2 * IPv6 over IPv4 tunnel device - Simple Internet Transition (SIT)
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
7 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
15 * Roger Venning <r.venning@telstra.com>: 6to4 support
16 * Nate Thompson <nate@thebog.net>: 6to4 support
17 * Fred Templin <fred.l.templin@boeing.com>: isatap support
20 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
22 #include <linux/module.h>
23 #include <linux/capability.h>
24 #include <linux/errno.h>
25 #include <linux/types.h>
26 #include <linux/socket.h>
27 #include <linux/sockios.h>
28 #include <linux/net.h>
29 #include <linux/in6.h>
30 #include <linux/netdevice.h>
31 #include <linux/if_arp.h>
32 #include <linux/icmp.h>
33 #include <linux/slab.h>
34 #include <asm/uaccess.h>
35 #include <linux/init.h>
36 #include <linux/netfilter_ipv4.h>
37 #include <linux/if_ether.h>
43 #include <net/protocol.h>
44 #include <net/transp_v6.h>
45 #include <net/ip6_fib.h>
46 #include <net/ip6_route.h>
47 #include <net/ndisc.h>
48 #include <net/addrconf.h>
52 #include <net/ip_tunnels.h>
53 #include <net/inet_ecn.h>
55 #include <net/dsfield.h>
56 #include <net/net_namespace.h>
57 #include <net/netns/generic.h>
60 This version of net/ipv6/sit.c is cloned of net/ipv4/ip_gre.c
62 For comments look at net/ipv4/ip_gre.c --ANK
66 #define HASH(addr) (((__force u32)addr^((__force u32)addr>>4))&0xF)
68 static bool log_ecn_error
= true;
69 module_param(log_ecn_error
, bool, 0644);
70 MODULE_PARM_DESC(log_ecn_error
, "Log packets received with corrupted ECN");
72 static int ipip6_tunnel_init(struct net_device
*dev
);
73 static void ipip6_tunnel_setup(struct net_device
*dev
);
74 static void ipip6_dev_free(struct net_device
*dev
);
75 static bool check_6rd(struct ip_tunnel
*tunnel
, const struct in6_addr
*v6dst
,
77 static struct rtnl_link_ops sit_link_ops __read_mostly
;
79 static int sit_net_id __read_mostly
;
81 struct ip_tunnel __rcu
*tunnels_r_l
[HASH_SIZE
];
82 struct ip_tunnel __rcu
*tunnels_r
[HASH_SIZE
];
83 struct ip_tunnel __rcu
*tunnels_l
[HASH_SIZE
];
84 struct ip_tunnel __rcu
*tunnels_wc
[1];
85 struct ip_tunnel __rcu
**tunnels
[4];
87 struct net_device
*fb_tunnel_dev
;
91 * Must be invoked with rcu_read_lock
93 static struct ip_tunnel
*ipip6_tunnel_lookup(struct net
*net
,
94 struct net_device
*dev
, __be32 remote
, __be32 local
)
96 unsigned int h0
= HASH(remote
);
97 unsigned int h1
= HASH(local
);
99 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
101 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_r_l
[h0
^ h1
]) {
102 if (local
== t
->parms
.iph
.saddr
&&
103 remote
== t
->parms
.iph
.daddr
&&
104 (!dev
|| !t
->parms
.link
|| dev
->ifindex
== t
->parms
.link
) &&
105 (t
->dev
->flags
& IFF_UP
))
108 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_r
[h0
]) {
109 if (remote
== t
->parms
.iph
.daddr
&&
110 (!dev
|| !t
->parms
.link
|| dev
->ifindex
== t
->parms
.link
) &&
111 (t
->dev
->flags
& IFF_UP
))
114 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_l
[h1
]) {
115 if (local
== t
->parms
.iph
.saddr
&&
116 (!dev
|| !t
->parms
.link
|| dev
->ifindex
== t
->parms
.link
) &&
117 (t
->dev
->flags
& IFF_UP
))
120 t
= rcu_dereference(sitn
->tunnels_wc
[0]);
121 if (t
&& (t
->dev
->flags
& IFF_UP
))
126 static struct ip_tunnel __rcu
**__ipip6_bucket(struct sit_net
*sitn
,
127 struct ip_tunnel_parm
*parms
)
129 __be32 remote
= parms
->iph
.daddr
;
130 __be32 local
= parms
->iph
.saddr
;
142 return &sitn
->tunnels
[prio
][h
];
145 static inline struct ip_tunnel __rcu
**ipip6_bucket(struct sit_net
*sitn
,
148 return __ipip6_bucket(sitn
, &t
->parms
);
151 static void ipip6_tunnel_unlink(struct sit_net
*sitn
, struct ip_tunnel
*t
)
153 struct ip_tunnel __rcu
**tp
;
154 struct ip_tunnel
*iter
;
156 for (tp
= ipip6_bucket(sitn
, t
);
157 (iter
= rtnl_dereference(*tp
)) != NULL
;
160 rcu_assign_pointer(*tp
, t
->next
);
166 static void ipip6_tunnel_link(struct sit_net
*sitn
, struct ip_tunnel
*t
)
168 struct ip_tunnel __rcu
**tp
= ipip6_bucket(sitn
, t
);
170 rcu_assign_pointer(t
->next
, rtnl_dereference(*tp
));
171 rcu_assign_pointer(*tp
, t
);
174 static void ipip6_tunnel_clone_6rd(struct net_device
*dev
, struct sit_net
*sitn
)
176 #ifdef CONFIG_IPV6_SIT_6RD
177 struct ip_tunnel
*t
= netdev_priv(dev
);
179 if (t
->dev
== sitn
->fb_tunnel_dev
) {
180 ipv6_addr_set(&t
->ip6rd
.prefix
, htonl(0x20020000), 0, 0, 0);
181 t
->ip6rd
.relay_prefix
= 0;
182 t
->ip6rd
.prefixlen
= 16;
183 t
->ip6rd
.relay_prefixlen
= 0;
185 struct ip_tunnel
*t0
= netdev_priv(sitn
->fb_tunnel_dev
);
186 memcpy(&t
->ip6rd
, &t0
->ip6rd
, sizeof(t
->ip6rd
));
191 static int ipip6_tunnel_create(struct net_device
*dev
)
193 struct ip_tunnel
*t
= netdev_priv(dev
);
194 struct net
*net
= dev_net(dev
);
195 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
198 memcpy(dev
->dev_addr
, &t
->parms
.iph
.saddr
, 4);
199 memcpy(dev
->broadcast
, &t
->parms
.iph
.daddr
, 4);
201 if ((__force u16
)t
->parms
.i_flags
& SIT_ISATAP
)
202 dev
->priv_flags
|= IFF_ISATAP
;
204 dev
->rtnl_link_ops
= &sit_link_ops
;
206 err
= register_netdevice(dev
);
210 ipip6_tunnel_clone_6rd(dev
, sitn
);
214 ipip6_tunnel_link(sitn
, t
);
221 static struct ip_tunnel
*ipip6_tunnel_locate(struct net
*net
,
222 struct ip_tunnel_parm
*parms
, int create
)
224 __be32 remote
= parms
->iph
.daddr
;
225 __be32 local
= parms
->iph
.saddr
;
226 struct ip_tunnel
*t
, *nt
;
227 struct ip_tunnel __rcu
**tp
;
228 struct net_device
*dev
;
230 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
232 for (tp
= __ipip6_bucket(sitn
, parms
);
233 (t
= rtnl_dereference(*tp
)) != NULL
;
235 if (local
== t
->parms
.iph
.saddr
&&
236 remote
== t
->parms
.iph
.daddr
&&
237 parms
->link
== t
->parms
.link
) {
248 strlcpy(name
, parms
->name
, IFNAMSIZ
);
250 strcpy(name
, "sit%d");
252 dev
= alloc_netdev(sizeof(*t
), name
, NET_NAME_UNKNOWN
,
257 dev_net_set(dev
, net
);
259 nt
= netdev_priv(dev
);
262 if (ipip6_tunnel_create(dev
) < 0)
273 #define for_each_prl_rcu(start) \
274 for (prl = rcu_dereference(start); \
276 prl = rcu_dereference(prl->next))
278 static struct ip_tunnel_prl_entry
*
279 __ipip6_tunnel_locate_prl(struct ip_tunnel
*t
, __be32 addr
)
281 struct ip_tunnel_prl_entry
*prl
;
283 for_each_prl_rcu(t
->prl
)
284 if (prl
->addr
== addr
)
290 static int ipip6_tunnel_get_prl(struct ip_tunnel
*t
,
291 struct ip_tunnel_prl __user
*a
)
293 struct ip_tunnel_prl kprl
, *kp
;
294 struct ip_tunnel_prl_entry
*prl
;
295 unsigned int cmax
, c
= 0, ca
, len
;
298 if (copy_from_user(&kprl
, a
, sizeof(kprl
)))
300 cmax
= kprl
.datalen
/ sizeof(kprl
);
301 if (cmax
> 1 && kprl
.addr
!= htonl(INADDR_ANY
))
304 /* For simple GET or for root users,
305 * we try harder to allocate.
307 kp
= (cmax
<= 1 || capable(CAP_NET_ADMIN
)) ?
308 kcalloc(cmax
, sizeof(*kp
), GFP_KERNEL
) :
313 ca
= t
->prl_count
< cmax
? t
->prl_count
: cmax
;
316 /* We don't try hard to allocate much memory for
318 * For root users, retry allocating enough memory for
321 kp
= kcalloc(ca
, sizeof(*kp
), GFP_ATOMIC
);
329 for_each_prl_rcu(t
->prl
) {
332 if (kprl
.addr
!= htonl(INADDR_ANY
) && prl
->addr
!= kprl
.addr
)
334 kp
[c
].addr
= prl
->addr
;
335 kp
[c
].flags
= prl
->flags
;
337 if (kprl
.addr
!= htonl(INADDR_ANY
))
343 len
= sizeof(*kp
) * c
;
345 if ((len
&& copy_to_user(a
+ 1, kp
, len
)) || put_user(len
, &a
->datalen
))
354 ipip6_tunnel_add_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
, int chg
)
356 struct ip_tunnel_prl_entry
*p
;
359 if (a
->addr
== htonl(INADDR_ANY
))
364 for (p
= rtnl_dereference(t
->prl
); p
; p
= rtnl_dereference(p
->next
)) {
365 if (p
->addr
== a
->addr
) {
380 p
= kzalloc(sizeof(struct ip_tunnel_prl_entry
), GFP_KERNEL
);
390 rcu_assign_pointer(t
->prl
, p
);
395 static void prl_list_destroy_rcu(struct rcu_head
*head
)
397 struct ip_tunnel_prl_entry
*p
, *n
;
399 p
= container_of(head
, struct ip_tunnel_prl_entry
, rcu_head
);
401 n
= rcu_dereference_protected(p
->next
, 1);
408 ipip6_tunnel_del_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
)
410 struct ip_tunnel_prl_entry
*x
;
411 struct ip_tunnel_prl_entry __rcu
**p
;
416 if (a
&& a
->addr
!= htonl(INADDR_ANY
)) {
418 (x
= rtnl_dereference(*p
)) != NULL
;
420 if (x
->addr
== a
->addr
) {
422 kfree_rcu(x
, rcu_head
);
429 x
= rtnl_dereference(t
->prl
);
432 call_rcu(&x
->rcu_head
, prl_list_destroy_rcu
);
441 isatap_chksrc(struct sk_buff
*skb
, const struct iphdr
*iph
, struct ip_tunnel
*t
)
443 struct ip_tunnel_prl_entry
*p
;
447 p
= __ipip6_tunnel_locate_prl(t
, iph
->saddr
);
449 if (p
->flags
& PRL_DEFAULT
)
450 skb
->ndisc_nodetype
= NDISC_NODETYPE_DEFAULT
;
452 skb
->ndisc_nodetype
= NDISC_NODETYPE_NODEFAULT
;
454 const struct in6_addr
*addr6
= &ipv6_hdr(skb
)->saddr
;
456 if (ipv6_addr_is_isatap(addr6
) &&
457 (addr6
->s6_addr32
[3] == iph
->saddr
) &&
458 ipv6_chk_prefix(addr6
, t
->dev
))
459 skb
->ndisc_nodetype
= NDISC_NODETYPE_HOST
;
467 static void ipip6_tunnel_uninit(struct net_device
*dev
)
469 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
470 struct sit_net
*sitn
= net_generic(tunnel
->net
, sit_net_id
);
472 if (dev
== sitn
->fb_tunnel_dev
) {
473 RCU_INIT_POINTER(sitn
->tunnels_wc
[0], NULL
);
475 ipip6_tunnel_unlink(sitn
, tunnel
);
476 ipip6_tunnel_del_prl(tunnel
, NULL
);
478 dst_cache_reset(&tunnel
->dst_cache
);
482 /* Generate icmpv6 with type/code ICMPV6_DEST_UNREACH/ICMPV6_ADDR_UNREACH
483 * if sufficient data bytes are available
485 static int ipip6_err_gen_icmpv6_unreach(struct sk_buff
*skb
)
487 int ihl
= ((const struct iphdr
*)skb
->data
)->ihl
*4;
489 struct sk_buff
*skb2
;
491 if (!pskb_may_pull(skb
, ihl
+ sizeof(struct ipv6hdr
) + 8))
494 skb2
= skb_clone(skb
, GFP_ATOMIC
);
501 skb_reset_network_header(skb2
);
503 rt
= rt6_lookup(dev_net(skb
->dev
), &ipv6_hdr(skb2
)->saddr
, NULL
, 0, 0);
505 if (rt
&& rt
->dst
.dev
)
506 skb2
->dev
= rt
->dst
.dev
;
508 icmpv6_send(skb2
, ICMPV6_DEST_UNREACH
, ICMPV6_ADDR_UNREACH
, 0);
518 static int ipip6_err(struct sk_buff
*skb
, u32 info
)
520 const struct iphdr
*iph
= (const struct iphdr
*)skb
->data
;
521 const int type
= icmp_hdr(skb
)->type
;
522 const int code
= icmp_hdr(skb
)->code
;
528 case ICMP_PARAMETERPROB
:
531 case ICMP_DEST_UNREACH
:
534 /* Impossible event. */
537 /* All others are translated to HOST_UNREACH.
538 rfc2003 contains "deep thoughts" about NET_UNREACH,
539 I believe they are just ether pollution. --ANK
544 case ICMP_TIME_EXCEEDED
:
545 if (code
!= ICMP_EXC_TTL
)
554 t
= ipip6_tunnel_lookup(dev_net(skb
->dev
),
561 if (type
== ICMP_DEST_UNREACH
&& code
== ICMP_FRAG_NEEDED
) {
562 ipv4_update_pmtu(skb
, dev_net(skb
->dev
), info
,
563 t
->parms
.link
, 0, IPPROTO_IPV6
, 0);
567 if (type
== ICMP_REDIRECT
) {
568 ipv4_redirect(skb
, dev_net(skb
->dev
), t
->parms
.link
, 0,
574 if (t
->parms
.iph
.daddr
== 0)
578 if (!ipip6_err_gen_icmpv6_unreach(skb
))
581 if (t
->parms
.iph
.ttl
== 0 && type
== ICMP_TIME_EXCEEDED
)
584 if (time_before(jiffies
, t
->err_time
+ IPTUNNEL_ERR_TIMEO
))
588 t
->err_time
= jiffies
;
593 static inline bool is_spoofed_6rd(struct ip_tunnel
*tunnel
, const __be32 v4addr
,
594 const struct in6_addr
*v6addr
)
597 if (check_6rd(tunnel
, v6addr
, &v4embed
) && v4addr
!= v4embed
)
602 /* Checks if an address matches an address on the tunnel interface.
603 * Used to detect the NAT of proto 41 packets and let them pass spoofing test.
605 * This function is called after we considered the packet as spoofed
607 * We may have a router that is doing NAT for proto 41 packets
608 * for an internal station. Destination a.a.a.a/PREFIX:bbbb:bbbb
609 * will be translated to n.n.n.n/PREFIX:bbbb:bbbb. And is_spoofed_6rd
610 * function will return true, dropping the packet.
611 * But, we can still check if is spoofed against the IP
612 * addresses associated with the interface.
614 static bool only_dnatted(const struct ip_tunnel
*tunnel
,
615 const struct in6_addr
*v6dst
)
619 #ifdef CONFIG_IPV6_SIT_6RD
620 prefix_len
= tunnel
->ip6rd
.prefixlen
+ 32
621 - tunnel
->ip6rd
.relay_prefixlen
;
625 return ipv6_chk_custom_prefix(v6dst
, prefix_len
, tunnel
->dev
);
628 /* Returns true if a packet is spoofed */
629 static bool packet_is_spoofed(struct sk_buff
*skb
,
630 const struct iphdr
*iph
,
631 struct ip_tunnel
*tunnel
)
633 const struct ipv6hdr
*ipv6h
;
635 if (tunnel
->dev
->priv_flags
& IFF_ISATAP
) {
636 if (!isatap_chksrc(skb
, iph
, tunnel
))
642 if (tunnel
->dev
->flags
& IFF_POINTOPOINT
)
645 ipv6h
= ipv6_hdr(skb
);
647 if (unlikely(is_spoofed_6rd(tunnel
, iph
->saddr
, &ipv6h
->saddr
))) {
648 net_warn_ratelimited("Src spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
649 &iph
->saddr
, &ipv6h
->saddr
,
650 &iph
->daddr
, &ipv6h
->daddr
);
654 if (likely(!is_spoofed_6rd(tunnel
, iph
->daddr
, &ipv6h
->daddr
)))
657 if (only_dnatted(tunnel
, &ipv6h
->daddr
))
660 net_warn_ratelimited("Dst spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
661 &iph
->saddr
, &ipv6h
->saddr
,
662 &iph
->daddr
, &ipv6h
->daddr
);
666 static int ipip6_rcv(struct sk_buff
*skb
)
668 const struct iphdr
*iph
= ip_hdr(skb
);
669 struct ip_tunnel
*tunnel
;
672 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
673 iph
->saddr
, iph
->daddr
);
675 struct pcpu_sw_netstats
*tstats
;
677 if (tunnel
->parms
.iph
.protocol
!= IPPROTO_IPV6
&&
678 tunnel
->parms
.iph
.protocol
!= 0)
681 skb
->mac_header
= skb
->network_header
;
682 skb_reset_network_header(skb
);
683 IPCB(skb
)->flags
= 0;
684 skb
->dev
= tunnel
->dev
;
686 if (packet_is_spoofed(skb
, iph
, tunnel
)) {
687 tunnel
->dev
->stats
.rx_errors
++;
691 if (iptunnel_pull_header(skb
, 0, htons(ETH_P_IPV6
),
692 !net_eq(tunnel
->net
, dev_net(tunnel
->dev
))))
695 err
= IP_ECN_decapsulate(iph
, skb
);
698 net_info_ratelimited("non-ECT from %pI4 with TOS=%#x\n",
699 &iph
->saddr
, iph
->tos
);
701 ++tunnel
->dev
->stats
.rx_frame_errors
;
702 ++tunnel
->dev
->stats
.rx_errors
;
707 tstats
= this_cpu_ptr(tunnel
->dev
->tstats
);
708 u64_stats_update_begin(&tstats
->syncp
);
709 tstats
->rx_packets
++;
710 tstats
->rx_bytes
+= skb
->len
;
711 u64_stats_update_end(&tstats
->syncp
);
718 /* no tunnel matched, let upstream know, ipsec may handle it */
725 static const struct tnl_ptk_info tpi
= {
726 /* no tunnel info required for ipip. */
727 .proto
= htons(ETH_P_IP
),
730 static int ipip_rcv(struct sk_buff
*skb
)
732 const struct iphdr
*iph
;
733 struct ip_tunnel
*tunnel
;
736 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
737 iph
->saddr
, iph
->daddr
);
739 if (tunnel
->parms
.iph
.protocol
!= IPPROTO_IPIP
&&
740 tunnel
->parms
.iph
.protocol
!= 0)
743 if (!xfrm4_policy_check(NULL
, XFRM_POLICY_IN
, skb
))
745 if (iptunnel_pull_header(skb
, 0, tpi
.proto
, false))
747 return ip_tunnel_rcv(tunnel
, skb
, &tpi
, NULL
, log_ecn_error
);
758 * If the IPv6 address comes from 6rd / 6to4 (RFC 3056) addr space this function
759 * stores the embedded IPv4 address in v4dst and returns true.
761 static bool check_6rd(struct ip_tunnel
*tunnel
, const struct in6_addr
*v6dst
,
764 #ifdef CONFIG_IPV6_SIT_6RD
765 if (ipv6_prefix_equal(v6dst
, &tunnel
->ip6rd
.prefix
,
766 tunnel
->ip6rd
.prefixlen
)) {
767 unsigned int pbw0
, pbi0
;
771 pbw0
= tunnel
->ip6rd
.prefixlen
>> 5;
772 pbi0
= tunnel
->ip6rd
.prefixlen
& 0x1f;
774 d
= (ntohl(v6dst
->s6_addr32
[pbw0
]) << pbi0
) >>
775 tunnel
->ip6rd
.relay_prefixlen
;
777 pbi1
= pbi0
- tunnel
->ip6rd
.relay_prefixlen
;
779 d
|= ntohl(v6dst
->s6_addr32
[pbw0
+ 1]) >>
782 *v4dst
= tunnel
->ip6rd
.relay_prefix
| htonl(d
);
786 if (v6dst
->s6_addr16
[0] == htons(0x2002)) {
787 /* 6to4 v6 addr has 16 bits prefix, 32 v4addr, 16 SLA, ... */
788 memcpy(v4dst
, &v6dst
->s6_addr16
[1], 4);
795 static inline __be32
try_6rd(struct ip_tunnel
*tunnel
,
796 const struct in6_addr
*v6dst
)
799 check_6rd(tunnel
, v6dst
, &dst
);
804 * This function assumes it is being called from dev_queue_xmit()
805 * and that skb is filled properly by that function.
808 static netdev_tx_t
ipip6_tunnel_xmit(struct sk_buff
*skb
,
809 struct net_device
*dev
)
811 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
812 const struct iphdr
*tiph
= &tunnel
->parms
.iph
;
813 const struct ipv6hdr
*iph6
= ipv6_hdr(skb
);
814 u8 tos
= tunnel
->parms
.iph
.tos
;
815 __be16 df
= tiph
->frag_off
;
816 struct rtable
*rt
; /* Route to the other host */
817 struct net_device
*tdev
; /* Device to other host */
818 unsigned int max_headroom
; /* The extra header space needed */
819 __be32 dst
= tiph
->daddr
;
822 const struct in6_addr
*addr6
;
825 u8 protocol
= IPPROTO_IPV6
;
826 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
828 if (skb
->protocol
!= htons(ETH_P_IPV6
))
832 tos
= ipv6_get_dsfield(iph6
);
834 /* ISATAP (RFC4214) - must come before 6to4 */
835 if (dev
->priv_flags
& IFF_ISATAP
) {
836 struct neighbour
*neigh
= NULL
;
837 bool do_tx_error
= false;
840 neigh
= dst_neigh_lookup(skb_dst(skb
), &iph6
->daddr
);
843 net_dbg_ratelimited("nexthop == NULL\n");
847 addr6
= (const struct in6_addr
*)&neigh
->primary_key
;
848 addr_type
= ipv6_addr_type(addr6
);
850 if ((addr_type
& IPV6_ADDR_UNICAST
) &&
851 ipv6_addr_is_isatap(addr6
))
852 dst
= addr6
->s6_addr32
[3];
856 neigh_release(neigh
);
862 dst
= try_6rd(tunnel
, &iph6
->daddr
);
865 struct neighbour
*neigh
= NULL
;
866 bool do_tx_error
= false;
869 neigh
= dst_neigh_lookup(skb_dst(skb
), &iph6
->daddr
);
872 net_dbg_ratelimited("nexthop == NULL\n");
876 addr6
= (const struct in6_addr
*)&neigh
->primary_key
;
877 addr_type
= ipv6_addr_type(addr6
);
879 if (addr_type
== IPV6_ADDR_ANY
) {
880 addr6
= &ipv6_hdr(skb
)->daddr
;
881 addr_type
= ipv6_addr_type(addr6
);
884 if ((addr_type
& IPV6_ADDR_COMPATv4
) != 0)
885 dst
= addr6
->s6_addr32
[3];
889 neigh_release(neigh
);
894 rt
= ip_route_output_ports(tunnel
->net
, &fl4
, NULL
,
897 IPPROTO_IPV6
, RT_TOS(tos
),
900 dev
->stats
.tx_carrier_errors
++;
903 if (rt
->rt_type
!= RTN_UNICAST
) {
905 dev
->stats
.tx_carrier_errors
++;
912 dev
->stats
.collisions
++;
916 skb
= iptunnel_handle_offloads(skb
, SKB_GSO_SIT
);
923 mtu
= dst_mtu(&rt
->dst
) - t_hlen
;
926 dev
->stats
.collisions
++;
931 if (mtu
< IPV6_MIN_MTU
) {
936 if (tunnel
->parms
.iph
.daddr
&& skb_dst(skb
))
937 skb_dst(skb
)->ops
->update_pmtu(skb_dst(skb
), NULL
, skb
, mtu
);
939 if (skb
->len
> mtu
&& !skb_is_gso(skb
)) {
940 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
946 if (tunnel
->err_count
> 0) {
947 if (time_before(jiffies
,
948 tunnel
->err_time
+ IPTUNNEL_ERR_TIMEO
)) {
950 dst_link_failure(skb
);
952 tunnel
->err_count
= 0;
956 * Okay, now see if we can stuff it in the buffer as-is.
958 max_headroom
= LL_RESERVED_SPACE(tdev
) + t_hlen
;
960 if (skb_headroom(skb
) < max_headroom
|| skb_shared(skb
) ||
961 (skb_cloned(skb
) && !skb_clone_writable(skb
, 0))) {
962 struct sk_buff
*new_skb
= skb_realloc_headroom(skb
, max_headroom
);
965 dev
->stats
.tx_dropped
++;
970 skb_set_owner_w(new_skb
, skb
->sk
);
973 iph6
= ipv6_hdr(skb
);
977 ttl
= iph6
->hop_limit
;
978 tos
= INET_ECN_encapsulate(tos
, ipv6_get_dsfield(iph6
));
980 if (ip_tunnel_encap(skb
, tunnel
, &protocol
, &fl4
) < 0) {
985 skb_set_inner_ipproto(skb
, IPPROTO_IPV6
);
987 iptunnel_xmit(NULL
, rt
, skb
, fl4
.saddr
, fl4
.daddr
, protocol
, tos
, ttl
,
988 df
, !net_eq(tunnel
->net
, dev_net(dev
)));
992 dst_link_failure(skb
);
996 dev
->stats
.tx_errors
++;
1000 static netdev_tx_t
ipip_tunnel_xmit(struct sk_buff
*skb
, struct net_device
*dev
)
1002 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1003 const struct iphdr
*tiph
= &tunnel
->parms
.iph
;
1005 skb
= iptunnel_handle_offloads(skb
, SKB_GSO_IPIP
);
1009 skb_set_inner_ipproto(skb
, IPPROTO_IPIP
);
1011 ip_tunnel_xmit(skb
, dev
, tiph
, IPPROTO_IPIP
);
1012 return NETDEV_TX_OK
;
1014 dev
->stats
.tx_errors
++;
1015 return NETDEV_TX_OK
;
1018 static netdev_tx_t
sit_tunnel_xmit(struct sk_buff
*skb
,
1019 struct net_device
*dev
)
1021 switch (skb
->protocol
) {
1022 case htons(ETH_P_IP
):
1023 ipip_tunnel_xmit(skb
, dev
);
1025 case htons(ETH_P_IPV6
):
1026 ipip6_tunnel_xmit(skb
, dev
);
1032 return NETDEV_TX_OK
;
1035 dev
->stats
.tx_errors
++;
1037 return NETDEV_TX_OK
;
1041 static void ipip6_tunnel_bind_dev(struct net_device
*dev
)
1043 struct net_device
*tdev
= NULL
;
1044 struct ip_tunnel
*tunnel
;
1045 const struct iphdr
*iph
;
1048 tunnel
= netdev_priv(dev
);
1049 iph
= &tunnel
->parms
.iph
;
1052 struct rtable
*rt
= ip_route_output_ports(tunnel
->net
, &fl4
,
1054 iph
->daddr
, iph
->saddr
,
1058 tunnel
->parms
.link
);
1064 dev
->flags
|= IFF_POINTOPOINT
;
1067 if (!tdev
&& tunnel
->parms
.link
)
1068 tdev
= __dev_get_by_index(tunnel
->net
, tunnel
->parms
.link
);
1071 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1073 dev
->hard_header_len
= tdev
->hard_header_len
+ sizeof(struct iphdr
);
1074 dev
->mtu
= tdev
->mtu
- t_hlen
;
1075 if (dev
->mtu
< IPV6_MIN_MTU
)
1076 dev
->mtu
= IPV6_MIN_MTU
;
1080 static void ipip6_tunnel_update(struct ip_tunnel
*t
, struct ip_tunnel_parm
*p
)
1082 struct net
*net
= t
->net
;
1083 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1085 ipip6_tunnel_unlink(sitn
, t
);
1087 t
->parms
.iph
.saddr
= p
->iph
.saddr
;
1088 t
->parms
.iph
.daddr
= p
->iph
.daddr
;
1089 memcpy(t
->dev
->dev_addr
, &p
->iph
.saddr
, 4);
1090 memcpy(t
->dev
->broadcast
, &p
->iph
.daddr
, 4);
1091 ipip6_tunnel_link(sitn
, t
);
1092 t
->parms
.iph
.ttl
= p
->iph
.ttl
;
1093 t
->parms
.iph
.tos
= p
->iph
.tos
;
1094 if (t
->parms
.link
!= p
->link
) {
1095 t
->parms
.link
= p
->link
;
1096 ipip6_tunnel_bind_dev(t
->dev
);
1098 dst_cache_reset(&t
->dst_cache
);
1099 netdev_state_change(t
->dev
);
1102 #ifdef CONFIG_IPV6_SIT_6RD
1103 static int ipip6_tunnel_update_6rd(struct ip_tunnel
*t
,
1104 struct ip_tunnel_6rd
*ip6rd
)
1106 struct in6_addr prefix
;
1107 __be32 relay_prefix
;
1109 if (ip6rd
->relay_prefixlen
> 32 ||
1110 ip6rd
->prefixlen
+ (32 - ip6rd
->relay_prefixlen
) > 64)
1113 ipv6_addr_prefix(&prefix
, &ip6rd
->prefix
, ip6rd
->prefixlen
);
1114 if (!ipv6_addr_equal(&prefix
, &ip6rd
->prefix
))
1116 if (ip6rd
->relay_prefixlen
)
1117 relay_prefix
= ip6rd
->relay_prefix
&
1118 htonl(0xffffffffUL
<<
1119 (32 - ip6rd
->relay_prefixlen
));
1122 if (relay_prefix
!= ip6rd
->relay_prefix
)
1125 t
->ip6rd
.prefix
= prefix
;
1126 t
->ip6rd
.relay_prefix
= relay_prefix
;
1127 t
->ip6rd
.prefixlen
= ip6rd
->prefixlen
;
1128 t
->ip6rd
.relay_prefixlen
= ip6rd
->relay_prefixlen
;
1129 dst_cache_reset(&t
->dst_cache
);
1130 netdev_state_change(t
->dev
);
1136 ipip6_tunnel_ioctl(struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
1139 struct ip_tunnel_parm p
;
1140 struct ip_tunnel_prl prl
;
1141 struct ip_tunnel
*t
= netdev_priv(dev
);
1142 struct net
*net
= t
->net
;
1143 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1144 #ifdef CONFIG_IPV6_SIT_6RD
1145 struct ip_tunnel_6rd ip6rd
;
1150 #ifdef CONFIG_IPV6_SIT_6RD
1153 if (dev
== sitn
->fb_tunnel_dev
) {
1154 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
))) {
1158 t
= ipip6_tunnel_locate(net
, &p
, 0);
1160 t
= netdev_priv(dev
);
1164 if (cmd
== SIOCGETTUNNEL
) {
1165 memcpy(&p
, &t
->parms
, sizeof(p
));
1166 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
,
1169 #ifdef CONFIG_IPV6_SIT_6RD
1171 ip6rd
.prefix
= t
->ip6rd
.prefix
;
1172 ip6rd
.relay_prefix
= t
->ip6rd
.relay_prefix
;
1173 ip6rd
.prefixlen
= t
->ip6rd
.prefixlen
;
1174 ip6rd
.relay_prefixlen
= t
->ip6rd
.relay_prefixlen
;
1175 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &ip6rd
,
1186 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1190 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
1194 if (p
.iph
.protocol
!= IPPROTO_IPV6
&&
1195 p
.iph
.protocol
!= IPPROTO_IPIP
&&
1196 p
.iph
.protocol
!= 0)
1198 if (p
.iph
.version
!= 4 ||
1199 p
.iph
.ihl
!= 5 || (p
.iph
.frag_off
&htons(~IP_DF
)))
1202 p
.iph
.frag_off
|= htons(IP_DF
);
1204 t
= ipip6_tunnel_locate(net
, &p
, cmd
== SIOCADDTUNNEL
);
1206 if (dev
!= sitn
->fb_tunnel_dev
&& cmd
== SIOCCHGTUNNEL
) {
1208 if (t
->dev
!= dev
) {
1213 if (((dev
->flags
&IFF_POINTOPOINT
) && !p
.iph
.daddr
) ||
1214 (!(dev
->flags
&IFF_POINTOPOINT
) && p
.iph
.daddr
)) {
1218 t
= netdev_priv(dev
);
1221 ipip6_tunnel_update(t
, &p
);
1226 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &t
->parms
, sizeof(p
)))
1229 err
= (cmd
== SIOCADDTUNNEL
? -ENOBUFS
: -ENOENT
);
1234 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1237 if (dev
== sitn
->fb_tunnel_dev
) {
1239 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
1242 t
= ipip6_tunnel_locate(net
, &p
, 0);
1246 if (t
== netdev_priv(sitn
->fb_tunnel_dev
))
1250 unregister_netdevice(dev
);
1256 if (dev
== sitn
->fb_tunnel_dev
)
1258 err
= ipip6_tunnel_get_prl(t
, ifr
->ifr_ifru
.ifru_data
);
1265 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1268 if (dev
== sitn
->fb_tunnel_dev
)
1271 if (copy_from_user(&prl
, ifr
->ifr_ifru
.ifru_data
, sizeof(prl
)))
1276 err
= ipip6_tunnel_del_prl(t
, &prl
);
1280 err
= ipip6_tunnel_add_prl(t
, &prl
, cmd
== SIOCCHGPRL
);
1283 dst_cache_reset(&t
->dst_cache
);
1284 netdev_state_change(dev
);
1287 #ifdef CONFIG_IPV6_SIT_6RD
1292 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1296 if (copy_from_user(&ip6rd
, ifr
->ifr_ifru
.ifru_data
,
1300 if (cmd
!= SIOCDEL6RD
) {
1301 err
= ipip6_tunnel_update_6rd(t
, &ip6rd
);
1305 ipip6_tunnel_clone_6rd(dev
, sitn
);
1319 static int ipip6_tunnel_change_mtu(struct net_device
*dev
, int new_mtu
)
1321 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1322 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1324 if (new_mtu
< IPV6_MIN_MTU
|| new_mtu
> 0xFFF8 - t_hlen
)
1330 static const struct net_device_ops ipip6_netdev_ops
= {
1331 .ndo_init
= ipip6_tunnel_init
,
1332 .ndo_uninit
= ipip6_tunnel_uninit
,
1333 .ndo_start_xmit
= sit_tunnel_xmit
,
1334 .ndo_do_ioctl
= ipip6_tunnel_ioctl
,
1335 .ndo_change_mtu
= ipip6_tunnel_change_mtu
,
1336 .ndo_get_stats64
= ip_tunnel_get_stats64
,
1337 .ndo_get_iflink
= ip_tunnel_get_iflink
,
1340 static void ipip6_dev_free(struct net_device
*dev
)
1342 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1344 dst_cache_destroy(&tunnel
->dst_cache
);
1345 free_percpu(dev
->tstats
);
1349 #define SIT_FEATURES (NETIF_F_SG | \
1350 NETIF_F_FRAGLIST | \
1352 NETIF_F_GSO_SOFTWARE | \
1355 static void ipip6_tunnel_setup(struct net_device
*dev
)
1357 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1358 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1360 dev
->netdev_ops
= &ipip6_netdev_ops
;
1361 dev
->destructor
= ipip6_dev_free
;
1363 dev
->type
= ARPHRD_SIT
;
1364 dev
->hard_header_len
= LL_MAX_HEADER
+ t_hlen
;
1365 dev
->mtu
= ETH_DATA_LEN
- t_hlen
;
1366 dev
->flags
= IFF_NOARP
;
1367 netif_keep_dst(dev
);
1369 dev
->features
|= NETIF_F_LLTX
;
1370 dev
->features
|= SIT_FEATURES
;
1371 dev
->hw_features
|= SIT_FEATURES
;
1374 static int ipip6_tunnel_init(struct net_device
*dev
)
1376 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1380 tunnel
->net
= dev_net(dev
);
1381 strcpy(tunnel
->parms
.name
, dev
->name
);
1383 ipip6_tunnel_bind_dev(dev
);
1384 dev
->tstats
= netdev_alloc_pcpu_stats(struct pcpu_sw_netstats
);
1388 err
= dst_cache_init(&tunnel
->dst_cache
, GFP_KERNEL
);
1390 free_percpu(dev
->tstats
);
1397 static void __net_init
ipip6_fb_tunnel_init(struct net_device
*dev
)
1399 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1400 struct iphdr
*iph
= &tunnel
->parms
.iph
;
1401 struct net
*net
= dev_net(dev
);
1402 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1405 iph
->protocol
= IPPROTO_IPV6
;
1410 rcu_assign_pointer(sitn
->tunnels_wc
[0], tunnel
);
1413 static int ipip6_validate(struct nlattr
*tb
[], struct nlattr
*data
[])
1417 if (!data
|| !data
[IFLA_IPTUN_PROTO
])
1420 proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1421 if (proto
!= IPPROTO_IPV6
&&
1422 proto
!= IPPROTO_IPIP
&&
1429 static void ipip6_netlink_parms(struct nlattr
*data
[],
1430 struct ip_tunnel_parm
*parms
)
1432 memset(parms
, 0, sizeof(*parms
));
1434 parms
->iph
.version
= 4;
1435 parms
->iph
.protocol
= IPPROTO_IPV6
;
1437 parms
->iph
.ttl
= 64;
1442 if (data
[IFLA_IPTUN_LINK
])
1443 parms
->link
= nla_get_u32(data
[IFLA_IPTUN_LINK
]);
1445 if (data
[IFLA_IPTUN_LOCAL
])
1446 parms
->iph
.saddr
= nla_get_be32(data
[IFLA_IPTUN_LOCAL
]);
1448 if (data
[IFLA_IPTUN_REMOTE
])
1449 parms
->iph
.daddr
= nla_get_be32(data
[IFLA_IPTUN_REMOTE
]);
1451 if (data
[IFLA_IPTUN_TTL
]) {
1452 parms
->iph
.ttl
= nla_get_u8(data
[IFLA_IPTUN_TTL
]);
1454 parms
->iph
.frag_off
= htons(IP_DF
);
1457 if (data
[IFLA_IPTUN_TOS
])
1458 parms
->iph
.tos
= nla_get_u8(data
[IFLA_IPTUN_TOS
]);
1460 if (!data
[IFLA_IPTUN_PMTUDISC
] || nla_get_u8(data
[IFLA_IPTUN_PMTUDISC
]))
1461 parms
->iph
.frag_off
= htons(IP_DF
);
1463 if (data
[IFLA_IPTUN_FLAGS
])
1464 parms
->i_flags
= nla_get_be16(data
[IFLA_IPTUN_FLAGS
]);
1466 if (data
[IFLA_IPTUN_PROTO
])
1467 parms
->iph
.protocol
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1471 /* This function returns true when ENCAP attributes are present in the nl msg */
1472 static bool ipip6_netlink_encap_parms(struct nlattr
*data
[],
1473 struct ip_tunnel_encap
*ipencap
)
1477 memset(ipencap
, 0, sizeof(*ipencap
));
1482 if (data
[IFLA_IPTUN_ENCAP_TYPE
]) {
1484 ipencap
->type
= nla_get_u16(data
[IFLA_IPTUN_ENCAP_TYPE
]);
1487 if (data
[IFLA_IPTUN_ENCAP_FLAGS
]) {
1489 ipencap
->flags
= nla_get_u16(data
[IFLA_IPTUN_ENCAP_FLAGS
]);
1492 if (data
[IFLA_IPTUN_ENCAP_SPORT
]) {
1494 ipencap
->sport
= nla_get_be16(data
[IFLA_IPTUN_ENCAP_SPORT
]);
1497 if (data
[IFLA_IPTUN_ENCAP_DPORT
]) {
1499 ipencap
->dport
= nla_get_be16(data
[IFLA_IPTUN_ENCAP_DPORT
]);
1505 #ifdef CONFIG_IPV6_SIT_6RD
1506 /* This function returns true when 6RD attributes are present in the nl msg */
1507 static bool ipip6_netlink_6rd_parms(struct nlattr
*data
[],
1508 struct ip_tunnel_6rd
*ip6rd
)
1511 memset(ip6rd
, 0, sizeof(*ip6rd
));
1516 if (data
[IFLA_IPTUN_6RD_PREFIX
]) {
1518 ip6rd
->prefix
= nla_get_in6_addr(data
[IFLA_IPTUN_6RD_PREFIX
]);
1521 if (data
[IFLA_IPTUN_6RD_RELAY_PREFIX
]) {
1523 ip6rd
->relay_prefix
=
1524 nla_get_be32(data
[IFLA_IPTUN_6RD_RELAY_PREFIX
]);
1527 if (data
[IFLA_IPTUN_6RD_PREFIXLEN
]) {
1529 ip6rd
->prefixlen
= nla_get_u16(data
[IFLA_IPTUN_6RD_PREFIXLEN
]);
1532 if (data
[IFLA_IPTUN_6RD_RELAY_PREFIXLEN
]) {
1534 ip6rd
->relay_prefixlen
=
1535 nla_get_u16(data
[IFLA_IPTUN_6RD_RELAY_PREFIXLEN
]);
1542 static int ipip6_newlink(struct net
*src_net
, struct net_device
*dev
,
1543 struct nlattr
*tb
[], struct nlattr
*data
[])
1545 struct net
*net
= dev_net(dev
);
1546 struct ip_tunnel
*nt
;
1547 struct ip_tunnel_encap ipencap
;
1548 #ifdef CONFIG_IPV6_SIT_6RD
1549 struct ip_tunnel_6rd ip6rd
;
1553 nt
= netdev_priv(dev
);
1555 if (ipip6_netlink_encap_parms(data
, &ipencap
)) {
1556 err
= ip_tunnel_encap_setup(nt
, &ipencap
);
1561 ipip6_netlink_parms(data
, &nt
->parms
);
1563 if (ipip6_tunnel_locate(net
, &nt
->parms
, 0))
1566 err
= ipip6_tunnel_create(dev
);
1570 #ifdef CONFIG_IPV6_SIT_6RD
1571 if (ipip6_netlink_6rd_parms(data
, &ip6rd
))
1572 err
= ipip6_tunnel_update_6rd(nt
, &ip6rd
);
1578 static int ipip6_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
1579 struct nlattr
*data
[])
1581 struct ip_tunnel
*t
= netdev_priv(dev
);
1582 struct ip_tunnel_parm p
;
1583 struct ip_tunnel_encap ipencap
;
1584 struct net
*net
= t
->net
;
1585 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1586 #ifdef CONFIG_IPV6_SIT_6RD
1587 struct ip_tunnel_6rd ip6rd
;
1591 if (dev
== sitn
->fb_tunnel_dev
)
1594 if (ipip6_netlink_encap_parms(data
, &ipencap
)) {
1595 err
= ip_tunnel_encap_setup(t
, &ipencap
);
1600 ipip6_netlink_parms(data
, &p
);
1602 if (((dev
->flags
& IFF_POINTOPOINT
) && !p
.iph
.daddr
) ||
1603 (!(dev
->flags
& IFF_POINTOPOINT
) && p
.iph
.daddr
))
1606 t
= ipip6_tunnel_locate(net
, &p
, 0);
1612 t
= netdev_priv(dev
);
1614 ipip6_tunnel_update(t
, &p
);
1616 #ifdef CONFIG_IPV6_SIT_6RD
1617 if (ipip6_netlink_6rd_parms(data
, &ip6rd
))
1618 return ipip6_tunnel_update_6rd(t
, &ip6rd
);
1624 static size_t ipip6_get_size(const struct net_device
*dev
)
1627 /* IFLA_IPTUN_LINK */
1629 /* IFLA_IPTUN_LOCAL */
1631 /* IFLA_IPTUN_REMOTE */
1633 /* IFLA_IPTUN_TTL */
1635 /* IFLA_IPTUN_TOS */
1637 /* IFLA_IPTUN_PMTUDISC */
1639 /* IFLA_IPTUN_FLAGS */
1641 /* IFLA_IPTUN_PROTO */
1643 #ifdef CONFIG_IPV6_SIT_6RD
1644 /* IFLA_IPTUN_6RD_PREFIX */
1645 nla_total_size(sizeof(struct in6_addr
)) +
1646 /* IFLA_IPTUN_6RD_RELAY_PREFIX */
1648 /* IFLA_IPTUN_6RD_PREFIXLEN */
1650 /* IFLA_IPTUN_6RD_RELAY_PREFIXLEN */
1653 /* IFLA_IPTUN_ENCAP_TYPE */
1655 /* IFLA_IPTUN_ENCAP_FLAGS */
1657 /* IFLA_IPTUN_ENCAP_SPORT */
1659 /* IFLA_IPTUN_ENCAP_DPORT */
1664 static int ipip6_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
1666 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1667 struct ip_tunnel_parm
*parm
= &tunnel
->parms
;
1669 if (nla_put_u32(skb
, IFLA_IPTUN_LINK
, parm
->link
) ||
1670 nla_put_in_addr(skb
, IFLA_IPTUN_LOCAL
, parm
->iph
.saddr
) ||
1671 nla_put_in_addr(skb
, IFLA_IPTUN_REMOTE
, parm
->iph
.daddr
) ||
1672 nla_put_u8(skb
, IFLA_IPTUN_TTL
, parm
->iph
.ttl
) ||
1673 nla_put_u8(skb
, IFLA_IPTUN_TOS
, parm
->iph
.tos
) ||
1674 nla_put_u8(skb
, IFLA_IPTUN_PMTUDISC
,
1675 !!(parm
->iph
.frag_off
& htons(IP_DF
))) ||
1676 nla_put_u8(skb
, IFLA_IPTUN_PROTO
, parm
->iph
.protocol
) ||
1677 nla_put_be16(skb
, IFLA_IPTUN_FLAGS
, parm
->i_flags
))
1678 goto nla_put_failure
;
1680 #ifdef CONFIG_IPV6_SIT_6RD
1681 if (nla_put_in6_addr(skb
, IFLA_IPTUN_6RD_PREFIX
,
1682 &tunnel
->ip6rd
.prefix
) ||
1683 nla_put_in_addr(skb
, IFLA_IPTUN_6RD_RELAY_PREFIX
,
1684 tunnel
->ip6rd
.relay_prefix
) ||
1685 nla_put_u16(skb
, IFLA_IPTUN_6RD_PREFIXLEN
,
1686 tunnel
->ip6rd
.prefixlen
) ||
1687 nla_put_u16(skb
, IFLA_IPTUN_6RD_RELAY_PREFIXLEN
,
1688 tunnel
->ip6rd
.relay_prefixlen
))
1689 goto nla_put_failure
;
1692 if (nla_put_u16(skb
, IFLA_IPTUN_ENCAP_TYPE
,
1693 tunnel
->encap
.type
) ||
1694 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_SPORT
,
1695 tunnel
->encap
.sport
) ||
1696 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_DPORT
,
1697 tunnel
->encap
.dport
) ||
1698 nla_put_u16(skb
, IFLA_IPTUN_ENCAP_FLAGS
,
1699 tunnel
->encap
.flags
))
1700 goto nla_put_failure
;
1708 static const struct nla_policy ipip6_policy
[IFLA_IPTUN_MAX
+ 1] = {
1709 [IFLA_IPTUN_LINK
] = { .type
= NLA_U32
},
1710 [IFLA_IPTUN_LOCAL
] = { .type
= NLA_U32
},
1711 [IFLA_IPTUN_REMOTE
] = { .type
= NLA_U32
},
1712 [IFLA_IPTUN_TTL
] = { .type
= NLA_U8
},
1713 [IFLA_IPTUN_TOS
] = { .type
= NLA_U8
},
1714 [IFLA_IPTUN_PMTUDISC
] = { .type
= NLA_U8
},
1715 [IFLA_IPTUN_FLAGS
] = { .type
= NLA_U16
},
1716 [IFLA_IPTUN_PROTO
] = { .type
= NLA_U8
},
1717 #ifdef CONFIG_IPV6_SIT_6RD
1718 [IFLA_IPTUN_6RD_PREFIX
] = { .len
= sizeof(struct in6_addr
) },
1719 [IFLA_IPTUN_6RD_RELAY_PREFIX
] = { .type
= NLA_U32
},
1720 [IFLA_IPTUN_6RD_PREFIXLEN
] = { .type
= NLA_U16
},
1721 [IFLA_IPTUN_6RD_RELAY_PREFIXLEN
] = { .type
= NLA_U16
},
1723 [IFLA_IPTUN_ENCAP_TYPE
] = { .type
= NLA_U16
},
1724 [IFLA_IPTUN_ENCAP_FLAGS
] = { .type
= NLA_U16
},
1725 [IFLA_IPTUN_ENCAP_SPORT
] = { .type
= NLA_U16
},
1726 [IFLA_IPTUN_ENCAP_DPORT
] = { .type
= NLA_U16
},
1729 static void ipip6_dellink(struct net_device
*dev
, struct list_head
*head
)
1731 struct net
*net
= dev_net(dev
);
1732 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1734 if (dev
!= sitn
->fb_tunnel_dev
)
1735 unregister_netdevice_queue(dev
, head
);
1738 static struct rtnl_link_ops sit_link_ops __read_mostly
= {
1740 .maxtype
= IFLA_IPTUN_MAX
,
1741 .policy
= ipip6_policy
,
1742 .priv_size
= sizeof(struct ip_tunnel
),
1743 .setup
= ipip6_tunnel_setup
,
1744 .validate
= ipip6_validate
,
1745 .newlink
= ipip6_newlink
,
1746 .changelink
= ipip6_changelink
,
1747 .get_size
= ipip6_get_size
,
1748 .fill_info
= ipip6_fill_info
,
1749 .dellink
= ipip6_dellink
,
1750 .get_link_net
= ip_tunnel_get_link_net
,
1753 static struct xfrm_tunnel sit_handler __read_mostly
= {
1754 .handler
= ipip6_rcv
,
1755 .err_handler
= ipip6_err
,
1759 static struct xfrm_tunnel ipip_handler __read_mostly
= {
1760 .handler
= ipip_rcv
,
1761 .err_handler
= ipip6_err
,
1765 static void __net_exit
sit_destroy_tunnels(struct net
*net
,
1766 struct list_head
*head
)
1768 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1769 struct net_device
*dev
, *aux
;
1772 for_each_netdev_safe(net
, dev
, aux
)
1773 if (dev
->rtnl_link_ops
== &sit_link_ops
)
1774 unregister_netdevice_queue(dev
, head
);
1776 for (prio
= 1; prio
< 4; prio
++) {
1778 for (h
= 0; h
< HASH_SIZE
; h
++) {
1779 struct ip_tunnel
*t
;
1781 t
= rtnl_dereference(sitn
->tunnels
[prio
][h
]);
1783 /* If dev is in the same netns, it has already
1784 * been added to the list by the previous loop.
1786 if (!net_eq(dev_net(t
->dev
), net
))
1787 unregister_netdevice_queue(t
->dev
,
1789 t
= rtnl_dereference(t
->next
);
1795 static int __net_init
sit_init_net(struct net
*net
)
1797 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1798 struct ip_tunnel
*t
;
1801 sitn
->tunnels
[0] = sitn
->tunnels_wc
;
1802 sitn
->tunnels
[1] = sitn
->tunnels_l
;
1803 sitn
->tunnels
[2] = sitn
->tunnels_r
;
1804 sitn
->tunnels
[3] = sitn
->tunnels_r_l
;
1806 sitn
->fb_tunnel_dev
= alloc_netdev(sizeof(struct ip_tunnel
), "sit0",
1808 ipip6_tunnel_setup
);
1809 if (!sitn
->fb_tunnel_dev
) {
1813 dev_net_set(sitn
->fb_tunnel_dev
, net
);
1814 sitn
->fb_tunnel_dev
->rtnl_link_ops
= &sit_link_ops
;
1815 /* FB netdevice is special: we have one, and only one per netns.
1816 * Allowing to move it to another netns is clearly unsafe.
1818 sitn
->fb_tunnel_dev
->features
|= NETIF_F_NETNS_LOCAL
;
1820 err
= register_netdev(sitn
->fb_tunnel_dev
);
1824 ipip6_tunnel_clone_6rd(sitn
->fb_tunnel_dev
, sitn
);
1825 ipip6_fb_tunnel_init(sitn
->fb_tunnel_dev
);
1827 t
= netdev_priv(sitn
->fb_tunnel_dev
);
1829 strcpy(t
->parms
.name
, sitn
->fb_tunnel_dev
->name
);
1833 ipip6_dev_free(sitn
->fb_tunnel_dev
);
1838 static void __net_exit
sit_exit_net(struct net
*net
)
1843 sit_destroy_tunnels(net
, &list
);
1844 unregister_netdevice_many(&list
);
1848 static struct pernet_operations sit_net_ops
= {
1849 .init
= sit_init_net
,
1850 .exit
= sit_exit_net
,
1852 .size
= sizeof(struct sit_net
),
1855 static void __exit
sit_cleanup(void)
1857 rtnl_link_unregister(&sit_link_ops
);
1858 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1859 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
1861 unregister_pernet_device(&sit_net_ops
);
1862 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1865 static int __init
sit_init(void)
1869 pr_info("IPv6 over IPv4 tunneling driver\n");
1871 err
= register_pernet_device(&sit_net_ops
);
1874 err
= xfrm4_tunnel_register(&sit_handler
, AF_INET6
);
1876 pr_info("%s: can't register ip6ip4\n", __func__
);
1877 goto xfrm_tunnel_failed
;
1879 err
= xfrm4_tunnel_register(&ipip_handler
, AF_INET
);
1881 pr_info("%s: can't register ip4ip4\n", __func__
);
1882 goto xfrm_tunnel4_failed
;
1884 err
= rtnl_link_register(&sit_link_ops
);
1886 goto rtnl_link_failed
;
1892 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
1893 xfrm_tunnel4_failed
:
1894 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1896 unregister_pernet_device(&sit_net_ops
);
1900 module_init(sit_init
);
1901 module_exit(sit_cleanup
);
1902 MODULE_LICENSE("GPL");
1903 MODULE_ALIAS_RTNL_LINK("sit");
1904 MODULE_ALIAS_NETDEV("sit0");