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 #include <linux/module.h>
21 #include <linux/capability.h>
22 #include <linux/errno.h>
23 #include <linux/types.h>
24 #include <linux/socket.h>
25 #include <linux/sockios.h>
26 #include <linux/net.h>
27 #include <linux/in6.h>
28 #include <linux/netdevice.h>
29 #include <linux/if_arp.h>
30 #include <linux/icmp.h>
31 #include <asm/uaccess.h>
32 #include <linux/init.h>
33 #include <linux/netfilter_ipv4.h>
34 #include <linux/if_ether.h>
40 #include <net/protocol.h>
41 #include <net/transp_v6.h>
42 #include <net/ip6_fib.h>
43 #include <net/ip6_route.h>
44 #include <net/ndisc.h>
45 #include <net/addrconf.h>
50 #include <net/inet_ecn.h>
52 #include <net/dsfield.h>
53 #include <net/net_namespace.h>
54 #include <net/netns/generic.h>
57 This version of net/ipv6/sit.c is cloned of net/ipv4/ip_gre.c
59 For comments look at net/ipv4/ip_gre.c --ANK
63 #define HASH(addr) (((__force u32)addr^((__force u32)addr>>4))&0xF)
65 static void ipip6_fb_tunnel_init(struct net_device
*dev
);
66 static void ipip6_tunnel_init(struct net_device
*dev
);
67 static void ipip6_tunnel_setup(struct net_device
*dev
);
69 static int sit_net_id
;
71 struct ip_tunnel
*tunnels_r_l
[HASH_SIZE
];
72 struct ip_tunnel
*tunnels_r
[HASH_SIZE
];
73 struct ip_tunnel
*tunnels_l
[HASH_SIZE
];
74 struct ip_tunnel
*tunnels_wc
[1];
75 struct ip_tunnel
**tunnels
[4];
77 struct net_device
*fb_tunnel_dev
;
80 static DEFINE_RWLOCK(ipip6_lock
);
82 static struct ip_tunnel
* ipip6_tunnel_lookup(struct net
*net
,
83 struct net_device
*dev
, __be32 remote
, __be32 local
)
85 unsigned h0
= HASH(remote
);
86 unsigned h1
= HASH(local
);
88 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
90 for (t
= sitn
->tunnels_r_l
[h0
^h1
]; t
; t
= t
->next
) {
91 if (local
== t
->parms
.iph
.saddr
&&
92 remote
== t
->parms
.iph
.daddr
&&
93 (!dev
|| !t
->parms
.link
|| dev
->iflink
== t
->parms
.link
) &&
94 (t
->dev
->flags
& IFF_UP
))
97 for (t
= sitn
->tunnels_r
[h0
]; t
; t
= t
->next
) {
98 if (remote
== t
->parms
.iph
.daddr
&&
99 (!dev
|| !t
->parms
.link
|| dev
->iflink
== t
->parms
.link
) &&
100 (t
->dev
->flags
& IFF_UP
))
103 for (t
= sitn
->tunnels_l
[h1
]; t
; t
= t
->next
) {
104 if (local
== t
->parms
.iph
.saddr
&&
105 (!dev
|| !t
->parms
.link
|| dev
->iflink
== t
->parms
.link
) &&
106 (t
->dev
->flags
& IFF_UP
))
109 t
= sitn
->tunnels_wc
[0];
110 if ((t
!= NULL
) && (t
->dev
->flags
& IFF_UP
))
115 static struct ip_tunnel
**__ipip6_bucket(struct sit_net
*sitn
,
116 struct ip_tunnel_parm
*parms
)
118 __be32 remote
= parms
->iph
.daddr
;
119 __be32 local
= parms
->iph
.saddr
;
131 return &sitn
->tunnels
[prio
][h
];
134 static inline struct ip_tunnel
**ipip6_bucket(struct sit_net
*sitn
,
137 return __ipip6_bucket(sitn
, &t
->parms
);
140 static void ipip6_tunnel_unlink(struct sit_net
*sitn
, struct ip_tunnel
*t
)
142 struct ip_tunnel
**tp
;
144 for (tp
= ipip6_bucket(sitn
, t
); *tp
; tp
= &(*tp
)->next
) {
146 write_lock_bh(&ipip6_lock
);
148 write_unlock_bh(&ipip6_lock
);
154 static void ipip6_tunnel_link(struct sit_net
*sitn
, struct ip_tunnel
*t
)
156 struct ip_tunnel
**tp
= ipip6_bucket(sitn
, t
);
159 write_lock_bh(&ipip6_lock
);
161 write_unlock_bh(&ipip6_lock
);
164 static void ipip6_tunnel_clone_6rd(struct net_device
*dev
, struct sit_net
*sitn
)
166 #ifdef CONFIG_IPV6_SIT_6RD
167 struct ip_tunnel
*t
= netdev_priv(dev
);
169 if (t
->dev
== sitn
->fb_tunnel_dev
) {
170 ipv6_addr_set(&t
->ip6rd
.prefix
, htonl(0x20020000), 0, 0, 0);
171 t
->ip6rd
.relay_prefix
= 0;
172 t
->ip6rd
.prefixlen
= 16;
173 t
->ip6rd
.relay_prefixlen
= 0;
175 struct ip_tunnel
*t0
= netdev_priv(sitn
->fb_tunnel_dev
);
176 memcpy(&t
->ip6rd
, &t0
->ip6rd
, sizeof(t
->ip6rd
));
181 static struct ip_tunnel
* ipip6_tunnel_locate(struct net
*net
,
182 struct ip_tunnel_parm
*parms
, int create
)
184 __be32 remote
= parms
->iph
.daddr
;
185 __be32 local
= parms
->iph
.saddr
;
186 struct ip_tunnel
*t
, **tp
, *nt
;
187 struct net_device
*dev
;
189 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
191 for (tp
= __ipip6_bucket(sitn
, parms
); (t
= *tp
) != NULL
; tp
= &t
->next
) {
192 if (local
== t
->parms
.iph
.saddr
&&
193 remote
== t
->parms
.iph
.daddr
&&
194 parms
->link
== t
->parms
.link
) {
205 strlcpy(name
, parms
->name
, IFNAMSIZ
);
207 sprintf(name
, "sit%%d");
209 dev
= alloc_netdev(sizeof(*t
), name
, ipip6_tunnel_setup
);
213 dev_net_set(dev
, net
);
215 if (strchr(name
, '%')) {
216 if (dev_alloc_name(dev
, name
) < 0)
220 nt
= netdev_priv(dev
);
223 ipip6_tunnel_init(dev
);
224 ipip6_tunnel_clone_6rd(dev
, sitn
);
226 if (parms
->i_flags
& SIT_ISATAP
)
227 dev
->priv_flags
|= IFF_ISATAP
;
229 if (register_netdevice(dev
) < 0)
234 ipip6_tunnel_link(sitn
, nt
);
243 static DEFINE_SPINLOCK(ipip6_prl_lock
);
245 #define for_each_prl_rcu(start) \
246 for (prl = rcu_dereference(start); \
248 prl = rcu_dereference(prl->next))
250 static struct ip_tunnel_prl_entry
*
251 __ipip6_tunnel_locate_prl(struct ip_tunnel
*t
, __be32 addr
)
253 struct ip_tunnel_prl_entry
*prl
;
255 for_each_prl_rcu(t
->prl
)
256 if (prl
->addr
== addr
)
262 static int ipip6_tunnel_get_prl(struct ip_tunnel
*t
,
263 struct ip_tunnel_prl __user
*a
)
265 struct ip_tunnel_prl kprl
, *kp
;
266 struct ip_tunnel_prl_entry
*prl
;
267 unsigned int cmax
, c
= 0, ca
, len
;
270 if (copy_from_user(&kprl
, a
, sizeof(kprl
)))
272 cmax
= kprl
.datalen
/ sizeof(kprl
);
273 if (cmax
> 1 && kprl
.addr
!= htonl(INADDR_ANY
))
276 /* For simple GET or for root users,
277 * we try harder to allocate.
279 kp
= (cmax
<= 1 || capable(CAP_NET_ADMIN
)) ?
280 kcalloc(cmax
, sizeof(*kp
), GFP_KERNEL
) :
285 ca
= t
->prl_count
< cmax
? t
->prl_count
: cmax
;
288 /* We don't try hard to allocate much memory for
290 * For root users, retry allocating enough memory for
293 kp
= kcalloc(ca
, sizeof(*kp
), GFP_ATOMIC
);
301 for_each_prl_rcu(t
->prl
) {
304 if (kprl
.addr
!= htonl(INADDR_ANY
) && prl
->addr
!= kprl
.addr
)
306 kp
[c
].addr
= prl
->addr
;
307 kp
[c
].flags
= prl
->flags
;
309 if (kprl
.addr
!= htonl(INADDR_ANY
))
315 len
= sizeof(*kp
) * c
;
317 if ((len
&& copy_to_user(a
+ 1, kp
, len
)) || put_user(len
, &a
->datalen
))
326 ipip6_tunnel_add_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
, int chg
)
328 struct ip_tunnel_prl_entry
*p
;
331 if (a
->addr
== htonl(INADDR_ANY
))
334 spin_lock(&ipip6_prl_lock
);
336 for (p
= t
->prl
; p
; p
= p
->next
) {
337 if (p
->addr
== a
->addr
) {
352 p
= kzalloc(sizeof(struct ip_tunnel_prl_entry
), GFP_KERNEL
);
358 INIT_RCU_HEAD(&p
->rcu_head
);
363 rcu_assign_pointer(t
->prl
, p
);
365 spin_unlock(&ipip6_prl_lock
);
369 static void prl_entry_destroy_rcu(struct rcu_head
*head
)
371 kfree(container_of(head
, struct ip_tunnel_prl_entry
, rcu_head
));
374 static void prl_list_destroy_rcu(struct rcu_head
*head
)
376 struct ip_tunnel_prl_entry
*p
, *n
;
378 p
= container_of(head
, struct ip_tunnel_prl_entry
, rcu_head
);
387 ipip6_tunnel_del_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
)
389 struct ip_tunnel_prl_entry
*x
, **p
;
392 spin_lock(&ipip6_prl_lock
);
394 if (a
&& a
->addr
!= htonl(INADDR_ANY
)) {
395 for (p
= &t
->prl
; *p
; p
= &(*p
)->next
) {
396 if ((*p
)->addr
== a
->addr
) {
399 call_rcu(&x
->rcu_head
, prl_entry_destroy_rcu
);
409 call_rcu(&x
->rcu_head
, prl_list_destroy_rcu
);
414 spin_unlock(&ipip6_prl_lock
);
419 isatap_chksrc(struct sk_buff
*skb
, struct iphdr
*iph
, struct ip_tunnel
*t
)
421 struct ip_tunnel_prl_entry
*p
;
425 p
= __ipip6_tunnel_locate_prl(t
, iph
->saddr
);
427 if (p
->flags
& PRL_DEFAULT
)
428 skb
->ndisc_nodetype
= NDISC_NODETYPE_DEFAULT
;
430 skb
->ndisc_nodetype
= NDISC_NODETYPE_NODEFAULT
;
432 struct in6_addr
*addr6
= &ipv6_hdr(skb
)->saddr
;
433 if (ipv6_addr_is_isatap(addr6
) &&
434 (addr6
->s6_addr32
[3] == iph
->saddr
) &&
435 ipv6_chk_prefix(addr6
, t
->dev
))
436 skb
->ndisc_nodetype
= NDISC_NODETYPE_HOST
;
444 static void ipip6_tunnel_uninit(struct net_device
*dev
)
446 struct net
*net
= dev_net(dev
);
447 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
449 if (dev
== sitn
->fb_tunnel_dev
) {
450 write_lock_bh(&ipip6_lock
);
451 sitn
->tunnels_wc
[0] = NULL
;
452 write_unlock_bh(&ipip6_lock
);
455 ipip6_tunnel_unlink(sitn
, netdev_priv(dev
));
456 ipip6_tunnel_del_prl(netdev_priv(dev
), NULL
);
462 static int ipip6_err(struct sk_buff
*skb
, u32 info
)
465 /* All the routers (except for Linux) return only
466 8 bytes of packet payload. It means, that precise relaying of
467 ICMP in the real Internet is absolutely infeasible.
469 struct iphdr
*iph
= (struct iphdr
*)skb
->data
;
470 const int type
= icmp_hdr(skb
)->type
;
471 const int code
= icmp_hdr(skb
)->code
;
477 case ICMP_PARAMETERPROB
:
480 case ICMP_DEST_UNREACH
:
483 case ICMP_PORT_UNREACH
:
484 /* Impossible event. */
486 case ICMP_FRAG_NEEDED
:
487 /* Soft state for pmtu is maintained by IP core. */
490 /* All others are translated to HOST_UNREACH.
491 rfc2003 contains "deep thoughts" about NET_UNREACH,
492 I believe they are just ether pollution. --ANK
497 case ICMP_TIME_EXCEEDED
:
498 if (code
!= ICMP_EXC_TTL
)
505 read_lock(&ipip6_lock
);
506 t
= ipip6_tunnel_lookup(dev_net(skb
->dev
),
510 if (t
== NULL
|| t
->parms
.iph
.daddr
== 0)
514 if (t
->parms
.iph
.ttl
== 0 && type
== ICMP_TIME_EXCEEDED
)
517 if (time_before(jiffies
, t
->err_time
+ IPTUNNEL_ERR_TIMEO
))
521 t
->err_time
= jiffies
;
523 read_unlock(&ipip6_lock
);
527 static inline void ipip6_ecn_decapsulate(struct iphdr
*iph
, struct sk_buff
*skb
)
529 if (INET_ECN_is_ce(iph
->tos
))
530 IP6_ECN_set_ce(ipv6_hdr(skb
));
533 static int ipip6_rcv(struct sk_buff
*skb
)
536 struct ip_tunnel
*tunnel
;
538 if (!pskb_may_pull(skb
, sizeof(struct ipv6hdr
)))
543 read_lock(&ipip6_lock
);
544 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
545 iph
->saddr
, iph
->daddr
);
546 if (tunnel
!= NULL
) {
548 skb
->mac_header
= skb
->network_header
;
549 skb_reset_network_header(skb
);
550 IPCB(skb
)->flags
= 0;
551 skb
->protocol
= htons(ETH_P_IPV6
);
552 skb
->pkt_type
= PACKET_HOST
;
554 if ((tunnel
->dev
->priv_flags
& IFF_ISATAP
) &&
555 !isatap_chksrc(skb
, iph
, tunnel
)) {
556 tunnel
->dev
->stats
.rx_errors
++;
557 read_unlock(&ipip6_lock
);
561 tunnel
->dev
->stats
.rx_packets
++;
562 tunnel
->dev
->stats
.rx_bytes
+= skb
->len
;
563 skb
->dev
= tunnel
->dev
;
566 ipip6_ecn_decapsulate(iph
, skb
);
568 read_unlock(&ipip6_lock
);
572 icmp_send(skb
, ICMP_DEST_UNREACH
, ICMP_PORT_UNREACH
, 0);
573 read_unlock(&ipip6_lock
);
580 * Returns the embedded IPv4 address if the IPv6 address
581 * comes from 6rd / 6to4 (RFC 3056) addr space.
584 __be32
try_6rd(struct in6_addr
*v6dst
, struct ip_tunnel
*tunnel
)
588 #ifdef CONFIG_IPV6_SIT_6RD
589 if (ipv6_prefix_equal(v6dst
, &tunnel
->ip6rd
.prefix
,
590 tunnel
->ip6rd
.prefixlen
)) {
595 pbw0
= tunnel
->ip6rd
.prefixlen
>> 5;
596 pbi0
= tunnel
->ip6rd
.prefixlen
& 0x1f;
598 d
= (ntohl(v6dst
->s6_addr32
[pbw0
]) << pbi0
) >>
599 tunnel
->ip6rd
.relay_prefixlen
;
601 pbi1
= pbi0
- tunnel
->ip6rd
.relay_prefixlen
;
603 d
|= ntohl(v6dst
->s6_addr32
[pbw0
+ 1]) >>
606 dst
= tunnel
->ip6rd
.relay_prefix
| htonl(d
);
609 if (v6dst
->s6_addr16
[0] == htons(0x2002)) {
610 /* 6to4 v6 addr has 16 bits prefix, 32 v4addr, 16 SLA, ... */
611 memcpy(&dst
, &v6dst
->s6_addr16
[1], 4);
618 * This function assumes it is being called from dev_queue_xmit()
619 * and that skb is filled properly by that function.
622 static netdev_tx_t
ipip6_tunnel_xmit(struct sk_buff
*skb
,
623 struct net_device
*dev
)
625 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
626 struct net_device_stats
*stats
= &dev
->stats
;
627 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, 0);
628 struct iphdr
*tiph
= &tunnel
->parms
.iph
;
629 struct ipv6hdr
*iph6
= ipv6_hdr(skb
);
630 u8 tos
= tunnel
->parms
.iph
.tos
;
631 struct rtable
*rt
; /* Route to the other host */
632 struct net_device
*tdev
; /* Device to other host */
633 struct iphdr
*iph
; /* Our new IP header */
634 unsigned int max_headroom
; /* The extra header space needed */
635 __be32 dst
= tiph
->daddr
;
637 struct in6_addr
*addr6
;
640 if (skb
->protocol
!= htons(ETH_P_IPV6
))
643 /* ISATAP (RFC4214) - must come before 6to4 */
644 if (dev
->priv_flags
& IFF_ISATAP
) {
645 struct neighbour
*neigh
= NULL
;
648 neigh
= skb_dst(skb
)->neighbour
;
652 printk(KERN_DEBUG
"sit: nexthop == NULL\n");
656 addr6
= (struct in6_addr
*)&neigh
->primary_key
;
657 addr_type
= ipv6_addr_type(addr6
);
659 if ((addr_type
& IPV6_ADDR_UNICAST
) &&
660 ipv6_addr_is_isatap(addr6
))
661 dst
= addr6
->s6_addr32
[3];
667 dst
= try_6rd(&iph6
->daddr
, tunnel
);
670 struct neighbour
*neigh
= NULL
;
673 neigh
= skb_dst(skb
)->neighbour
;
677 printk(KERN_DEBUG
"sit: nexthop == NULL\n");
681 addr6
= (struct in6_addr
*)&neigh
->primary_key
;
682 addr_type
= ipv6_addr_type(addr6
);
684 if (addr_type
== IPV6_ADDR_ANY
) {
685 addr6
= &ipv6_hdr(skb
)->daddr
;
686 addr_type
= ipv6_addr_type(addr6
);
689 if ((addr_type
& IPV6_ADDR_COMPATv4
) == 0)
692 dst
= addr6
->s6_addr32
[3];
696 struct flowi fl
= { .nl_u
= { .ip4_u
=
698 .saddr
= tiph
->saddr
,
699 .tos
= RT_TOS(tos
) } },
700 .oif
= tunnel
->parms
.link
,
701 .proto
= IPPROTO_IPV6
};
702 if (ip_route_output_key(dev_net(dev
), &rt
, &fl
)) {
703 stats
->tx_carrier_errors
++;
707 if (rt
->rt_type
!= RTN_UNICAST
) {
709 stats
->tx_carrier_errors
++;
712 tdev
= rt
->u
.dst
.dev
;
721 mtu
= dst_mtu(&rt
->u
.dst
) - sizeof(struct iphdr
);
723 mtu
= skb_dst(skb
) ? dst_mtu(skb_dst(skb
)) : dev
->mtu
;
730 if (mtu
< IPV6_MIN_MTU
)
732 if (tunnel
->parms
.iph
.daddr
&& skb_dst(skb
))
733 skb_dst(skb
)->ops
->update_pmtu(skb_dst(skb
), mtu
);
735 if (skb
->len
> mtu
) {
736 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
, dev
);
741 if (tunnel
->err_count
> 0) {
742 if (time_before(jiffies
,
743 tunnel
->err_time
+ IPTUNNEL_ERR_TIMEO
)) {
745 dst_link_failure(skb
);
747 tunnel
->err_count
= 0;
751 * Okay, now see if we can stuff it in the buffer as-is.
753 max_headroom
= LL_RESERVED_SPACE(tdev
)+sizeof(struct iphdr
);
755 if (skb_headroom(skb
) < max_headroom
|| skb_shared(skb
) ||
756 (skb_cloned(skb
) && !skb_clone_writable(skb
, 0))) {
757 struct sk_buff
*new_skb
= skb_realloc_headroom(skb
, max_headroom
);
765 skb_set_owner_w(new_skb
, skb
->sk
);
768 iph6
= ipv6_hdr(skb
);
771 skb
->transport_header
= skb
->network_header
;
772 skb_push(skb
, sizeof(struct iphdr
));
773 skb_reset_network_header(skb
);
774 memset(&(IPCB(skb
)->opt
), 0, sizeof(IPCB(skb
)->opt
));
775 IPCB(skb
)->flags
= 0;
777 skb_dst_set(skb
, &rt
->u
.dst
);
780 * Push down and install the IPIP header.
785 iph
->ihl
= sizeof(struct iphdr
)>>2;
786 if (mtu
> IPV6_MIN_MTU
)
787 iph
->frag_off
= tiph
->frag_off
;
791 iph
->protocol
= IPPROTO_IPV6
;
792 iph
->tos
= INET_ECN_encapsulate(tos
, ipv6_get_dsfield(iph6
));
793 iph
->daddr
= rt
->rt_dst
;
794 iph
->saddr
= rt
->rt_src
;
796 if ((iph
->ttl
= tiph
->ttl
) == 0)
797 iph
->ttl
= iph6
->hop_limit
;
805 dst_link_failure(skb
);
812 static void ipip6_tunnel_bind_dev(struct net_device
*dev
)
814 struct net_device
*tdev
= NULL
;
815 struct ip_tunnel
*tunnel
;
818 tunnel
= netdev_priv(dev
);
819 iph
= &tunnel
->parms
.iph
;
822 struct flowi fl
= { .nl_u
= { .ip4_u
=
823 { .daddr
= iph
->daddr
,
825 .tos
= RT_TOS(iph
->tos
) } },
826 .oif
= tunnel
->parms
.link
,
827 .proto
= IPPROTO_IPV6
};
829 if (!ip_route_output_key(dev_net(dev
), &rt
, &fl
)) {
830 tdev
= rt
->u
.dst
.dev
;
833 dev
->flags
|= IFF_POINTOPOINT
;
836 if (!tdev
&& tunnel
->parms
.link
)
837 tdev
= __dev_get_by_index(dev_net(dev
), tunnel
->parms
.link
);
840 dev
->hard_header_len
= tdev
->hard_header_len
+ sizeof(struct iphdr
);
841 dev
->mtu
= tdev
->mtu
- sizeof(struct iphdr
);
842 if (dev
->mtu
< IPV6_MIN_MTU
)
843 dev
->mtu
= IPV6_MIN_MTU
;
845 dev
->iflink
= tunnel
->parms
.link
;
849 ipip6_tunnel_ioctl (struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
852 struct ip_tunnel_parm p
;
853 struct ip_tunnel_prl prl
;
855 struct net
*net
= dev_net(dev
);
856 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
857 #ifdef CONFIG_IPV6_SIT_6RD
858 struct ip_tunnel_6rd ip6rd
;
863 #ifdef CONFIG_IPV6_SIT_6RD
867 if (dev
== sitn
->fb_tunnel_dev
) {
868 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
))) {
872 t
= ipip6_tunnel_locate(net
, &p
, 0);
875 t
= netdev_priv(dev
);
878 if (cmd
== SIOCGETTUNNEL
) {
879 memcpy(&p
, &t
->parms
, sizeof(p
));
880 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
,
883 #ifdef CONFIG_IPV6_SIT_6RD
885 ipv6_addr_copy(&ip6rd
.prefix
, &t
->ip6rd
.prefix
);
886 ip6rd
.relay_prefix
= t
->ip6rd
.relay_prefix
;
887 ip6rd
.prefixlen
= t
->ip6rd
.prefixlen
;
888 ip6rd
.relay_prefixlen
= t
->ip6rd
.relay_prefixlen
;
889 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &ip6rd
,
900 if (!capable(CAP_NET_ADMIN
))
904 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
908 if (p
.iph
.version
!= 4 || p
.iph
.protocol
!= IPPROTO_IPV6
||
909 p
.iph
.ihl
!= 5 || (p
.iph
.frag_off
&htons(~IP_DF
)))
912 p
.iph
.frag_off
|= htons(IP_DF
);
914 t
= ipip6_tunnel_locate(net
, &p
, cmd
== SIOCADDTUNNEL
);
916 if (dev
!= sitn
->fb_tunnel_dev
&& cmd
== SIOCCHGTUNNEL
) {
923 if (((dev
->flags
&IFF_POINTOPOINT
) && !p
.iph
.daddr
) ||
924 (!(dev
->flags
&IFF_POINTOPOINT
) && p
.iph
.daddr
)) {
928 t
= netdev_priv(dev
);
929 ipip6_tunnel_unlink(sitn
, t
);
930 t
->parms
.iph
.saddr
= p
.iph
.saddr
;
931 t
->parms
.iph
.daddr
= p
.iph
.daddr
;
932 memcpy(dev
->dev_addr
, &p
.iph
.saddr
, 4);
933 memcpy(dev
->broadcast
, &p
.iph
.daddr
, 4);
934 ipip6_tunnel_link(sitn
, t
);
935 netdev_state_change(dev
);
941 if (cmd
== SIOCCHGTUNNEL
) {
942 t
->parms
.iph
.ttl
= p
.iph
.ttl
;
943 t
->parms
.iph
.tos
= p
.iph
.tos
;
944 if (t
->parms
.link
!= p
.link
) {
945 t
->parms
.link
= p
.link
;
946 ipip6_tunnel_bind_dev(dev
);
947 netdev_state_change(dev
);
950 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &t
->parms
, sizeof(p
)))
953 err
= (cmd
== SIOCADDTUNNEL
? -ENOBUFS
: -ENOENT
);
958 if (!capable(CAP_NET_ADMIN
))
961 if (dev
== sitn
->fb_tunnel_dev
) {
963 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
966 if ((t
= ipip6_tunnel_locate(net
, &p
, 0)) == NULL
)
969 if (t
== netdev_priv(sitn
->fb_tunnel_dev
))
973 unregister_netdevice(dev
);
979 if (dev
== sitn
->fb_tunnel_dev
)
982 if (!(t
= netdev_priv(dev
)))
984 err
= ipip6_tunnel_get_prl(t
, ifr
->ifr_ifru
.ifru_data
);
991 if (!capable(CAP_NET_ADMIN
))
994 if (dev
== sitn
->fb_tunnel_dev
)
997 if (copy_from_user(&prl
, ifr
->ifr_ifru
.ifru_data
, sizeof(prl
)))
1000 if (!(t
= netdev_priv(dev
)))
1005 err
= ipip6_tunnel_del_prl(t
, &prl
);
1009 err
= ipip6_tunnel_add_prl(t
, &prl
, cmd
== SIOCCHGPRL
);
1012 netdev_state_change(dev
);
1015 #ifdef CONFIG_IPV6_SIT_6RD
1020 if (!capable(CAP_NET_ADMIN
))
1024 if (copy_from_user(&ip6rd
, ifr
->ifr_ifru
.ifru_data
,
1028 t
= netdev_priv(dev
);
1030 if (cmd
!= SIOCDEL6RD
) {
1031 struct in6_addr prefix
;
1032 __be32 relay_prefix
;
1035 if (ip6rd
.relay_prefixlen
> 32 ||
1036 ip6rd
.prefixlen
+ (32 - ip6rd
.relay_prefixlen
) > 64)
1039 ipv6_addr_prefix(&prefix
, &ip6rd
.prefix
,
1041 if (!ipv6_addr_equal(&prefix
, &ip6rd
.prefix
))
1043 if (ip6rd
.relay_prefixlen
)
1044 relay_prefix
= ip6rd
.relay_prefix
&
1045 htonl(0xffffffffUL
<<
1046 (32 - ip6rd
.relay_prefixlen
));
1049 if (relay_prefix
!= ip6rd
.relay_prefix
)
1052 ipv6_addr_copy(&t
->ip6rd
.prefix
, &prefix
);
1053 t
->ip6rd
.relay_prefix
= relay_prefix
;
1054 t
->ip6rd
.prefixlen
= ip6rd
.prefixlen
;
1055 t
->ip6rd
.relay_prefixlen
= ip6rd
.relay_prefixlen
;
1057 ipip6_tunnel_clone_6rd(dev
, sitn
);
1071 static int ipip6_tunnel_change_mtu(struct net_device
*dev
, int new_mtu
)
1073 if (new_mtu
< IPV6_MIN_MTU
|| new_mtu
> 0xFFF8 - sizeof(struct iphdr
))
1079 static const struct net_device_ops ipip6_netdev_ops
= {
1080 .ndo_uninit
= ipip6_tunnel_uninit
,
1081 .ndo_start_xmit
= ipip6_tunnel_xmit
,
1082 .ndo_do_ioctl
= ipip6_tunnel_ioctl
,
1083 .ndo_change_mtu
= ipip6_tunnel_change_mtu
,
1086 static void ipip6_tunnel_setup(struct net_device
*dev
)
1088 dev
->netdev_ops
= &ipip6_netdev_ops
;
1089 dev
->destructor
= free_netdev
;
1091 dev
->type
= ARPHRD_SIT
;
1092 dev
->hard_header_len
= LL_MAX_HEADER
+ sizeof(struct iphdr
);
1093 dev
->mtu
= ETH_DATA_LEN
- sizeof(struct iphdr
);
1094 dev
->flags
= IFF_NOARP
;
1095 dev
->priv_flags
&= ~IFF_XMIT_DST_RELEASE
;
1098 dev
->features
|= NETIF_F_NETNS_LOCAL
;
1101 static void ipip6_tunnel_init(struct net_device
*dev
)
1103 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1106 strcpy(tunnel
->parms
.name
, dev
->name
);
1108 memcpy(dev
->dev_addr
, &tunnel
->parms
.iph
.saddr
, 4);
1109 memcpy(dev
->broadcast
, &tunnel
->parms
.iph
.daddr
, 4);
1111 ipip6_tunnel_bind_dev(dev
);
1114 static void ipip6_fb_tunnel_init(struct net_device
*dev
)
1116 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1117 struct iphdr
*iph
= &tunnel
->parms
.iph
;
1118 struct net
*net
= dev_net(dev
);
1119 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1122 strcpy(tunnel
->parms
.name
, dev
->name
);
1125 iph
->protocol
= IPPROTO_IPV6
;
1130 sitn
->tunnels_wc
[0] = tunnel
;
1133 static struct xfrm_tunnel sit_handler
= {
1134 .handler
= ipip6_rcv
,
1135 .err_handler
= ipip6_err
,
1139 static void sit_destroy_tunnels(struct sit_net
*sitn
)
1143 for (prio
= 1; prio
< 4; prio
++) {
1145 for (h
= 0; h
< HASH_SIZE
; h
++) {
1146 struct ip_tunnel
*t
;
1147 while ((t
= sitn
->tunnels
[prio
][h
]) != NULL
)
1148 unregister_netdevice(t
->dev
);
1153 static int sit_init_net(struct net
*net
)
1156 struct sit_net
*sitn
;
1159 sitn
= kzalloc(sizeof(struct sit_net
), GFP_KERNEL
);
1163 err
= net_assign_generic(net
, sit_net_id
, sitn
);
1167 sitn
->tunnels
[0] = sitn
->tunnels_wc
;
1168 sitn
->tunnels
[1] = sitn
->tunnels_l
;
1169 sitn
->tunnels
[2] = sitn
->tunnels_r
;
1170 sitn
->tunnels
[3] = sitn
->tunnels_r_l
;
1172 sitn
->fb_tunnel_dev
= alloc_netdev(sizeof(struct ip_tunnel
), "sit0",
1173 ipip6_tunnel_setup
);
1174 if (!sitn
->fb_tunnel_dev
) {
1178 dev_net_set(sitn
->fb_tunnel_dev
, net
);
1180 ipip6_fb_tunnel_init(sitn
->fb_tunnel_dev
);
1181 ipip6_tunnel_clone_6rd(sitn
->fb_tunnel_dev
, sitn
);
1183 if ((err
= register_netdev(sitn
->fb_tunnel_dev
)))
1189 dev_put(sitn
->fb_tunnel_dev
);
1190 free_netdev(sitn
->fb_tunnel_dev
);
1199 static void sit_exit_net(struct net
*net
)
1201 struct sit_net
*sitn
;
1203 sitn
= net_generic(net
, sit_net_id
);
1205 sit_destroy_tunnels(sitn
);
1206 unregister_netdevice(sitn
->fb_tunnel_dev
);
1211 static struct pernet_operations sit_net_ops
= {
1212 .init
= sit_init_net
,
1213 .exit
= sit_exit_net
,
1216 static void __exit
sit_cleanup(void)
1218 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1220 unregister_pernet_gen_device(sit_net_id
, &sit_net_ops
);
1221 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1224 static int __init
sit_init(void)
1228 printk(KERN_INFO
"IPv6 over IPv4 tunneling driver\n");
1230 if (xfrm4_tunnel_register(&sit_handler
, AF_INET6
) < 0) {
1231 printk(KERN_INFO
"sit init: Can't add protocol\n");
1235 err
= register_pernet_gen_device(&sit_net_id
, &sit_net_ops
);
1237 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1242 module_init(sit_init
);
1243 module_exit(sit_cleanup
);
1244 MODULE_LICENSE("GPL");
1245 MODULE_ALIAS("sit0");