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/net.h>
33 #include <linux/in6.h>
34 #include <linux/netdevice.h>
35 #include <linux/if_arp.h>
36 #include <linux/icmpv6.h>
37 #include <linux/init.h>
38 #include <linux/route.h>
39 #include <linux/rtnetlink.h>
40 #include <linux/netfilter_ipv6.h>
41 #include <linux/slab.h>
42 #include <linux/hash.h>
43 #include <linux/etherdevice.h>
45 #include <linux/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>
60 #include <net/dst_metadata.h>
62 MODULE_AUTHOR("Ville Nuorvala");
63 MODULE_DESCRIPTION("IPv6 tunneling device");
64 MODULE_LICENSE("GPL");
65 MODULE_ALIAS_RTNL_LINK("ip6tnl");
66 MODULE_ALIAS_NETDEV("ip6tnl0");
68 #define IP6_TUNNEL_HASH_SIZE_SHIFT 5
69 #define IP6_TUNNEL_HASH_SIZE (1 << IP6_TUNNEL_HASH_SIZE_SHIFT)
71 static bool log_ecn_error
= true;
72 module_param(log_ecn_error
, bool, 0644);
73 MODULE_PARM_DESC(log_ecn_error
, "Log packets received with corrupted ECN");
75 static u32
HASH(const struct in6_addr
*addr1
, const struct in6_addr
*addr2
)
77 u32 hash
= ipv6_addr_hash(addr1
) ^ ipv6_addr_hash(addr2
);
79 return hash_32(hash
, IP6_TUNNEL_HASH_SIZE_SHIFT
);
82 static int ip6_tnl_dev_init(struct net_device
*dev
);
83 static void ip6_tnl_dev_setup(struct net_device
*dev
);
84 static struct rtnl_link_ops ip6_link_ops __read_mostly
;
86 static unsigned int ip6_tnl_net_id __read_mostly
;
88 /* the IPv6 tunnel fallback device */
89 struct net_device
*fb_tnl_dev
;
90 /* lists for storing tunnels in use */
91 struct ip6_tnl __rcu
*tnls_r_l
[IP6_TUNNEL_HASH_SIZE
];
92 struct ip6_tnl __rcu
*tnls_wc
[1];
93 struct ip6_tnl __rcu
**tnls
[2];
94 struct ip6_tnl __rcu
*collect_md_tun
;
97 static struct net_device_stats
*ip6_get_stats(struct net_device
*dev
)
99 struct pcpu_sw_netstats tmp
, sum
= { 0 };
102 for_each_possible_cpu(i
) {
104 const struct pcpu_sw_netstats
*tstats
=
105 per_cpu_ptr(dev
->tstats
, i
);
108 start
= u64_stats_fetch_begin_irq(&tstats
->syncp
);
109 tmp
.rx_packets
= tstats
->rx_packets
;
110 tmp
.rx_bytes
= tstats
->rx_bytes
;
111 tmp
.tx_packets
= tstats
->tx_packets
;
112 tmp
.tx_bytes
= tstats
->tx_bytes
;
113 } while (u64_stats_fetch_retry_irq(&tstats
->syncp
, start
));
115 sum
.rx_packets
+= tmp
.rx_packets
;
116 sum
.rx_bytes
+= tmp
.rx_bytes
;
117 sum
.tx_packets
+= tmp
.tx_packets
;
118 sum
.tx_bytes
+= tmp
.tx_bytes
;
120 dev
->stats
.rx_packets
= sum
.rx_packets
;
121 dev
->stats
.rx_bytes
= sum
.rx_bytes
;
122 dev
->stats
.tx_packets
= sum
.tx_packets
;
123 dev
->stats
.tx_bytes
= sum
.tx_bytes
;
128 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
129 * @remote: the address of the tunnel exit-point
130 * @local: the address of the tunnel entry-point
133 * tunnel matching given end-points if found,
134 * else fallback tunnel if its device is up,
138 #define for_each_ip6_tunnel_rcu(start) \
139 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
141 static struct ip6_tnl
*
142 ip6_tnl_lookup(struct net
*net
, const struct in6_addr
*remote
, const struct in6_addr
*local
)
144 unsigned int hash
= HASH(remote
, local
);
146 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
149 for_each_ip6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
150 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
151 ipv6_addr_equal(remote
, &t
->parms
.raddr
) &&
152 (t
->dev
->flags
& IFF_UP
))
156 memset(&any
, 0, sizeof(any
));
157 hash
= HASH(&any
, local
);
158 for_each_ip6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
159 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
160 ipv6_addr_any(&t
->parms
.raddr
) &&
161 (t
->dev
->flags
& IFF_UP
))
165 hash
= HASH(remote
, &any
);
166 for_each_ip6_tunnel_rcu(ip6n
->tnls_r_l
[hash
]) {
167 if (ipv6_addr_equal(remote
, &t
->parms
.raddr
) &&
168 ipv6_addr_any(&t
->parms
.laddr
) &&
169 (t
->dev
->flags
& IFF_UP
))
173 t
= rcu_dereference(ip6n
->collect_md_tun
);
177 t
= rcu_dereference(ip6n
->tnls_wc
[0]);
178 if (t
&& (t
->dev
->flags
& IFF_UP
))
185 * ip6_tnl_bucket - get head of list matching given tunnel parameters
186 * @p: parameters containing tunnel end-points
189 * ip6_tnl_bucket() returns the head of the list matching the
190 * &struct in6_addr entries laddr and raddr in @p.
192 * Return: head of IPv6 tunnel list
195 static struct ip6_tnl __rcu
**
196 ip6_tnl_bucket(struct ip6_tnl_net
*ip6n
, const struct __ip6_tnl_parm
*p
)
198 const struct in6_addr
*remote
= &p
->raddr
;
199 const struct in6_addr
*local
= &p
->laddr
;
203 if (!ipv6_addr_any(remote
) || !ipv6_addr_any(local
)) {
205 h
= HASH(remote
, local
);
207 return &ip6n
->tnls
[prio
][h
];
211 * ip6_tnl_link - add tunnel to hash table
212 * @t: tunnel to be added
216 ip6_tnl_link(struct ip6_tnl_net
*ip6n
, struct ip6_tnl
*t
)
218 struct ip6_tnl __rcu
**tp
= ip6_tnl_bucket(ip6n
, &t
->parms
);
220 if (t
->parms
.collect_md
)
221 rcu_assign_pointer(ip6n
->collect_md_tun
, t
);
222 rcu_assign_pointer(t
->next
, rtnl_dereference(*tp
));
223 rcu_assign_pointer(*tp
, t
);
227 * ip6_tnl_unlink - remove tunnel from hash table
228 * @t: tunnel to be removed
232 ip6_tnl_unlink(struct ip6_tnl_net
*ip6n
, struct ip6_tnl
*t
)
234 struct ip6_tnl __rcu
**tp
;
235 struct ip6_tnl
*iter
;
237 if (t
->parms
.collect_md
)
238 rcu_assign_pointer(ip6n
->collect_md_tun
, NULL
);
240 for (tp
= ip6_tnl_bucket(ip6n
, &t
->parms
);
241 (iter
= rtnl_dereference(*tp
)) != NULL
;
244 rcu_assign_pointer(*tp
, t
->next
);
250 static void ip6_dev_free(struct net_device
*dev
)
252 struct ip6_tnl
*t
= netdev_priv(dev
);
254 gro_cells_destroy(&t
->gro_cells
);
255 dst_cache_destroy(&t
->dst_cache
);
256 free_percpu(dev
->tstats
);
259 static int ip6_tnl_create2(struct net_device
*dev
)
261 struct ip6_tnl
*t
= netdev_priv(dev
);
262 struct net
*net
= dev_net(dev
);
263 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
266 t
= netdev_priv(dev
);
268 dev
->rtnl_link_ops
= &ip6_link_ops
;
269 err
= register_netdevice(dev
);
273 strcpy(t
->parms
.name
, dev
->name
);
276 ip6_tnl_link(ip6n
, t
);
284 * ip6_tnl_create - create a new tunnel
285 * @p: tunnel parameters
286 * @pt: pointer to new tunnel
289 * Create tunnel matching given parameters.
292 * created tunnel or error pointer
295 static struct ip6_tnl
*ip6_tnl_create(struct net
*net
, struct __ip6_tnl_parm
*p
)
297 struct net_device
*dev
;
303 strlcpy(name
, p
->name
, IFNAMSIZ
);
305 sprintf(name
, "ip6tnl%%d");
307 dev
= alloc_netdev(sizeof(*t
), name
, NET_NAME_UNKNOWN
,
312 dev_net_set(dev
, net
);
314 t
= netdev_priv(dev
);
316 t
->net
= dev_net(dev
);
317 err
= ip6_tnl_create2(dev
);
330 * ip6_tnl_locate - find or create tunnel matching given parameters
331 * @p: tunnel parameters
332 * @create: != 0 if allowed to create new tunnel if no match found
335 * ip6_tnl_locate() first tries to locate an existing tunnel
336 * based on @parms. If this is unsuccessful, but @create is set a new
337 * tunnel device is created and registered for use.
340 * matching tunnel or error pointer
343 static struct ip6_tnl
*ip6_tnl_locate(struct net
*net
,
344 struct __ip6_tnl_parm
*p
, int create
)
346 const struct in6_addr
*remote
= &p
->raddr
;
347 const struct in6_addr
*local
= &p
->laddr
;
348 struct ip6_tnl __rcu
**tp
;
350 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
352 for (tp
= ip6_tnl_bucket(ip6n
, p
);
353 (t
= rtnl_dereference(*tp
)) != NULL
;
355 if (ipv6_addr_equal(local
, &t
->parms
.laddr
) &&
356 ipv6_addr_equal(remote
, &t
->parms
.raddr
)) {
358 return ERR_PTR(-EEXIST
);
364 return ERR_PTR(-ENODEV
);
365 return ip6_tnl_create(net
, p
);
369 * ip6_tnl_dev_uninit - tunnel device uninitializer
370 * @dev: the device to be destroyed
373 * ip6_tnl_dev_uninit() removes tunnel from its list
377 ip6_tnl_dev_uninit(struct net_device
*dev
)
379 struct ip6_tnl
*t
= netdev_priv(dev
);
380 struct net
*net
= t
->net
;
381 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
383 if (dev
== ip6n
->fb_tnl_dev
)
384 RCU_INIT_POINTER(ip6n
->tnls_wc
[0], NULL
);
386 ip6_tnl_unlink(ip6n
, t
);
387 dst_cache_reset(&t
->dst_cache
);
392 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
393 * @skb: received socket buffer
396 * 0 if none was found,
397 * else index to encapsulation limit
400 __u16
ip6_tnl_parse_tlv_enc_lim(struct sk_buff
*skb
, __u8
*raw
)
402 const struct ipv6hdr
*ipv6h
= (const struct ipv6hdr
*)raw
;
403 unsigned int nhoff
= raw
- skb
->data
;
404 unsigned int off
= nhoff
+ sizeof(*ipv6h
);
405 u8 next
, nexthdr
= ipv6h
->nexthdr
;
407 while (ipv6_ext_hdr(nexthdr
) && nexthdr
!= NEXTHDR_NONE
) {
408 struct ipv6_opt_hdr
*hdr
;
411 if (!pskb_may_pull(skb
, off
+ sizeof(*hdr
)))
414 hdr
= (struct ipv6_opt_hdr
*)(skb
->data
+ off
);
415 if (nexthdr
== NEXTHDR_FRAGMENT
) {
416 struct frag_hdr
*frag_hdr
= (struct frag_hdr
*) hdr
;
417 if (frag_hdr
->frag_off
)
420 } else if (nexthdr
== NEXTHDR_AUTH
) {
421 optlen
= (hdr
->hdrlen
+ 2) << 2;
423 optlen
= ipv6_optlen(hdr
);
425 /* cache hdr->nexthdr, since pskb_may_pull() might
429 if (nexthdr
== NEXTHDR_DEST
) {
432 /* Remember : hdr is no longer valid at this point. */
433 if (!pskb_may_pull(skb
, off
+ optlen
))
437 struct ipv6_tlv_tnl_enc_lim
*tel
;
439 /* No more room for encapsulation limit */
440 if (i
+ sizeof(*tel
) > optlen
)
443 tel
= (struct ipv6_tlv_tnl_enc_lim
*)(skb
->data
+ off
+ i
);
444 /* return index of option if found and valid */
445 if (tel
->type
== IPV6_TLV_TNL_ENCAP_LIMIT
&&
447 return i
+ off
- nhoff
;
448 /* else jump to next option */
450 i
+= tel
->length
+ 2;
460 EXPORT_SYMBOL(ip6_tnl_parse_tlv_enc_lim
);
463 * ip6_tnl_err - tunnel error handler
466 * ip6_tnl_err() should handle errors in the tunnel according
467 * to the specifications in RFC 2473.
471 ip6_tnl_err(struct sk_buff
*skb
, __u8 ipproto
, struct inet6_skb_parm
*opt
,
472 u8
*type
, u8
*code
, int *msg
, __u32
*info
, int offset
)
474 const struct ipv6hdr
*ipv6h
= (const struct ipv6hdr
*) skb
->data
;
477 u8 rel_type
= ICMPV6_DEST_UNREACH
;
478 u8 rel_code
= ICMPV6_ADDR_UNREACH
;
484 /* If the packet doesn't contain the original IPv6 header we are
485 in trouble since we might need the source address for further
486 processing of the error. */
489 t
= ip6_tnl_lookup(dev_net(skb
->dev
), &ipv6h
->daddr
, &ipv6h
->saddr
);
493 tproto
= ACCESS_ONCE(t
->parms
.proto
);
494 if (tproto
!= ipproto
&& tproto
!= 0)
501 struct ipv6_tlv_tnl_enc_lim
*tel
;
503 case ICMPV6_DEST_UNREACH
:
504 net_dbg_ratelimited("%s: Path to destination invalid or inactive!\n",
508 case ICMPV6_TIME_EXCEED
:
509 if ((*code
) == ICMPV6_EXC_HOPLIMIT
) {
510 net_dbg_ratelimited("%s: Too small hop limit or routing loop in tunnel!\n",
515 case ICMPV6_PARAMPROB
:
517 if ((*code
) == ICMPV6_HDR_FIELD
)
518 teli
= ip6_tnl_parse_tlv_enc_lim(skb
, skb
->data
);
520 if (teli
&& teli
== *info
- 2) {
521 tel
= (struct ipv6_tlv_tnl_enc_lim
*) &skb
->data
[teli
];
522 if (tel
->encap_limit
== 0) {
523 net_dbg_ratelimited("%s: Too small encapsulation limit or routing loop in tunnel!\n",
528 net_dbg_ratelimited("%s: Recipient unable to parse tunneled packet!\n",
532 case ICMPV6_PKT_TOOBIG
:
533 mtu
= *info
- offset
;
534 if (mtu
< IPV6_MIN_MTU
)
538 len
= sizeof(*ipv6h
) + ntohs(ipv6h
->payload_len
);
540 rel_type
= ICMPV6_PKT_TOOBIG
;
559 ip4ip6_err(struct sk_buff
*skb
, struct inet6_skb_parm
*opt
,
560 u8 type
, u8 code
, int offset
, __be32 info
)
565 __u32 rel_info
= ntohl(info
);
567 struct sk_buff
*skb2
;
568 const struct iphdr
*eiph
;
572 err
= ip6_tnl_err(skb
, IPPROTO_IPIP
, opt
, &rel_type
, &rel_code
,
573 &rel_msg
, &rel_info
, offset
);
581 case ICMPV6_DEST_UNREACH
:
582 if (rel_code
!= ICMPV6_ADDR_UNREACH
)
584 rel_type
= ICMP_DEST_UNREACH
;
585 rel_code
= ICMP_HOST_UNREACH
;
587 case ICMPV6_PKT_TOOBIG
:
590 rel_type
= ICMP_DEST_UNREACH
;
591 rel_code
= ICMP_FRAG_NEEDED
;
594 rel_type
= ICMP_REDIRECT
;
595 rel_code
= ICMP_REDIR_HOST
;
600 if (!pskb_may_pull(skb
, offset
+ sizeof(struct iphdr
)))
603 skb2
= skb_clone(skb
, GFP_ATOMIC
);
609 skb_pull(skb2
, offset
);
610 skb_reset_network_header(skb2
);
613 /* Try to guess incoming interface */
614 rt
= ip_route_output_ports(dev_net(skb
->dev
), &fl4
, NULL
,
617 IPPROTO_IPIP
, RT_TOS(eiph
->tos
), 0);
621 skb2
->dev
= rt
->dst
.dev
;
623 /* route "incoming" packet */
624 if (rt
->rt_flags
& RTCF_LOCAL
) {
627 rt
= ip_route_output_ports(dev_net(skb
->dev
), &fl4
, NULL
,
628 eiph
->daddr
, eiph
->saddr
,
631 RT_TOS(eiph
->tos
), 0);
633 rt
->dst
.dev
->type
!= ARPHRD_TUNNEL
) {
638 skb_dst_set(skb2
, &rt
->dst
);
641 if (ip_route_input(skb2
, eiph
->daddr
, eiph
->saddr
, eiph
->tos
,
643 skb_dst(skb2
)->dev
->type
!= ARPHRD_TUNNEL
)
647 /* change mtu on this route */
648 if (rel_type
== ICMP_DEST_UNREACH
&& rel_code
== ICMP_FRAG_NEEDED
) {
649 if (rel_info
> dst_mtu(skb_dst(skb2
)))
652 skb_dst(skb2
)->ops
->update_pmtu(skb_dst(skb2
), NULL
, skb2
, rel_info
);
654 if (rel_type
== ICMP_REDIRECT
)
655 skb_dst(skb2
)->ops
->redirect(skb_dst(skb2
), NULL
, skb2
);
657 icmp_send(skb2
, rel_type
, rel_code
, htonl(rel_info
));
665 ip6ip6_err(struct sk_buff
*skb
, struct inet6_skb_parm
*opt
,
666 u8 type
, u8 code
, int offset
, __be32 info
)
671 __u32 rel_info
= ntohl(info
);
674 err
= ip6_tnl_err(skb
, IPPROTO_IPV6
, opt
, &rel_type
, &rel_code
,
675 &rel_msg
, &rel_info
, offset
);
679 if (rel_msg
&& pskb_may_pull(skb
, offset
+ sizeof(struct ipv6hdr
))) {
681 struct sk_buff
*skb2
= skb_clone(skb
, GFP_ATOMIC
);
687 skb_pull(skb2
, offset
);
688 skb_reset_network_header(skb2
);
690 /* Try to guess incoming interface */
691 rt
= rt6_lookup(dev_net(skb
->dev
), &ipv6_hdr(skb2
)->saddr
,
694 if (rt
&& rt
->dst
.dev
)
695 skb2
->dev
= rt
->dst
.dev
;
697 icmpv6_send(skb2
, rel_type
, rel_code
, rel_info
);
707 static int ip4ip6_dscp_ecn_decapsulate(const struct ip6_tnl
*t
,
708 const struct ipv6hdr
*ipv6h
,
711 __u8 dsfield
= ipv6_get_dsfield(ipv6h
) & ~INET_ECN_MASK
;
713 if (t
->parms
.flags
& IP6_TNL_F_RCV_DSCP_COPY
)
714 ipv4_change_dsfield(ip_hdr(skb
), INET_ECN_MASK
, dsfield
);
716 return IP6_ECN_decapsulate(ipv6h
, skb
);
719 static int ip6ip6_dscp_ecn_decapsulate(const struct ip6_tnl
*t
,
720 const struct ipv6hdr
*ipv6h
,
723 if (t
->parms
.flags
& IP6_TNL_F_RCV_DSCP_COPY
)
724 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h
), ipv6_hdr(skb
));
726 return IP6_ECN_decapsulate(ipv6h
, skb
);
729 __u32
ip6_tnl_get_cap(struct ip6_tnl
*t
,
730 const struct in6_addr
*laddr
,
731 const struct in6_addr
*raddr
)
733 struct __ip6_tnl_parm
*p
= &t
->parms
;
734 int ltype
= ipv6_addr_type(laddr
);
735 int rtype
= ipv6_addr_type(raddr
);
738 if (ltype
== IPV6_ADDR_ANY
|| rtype
== IPV6_ADDR_ANY
) {
739 flags
= IP6_TNL_F_CAP_PER_PACKET
;
740 } else if (ltype
& (IPV6_ADDR_UNICAST
|IPV6_ADDR_MULTICAST
) &&
741 rtype
& (IPV6_ADDR_UNICAST
|IPV6_ADDR_MULTICAST
) &&
742 !((ltype
|rtype
) & IPV6_ADDR_LOOPBACK
) &&
743 (!((ltype
|rtype
) & IPV6_ADDR_LINKLOCAL
) || p
->link
)) {
744 if (ltype
&IPV6_ADDR_UNICAST
)
745 flags
|= IP6_TNL_F_CAP_XMIT
;
746 if (rtype
&IPV6_ADDR_UNICAST
)
747 flags
|= IP6_TNL_F_CAP_RCV
;
751 EXPORT_SYMBOL(ip6_tnl_get_cap
);
753 /* called with rcu_read_lock() */
754 int ip6_tnl_rcv_ctl(struct ip6_tnl
*t
,
755 const struct in6_addr
*laddr
,
756 const struct in6_addr
*raddr
)
758 struct __ip6_tnl_parm
*p
= &t
->parms
;
760 struct net
*net
= t
->net
;
762 if ((p
->flags
& IP6_TNL_F_CAP_RCV
) ||
763 ((p
->flags
& IP6_TNL_F_CAP_PER_PACKET
) &&
764 (ip6_tnl_get_cap(t
, laddr
, raddr
) & IP6_TNL_F_CAP_RCV
))) {
765 struct net_device
*ldev
= NULL
;
768 ldev
= dev_get_by_index_rcu(net
, p
->link
);
770 if ((ipv6_addr_is_multicast(laddr
) ||
771 likely(ipv6_chk_addr(net
, laddr
, ldev
, 0))) &&
772 likely(!ipv6_chk_addr(net
, raddr
, NULL
, 0)))
777 EXPORT_SYMBOL_GPL(ip6_tnl_rcv_ctl
);
779 static int __ip6_tnl_rcv(struct ip6_tnl
*tunnel
, struct sk_buff
*skb
,
780 const struct tnl_ptk_info
*tpi
,
781 struct metadata_dst
*tun_dst
,
782 int (*dscp_ecn_decapsulate
)(const struct ip6_tnl
*t
,
783 const struct ipv6hdr
*ipv6h
,
784 struct sk_buff
*skb
),
787 struct pcpu_sw_netstats
*tstats
;
788 const struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
791 if ((!(tpi
->flags
& TUNNEL_CSUM
) &&
792 (tunnel
->parms
.i_flags
& TUNNEL_CSUM
)) ||
793 ((tpi
->flags
& TUNNEL_CSUM
) &&
794 !(tunnel
->parms
.i_flags
& TUNNEL_CSUM
))) {
795 tunnel
->dev
->stats
.rx_crc_errors
++;
796 tunnel
->dev
->stats
.rx_errors
++;
800 if (tunnel
->parms
.i_flags
& TUNNEL_SEQ
) {
801 if (!(tpi
->flags
& TUNNEL_SEQ
) ||
803 (s32
)(ntohl(tpi
->seq
) - tunnel
->i_seqno
) < 0)) {
804 tunnel
->dev
->stats
.rx_fifo_errors
++;
805 tunnel
->dev
->stats
.rx_errors
++;
808 tunnel
->i_seqno
= ntohl(tpi
->seq
) + 1;
811 skb
->protocol
= tpi
->proto
;
813 /* Warning: All skb pointers will be invalidated! */
814 if (tunnel
->dev
->type
== ARPHRD_ETHER
) {
815 if (!pskb_may_pull(skb
, ETH_HLEN
)) {
816 tunnel
->dev
->stats
.rx_length_errors
++;
817 tunnel
->dev
->stats
.rx_errors
++;
821 ipv6h
= ipv6_hdr(skb
);
822 skb
->protocol
= eth_type_trans(skb
, tunnel
->dev
);
823 skb_postpull_rcsum(skb
, eth_hdr(skb
), ETH_HLEN
);
825 skb
->dev
= tunnel
->dev
;
828 skb_reset_network_header(skb
);
829 memset(skb
->cb
, 0, sizeof(struct inet6_skb_parm
));
831 __skb_tunnel_rx(skb
, tunnel
->dev
, tunnel
->net
);
833 err
= dscp_ecn_decapsulate(tunnel
, ipv6h
, skb
);
836 net_info_ratelimited("non-ECT from %pI6 with DS=%#x\n",
838 ipv6_get_dsfield(ipv6h
));
840 ++tunnel
->dev
->stats
.rx_frame_errors
;
841 ++tunnel
->dev
->stats
.rx_errors
;
846 tstats
= this_cpu_ptr(tunnel
->dev
->tstats
);
847 u64_stats_update_begin(&tstats
->syncp
);
848 tstats
->rx_packets
++;
849 tstats
->rx_bytes
+= skb
->len
;
850 u64_stats_update_end(&tstats
->syncp
);
852 skb_scrub_packet(skb
, !net_eq(tunnel
->net
, dev_net(tunnel
->dev
)));
855 skb_dst_set(skb
, (struct dst_entry
*)tun_dst
);
857 gro_cells_receive(&tunnel
->gro_cells
, skb
);
862 dst_release((struct dst_entry
*)tun_dst
);
867 int ip6_tnl_rcv(struct ip6_tnl
*t
, struct sk_buff
*skb
,
868 const struct tnl_ptk_info
*tpi
,
869 struct metadata_dst
*tun_dst
,
872 return __ip6_tnl_rcv(t
, skb
, tpi
, NULL
, ip6ip6_dscp_ecn_decapsulate
,
875 EXPORT_SYMBOL(ip6_tnl_rcv
);
877 static const struct tnl_ptk_info tpi_v6
= {
878 /* no tunnel info required for ipxip6. */
879 .proto
= htons(ETH_P_IPV6
),
882 static const struct tnl_ptk_info tpi_v4
= {
883 /* no tunnel info required for ipxip6. */
884 .proto
= htons(ETH_P_IP
),
887 static int ipxip6_rcv(struct sk_buff
*skb
, u8 ipproto
,
888 const struct tnl_ptk_info
*tpi
,
889 int (*dscp_ecn_decapsulate
)(const struct ip6_tnl
*t
,
890 const struct ipv6hdr
*ipv6h
,
891 struct sk_buff
*skb
))
894 const struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
895 struct metadata_dst
*tun_dst
= NULL
;
899 t
= ip6_tnl_lookup(dev_net(skb
->dev
), &ipv6h
->saddr
, &ipv6h
->daddr
);
902 u8 tproto
= ACCESS_ONCE(t
->parms
.proto
);
904 if (tproto
!= ipproto
&& tproto
!= 0)
906 if (!xfrm6_policy_check(NULL
, XFRM_POLICY_IN
, skb
))
908 if (!ip6_tnl_rcv_ctl(t
, &ipv6h
->daddr
, &ipv6h
->saddr
))
910 if (iptunnel_pull_header(skb
, 0, tpi
->proto
, false))
912 if (t
->parms
.collect_md
) {
913 tun_dst
= ipv6_tun_rx_dst(skb
, 0, 0, 0);
917 ret
= __ip6_tnl_rcv(t
, skb
, tpi
, tun_dst
, dscp_ecn_decapsulate
,
931 static int ip4ip6_rcv(struct sk_buff
*skb
)
933 return ipxip6_rcv(skb
, IPPROTO_IPIP
, &tpi_v4
,
934 ip4ip6_dscp_ecn_decapsulate
);
937 static int ip6ip6_rcv(struct sk_buff
*skb
)
939 return ipxip6_rcv(skb
, IPPROTO_IPV6
, &tpi_v6
,
940 ip6ip6_dscp_ecn_decapsulate
);
943 struct ipv6_tel_txoption
{
944 struct ipv6_txoptions ops
;
948 static void init_tel_txopt(struct ipv6_tel_txoption
*opt
, __u8 encap_limit
)
950 memset(opt
, 0, sizeof(struct ipv6_tel_txoption
));
952 opt
->dst_opt
[2] = IPV6_TLV_TNL_ENCAP_LIMIT
;
954 opt
->dst_opt
[4] = encap_limit
;
955 opt
->dst_opt
[5] = IPV6_TLV_PADN
;
958 opt
->ops
.dst1opt
= (struct ipv6_opt_hdr
*) opt
->dst_opt
;
959 opt
->ops
.opt_nflen
= 8;
963 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
964 * @t: the outgoing tunnel device
965 * @hdr: IPv6 header from the incoming packet
968 * Avoid trivial tunneling loop by checking that tunnel exit-point
969 * doesn't match source of incoming packet.
977 ip6_tnl_addr_conflict(const struct ip6_tnl
*t
, const struct ipv6hdr
*hdr
)
979 return ipv6_addr_equal(&t
->parms
.raddr
, &hdr
->saddr
);
982 int ip6_tnl_xmit_ctl(struct ip6_tnl
*t
,
983 const struct in6_addr
*laddr
,
984 const struct in6_addr
*raddr
)
986 struct __ip6_tnl_parm
*p
= &t
->parms
;
988 struct net
*net
= t
->net
;
990 if ((p
->flags
& IP6_TNL_F_CAP_XMIT
) ||
991 ((p
->flags
& IP6_TNL_F_CAP_PER_PACKET
) &&
992 (ip6_tnl_get_cap(t
, laddr
, raddr
) & IP6_TNL_F_CAP_XMIT
))) {
993 struct net_device
*ldev
= NULL
;
997 ldev
= dev_get_by_index_rcu(net
, p
->link
);
999 if (unlikely(!ipv6_chk_addr(net
, laddr
, ldev
, 0)))
1000 pr_warn("%s xmit: Local address not yet configured!\n",
1002 else if (!ipv6_addr_is_multicast(raddr
) &&
1003 unlikely(ipv6_chk_addr(net
, raddr
, NULL
, 0)))
1004 pr_warn("%s xmit: Routing loop! Remote address found on this node!\n",
1012 EXPORT_SYMBOL_GPL(ip6_tnl_xmit_ctl
);
1015 * ip6_tnl_xmit - encapsulate packet and send
1016 * @skb: the outgoing socket buffer
1017 * @dev: the outgoing tunnel device
1018 * @dsfield: dscp code for outer header
1019 * @fl6: flow of tunneled packet
1020 * @encap_limit: encapsulation limit
1021 * @pmtu: Path MTU is stored if packet is too big
1022 * @proto: next header value
1025 * Build new header and do some sanity checks on the packet before sending
1031 * %-EMSGSIZE message too big. return mtu in this case.
1034 int ip6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
, __u8 dsfield
,
1035 struct flowi6
*fl6
, int encap_limit
, __u32
*pmtu
,
1038 struct ip6_tnl
*t
= netdev_priv(dev
);
1039 struct net
*net
= t
->net
;
1040 struct net_device_stats
*stats
= &t
->dev
->stats
;
1041 struct ipv6hdr
*ipv6h
;
1042 struct ipv6_tel_txoption opt
;
1043 struct dst_entry
*dst
= NULL
, *ndst
= NULL
;
1044 struct net_device
*tdev
;
1046 unsigned int eth_hlen
= t
->dev
->type
== ARPHRD_ETHER
? ETH_HLEN
: 0;
1047 unsigned int psh_hlen
= sizeof(struct ipv6hdr
) + t
->encap_hlen
;
1048 unsigned int max_headroom
= psh_hlen
;
1049 bool use_cache
= false;
1053 if (t
->parms
.collect_md
) {
1054 hop_limit
= skb_tunnel_info(skb
)->key
.ttl
;
1057 hop_limit
= t
->parms
.hop_limit
;
1061 if (ipv6_addr_any(&t
->parms
.raddr
)) {
1062 if (skb
->protocol
== htons(ETH_P_IPV6
)) {
1063 struct in6_addr
*addr6
;
1064 struct neighbour
*neigh
;
1068 goto tx_err_link_failure
;
1070 neigh
= dst_neigh_lookup(skb_dst(skb
),
1071 &ipv6_hdr(skb
)->daddr
);
1073 goto tx_err_link_failure
;
1075 addr6
= (struct in6_addr
*)&neigh
->primary_key
;
1076 addr_type
= ipv6_addr_type(addr6
);
1078 if (addr_type
== IPV6_ADDR_ANY
)
1079 addr6
= &ipv6_hdr(skb
)->daddr
;
1081 memcpy(&fl6
->daddr
, addr6
, sizeof(fl6
->daddr
));
1082 neigh_release(neigh
);
1084 } else if (!(t
->parms
.flags
&
1085 (IP6_TNL_F_USE_ORIG_TCLASS
| IP6_TNL_F_USE_ORIG_FWMARK
))) {
1086 /* enable the cache only only if the routing decision does
1087 * not depend on the current inner header value
1093 dst
= dst_cache_get(&t
->dst_cache
);
1095 if (!ip6_tnl_xmit_ctl(t
, &fl6
->saddr
, &fl6
->daddr
))
1096 goto tx_err_link_failure
;
1100 /* add dsfield to flowlabel for route lookup */
1101 fl6
->flowlabel
= ip6_make_flowinfo(dsfield
, fl6
->flowlabel
);
1103 dst
= ip6_route_output(net
, NULL
, fl6
);
1106 goto tx_err_link_failure
;
1107 dst
= xfrm_lookup(net
, dst
, flowi6_to_flowi(fl6
), NULL
, 0);
1111 goto tx_err_link_failure
;
1113 if (t
->parms
.collect_md
&&
1114 ipv6_dev_get_saddr(net
, ip6_dst_idev(dst
)->dev
,
1115 &fl6
->daddr
, 0, &fl6
->saddr
))
1116 goto tx_err_link_failure
;
1123 stats
->collisions
++;
1124 net_warn_ratelimited("%s: Local routing loop detected!\n",
1126 goto tx_err_dst_release
;
1128 mtu
= dst_mtu(dst
) - eth_hlen
- psh_hlen
- t
->tun_hlen
;
1129 if (encap_limit
>= 0) {
1133 if (mtu
< IPV6_MIN_MTU
)
1135 if (skb_dst(skb
) && !t
->parms
.collect_md
)
1136 skb_dst(skb
)->ops
->update_pmtu(skb_dst(skb
), NULL
, skb
, mtu
);
1137 if (skb
->len
- t
->tun_hlen
- eth_hlen
> mtu
&& !skb_is_gso(skb
)) {
1140 goto tx_err_dst_release
;
1143 if (t
->err_count
> 0) {
1144 if (time_before(jiffies
,
1145 t
->err_time
+ IP6TUNNEL_ERR_TIMEO
)) {
1148 dst_link_failure(skb
);
1154 skb_scrub_packet(skb
, !net_eq(t
->net
, dev_net(dev
)));
1157 * Okay, now see if we can stuff it in the buffer as-is.
1159 max_headroom
+= LL_RESERVED_SPACE(tdev
);
1161 if (skb_headroom(skb
) < max_headroom
|| skb_shared(skb
) ||
1162 (skb_cloned(skb
) && !skb_clone_writable(skb
, 0))) {
1163 struct sk_buff
*new_skb
;
1165 new_skb
= skb_realloc_headroom(skb
, max_headroom
);
1167 goto tx_err_dst_release
;
1170 skb_set_owner_w(new_skb
, skb
->sk
);
1175 if (t
->parms
.collect_md
) {
1176 if (t
->encap
.type
!= TUNNEL_ENCAP_NONE
)
1177 goto tx_err_dst_release
;
1179 if (use_cache
&& ndst
)
1180 dst_cache_set_ip6(&t
->dst_cache
, ndst
, &fl6
->saddr
);
1182 skb_dst_set(skb
, dst
);
1184 if (encap_limit
>= 0) {
1185 init_tel_txopt(&opt
, encap_limit
);
1186 ipv6_push_frag_opts(skb
, &opt
.ops
, &proto
);
1189 /* Calculate max headroom for all the headers and adjust
1190 * needed_headroom if necessary.
1192 max_headroom
= LL_RESERVED_SPACE(dst
->dev
) + sizeof(struct ipv6hdr
)
1193 + dst
->header_len
+ t
->hlen
;
1194 if (max_headroom
> dev
->needed_headroom
)
1195 dev
->needed_headroom
= max_headroom
;
1197 err
= ip6_tnl_encap(skb
, t
, &proto
, fl6
);
1201 skb_push(skb
, sizeof(struct ipv6hdr
));
1202 skb_reset_network_header(skb
);
1203 ipv6h
= ipv6_hdr(skb
);
1204 ip6_flow_hdr(ipv6h
, dsfield
,
1205 ip6_make_flowlabel(net
, skb
, fl6
->flowlabel
, true, fl6
));
1206 ipv6h
->hop_limit
= hop_limit
;
1207 ipv6h
->nexthdr
= proto
;
1208 ipv6h
->saddr
= fl6
->saddr
;
1209 ipv6h
->daddr
= fl6
->daddr
;
1210 ip6tunnel_xmit(NULL
, skb
, dev
);
1212 tx_err_link_failure
:
1213 stats
->tx_carrier_errors
++;
1214 dst_link_failure(skb
);
1219 EXPORT_SYMBOL(ip6_tnl_xmit
);
1222 ip4ip6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1224 struct ip6_tnl
*t
= netdev_priv(dev
);
1225 const struct iphdr
*iph
= ip_hdr(skb
);
1226 int encap_limit
= -1;
1233 memset(&(IPCB(skb
)->opt
), 0, sizeof(IPCB(skb
)->opt
));
1235 tproto
= ACCESS_ONCE(t
->parms
.proto
);
1236 if (tproto
!= IPPROTO_IPIP
&& tproto
!= 0)
1239 if (t
->parms
.collect_md
) {
1240 struct ip_tunnel_info
*tun_info
;
1241 const struct ip_tunnel_key
*key
;
1243 tun_info
= skb_tunnel_info(skb
);
1244 if (unlikely(!tun_info
|| !(tun_info
->mode
& IP_TUNNEL_INFO_TX
) ||
1245 ip_tunnel_info_af(tun_info
) != AF_INET6
))
1247 key
= &tun_info
->key
;
1248 memset(&fl6
, 0, sizeof(fl6
));
1249 fl6
.flowi6_proto
= IPPROTO_IPIP
;
1250 fl6
.daddr
= key
->u
.ipv6
.dst
;
1251 fl6
.flowlabel
= key
->label
;
1254 if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1255 encap_limit
= t
->parms
.encap_limit
;
1257 memcpy(&fl6
, &t
->fl
.u
.ip6
, sizeof(fl6
));
1258 fl6
.flowi6_proto
= IPPROTO_IPIP
;
1260 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_TCLASS
)
1261 dsfield
= ipv4_get_dsfield(iph
);
1263 dsfield
= ip6_tclass(t
->parms
.flowinfo
);
1264 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_FWMARK
)
1265 fl6
.flowi6_mark
= skb
->mark
;
1267 fl6
.flowi6_mark
= t
->parms
.fwmark
;
1270 fl6
.flowi6_uid
= sock_net_uid(dev_net(dev
), NULL
);
1272 if (iptunnel_handle_offloads(skb
, SKB_GSO_IPXIP6
))
1275 dsfield
= INET_ECN_encapsulate(dsfield
, ipv4_get_dsfield(iph
));
1277 skb_set_inner_ipproto(skb
, IPPROTO_IPIP
);
1279 err
= ip6_tnl_xmit(skb
, dev
, dsfield
, &fl6
, encap_limit
, &mtu
,
1282 /* XXX: send ICMP error even if DF is not set. */
1283 if (err
== -EMSGSIZE
)
1284 icmp_send(skb
, ICMP_DEST_UNREACH
, ICMP_FRAG_NEEDED
,
1293 ip6ip6_tnl_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1295 struct ip6_tnl
*t
= netdev_priv(dev
);
1296 struct ipv6hdr
*ipv6h
= ipv6_hdr(skb
);
1297 int encap_limit
= -1;
1305 tproto
= ACCESS_ONCE(t
->parms
.proto
);
1306 if ((tproto
!= IPPROTO_IPV6
&& tproto
!= 0) ||
1307 ip6_tnl_addr_conflict(t
, ipv6h
))
1310 if (t
->parms
.collect_md
) {
1311 struct ip_tunnel_info
*tun_info
;
1312 const struct ip_tunnel_key
*key
;
1314 tun_info
= skb_tunnel_info(skb
);
1315 if (unlikely(!tun_info
|| !(tun_info
->mode
& IP_TUNNEL_INFO_TX
) ||
1316 ip_tunnel_info_af(tun_info
) != AF_INET6
))
1318 key
= &tun_info
->key
;
1319 memset(&fl6
, 0, sizeof(fl6
));
1320 fl6
.flowi6_proto
= IPPROTO_IPV6
;
1321 fl6
.daddr
= key
->u
.ipv6
.dst
;
1322 fl6
.flowlabel
= key
->label
;
1325 offset
= ip6_tnl_parse_tlv_enc_lim(skb
, skb_network_header(skb
));
1326 /* ip6_tnl_parse_tlv_enc_lim() might have reallocated skb->head */
1327 ipv6h
= ipv6_hdr(skb
);
1329 struct ipv6_tlv_tnl_enc_lim
*tel
;
1331 tel
= (void *)&skb_network_header(skb
)[offset
];
1332 if (tel
->encap_limit
== 0) {
1333 icmpv6_send(skb
, ICMPV6_PARAMPROB
,
1334 ICMPV6_HDR_FIELD
, offset
+ 2);
1337 encap_limit
= tel
->encap_limit
- 1;
1338 } else if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
)) {
1339 encap_limit
= t
->parms
.encap_limit
;
1342 memcpy(&fl6
, &t
->fl
.u
.ip6
, sizeof(fl6
));
1343 fl6
.flowi6_proto
= IPPROTO_IPV6
;
1345 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_TCLASS
)
1346 dsfield
= ipv6_get_dsfield(ipv6h
);
1348 dsfield
= ip6_tclass(t
->parms
.flowinfo
);
1349 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_FLOWLABEL
)
1350 fl6
.flowlabel
|= ip6_flowlabel(ipv6h
);
1351 if (t
->parms
.flags
& IP6_TNL_F_USE_ORIG_FWMARK
)
1352 fl6
.flowi6_mark
= skb
->mark
;
1354 fl6
.flowi6_mark
= t
->parms
.fwmark
;
1357 fl6
.flowi6_uid
= sock_net_uid(dev_net(dev
), NULL
);
1359 if (iptunnel_handle_offloads(skb
, SKB_GSO_IPXIP6
))
1362 dsfield
= INET_ECN_encapsulate(dsfield
, ipv6_get_dsfield(ipv6h
));
1364 skb_set_inner_ipproto(skb
, IPPROTO_IPV6
);
1366 err
= ip6_tnl_xmit(skb
, dev
, dsfield
, &fl6
, encap_limit
, &mtu
,
1369 if (err
== -EMSGSIZE
)
1370 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
1378 ip6_tnl_start_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1380 struct ip6_tnl
*t
= netdev_priv(dev
);
1381 struct net_device_stats
*stats
= &t
->dev
->stats
;
1384 switch (skb
->protocol
) {
1385 case htons(ETH_P_IP
):
1386 ret
= ip4ip6_tnl_xmit(skb
, dev
);
1388 case htons(ETH_P_IPV6
):
1389 ret
= ip6ip6_tnl_xmit(skb
, dev
);
1398 return NETDEV_TX_OK
;
1402 stats
->tx_dropped
++;
1404 return NETDEV_TX_OK
;
1407 static void ip6_tnl_link_config(struct ip6_tnl
*t
)
1409 struct net_device
*dev
= t
->dev
;
1410 struct __ip6_tnl_parm
*p
= &t
->parms
;
1411 struct flowi6
*fl6
= &t
->fl
.u
.ip6
;
1414 memcpy(dev
->dev_addr
, &p
->laddr
, sizeof(struct in6_addr
));
1415 memcpy(dev
->broadcast
, &p
->raddr
, sizeof(struct in6_addr
));
1417 /* Set up flowi template */
1418 fl6
->saddr
= p
->laddr
;
1419 fl6
->daddr
= p
->raddr
;
1420 fl6
->flowi6_oif
= p
->link
;
1423 if (!(p
->flags
&IP6_TNL_F_USE_ORIG_TCLASS
))
1424 fl6
->flowlabel
|= IPV6_TCLASS_MASK
& p
->flowinfo
;
1425 if (!(p
->flags
&IP6_TNL_F_USE_ORIG_FLOWLABEL
))
1426 fl6
->flowlabel
|= IPV6_FLOWLABEL_MASK
& p
->flowinfo
;
1428 p
->flags
&= ~(IP6_TNL_F_CAP_XMIT
|IP6_TNL_F_CAP_RCV
|IP6_TNL_F_CAP_PER_PACKET
);
1429 p
->flags
|= ip6_tnl_get_cap(t
, &p
->laddr
, &p
->raddr
);
1431 if (p
->flags
&IP6_TNL_F_CAP_XMIT
&& p
->flags
&IP6_TNL_F_CAP_RCV
)
1432 dev
->flags
|= IFF_POINTOPOINT
;
1434 dev
->flags
&= ~IFF_POINTOPOINT
;
1437 t
->hlen
= t
->encap_hlen
+ t
->tun_hlen
;
1438 t_hlen
= t
->hlen
+ sizeof(struct ipv6hdr
);
1440 if (p
->flags
& IP6_TNL_F_CAP_XMIT
) {
1441 int strict
= (ipv6_addr_type(&p
->raddr
) &
1442 (IPV6_ADDR_MULTICAST
|IPV6_ADDR_LINKLOCAL
));
1444 struct rt6_info
*rt
= rt6_lookup(t
->net
,
1445 &p
->raddr
, &p
->laddr
,
1452 dev
->hard_header_len
= rt
->dst
.dev
->hard_header_len
+
1455 dev
->mtu
= rt
->dst
.dev
->mtu
- t_hlen
;
1456 if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1459 if (dev
->mtu
< IPV6_MIN_MTU
)
1460 dev
->mtu
= IPV6_MIN_MTU
;
1467 * ip6_tnl_change - update the tunnel parameters
1468 * @t: tunnel to be changed
1469 * @p: tunnel configuration parameters
1472 * ip6_tnl_change() updates the tunnel parameters
1476 ip6_tnl_change(struct ip6_tnl
*t
, const struct __ip6_tnl_parm
*p
)
1478 t
->parms
.laddr
= p
->laddr
;
1479 t
->parms
.raddr
= p
->raddr
;
1480 t
->parms
.flags
= p
->flags
;
1481 t
->parms
.hop_limit
= p
->hop_limit
;
1482 t
->parms
.encap_limit
= p
->encap_limit
;
1483 t
->parms
.flowinfo
= p
->flowinfo
;
1484 t
->parms
.link
= p
->link
;
1485 t
->parms
.proto
= p
->proto
;
1486 t
->parms
.fwmark
= p
->fwmark
;
1487 dst_cache_reset(&t
->dst_cache
);
1488 ip6_tnl_link_config(t
);
1492 static int ip6_tnl_update(struct ip6_tnl
*t
, struct __ip6_tnl_parm
*p
)
1494 struct net
*net
= t
->net
;
1495 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1498 ip6_tnl_unlink(ip6n
, t
);
1500 err
= ip6_tnl_change(t
, p
);
1501 ip6_tnl_link(ip6n
, t
);
1502 netdev_state_change(t
->dev
);
1506 static int ip6_tnl0_update(struct ip6_tnl
*t
, struct __ip6_tnl_parm
*p
)
1508 /* for default tnl0 device allow to change only the proto */
1509 t
->parms
.proto
= p
->proto
;
1510 netdev_state_change(t
->dev
);
1515 ip6_tnl_parm_from_user(struct __ip6_tnl_parm
*p
, const struct ip6_tnl_parm
*u
)
1517 p
->laddr
= u
->laddr
;
1518 p
->raddr
= u
->raddr
;
1519 p
->flags
= u
->flags
;
1520 p
->hop_limit
= u
->hop_limit
;
1521 p
->encap_limit
= u
->encap_limit
;
1522 p
->flowinfo
= u
->flowinfo
;
1524 p
->proto
= u
->proto
;
1525 memcpy(p
->name
, u
->name
, sizeof(u
->name
));
1529 ip6_tnl_parm_to_user(struct ip6_tnl_parm
*u
, const struct __ip6_tnl_parm
*p
)
1531 u
->laddr
= p
->laddr
;
1532 u
->raddr
= p
->raddr
;
1533 u
->flags
= p
->flags
;
1534 u
->hop_limit
= p
->hop_limit
;
1535 u
->encap_limit
= p
->encap_limit
;
1536 u
->flowinfo
= p
->flowinfo
;
1538 u
->proto
= p
->proto
;
1539 memcpy(u
->name
, p
->name
, sizeof(u
->name
));
1543 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1544 * @dev: virtual device associated with tunnel
1545 * @ifr: parameters passed from userspace
1546 * @cmd: command to be performed
1549 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1552 * The possible commands are the following:
1553 * %SIOCGETTUNNEL: get tunnel parameters for device
1554 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1555 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1556 * %SIOCDELTUNNEL: delete tunnel
1558 * The fallback device "ip6tnl0", created during module
1559 * initialization, can be used for creating other tunnel devices.
1563 * %-EFAULT if unable to copy data to or from userspace,
1564 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1565 * %-EINVAL if passed tunnel parameters are invalid,
1566 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1567 * %-ENODEV if attempting to change or delete a nonexisting device
1571 ip6_tnl_ioctl(struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
1574 struct ip6_tnl_parm p
;
1575 struct __ip6_tnl_parm p1
;
1576 struct ip6_tnl
*t
= netdev_priv(dev
);
1577 struct net
*net
= t
->net
;
1578 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1580 memset(&p1
, 0, sizeof(p1
));
1584 if (dev
== ip6n
->fb_tnl_dev
) {
1585 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
))) {
1589 ip6_tnl_parm_from_user(&p1
, &p
);
1590 t
= ip6_tnl_locate(net
, &p1
, 0);
1592 t
= netdev_priv(dev
);
1594 memset(&p
, 0, sizeof(p
));
1596 ip6_tnl_parm_to_user(&p
, &t
->parms
);
1597 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof(p
))) {
1604 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1607 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
1610 if (p
.proto
!= IPPROTO_IPV6
&& p
.proto
!= IPPROTO_IPIP
&&
1613 ip6_tnl_parm_from_user(&p1
, &p
);
1614 t
= ip6_tnl_locate(net
, &p1
, cmd
== SIOCADDTUNNEL
);
1615 if (cmd
== SIOCCHGTUNNEL
) {
1617 if (t
->dev
!= dev
) {
1622 t
= netdev_priv(dev
);
1623 if (dev
== ip6n
->fb_tnl_dev
)
1624 err
= ip6_tnl0_update(t
, &p1
);
1626 err
= ip6_tnl_update(t
, &p1
);
1630 ip6_tnl_parm_to_user(&p
, &t
->parms
);
1631 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
, sizeof(p
)))
1640 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1643 if (dev
== ip6n
->fb_tnl_dev
) {
1645 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
1648 ip6_tnl_parm_from_user(&p1
, &p
);
1649 t
= ip6_tnl_locate(net
, &p1
, 0);
1653 if (t
->dev
== ip6n
->fb_tnl_dev
)
1658 unregister_netdevice(dev
);
1667 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1668 * @dev: virtual device associated with tunnel
1669 * @new_mtu: the new mtu
1673 * %-EINVAL if mtu too small
1676 int ip6_tnl_change_mtu(struct net_device
*dev
, int new_mtu
)
1678 struct ip6_tnl
*tnl
= netdev_priv(dev
);
1680 if (tnl
->parms
.proto
== IPPROTO_IPIP
) {
1681 if (new_mtu
< ETH_MIN_MTU
)
1684 if (new_mtu
< IPV6_MIN_MTU
)
1687 if (new_mtu
> 0xFFF8 - dev
->hard_header_len
)
1692 EXPORT_SYMBOL(ip6_tnl_change_mtu
);
1694 int ip6_tnl_get_iflink(const struct net_device
*dev
)
1696 struct ip6_tnl
*t
= netdev_priv(dev
);
1698 return t
->parms
.link
;
1700 EXPORT_SYMBOL(ip6_tnl_get_iflink
);
1702 int ip6_tnl_encap_add_ops(const struct ip6_tnl_encap_ops
*ops
,
1705 if (num
>= MAX_IPTUN_ENCAP_OPS
)
1708 return !cmpxchg((const struct ip6_tnl_encap_ops
**)
1709 &ip6tun_encaps
[num
],
1710 NULL
, ops
) ? 0 : -1;
1712 EXPORT_SYMBOL(ip6_tnl_encap_add_ops
);
1714 int ip6_tnl_encap_del_ops(const struct ip6_tnl_encap_ops
*ops
,
1719 if (num
>= MAX_IPTUN_ENCAP_OPS
)
1722 ret
= (cmpxchg((const struct ip6_tnl_encap_ops
**)
1723 &ip6tun_encaps
[num
],
1724 ops
, NULL
) == ops
) ? 0 : -1;
1730 EXPORT_SYMBOL(ip6_tnl_encap_del_ops
);
1732 int ip6_tnl_encap_setup(struct ip6_tnl
*t
,
1733 struct ip_tunnel_encap
*ipencap
)
1737 memset(&t
->encap
, 0, sizeof(t
->encap
));
1739 hlen
= ip6_encap_hlen(ipencap
);
1743 t
->encap
.type
= ipencap
->type
;
1744 t
->encap
.sport
= ipencap
->sport
;
1745 t
->encap
.dport
= ipencap
->dport
;
1746 t
->encap
.flags
= ipencap
->flags
;
1748 t
->encap_hlen
= hlen
;
1749 t
->hlen
= t
->encap_hlen
+ t
->tun_hlen
;
1753 EXPORT_SYMBOL_GPL(ip6_tnl_encap_setup
);
1755 static const struct net_device_ops ip6_tnl_netdev_ops
= {
1756 .ndo_init
= ip6_tnl_dev_init
,
1757 .ndo_uninit
= ip6_tnl_dev_uninit
,
1758 .ndo_start_xmit
= ip6_tnl_start_xmit
,
1759 .ndo_do_ioctl
= ip6_tnl_ioctl
,
1760 .ndo_change_mtu
= ip6_tnl_change_mtu
,
1761 .ndo_get_stats
= ip6_get_stats
,
1762 .ndo_get_iflink
= ip6_tnl_get_iflink
,
1765 #define IPXIPX_FEATURES (NETIF_F_SG | \
1766 NETIF_F_FRAGLIST | \
1768 NETIF_F_GSO_SOFTWARE | \
1772 * ip6_tnl_dev_setup - setup virtual tunnel device
1773 * @dev: virtual device associated with tunnel
1776 * Initialize function pointers and device parameters
1779 static void ip6_tnl_dev_setup(struct net_device
*dev
)
1781 dev
->netdev_ops
= &ip6_tnl_netdev_ops
;
1782 dev
->needs_free_netdev
= true;
1783 dev
->priv_destructor
= ip6_dev_free
;
1785 dev
->type
= ARPHRD_TUNNEL6
;
1786 dev
->flags
|= IFF_NOARP
;
1787 dev
->addr_len
= sizeof(struct in6_addr
);
1788 dev
->features
|= NETIF_F_LLTX
;
1789 netif_keep_dst(dev
);
1791 dev
->features
|= IPXIPX_FEATURES
;
1792 dev
->hw_features
|= IPXIPX_FEATURES
;
1794 /* This perm addr will be used as interface identifier by IPv6 */
1795 dev
->addr_assign_type
= NET_ADDR_RANDOM
;
1796 eth_random_addr(dev
->perm_addr
);
1801 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1802 * @dev: virtual device associated with tunnel
1806 ip6_tnl_dev_init_gen(struct net_device
*dev
)
1808 struct ip6_tnl
*t
= netdev_priv(dev
);
1813 t
->net
= dev_net(dev
);
1814 dev
->tstats
= netdev_alloc_pcpu_stats(struct pcpu_sw_netstats
);
1818 ret
= dst_cache_init(&t
->dst_cache
, GFP_KERNEL
);
1822 ret
= gro_cells_init(&t
->gro_cells
, dev
);
1827 t
->hlen
= t
->encap_hlen
+ t
->tun_hlen
;
1828 t_hlen
= t
->hlen
+ sizeof(struct ipv6hdr
);
1830 dev
->type
= ARPHRD_TUNNEL6
;
1831 dev
->hard_header_len
= LL_MAX_HEADER
+ t_hlen
;
1832 dev
->mtu
= ETH_DATA_LEN
- t_hlen
;
1833 if (!(t
->parms
.flags
& IP6_TNL_F_IGN_ENCAP_LIMIT
))
1835 dev
->min_mtu
= ETH_MIN_MTU
;
1836 dev
->max_mtu
= 0xFFF8 - dev
->hard_header_len
;
1841 dst_cache_destroy(&t
->dst_cache
);
1843 free_percpu(dev
->tstats
);
1850 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1851 * @dev: virtual device associated with tunnel
1854 static int ip6_tnl_dev_init(struct net_device
*dev
)
1856 struct ip6_tnl
*t
= netdev_priv(dev
);
1857 int err
= ip6_tnl_dev_init_gen(dev
);
1861 ip6_tnl_link_config(t
);
1862 if (t
->parms
.collect_md
) {
1863 dev
->features
|= NETIF_F_NETNS_LOCAL
;
1864 netif_keep_dst(dev
);
1870 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1871 * @dev: fallback device
1876 static int __net_init
ip6_fb_tnl_dev_init(struct net_device
*dev
)
1878 struct ip6_tnl
*t
= netdev_priv(dev
);
1879 struct net
*net
= dev_net(dev
);
1880 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1882 t
->parms
.proto
= IPPROTO_IPV6
;
1885 rcu_assign_pointer(ip6n
->tnls_wc
[0], t
);
1889 static int ip6_tnl_validate(struct nlattr
*tb
[], struct nlattr
*data
[],
1890 struct netlink_ext_ack
*extack
)
1894 if (!data
|| !data
[IFLA_IPTUN_PROTO
])
1897 proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1898 if (proto
!= IPPROTO_IPV6
&&
1899 proto
!= IPPROTO_IPIP
&&
1906 static void ip6_tnl_netlink_parms(struct nlattr
*data
[],
1907 struct __ip6_tnl_parm
*parms
)
1909 memset(parms
, 0, sizeof(*parms
));
1914 if (data
[IFLA_IPTUN_LINK
])
1915 parms
->link
= nla_get_u32(data
[IFLA_IPTUN_LINK
]);
1917 if (data
[IFLA_IPTUN_LOCAL
])
1918 parms
->laddr
= nla_get_in6_addr(data
[IFLA_IPTUN_LOCAL
]);
1920 if (data
[IFLA_IPTUN_REMOTE
])
1921 parms
->raddr
= nla_get_in6_addr(data
[IFLA_IPTUN_REMOTE
]);
1923 if (data
[IFLA_IPTUN_TTL
])
1924 parms
->hop_limit
= nla_get_u8(data
[IFLA_IPTUN_TTL
]);
1926 if (data
[IFLA_IPTUN_ENCAP_LIMIT
])
1927 parms
->encap_limit
= nla_get_u8(data
[IFLA_IPTUN_ENCAP_LIMIT
]);
1929 if (data
[IFLA_IPTUN_FLOWINFO
])
1930 parms
->flowinfo
= nla_get_be32(data
[IFLA_IPTUN_FLOWINFO
]);
1932 if (data
[IFLA_IPTUN_FLAGS
])
1933 parms
->flags
= nla_get_u32(data
[IFLA_IPTUN_FLAGS
]);
1935 if (data
[IFLA_IPTUN_PROTO
])
1936 parms
->proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1938 if (data
[IFLA_IPTUN_COLLECT_METADATA
])
1939 parms
->collect_md
= true;
1941 if (data
[IFLA_IPTUN_FWMARK
])
1942 parms
->fwmark
= nla_get_u32(data
[IFLA_IPTUN_FWMARK
]);
1945 static bool ip6_tnl_netlink_encap_parms(struct nlattr
*data
[],
1946 struct ip_tunnel_encap
*ipencap
)
1950 memset(ipencap
, 0, sizeof(*ipencap
));
1955 if (data
[IFLA_IPTUN_ENCAP_TYPE
]) {
1957 ipencap
->type
= nla_get_u16(data
[IFLA_IPTUN_ENCAP_TYPE
]);
1960 if (data
[IFLA_IPTUN_ENCAP_FLAGS
]) {
1962 ipencap
->flags
= nla_get_u16(data
[IFLA_IPTUN_ENCAP_FLAGS
]);
1965 if (data
[IFLA_IPTUN_ENCAP_SPORT
]) {
1967 ipencap
->sport
= nla_get_be16(data
[IFLA_IPTUN_ENCAP_SPORT
]);
1970 if (data
[IFLA_IPTUN_ENCAP_DPORT
]) {
1972 ipencap
->dport
= nla_get_be16(data
[IFLA_IPTUN_ENCAP_DPORT
]);
1978 static int ip6_tnl_newlink(struct net
*src_net
, struct net_device
*dev
,
1979 struct nlattr
*tb
[], struct nlattr
*data
[],
1980 struct netlink_ext_ack
*extack
)
1982 struct net
*net
= dev_net(dev
);
1983 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
1984 struct ip6_tnl
*nt
, *t
;
1985 struct ip_tunnel_encap ipencap
;
1987 nt
= netdev_priv(dev
);
1989 if (ip6_tnl_netlink_encap_parms(data
, &ipencap
)) {
1990 int err
= ip6_tnl_encap_setup(nt
, &ipencap
);
1996 ip6_tnl_netlink_parms(data
, &nt
->parms
);
1998 if (nt
->parms
.collect_md
) {
1999 if (rtnl_dereference(ip6n
->collect_md_tun
))
2002 t
= ip6_tnl_locate(net
, &nt
->parms
, 0);
2007 return ip6_tnl_create2(dev
);
2010 static int ip6_tnl_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
2011 struct nlattr
*data
[],
2012 struct netlink_ext_ack
*extack
)
2014 struct ip6_tnl
*t
= netdev_priv(dev
);
2015 struct __ip6_tnl_parm p
;
2016 struct net
*net
= t
->net
;
2017 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
2018 struct ip_tunnel_encap ipencap
;
2020 if (dev
== ip6n
->fb_tnl_dev
)
2023 if (ip6_tnl_netlink_encap_parms(data
, &ipencap
)) {
2024 int err
= ip6_tnl_encap_setup(t
, &ipencap
);
2029 ip6_tnl_netlink_parms(data
, &p
);
2033 t
= ip6_tnl_locate(net
, &p
, 0);
2038 t
= netdev_priv(dev
);
2040 return ip6_tnl_update(t
, &p
);
2043 static void ip6_tnl_dellink(struct net_device
*dev
, struct list_head
*head
)
2045 struct net
*net
= dev_net(dev
);
2046 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
2048 if (dev
!= ip6n
->fb_tnl_dev
)
2049 unregister_netdevice_queue(dev
, head
);
2052 static size_t ip6_tnl_get_size(const struct net_device
*dev
)
2055 /* IFLA_IPTUN_LINK */
2057 /* IFLA_IPTUN_LOCAL */
2058 nla_total_size(sizeof(struct in6_addr
)) +
2059 /* IFLA_IPTUN_REMOTE */
2060 nla_total_size(sizeof(struct in6_addr
)) +
2061 /* IFLA_IPTUN_TTL */
2063 /* IFLA_IPTUN_ENCAP_LIMIT */
2065 /* IFLA_IPTUN_FLOWINFO */
2067 /* IFLA_IPTUN_FLAGS */
2069 /* IFLA_IPTUN_PROTO */
2071 /* IFLA_IPTUN_ENCAP_TYPE */
2073 /* IFLA_IPTUN_ENCAP_FLAGS */
2075 /* IFLA_IPTUN_ENCAP_SPORT */
2077 /* IFLA_IPTUN_ENCAP_DPORT */
2079 /* IFLA_IPTUN_COLLECT_METADATA */
2081 /* IFLA_IPTUN_FWMARK */
2086 static int ip6_tnl_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
2088 struct ip6_tnl
*tunnel
= netdev_priv(dev
);
2089 struct __ip6_tnl_parm
*parm
= &tunnel
->parms
;
2091 if (nla_put_u32(skb
, IFLA_IPTUN_LINK
, parm
->link
) ||
2092 nla_put_in6_addr(skb
, IFLA_IPTUN_LOCAL
, &parm
->laddr
) ||
2093 nla_put_in6_addr(skb
, IFLA_IPTUN_REMOTE
, &parm
->raddr
) ||
2094 nla_put_u8(skb
, IFLA_IPTUN_TTL
, parm
->hop_limit
) ||
2095 nla_put_u8(skb
, IFLA_IPTUN_ENCAP_LIMIT
, parm
->encap_limit
) ||
2096 nla_put_be32(skb
, IFLA_IPTUN_FLOWINFO
, parm
->flowinfo
) ||
2097 nla_put_u32(skb
, IFLA_IPTUN_FLAGS
, parm
->flags
) ||
2098 nla_put_u8(skb
, IFLA_IPTUN_PROTO
, parm
->proto
) ||
2099 nla_put_u32(skb
, IFLA_IPTUN_FWMARK
, parm
->fwmark
))
2100 goto nla_put_failure
;
2102 if (nla_put_u16(skb
, IFLA_IPTUN_ENCAP_TYPE
, tunnel
->encap
.type
) ||
2103 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_SPORT
, tunnel
->encap
.sport
) ||
2104 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_DPORT
, tunnel
->encap
.dport
) ||
2105 nla_put_u16(skb
, IFLA_IPTUN_ENCAP_FLAGS
, tunnel
->encap
.flags
))
2106 goto nla_put_failure
;
2108 if (parm
->collect_md
)
2109 if (nla_put_flag(skb
, IFLA_IPTUN_COLLECT_METADATA
))
2110 goto nla_put_failure
;
2118 struct net
*ip6_tnl_get_link_net(const struct net_device
*dev
)
2120 struct ip6_tnl
*tunnel
= netdev_priv(dev
);
2124 EXPORT_SYMBOL(ip6_tnl_get_link_net
);
2126 static const struct nla_policy ip6_tnl_policy
[IFLA_IPTUN_MAX
+ 1] = {
2127 [IFLA_IPTUN_LINK
] = { .type
= NLA_U32
},
2128 [IFLA_IPTUN_LOCAL
] = { .len
= sizeof(struct in6_addr
) },
2129 [IFLA_IPTUN_REMOTE
] = { .len
= sizeof(struct in6_addr
) },
2130 [IFLA_IPTUN_TTL
] = { .type
= NLA_U8
},
2131 [IFLA_IPTUN_ENCAP_LIMIT
] = { .type
= NLA_U8
},
2132 [IFLA_IPTUN_FLOWINFO
] = { .type
= NLA_U32
},
2133 [IFLA_IPTUN_FLAGS
] = { .type
= NLA_U32
},
2134 [IFLA_IPTUN_PROTO
] = { .type
= NLA_U8
},
2135 [IFLA_IPTUN_ENCAP_TYPE
] = { .type
= NLA_U16
},
2136 [IFLA_IPTUN_ENCAP_FLAGS
] = { .type
= NLA_U16
},
2137 [IFLA_IPTUN_ENCAP_SPORT
] = { .type
= NLA_U16
},
2138 [IFLA_IPTUN_ENCAP_DPORT
] = { .type
= NLA_U16
},
2139 [IFLA_IPTUN_COLLECT_METADATA
] = { .type
= NLA_FLAG
},
2140 [IFLA_IPTUN_FWMARK
] = { .type
= NLA_U32
},
2143 static struct rtnl_link_ops ip6_link_ops __read_mostly
= {
2145 .maxtype
= IFLA_IPTUN_MAX
,
2146 .policy
= ip6_tnl_policy
,
2147 .priv_size
= sizeof(struct ip6_tnl
),
2148 .setup
= ip6_tnl_dev_setup
,
2149 .validate
= ip6_tnl_validate
,
2150 .newlink
= ip6_tnl_newlink
,
2151 .changelink
= ip6_tnl_changelink
,
2152 .dellink
= ip6_tnl_dellink
,
2153 .get_size
= ip6_tnl_get_size
,
2154 .fill_info
= ip6_tnl_fill_info
,
2155 .get_link_net
= ip6_tnl_get_link_net
,
2158 static struct xfrm6_tunnel ip4ip6_handler __read_mostly
= {
2159 .handler
= ip4ip6_rcv
,
2160 .err_handler
= ip4ip6_err
,
2164 static struct xfrm6_tunnel ip6ip6_handler __read_mostly
= {
2165 .handler
= ip6ip6_rcv
,
2166 .err_handler
= ip6ip6_err
,
2170 static void __net_exit
ip6_tnl_destroy_tunnels(struct net
*net
)
2172 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
2173 struct net_device
*dev
, *aux
;
2178 for_each_netdev_safe(net
, dev
, aux
)
2179 if (dev
->rtnl_link_ops
== &ip6_link_ops
)
2180 unregister_netdevice_queue(dev
, &list
);
2182 for (h
= 0; h
< IP6_TUNNEL_HASH_SIZE
; h
++) {
2183 t
= rtnl_dereference(ip6n
->tnls_r_l
[h
]);
2185 /* If dev is in the same netns, it has already
2186 * been added to the list by the previous loop.
2188 if (!net_eq(dev_net(t
->dev
), net
))
2189 unregister_netdevice_queue(t
->dev
, &list
);
2190 t
= rtnl_dereference(t
->next
);
2194 unregister_netdevice_many(&list
);
2197 static int __net_init
ip6_tnl_init_net(struct net
*net
)
2199 struct ip6_tnl_net
*ip6n
= net_generic(net
, ip6_tnl_net_id
);
2200 struct ip6_tnl
*t
= NULL
;
2203 ip6n
->tnls
[0] = ip6n
->tnls_wc
;
2204 ip6n
->tnls
[1] = ip6n
->tnls_r_l
;
2207 ip6n
->fb_tnl_dev
= alloc_netdev(sizeof(struct ip6_tnl
), "ip6tnl0",
2208 NET_NAME_UNKNOWN
, ip6_tnl_dev_setup
);
2210 if (!ip6n
->fb_tnl_dev
)
2212 dev_net_set(ip6n
->fb_tnl_dev
, net
);
2213 ip6n
->fb_tnl_dev
->rtnl_link_ops
= &ip6_link_ops
;
2214 /* FB netdevice is special: we have one, and only one per netns.
2215 * Allowing to move it to another netns is clearly unsafe.
2217 ip6n
->fb_tnl_dev
->features
|= NETIF_F_NETNS_LOCAL
;
2219 err
= ip6_fb_tnl_dev_init(ip6n
->fb_tnl_dev
);
2223 err
= register_netdev(ip6n
->fb_tnl_dev
);
2227 t
= netdev_priv(ip6n
->fb_tnl_dev
);
2229 strcpy(t
->parms
.name
, ip6n
->fb_tnl_dev
->name
);
2233 free_netdev(ip6n
->fb_tnl_dev
);
2238 static void __net_exit
ip6_tnl_exit_net(struct net
*net
)
2241 ip6_tnl_destroy_tunnels(net
);
2245 static struct pernet_operations ip6_tnl_net_ops
= {
2246 .init
= ip6_tnl_init_net
,
2247 .exit
= ip6_tnl_exit_net
,
2248 .id
= &ip6_tnl_net_id
,
2249 .size
= sizeof(struct ip6_tnl_net
),
2253 * ip6_tunnel_init - register protocol and reserve needed resources
2255 * Return: 0 on success
2258 static int __init
ip6_tunnel_init(void)
2262 if (!ipv6_mod_enabled())
2265 err
= register_pernet_device(&ip6_tnl_net_ops
);
2269 err
= xfrm6_tunnel_register(&ip4ip6_handler
, AF_INET
);
2271 pr_err("%s: can't register ip4ip6\n", __func__
);
2275 err
= xfrm6_tunnel_register(&ip6ip6_handler
, AF_INET6
);
2277 pr_err("%s: can't register ip6ip6\n", __func__
);
2280 err
= rtnl_link_register(&ip6_link_ops
);
2282 goto rtnl_link_failed
;
2287 xfrm6_tunnel_deregister(&ip6ip6_handler
, AF_INET6
);
2289 xfrm6_tunnel_deregister(&ip4ip6_handler
, AF_INET
);
2291 unregister_pernet_device(&ip6_tnl_net_ops
);
2297 * ip6_tunnel_cleanup - free resources and unregister protocol
2300 static void __exit
ip6_tunnel_cleanup(void)
2302 rtnl_link_unregister(&ip6_link_ops
);
2303 if (xfrm6_tunnel_deregister(&ip4ip6_handler
, AF_INET
))
2304 pr_info("%s: can't deregister ip4ip6\n", __func__
);
2306 if (xfrm6_tunnel_deregister(&ip6ip6_handler
, AF_INET6
))
2307 pr_info("%s: can't deregister ip6ip6\n", __func__
);
2309 unregister_pernet_device(&ip6_tnl_net_ops
);
2312 module_init(ip6_tunnel_init
);
2313 module_exit(ip6_tunnel_cleanup
);