name
, p
->name
, sizeof(u
->name
));
1302 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1303 * @dev: virtual device associated with tunnel
1304 * @ifr: parameters passed from userspace
1305 * @cmd: command to be performed
1308 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1311 * The possible commands are the following:
1312 * %SIOCGETTUNNEL: get tunnel parameters for device
1313 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1314 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1315 * %SIOCDELTUNNEL: delete tunnel
1317 * The fallback device "ip6tnl0", created during module
1318 * initialization, can be used for creating other tunnel devices.
1322 * %-EFAULT if unable to copy data to or from userspace,
1323 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1324 * %-EINVAL if passed tunnel parameters are invalid,
1325 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1326 * %-ENODEV if attempting to change or delete a nonexisting device
1330 ip6_tnl_ioctl(struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
1333 struct ip6_tnl_parm p
;
1334 struct __ip6_tnl_parm p1
;
1335 struct ip6_tnl
*t
= NULL
;
1336 struct net
*net
= dev_net(dev
);
1337 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1341 if (dev
== ip6n
->fb_tnl_dev
) {
1342 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof (p
))) {
1346 ip6_tnl_parm_from_user(&p1
, &p
);
1347 t
= ip6_tnl_locate(net
, &p1
, 0);
1349 memset(&p
, 0, sizeof(p
));
1352 t
= netdev_priv(dev
);
1353 ip6_tnl_parm_to_user(&p
, &t
->parms
);
1354 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof (p
))) {
1361 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1364 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof (p
)))
1367 if (p
.proto
!= IPPROTO_IPV6
&& p
.proto
!= IPPROTO_IPIP
&&
1370 ip6_tnl_parm_from_user(&p1
, &p
);
1371 t
= ip6_tnl_locate(net
, &p1
, cmd
== SIOCADDTUNNEL
);
1372 if (dev
!= ip6n
->fb_tnl_dev
&& cmd
== SIOCCHGTUNNEL
) {
1374 if (t
->dev
!= dev
) {
1379 t
= netdev_priv(dev
);
1381 err
= ip6_tnl_update(t
, &p1
);
1385 ip6_tnl_parm_to_user(&p
, &t
->parms
);
1386 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof(p
)))
1390 err
= (cmd
== SIOCADDTUNNEL
? -ENOBUFS
: -ENOENT
);
1394 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1397 if (dev
== ip6n
->fb_tnl_dev
) {
1399 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof (p
)))
1402 ip6_tnl_parm_from_user(&p1
, &p
);
1403 t
= ip6_tnl_locate(net
, &p1
, 0);
1407 if (t
->dev
== ip6n
->fb_tnl_dev
)
1412 unregister_netdevice(dev
);
1421 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1422 * @dev: virtual device associated with tunnel
1423 * @new_mtu: the new mtu
1427 * %-EINVAL if mtu too small
1431 ip6_tnl_change_mtu(struct net_device
*dev
, int new_mtu
)
1433 if (new_mtu
< IPV6_MIN_MTU
) {
1441 static const struct net_device_ops ip6_tnl_netdev_ops
= {
1442 .ndo_init
= ip6_tnl_dev_init
,
1443 .ndo_uninit
= ip6_tnl_dev_uninit
,
1444 .ndo_start_xmit
= ip6_tnl_xmit
,
1445 .ndo_do_ioctl
= ip6_tnl_ioctl
,
1446 .ndo_change_mtu
= ip6_tnl_change_mtu
,
1447 .ndo_get_stats
= ip6_get_stats
,
1452 * ip6_tnl_dev_setup - setup virtual tunnel device
1453 * @dev: virtual device associated with tunnel
1456 * Initialize function pointers and device parameters
1459 static void ip6_tnl_dev_setup(struct net_device
*dev
)
1463 dev
->netdev_ops
= &ip6_tnl_netdev_ops
;
1464 dev
->destructor
= ip6_dev_free
;
1466 dev
->type
= ARPHRD_TUNNEL6
;
1467 dev
->hard_header_len
= LL_MAX_HEADER
+ sizeof (struct ipv6hdr
);
1468 dev
->mtu
= ETH_DATA_LEN
- sizeof (struct ipv6hdr
);
1469 t
= netdev_priv(dev
);
1470 if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1472 dev
->flags
|= IFF_NOARP
;
1473 dev
->addr_len
= sizeof(struct in6_addr
);
1474 dev
->features
|= NETIF_F_NETNS_LOCAL
;
1475 dev
->priv_flags
&= ~IFF_XMIT_DST_RELEASE
;
1480 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1481 * @dev: virtual device associated with tunnel
1485 ip6_tnl_dev_init_gen(struct net_device
*dev
)
1487 struct ip6_tnl
*t
= netdev_priv(dev
);
1490 dev
->tstats
= alloc_percpu(struct pcpu_tstats
);
1497 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1498 * @dev: virtual device associated with tunnel
1501 static int ip6_tnl_dev_init(struct net_device
*dev
)
1503 struct ip6_tnl
*t
= netdev_priv(dev
);
1504 int err
= ip6_tnl_dev_init_gen(dev
);
1508 ip6_tnl_link_config(t
);
1513 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1514 * @dev: fallback device
1519 static int __net_init
ip6_fb_tnl_dev_init(struct net_device
*dev
)
1521 struct ip6_tnl
*t
= netdev_priv(dev
);
1522 struct net
*net
= dev_net(dev
);
1523 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1525 t
->parms
.proto
= IPPROTO_IPV6
;
1528 rcu_assign_pointer(ip6n
->tnls_wc
[0], t
);
1532 static int ip6_tnl_validate(struct nlattr
*tb
[], struct nlattr
*data
[])
1536 if (!data
|| !data
[IFLA_IPTUN_PROTO
])
1539 proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1540 if (proto
!= IPPROTO_IPV6
&&
1541 proto
!= IPPROTO_IPIP
&&
1548 static void ip6_tnl_netlink_parms(struct nlattr
*data
[],
1549 struct __ip6_tnl_parm
*parms
)
1551 memset(parms
, 0, sizeof(*parms
));
1556 if (data
[IFLA_IPTUN_LINK
])
1557 parms
->link
= nla_get_u32(data
[IFLA_IPTUN_LINK
]);
1559 if (data
[IFLA_IPTUN_LOCAL
])
1560 nla_memcpy(&parms
->laddr
, data
[IFLA_IPTUN_LOCAL
],
1561 sizeof(struct in6_addr
));
1563 if (data
[IFLA_IPTUN_REMOTE
])
1564 nla_memcpy(&parms
->raddr
, data
[IFLA_IPTUN_REMOTE
],
1565 sizeof(struct in6_addr
));
1567 if (data
[IFLA_IPTUN_TTL
])
1568 parms
->hop_limit
= nla_get_u8(data
[IFLA_IPTUN_TTL
]);
1570 if (data
[IFLA_IPTUN_ENCAP_LIMIT
])
1571 parms
->encap_limit
= nla_get_u8(data
[IFLA_IPTUN_ENCAP_LIMIT
]);
1573 if (data
[IFLA_IPTUN_FLOWINFO
])
1574 parms
->flowinfo
= nla_get_be32(data
[IFLA_IPTUN_FLOWINFO
]);
1576 if (data
[IFLA_IPTUN_FLAGS
])
1577 parms
->flags
= nla_get_u32(data
[IFLA_IPTUN_FLAGS
]);
1579 if (data
[IFLA_IPTUN_PROTO
])
1580 parms
->proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1583 static int ip6_tnl_newlink(struct net
*src_net
, struct net_device
*dev
,
1584 struct nlattr
*tb
[], struct nlattr
*data
[])
1586 struct net
*net
= dev_net(dev
);
1589 nt
= netdev_priv(dev
);
1590 ip6_tnl_netlink_parms(data
, &nt
->parms
);
1592 if (ip6_tnl_locate(net
, &nt
->parms
, 0))
1595 return ip6_tnl_create2(dev
);
1598 static int ip6_tnl_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
1599 struct nlattr
*data
[])
1602 struct __ip6_tnl_parm p
;
1603 struct net
*net
= dev_net(dev
);
1604 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1606 if (dev
== ip6n
->fb_tnl_dev
)
1609 ip6_tnl_netlink_parms(data
, &p
);
1611 t
= ip6_tnl_locate(net
, &p
, 0);
1617 t
= netdev_priv(dev
);
1619 return ip6_tnl_update(t
, &p
);
1622 static void ip6_tnl_dellink(struct net_device
*dev
, struct list_head
*head
)
1624 struct net
*net
= dev_net(dev
);
1625 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1627 if (dev
!= ip6n
->fb_tnl_dev
)
1628 unregister_netdevice_queue(dev
, head
);
1631 static size_t ip6_tnl_get_size(const struct net_device
*dev
)
1634 /* IFLA_IPTUN_LINK */
1636 /* IFLA_IPTUN_LOCAL */
1637 nla_total_size(sizeof(struct in6_addr
)) +
1638 /* IFLA_IPTUN_REMOTE */
1639 nla_total_size(sizeof(struct in6_addr
)) +
1640 /* IFLA_IPTUN_TTL */
1642 /* IFLA_IPTUN_ENCAP_LIMIT */
1644 /* IFLA_IPTUN_FLOWINFO */
1646 /* IFLA_IPTUN_FLAGS */
1648 /* IFLA_IPTUN_PROTO */
1653 static int ip6_tnl_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
1655 struct ip6_tnl
*tunnel
= netdev_priv(dev
);
1656 struct __ip6_tnl_parm
*parm
= &tunnel
->parms
;
1658 if (nla_put_u32(skb
, IFLA_IPTUN_LINK
, parm
->link
) ||
1659 nla_put(skb
, IFLA_IPTUN_LOCAL
, sizeof(struct in6_addr
),
1661 nla_put(skb
, IFLA_IPTUN_REMOTE
, sizeof(struct in6_addr
),
1663 nla_put_u8(skb
, IFLA_IPTUN_TTL
, parm
->hop_limit
) ||
1664 nla_put_u8(skb
, IFLA_IPTUN_ENCAP_LIMIT
, parm
->encap_limit
) ||
1665 nla_put_be32(skb
, IFLA_IPTUN_FLOWINFO
, parm
->flowinfo
) ||
1666 nla_put_u32(skb
, IFLA_IPTUN_FLAGS
, parm
->flags
) ||
1667 nla_put_u8(skb
, IFLA_IPTUN_PROTO
, parm
->proto
))
1668 goto nla_put_failure
;
1675 static const struct nla_policy ip6_tnl_policy
[IFLA_IPTUN_MAX
+ 1] = {
1676 [IFLA_IPTUN_LINK
] = { .type
= NLA_U32
},
1677 [IFLA_IPTUN_LOCAL
] = { .len
= sizeof(struct in6_addr
) },
1678 [IFLA_IPTUN_REMOTE
] = { .len
= sizeof(struct in6_addr
) },
1679 [IFLA_IPTUN_TTL
] = { .type
= NLA_U8
},
1680 [IFLA_IPTUN_ENCAP_LIMIT
] = { .type
= NLA_U8
},
1681 [IFLA_IPTUN_FLOWINFO
] = { .type
= NLA_U32
},
1682 [IFLA_IPTUN_FLAGS
] = { .type
= NLA_U32
},
1683 [IFLA_IPTUN_PROTO
] = { .type
= NLA_U8
},
1686 static struct rtnl_link_ops ip6_link_ops __read_mostly
= {
1688 .maxtype
= IFLA_IPTUN_MAX
,
1689 .policy
= ip6_tnl_policy
,
1690 .priv_size
= sizeof(struct ip6_tnl
),
1691 .setup
= ip6_tnl_dev_setup
,
1692 .validate
= ip6_tnl_validate
,
1693 .newlink
= ip6_tnl_newlink
,
1694 .changelink
= ip6_tnl_changelink
,
1695 .dellink
= ip6_tnl_dellink
,
1696 .get_size
= ip6_tnl_get_size
,
1697 .fill_info
= ip6_tnl_fill_info
,
1700 static struct xfrm6_tunnel ip4ip6_handler __read_mostly
= {
1701 .handler
= ip4ip6_rcv
,
1702 .err_handler
= ip4ip6_err
,
1706 static struct xfrm6_tunnel ip6ip6_handler __read_mostly
= {
1707 .handler
= ip6ip6_rcv
,
1708 .err_handler
= ip6ip6_err
,
1712 static void __net_exit
ip6_tnl_destroy_tunnels(struct ip6_tnl_net
*ip6n
)
1718 for (h
= 0; h
< HASH_SIZE
; h
++) {
1719 t
= rtnl_dereference(ip6n
->tnls_r_l
[h
]);
1721 unregister_netdevice_queue(t
->dev
, &list
);
1722 t
= rtnl_dereference(t
->next
);
1726 t
= rtnl_dereference(ip6n
->tnls_wc
[0]);
1727 unregister_netdevice_queue(t
->dev
, &list
);
1728 unregister_netdevice_many(&list
);
1731 static int __net_init
ip6_tnl_init_net(struct net
*net
)
1733 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1734 struct ip6_tnl
*t
= NULL
;
1737 ip6n
->tnls
[0] = ip6n
->tnls_wc
;
1738 ip6n
->tnls
[1] = ip6n
->tnls_r_l
;
1741 ip6n
->fb_tnl_dev
= alloc_netdev(sizeof(struct ip6_tnl
), "ip6tnl0",
1744 if (!ip6n
->fb_tnl_dev
)
1746 dev_net_set(ip6n
->fb_tnl_dev
, net
);
1747 ip6n
->fb_tnl_dev
->rtnl_link_ops
= &ip6_link_ops
;
1749 err
= ip6_fb_tnl_dev_init(ip6n
->fb_tnl_dev
);
1753 err
= register_netdev(ip6n
->fb_tnl_dev
);
1757 t
= netdev_priv(ip6n
->fb_tnl_dev
);
1759 strcpy(t
->parms
.name
, ip6n
->fb_tnl_dev
->name
);
1763 ip6_dev_free(ip6n
->fb_tnl_dev
);
1768 static void __net_exit
ip6_tnl_exit_net(struct net
*net
)
1770 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1773 ip6_tnl_destroy_tunnels(ip6n
);
1777 static struct pernet_operations ip6_tnl_net_ops
= {
1778 .init
= ip6_tnl_init_net
,
1779 .exit
= ip6_tnl_exit_net
,
1780 .id
= &ip6_tnl_net_id
,
1781 .size
= sizeof(struct ip6_tnl_net
),
1785 * ip6_tunnel_init - register protocol and reserve needed resources
1787 * Return: 0 on success
1790 static int __init
ip6_tunnel_init(void)
1794 err
= register_pernet_device(&ip6_tnl_net_ops
);
1798 err
= xfrm6_tunnel_register(&ip4ip6_handler
, AF_INET
);
1800 pr_err("%s: can't register ip4ip6\n", __func__
);
1804 err
= xfrm6_tunnel_register(&ip6ip6_handler
, AF_INET6
);
1806 pr_err("%s: can't register ip6ip6\n", __func__
);
1809 err
= rtnl_link_register(&ip6_link_ops
);
1811 goto rtnl_link_failed
;
1816 xfrm6_tunnel_deregister(&ip6ip6_handler
, AF_INET6
);
1818 xfrm6_tunnel_deregister(&ip4ip6_handler
, AF_INET
);
1820 unregister_pernet_device(&ip6_tnl_net_ops
);
1826 * ip6_tunnel_cleanup - free resources and unregister protocol
1829 static void __exit
ip6_tunnel_cleanup(void)
1831 rtnl_link_unregister(&ip6_link_ops
);
1832 if (xfrm6_tunnel_deregister(&ip4ip6_handler
, AF_INET
))
1833 pr_info("%s: can't deregister ip4ip6\n", __func__
);
1835 if (xfrm6_tunnel_deregister(&ip6ip6_handler
, AF_INET6
))
1836 pr_info("%s: can't deregister ip6ip6\n", __func__
);
1838 unregister_pernet_device(&ip6_tnl_net_ops
);
1841 module_init(ip6_tunnel_init
);
1842 module_exit(ip6_tunnel_cleanup
);