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 <linux/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
65 #define IP6_SIT_HASH_SIZE 16
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 unsigned int sit_net_id __read_mostly
;
81 struct ip_tunnel __rcu
*tunnels_r_l
[IP6_SIT_HASH_SIZE
];
82 struct ip_tunnel __rcu
*tunnels_r
[IP6_SIT_HASH_SIZE
];
83 struct ip_tunnel __rcu
*tunnels_l
[IP6_SIT_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
,
95 __be32 remote
, __be32 local
,
98 unsigned int h0
= HASH(remote
);
99 unsigned int h1
= HASH(local
);
101 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
102 int ifindex
= dev
? dev
->ifindex
: 0;
104 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_r_l
[h0
^ h1
]) {
105 if (local
== t
->parms
.iph
.saddr
&&
106 remote
== t
->parms
.iph
.daddr
&&
107 (!dev
|| !t
->parms
.link
|| ifindex
== t
->parms
.link
||
108 sifindex
== t
->parms
.link
) &&
109 (t
->dev
->flags
& IFF_UP
))
112 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_r
[h0
]) {
113 if (remote
== t
->parms
.iph
.daddr
&&
114 (!dev
|| !t
->parms
.link
|| ifindex
== t
->parms
.link
||
115 sifindex
== t
->parms
.link
) &&
116 (t
->dev
->flags
& IFF_UP
))
119 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_l
[h1
]) {
120 if (local
== t
->parms
.iph
.saddr
&&
121 (!dev
|| !t
->parms
.link
|| ifindex
== t
->parms
.link
||
122 sifindex
== t
->parms
.link
) &&
123 (t
->dev
->flags
& IFF_UP
))
126 t
= rcu_dereference(sitn
->tunnels_wc
[0]);
127 if (t
&& (t
->dev
->flags
& IFF_UP
))
132 static struct ip_tunnel __rcu
**__ipip6_bucket(struct sit_net
*sitn
,
133 struct ip_tunnel_parm
*parms
)
135 __be32 remote
= parms
->iph
.daddr
;
136 __be32 local
= parms
->iph
.saddr
;
148 return &sitn
->tunnels
[prio
][h
];
151 static inline struct ip_tunnel __rcu
**ipip6_bucket(struct sit_net
*sitn
,
154 return __ipip6_bucket(sitn
, &t
->parms
);
157 static void ipip6_tunnel_unlink(struct sit_net
*sitn
, struct ip_tunnel
*t
)
159 struct ip_tunnel __rcu
**tp
;
160 struct ip_tunnel
*iter
;
162 for (tp
= ipip6_bucket(sitn
, t
);
163 (iter
= rtnl_dereference(*tp
)) != NULL
;
166 rcu_assign_pointer(*tp
, t
->next
);
172 static void ipip6_tunnel_link(struct sit_net
*sitn
, struct ip_tunnel
*t
)
174 struct ip_tunnel __rcu
**tp
= ipip6_bucket(sitn
, t
);
176 rcu_assign_pointer(t
->next
, rtnl_dereference(*tp
));
177 rcu_assign_pointer(*tp
, t
);
180 static void ipip6_tunnel_clone_6rd(struct net_device
*dev
, struct sit_net
*sitn
)
182 #ifdef CONFIG_IPV6_SIT_6RD
183 struct ip_tunnel
*t
= netdev_priv(dev
);
185 if (dev
== sitn
->fb_tunnel_dev
|| !sitn
->fb_tunnel_dev
) {
186 ipv6_addr_set(&t
->ip6rd
.prefix
, htonl(0x20020000), 0, 0, 0);
187 t
->ip6rd
.relay_prefix
= 0;
188 t
->ip6rd
.prefixlen
= 16;
189 t
->ip6rd
.relay_prefixlen
= 0;
191 struct ip_tunnel
*t0
= netdev_priv(sitn
->fb_tunnel_dev
);
192 memcpy(&t
->ip6rd
, &t0
->ip6rd
, sizeof(t
->ip6rd
));
197 static int ipip6_tunnel_create(struct net_device
*dev
)
199 struct ip_tunnel
*t
= netdev_priv(dev
);
200 struct net
*net
= dev_net(dev
);
201 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
204 memcpy(dev
->dev_addr
, &t
->parms
.iph
.saddr
, 4);
205 memcpy(dev
->broadcast
, &t
->parms
.iph
.daddr
, 4);
207 if ((__force u16
)t
->parms
.i_flags
& SIT_ISATAP
)
208 dev
->priv_flags
|= IFF_ISATAP
;
210 dev
->rtnl_link_ops
= &sit_link_ops
;
212 err
= register_netdevice(dev
);
216 ipip6_tunnel_clone_6rd(dev
, sitn
);
220 ipip6_tunnel_link(sitn
, t
);
227 static struct ip_tunnel
*ipip6_tunnel_locate(struct net
*net
,
228 struct ip_tunnel_parm
*parms
, int create
)
230 __be32 remote
= parms
->iph
.daddr
;
231 __be32 local
= parms
->iph
.saddr
;
232 struct ip_tunnel
*t
, *nt
;
233 struct ip_tunnel __rcu
**tp
;
234 struct net_device
*dev
;
236 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
238 for (tp
= __ipip6_bucket(sitn
, parms
);
239 (t
= rtnl_dereference(*tp
)) != NULL
;
241 if (local
== t
->parms
.iph
.saddr
&&
242 remote
== t
->parms
.iph
.daddr
&&
243 parms
->link
== t
->parms
.link
) {
253 if (parms
->name
[0]) {
254 if (!dev_valid_name(parms
->name
))
256 strlcpy(name
, parms
->name
, IFNAMSIZ
);
258 strcpy(name
, "sit%d");
260 dev
= alloc_netdev(sizeof(*t
), name
, NET_NAME_UNKNOWN
,
265 dev_net_set(dev
, net
);
267 nt
= netdev_priv(dev
);
270 if (ipip6_tunnel_create(dev
) < 0)
281 #define for_each_prl_rcu(start) \
282 for (prl = rcu_dereference(start); \
284 prl = rcu_dereference(prl->next))
286 static struct ip_tunnel_prl_entry
*
287 __ipip6_tunnel_locate_prl(struct ip_tunnel
*t
, __be32 addr
)
289 struct ip_tunnel_prl_entry
*prl
;
291 for_each_prl_rcu(t
->prl
)
292 if (prl
->addr
== addr
)
298 static int ipip6_tunnel_get_prl(struct ip_tunnel
*t
,
299 struct ip_tunnel_prl __user
*a
)
301 struct ip_tunnel_prl kprl
, *kp
;
302 struct ip_tunnel_prl_entry
*prl
;
303 unsigned int cmax
, c
= 0, ca
, len
;
306 if (copy_from_user(&kprl
, a
, sizeof(kprl
)))
308 cmax
= kprl
.datalen
/ sizeof(kprl
);
309 if (cmax
> 1 && kprl
.addr
!= htonl(INADDR_ANY
))
312 /* For simple GET or for root users,
313 * we try harder to allocate.
315 kp
= (cmax
<= 1 || capable(CAP_NET_ADMIN
)) ?
316 kcalloc(cmax
, sizeof(*kp
), GFP_KERNEL
| __GFP_NOWARN
) :
321 ca
= t
->prl_count
< cmax
? t
->prl_count
: cmax
;
324 /* We don't try hard to allocate much memory for
326 * For root users, retry allocating enough memory for
329 kp
= kcalloc(ca
, sizeof(*kp
), GFP_ATOMIC
);
337 for_each_prl_rcu(t
->prl
) {
340 if (kprl
.addr
!= htonl(INADDR_ANY
) && prl
->addr
!= kprl
.addr
)
342 kp
[c
].addr
= prl
->addr
;
343 kp
[c
].flags
= prl
->flags
;
345 if (kprl
.addr
!= htonl(INADDR_ANY
))
351 len
= sizeof(*kp
) * c
;
353 if ((len
&& copy_to_user(a
+ 1, kp
, len
)) || put_user(len
, &a
->datalen
))
362 ipip6_tunnel_add_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
, int chg
)
364 struct ip_tunnel_prl_entry
*p
;
367 if (a
->addr
== htonl(INADDR_ANY
))
372 for (p
= rtnl_dereference(t
->prl
); p
; p
= rtnl_dereference(p
->next
)) {
373 if (p
->addr
== a
->addr
) {
388 p
= kzalloc(sizeof(struct ip_tunnel_prl_entry
), GFP_KERNEL
);
398 rcu_assign_pointer(t
->prl
, p
);
403 static void prl_list_destroy_rcu(struct rcu_head
*head
)
405 struct ip_tunnel_prl_entry
*p
, *n
;
407 p
= container_of(head
, struct ip_tunnel_prl_entry
, rcu_head
);
409 n
= rcu_dereference_protected(p
->next
, 1);
416 ipip6_tunnel_del_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
)
418 struct ip_tunnel_prl_entry
*x
;
419 struct ip_tunnel_prl_entry __rcu
**p
;
424 if (a
&& a
->addr
!= htonl(INADDR_ANY
)) {
426 (x
= rtnl_dereference(*p
)) != NULL
;
428 if (x
->addr
== a
->addr
) {
430 kfree_rcu(x
, rcu_head
);
437 x
= rtnl_dereference(t
->prl
);
440 call_rcu(&x
->rcu_head
, prl_list_destroy_rcu
);
449 isatap_chksrc(struct sk_buff
*skb
, const struct iphdr
*iph
, struct ip_tunnel
*t
)
451 struct ip_tunnel_prl_entry
*p
;
455 p
= __ipip6_tunnel_locate_prl(t
, iph
->saddr
);
457 if (p
->flags
& PRL_DEFAULT
)
458 skb
->ndisc_nodetype
= NDISC_NODETYPE_DEFAULT
;
460 skb
->ndisc_nodetype
= NDISC_NODETYPE_NODEFAULT
;
462 const struct in6_addr
*addr6
= &ipv6_hdr(skb
)->saddr
;
464 if (ipv6_addr_is_isatap(addr6
) &&
465 (addr6
->s6_addr32
[3] == iph
->saddr
) &&
466 ipv6_chk_prefix(addr6
, t
->dev
))
467 skb
->ndisc_nodetype
= NDISC_NODETYPE_HOST
;
475 static void ipip6_tunnel_uninit(struct net_device
*dev
)
477 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
478 struct sit_net
*sitn
= net_generic(tunnel
->net
, sit_net_id
);
480 if (dev
== sitn
->fb_tunnel_dev
) {
481 RCU_INIT_POINTER(sitn
->tunnels_wc
[0], NULL
);
483 ipip6_tunnel_unlink(sitn
, tunnel
);
484 ipip6_tunnel_del_prl(tunnel
, NULL
);
486 dst_cache_reset(&tunnel
->dst_cache
);
490 static int ipip6_err(struct sk_buff
*skb
, u32 info
)
492 const struct iphdr
*iph
= (const struct iphdr
*)skb
->data
;
493 const int type
= icmp_hdr(skb
)->type
;
494 const int code
= icmp_hdr(skb
)->code
;
495 unsigned int data_len
= 0;
502 case ICMP_PARAMETERPROB
:
505 case ICMP_DEST_UNREACH
:
508 /* Impossible event. */
511 /* All others are translated to HOST_UNREACH.
512 rfc2003 contains "deep thoughts" about NET_UNREACH,
513 I believe they are just ether pollution. --ANK
518 case ICMP_TIME_EXCEEDED
:
519 if (code
!= ICMP_EXC_TTL
)
521 data_len
= icmp_hdr(skb
)->un
.reserved
[1] * 4; /* RFC 4884 4.1 */
529 sifindex
= netif_is_l3_master(skb
->dev
) ? IPCB(skb
)->iif
: 0;
530 t
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
531 iph
->daddr
, iph
->saddr
, sifindex
);
535 if (type
== ICMP_DEST_UNREACH
&& code
== ICMP_FRAG_NEEDED
) {
536 ipv4_update_pmtu(skb
, dev_net(skb
->dev
), info
,
537 t
->parms
.link
, 0, iph
->protocol
, 0);
541 if (type
== ICMP_REDIRECT
) {
542 ipv4_redirect(skb
, dev_net(skb
->dev
), t
->parms
.link
, 0,
549 if (__in6_dev_get(skb
->dev
) &&
550 !ip6_err_gen_icmpv6_unreach(skb
, iph
->ihl
* 4, type
, data_len
))
553 if (t
->parms
.iph
.daddr
== 0)
556 if (t
->parms
.iph
.ttl
== 0 && type
== ICMP_TIME_EXCEEDED
)
559 if (time_before(jiffies
, t
->err_time
+ IPTUNNEL_ERR_TIMEO
))
563 t
->err_time
= jiffies
;
568 static inline bool is_spoofed_6rd(struct ip_tunnel
*tunnel
, const __be32 v4addr
,
569 const struct in6_addr
*v6addr
)
572 if (check_6rd(tunnel
, v6addr
, &v4embed
) && v4addr
!= v4embed
)
577 /* Checks if an address matches an address on the tunnel interface.
578 * Used to detect the NAT of proto 41 packets and let them pass spoofing test.
580 * This function is called after we considered the packet as spoofed
582 * We may have a router that is doing NAT for proto 41 packets
583 * for an internal station. Destination a.a.a.a/PREFIX:bbbb:bbbb
584 * will be translated to n.n.n.n/PREFIX:bbbb:bbbb. And is_spoofed_6rd
585 * function will return true, dropping the packet.
586 * But, we can still check if is spoofed against the IP
587 * addresses associated with the interface.
589 static bool only_dnatted(const struct ip_tunnel
*tunnel
,
590 const struct in6_addr
*v6dst
)
594 #ifdef CONFIG_IPV6_SIT_6RD
595 prefix_len
= tunnel
->ip6rd
.prefixlen
+ 32
596 - tunnel
->ip6rd
.relay_prefixlen
;
600 return ipv6_chk_custom_prefix(v6dst
, prefix_len
, tunnel
->dev
);
603 /* Returns true if a packet is spoofed */
604 static bool packet_is_spoofed(struct sk_buff
*skb
,
605 const struct iphdr
*iph
,
606 struct ip_tunnel
*tunnel
)
608 const struct ipv6hdr
*ipv6h
;
610 if (tunnel
->dev
->priv_flags
& IFF_ISATAP
) {
611 if (!isatap_chksrc(skb
, iph
, tunnel
))
617 if (tunnel
->dev
->flags
& IFF_POINTOPOINT
)
620 ipv6h
= ipv6_hdr(skb
);
622 if (unlikely(is_spoofed_6rd(tunnel
, iph
->saddr
, &ipv6h
->saddr
))) {
623 net_warn_ratelimited("Src spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
624 &iph
->saddr
, &ipv6h
->saddr
,
625 &iph
->daddr
, &ipv6h
->daddr
);
629 if (likely(!is_spoofed_6rd(tunnel
, iph
->daddr
, &ipv6h
->daddr
)))
632 if (only_dnatted(tunnel
, &ipv6h
->daddr
))
635 net_warn_ratelimited("Dst spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
636 &iph
->saddr
, &ipv6h
->saddr
,
637 &iph
->daddr
, &ipv6h
->daddr
);
641 static int ipip6_rcv(struct sk_buff
*skb
)
643 const struct iphdr
*iph
= ip_hdr(skb
);
644 struct ip_tunnel
*tunnel
;
648 sifindex
= netif_is_l3_master(skb
->dev
) ? IPCB(skb
)->iif
: 0;
649 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
650 iph
->saddr
, iph
->daddr
, sifindex
);
652 struct pcpu_sw_netstats
*tstats
;
654 if (tunnel
->parms
.iph
.protocol
!= IPPROTO_IPV6
&&
655 tunnel
->parms
.iph
.protocol
!= 0)
658 skb
->mac_header
= skb
->network_header
;
659 skb_reset_network_header(skb
);
660 IPCB(skb
)->flags
= 0;
661 skb
->dev
= tunnel
->dev
;
663 if (packet_is_spoofed(skb
, iph
, tunnel
)) {
664 tunnel
->dev
->stats
.rx_errors
++;
668 if (iptunnel_pull_header(skb
, 0, htons(ETH_P_IPV6
),
669 !net_eq(tunnel
->net
, dev_net(tunnel
->dev
))))
672 /* skb can be uncloned in iptunnel_pull_header, so
673 * old iph is no longer valid
675 iph
= (const struct iphdr
*)skb_mac_header(skb
);
676 err
= IP_ECN_decapsulate(iph
, skb
);
679 net_info_ratelimited("non-ECT from %pI4 with TOS=%#x\n",
680 &iph
->saddr
, iph
->tos
);
682 ++tunnel
->dev
->stats
.rx_frame_errors
;
683 ++tunnel
->dev
->stats
.rx_errors
;
688 tstats
= this_cpu_ptr(tunnel
->dev
->tstats
);
689 u64_stats_update_begin(&tstats
->syncp
);
690 tstats
->rx_packets
++;
691 tstats
->rx_bytes
+= skb
->len
;
692 u64_stats_update_end(&tstats
->syncp
);
699 /* no tunnel matched, let upstream know, ipsec may handle it */
706 static const struct tnl_ptk_info ipip_tpi
= {
707 /* no tunnel info required for ipip. */
708 .proto
= htons(ETH_P_IP
),
711 #if IS_ENABLED(CONFIG_MPLS)
712 static const struct tnl_ptk_info mplsip_tpi
= {
713 /* no tunnel info required for mplsip. */
714 .proto
= htons(ETH_P_MPLS_UC
),
718 static int sit_tunnel_rcv(struct sk_buff
*skb
, u8 ipproto
)
720 const struct iphdr
*iph
;
721 struct ip_tunnel
*tunnel
;
724 sifindex
= netif_is_l3_master(skb
->dev
) ? IPCB(skb
)->iif
: 0;
727 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
728 iph
->saddr
, iph
->daddr
, sifindex
);
730 const struct tnl_ptk_info
*tpi
;
732 if (tunnel
->parms
.iph
.protocol
!= ipproto
&&
733 tunnel
->parms
.iph
.protocol
!= 0)
736 if (!xfrm4_policy_check(NULL
, XFRM_POLICY_IN
, skb
))
738 #if IS_ENABLED(CONFIG_MPLS)
739 if (ipproto
== IPPROTO_MPLS
)
744 if (iptunnel_pull_header(skb
, 0, tpi
->proto
, false))
746 return ip_tunnel_rcv(tunnel
, skb
, tpi
, NULL
, log_ecn_error
);
756 static int ipip_rcv(struct sk_buff
*skb
)
758 return sit_tunnel_rcv(skb
, IPPROTO_IPIP
);
761 #if IS_ENABLED(CONFIG_MPLS)
762 static int mplsip_rcv(struct sk_buff
*skb
)
764 return sit_tunnel_rcv(skb
, IPPROTO_MPLS
);
769 * If the IPv6 address comes from 6rd / 6to4 (RFC 3056) addr space this function
770 * stores the embedded IPv4 address in v4dst and returns true.
772 static bool check_6rd(struct ip_tunnel
*tunnel
, const struct in6_addr
*v6dst
,
775 #ifdef CONFIG_IPV6_SIT_6RD
776 if (ipv6_prefix_equal(v6dst
, &tunnel
->ip6rd
.prefix
,
777 tunnel
->ip6rd
.prefixlen
)) {
778 unsigned int pbw0
, pbi0
;
782 pbw0
= tunnel
->ip6rd
.prefixlen
>> 5;
783 pbi0
= tunnel
->ip6rd
.prefixlen
& 0x1f;
785 d
= tunnel
->ip6rd
.relay_prefixlen
< 32 ?
786 (ntohl(v6dst
->s6_addr32
[pbw0
]) << pbi0
) >>
787 tunnel
->ip6rd
.relay_prefixlen
: 0;
789 pbi1
= pbi0
- tunnel
->ip6rd
.relay_prefixlen
;
791 d
|= ntohl(v6dst
->s6_addr32
[pbw0
+ 1]) >>
794 *v4dst
= tunnel
->ip6rd
.relay_prefix
| htonl(d
);
798 if (v6dst
->s6_addr16
[0] == htons(0x2002)) {
799 /* 6to4 v6 addr has 16 bits prefix, 32 v4addr, 16 SLA, ... */
800 memcpy(v4dst
, &v6dst
->s6_addr16
[1], 4);
807 static inline __be32
try_6rd(struct ip_tunnel
*tunnel
,
808 const struct in6_addr
*v6dst
)
811 check_6rd(tunnel
, v6dst
, &dst
);
816 * This function assumes it is being called from dev_queue_xmit()
817 * and that skb is filled properly by that function.
820 static netdev_tx_t
ipip6_tunnel_xmit(struct sk_buff
*skb
,
821 struct net_device
*dev
)
823 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
824 const struct iphdr
*tiph
= &tunnel
->parms
.iph
;
825 const struct ipv6hdr
*iph6
= ipv6_hdr(skb
);
826 u8 tos
= tunnel
->parms
.iph
.tos
;
827 __be16 df
= tiph
->frag_off
;
828 struct rtable
*rt
; /* Route to the other host */
829 struct net_device
*tdev
; /* Device to other host */
830 unsigned int max_headroom
; /* The extra header space needed */
831 __be32 dst
= tiph
->daddr
;
834 const struct in6_addr
*addr6
;
837 u8 protocol
= IPPROTO_IPV6
;
838 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
841 tos
= ipv6_get_dsfield(iph6
);
843 /* ISATAP (RFC4214) - must come before 6to4 */
844 if (dev
->priv_flags
& IFF_ISATAP
) {
845 struct neighbour
*neigh
= NULL
;
846 bool do_tx_error
= false;
849 neigh
= dst_neigh_lookup(skb_dst(skb
), &iph6
->daddr
);
852 net_dbg_ratelimited("nexthop == NULL\n");
856 addr6
= (const struct in6_addr
*)&neigh
->primary_key
;
857 addr_type
= ipv6_addr_type(addr6
);
859 if ((addr_type
& IPV6_ADDR_UNICAST
) &&
860 ipv6_addr_is_isatap(addr6
))
861 dst
= addr6
->s6_addr32
[3];
865 neigh_release(neigh
);
871 dst
= try_6rd(tunnel
, &iph6
->daddr
);
874 struct neighbour
*neigh
= NULL
;
875 bool do_tx_error
= false;
878 neigh
= dst_neigh_lookup(skb_dst(skb
), &iph6
->daddr
);
881 net_dbg_ratelimited("nexthop == NULL\n");
885 addr6
= (const struct in6_addr
*)&neigh
->primary_key
;
886 addr_type
= ipv6_addr_type(addr6
);
888 if (addr_type
== IPV6_ADDR_ANY
) {
889 addr6
= &ipv6_hdr(skb
)->daddr
;
890 addr_type
= ipv6_addr_type(addr6
);
893 if ((addr_type
& IPV6_ADDR_COMPATv4
) != 0)
894 dst
= addr6
->s6_addr32
[3];
898 neigh_release(neigh
);
903 flowi4_init_output(&fl4
, tunnel
->parms
.link
, tunnel
->fwmark
,
904 RT_TOS(tos
), RT_SCOPE_UNIVERSE
, IPPROTO_IPV6
,
905 0, dst
, tiph
->saddr
, 0, 0,
906 sock_net_uid(tunnel
->net
, NULL
));
907 rt
= ip_route_output_flow(tunnel
->net
, &fl4
, NULL
);
910 dev
->stats
.tx_carrier_errors
++;
913 if (rt
->rt_type
!= RTN_UNICAST
) {
915 dev
->stats
.tx_carrier_errors
++;
922 dev
->stats
.collisions
++;
926 if (iptunnel_handle_offloads(skb
, SKB_GSO_IPXIP4
)) {
932 mtu
= dst_mtu(&rt
->dst
) - t_hlen
;
935 dev
->stats
.collisions
++;
940 if (mtu
< IPV6_MIN_MTU
) {
945 if (tunnel
->parms
.iph
.daddr
)
946 skb_dst_update_pmtu_no_confirm(skb
, mtu
);
948 if (skb
->len
> mtu
&& !skb_is_gso(skb
)) {
949 icmpv6_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
955 if (tunnel
->err_count
> 0) {
956 if (time_before(jiffies
,
957 tunnel
->err_time
+ IPTUNNEL_ERR_TIMEO
)) {
959 dst_link_failure(skb
);
961 tunnel
->err_count
= 0;
965 * Okay, now see if we can stuff it in the buffer as-is.
967 max_headroom
= LL_RESERVED_SPACE(tdev
) + t_hlen
;
969 if (skb_headroom(skb
) < max_headroom
|| skb_shared(skb
) ||
970 (skb_cloned(skb
) && !skb_clone_writable(skb
, 0))) {
971 struct sk_buff
*new_skb
= skb_realloc_headroom(skb
, max_headroom
);
974 dev
->stats
.tx_dropped
++;
979 skb_set_owner_w(new_skb
, skb
->sk
);
982 iph6
= ipv6_hdr(skb
);
986 ttl
= iph6
->hop_limit
;
987 tos
= INET_ECN_encapsulate(tos
, ipv6_get_dsfield(iph6
));
989 if (ip_tunnel_encap(skb
, tunnel
, &protocol
, &fl4
) < 0) {
994 skb_set_inner_ipproto(skb
, IPPROTO_IPV6
);
996 iptunnel_xmit(NULL
, rt
, skb
, fl4
.saddr
, fl4
.daddr
, protocol
, tos
, ttl
,
997 df
, !net_eq(tunnel
->net
, dev_net(dev
)));
1001 dst_link_failure(skb
);
1004 dev
->stats
.tx_errors
++;
1005 return NETDEV_TX_OK
;
1008 static netdev_tx_t
sit_tunnel_xmit__(struct sk_buff
*skb
,
1009 struct net_device
*dev
, u8 ipproto
)
1011 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1012 const struct iphdr
*tiph
= &tunnel
->parms
.iph
;
1014 if (iptunnel_handle_offloads(skb
, SKB_GSO_IPXIP4
))
1017 skb_set_inner_ipproto(skb
, ipproto
);
1019 ip_tunnel_xmit(skb
, dev
, tiph
, ipproto
);
1020 return NETDEV_TX_OK
;
1023 dev
->stats
.tx_errors
++;
1024 return NETDEV_TX_OK
;
1027 static netdev_tx_t
sit_tunnel_xmit(struct sk_buff
*skb
,
1028 struct net_device
*dev
)
1030 if (!pskb_inet_may_pull(skb
))
1033 switch (skb
->protocol
) {
1034 case htons(ETH_P_IP
):
1035 sit_tunnel_xmit__(skb
, dev
, IPPROTO_IPIP
);
1037 case htons(ETH_P_IPV6
):
1038 ipip6_tunnel_xmit(skb
, dev
);
1040 #if IS_ENABLED(CONFIG_MPLS)
1041 case htons(ETH_P_MPLS_UC
):
1042 sit_tunnel_xmit__(skb
, dev
, IPPROTO_MPLS
);
1049 return NETDEV_TX_OK
;
1052 dev
->stats
.tx_errors
++;
1054 return NETDEV_TX_OK
;
1058 static void ipip6_tunnel_bind_dev(struct net_device
*dev
)
1060 struct net_device
*tdev
= NULL
;
1061 struct ip_tunnel
*tunnel
;
1062 const struct iphdr
*iph
;
1065 tunnel
= netdev_priv(dev
);
1066 iph
= &tunnel
->parms
.iph
;
1069 struct rtable
*rt
= ip_route_output_ports(tunnel
->net
, &fl4
,
1071 iph
->daddr
, iph
->saddr
,
1075 tunnel
->parms
.link
);
1081 dev
->flags
|= IFF_POINTOPOINT
;
1084 if (!tdev
&& tunnel
->parms
.link
)
1085 tdev
= __dev_get_by_index(tunnel
->net
, tunnel
->parms
.link
);
1087 if (tdev
&& !netif_is_l3_master(tdev
)) {
1088 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1090 dev
->hard_header_len
= tdev
->hard_header_len
+ sizeof(struct iphdr
);
1091 dev
->mtu
= tdev
->mtu
- t_hlen
;
1092 if (dev
->mtu
< IPV6_MIN_MTU
)
1093 dev
->mtu
= IPV6_MIN_MTU
;
1097 static void ipip6_tunnel_update(struct ip_tunnel
*t
, struct ip_tunnel_parm
*p
,
1100 struct net
*net
= t
->net
;
1101 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1103 ipip6_tunnel_unlink(sitn
, t
);
1105 t
->parms
.iph
.saddr
= p
->iph
.saddr
;
1106 t
->parms
.iph
.daddr
= p
->iph
.daddr
;
1107 memcpy(t
->dev
->dev_addr
, &p
->iph
.saddr
, 4);
1108 memcpy(t
->dev
->broadcast
, &p
->iph
.daddr
, 4);
1109 ipip6_tunnel_link(sitn
, t
);
1110 t
->parms
.iph
.ttl
= p
->iph
.ttl
;
1111 t
->parms
.iph
.tos
= p
->iph
.tos
;
1112 t
->parms
.iph
.frag_off
= p
->iph
.frag_off
;
1113 if (t
->parms
.link
!= p
->link
|| t
->fwmark
!= fwmark
) {
1114 t
->parms
.link
= p
->link
;
1116 ipip6_tunnel_bind_dev(t
->dev
);
1118 dst_cache_reset(&t
->dst_cache
);
1119 netdev_state_change(t
->dev
);
1122 #ifdef CONFIG_IPV6_SIT_6RD
1123 static int ipip6_tunnel_update_6rd(struct ip_tunnel
*t
,
1124 struct ip_tunnel_6rd
*ip6rd
)
1126 struct in6_addr prefix
;
1127 __be32 relay_prefix
;
1129 if (ip6rd
->relay_prefixlen
> 32 ||
1130 ip6rd
->prefixlen
+ (32 - ip6rd
->relay_prefixlen
) > 64)
1133 ipv6_addr_prefix(&prefix
, &ip6rd
->prefix
, ip6rd
->prefixlen
);
1134 if (!ipv6_addr_equal(&prefix
, &ip6rd
->prefix
))
1136 if (ip6rd
->relay_prefixlen
)
1137 relay_prefix
= ip6rd
->relay_prefix
&
1138 htonl(0xffffffffUL
<<
1139 (32 - ip6rd
->relay_prefixlen
));
1142 if (relay_prefix
!= ip6rd
->relay_prefix
)
1145 t
->ip6rd
.prefix
= prefix
;
1146 t
->ip6rd
.relay_prefix
= relay_prefix
;
1147 t
->ip6rd
.prefixlen
= ip6rd
->prefixlen
;
1148 t
->ip6rd
.relay_prefixlen
= ip6rd
->relay_prefixlen
;
1149 dst_cache_reset(&t
->dst_cache
);
1150 netdev_state_change(t
->dev
);
1155 static bool ipip6_valid_ip_proto(u8 ipproto
)
1157 return ipproto
== IPPROTO_IPV6
||
1158 ipproto
== IPPROTO_IPIP
||
1159 #if IS_ENABLED(CONFIG_MPLS)
1160 ipproto
== IPPROTO_MPLS
||
1166 ipip6_tunnel_ioctl(struct net_device
*dev
, struct ifreq
*ifr
, int cmd
)
1169 struct ip_tunnel_parm p
;
1170 struct ip_tunnel_prl prl
;
1171 struct ip_tunnel
*t
= netdev_priv(dev
);
1172 struct net
*net
= t
->net
;
1173 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1174 #ifdef CONFIG_IPV6_SIT_6RD
1175 struct ip_tunnel_6rd ip6rd
;
1180 #ifdef CONFIG_IPV6_SIT_6RD
1183 if (dev
== sitn
->fb_tunnel_dev
) {
1184 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
))) {
1188 t
= ipip6_tunnel_locate(net
, &p
, 0);
1190 t
= netdev_priv(dev
);
1194 if (cmd
== SIOCGETTUNNEL
) {
1195 memcpy(&p
, &t
->parms
, sizeof(p
));
1196 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &p
,
1199 #ifdef CONFIG_IPV6_SIT_6RD
1201 ip6rd
.prefix
= t
->ip6rd
.prefix
;
1202 ip6rd
.relay_prefix
= t
->ip6rd
.relay_prefix
;
1203 ip6rd
.prefixlen
= t
->ip6rd
.prefixlen
;
1204 ip6rd
.relay_prefixlen
= t
->ip6rd
.relay_prefixlen
;
1205 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &ip6rd
,
1216 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1220 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
1224 if (!ipip6_valid_ip_proto(p
.iph
.protocol
))
1226 if (p
.iph
.version
!= 4 ||
1227 p
.iph
.ihl
!= 5 || (p
.iph
.frag_off
&htons(~IP_DF
)))
1230 p
.iph
.frag_off
|= htons(IP_DF
);
1232 t
= ipip6_tunnel_locate(net
, &p
, cmd
== SIOCADDTUNNEL
);
1234 if (dev
!= sitn
->fb_tunnel_dev
&& cmd
== SIOCCHGTUNNEL
) {
1236 if (t
->dev
!= dev
) {
1241 if (((dev
->flags
&IFF_POINTOPOINT
) && !p
.iph
.daddr
) ||
1242 (!(dev
->flags
&IFF_POINTOPOINT
) && p
.iph
.daddr
)) {
1246 t
= netdev_priv(dev
);
1249 ipip6_tunnel_update(t
, &p
, t
->fwmark
);
1254 if (copy_to_user(ifr
->ifr_ifru
.ifru_data
, &t
->parms
, sizeof(p
)))
1257 err
= (cmd
== SIOCADDTUNNEL
? -ENOBUFS
: -ENOENT
);
1262 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1265 if (dev
== sitn
->fb_tunnel_dev
) {
1267 if (copy_from_user(&p
, ifr
->ifr_ifru
.ifru_data
, sizeof(p
)))
1270 t
= ipip6_tunnel_locate(net
, &p
, 0);
1274 if (t
== netdev_priv(sitn
->fb_tunnel_dev
))
1278 unregister_netdevice(dev
);
1284 if (dev
== sitn
->fb_tunnel_dev
)
1286 err
= ipip6_tunnel_get_prl(t
, ifr
->ifr_ifru
.ifru_data
);
1293 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1296 if (dev
== sitn
->fb_tunnel_dev
)
1299 if (copy_from_user(&prl
, ifr
->ifr_ifru
.ifru_data
, sizeof(prl
)))
1304 err
= ipip6_tunnel_del_prl(t
, &prl
);
1308 err
= ipip6_tunnel_add_prl(t
, &prl
, cmd
== SIOCCHGPRL
);
1311 dst_cache_reset(&t
->dst_cache
);
1312 netdev_state_change(dev
);
1315 #ifdef CONFIG_IPV6_SIT_6RD
1320 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1324 if (copy_from_user(&ip6rd
, ifr
->ifr_ifru
.ifru_data
,
1328 if (cmd
!= SIOCDEL6RD
) {
1329 err
= ipip6_tunnel_update_6rd(t
, &ip6rd
);
1333 ipip6_tunnel_clone_6rd(dev
, sitn
);
1347 static const struct net_device_ops ipip6_netdev_ops
= {
1348 .ndo_init
= ipip6_tunnel_init
,
1349 .ndo_uninit
= ipip6_tunnel_uninit
,
1350 .ndo_start_xmit
= sit_tunnel_xmit
,
1351 .ndo_do_ioctl
= ipip6_tunnel_ioctl
,
1352 .ndo_get_stats64
= ip_tunnel_get_stats64
,
1353 .ndo_get_iflink
= ip_tunnel_get_iflink
,
1356 static void ipip6_dev_free(struct net_device
*dev
)
1358 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1360 dst_cache_destroy(&tunnel
->dst_cache
);
1361 free_percpu(dev
->tstats
);
1364 #define SIT_FEATURES (NETIF_F_SG | \
1365 NETIF_F_FRAGLIST | \
1367 NETIF_F_GSO_SOFTWARE | \
1370 static void ipip6_tunnel_setup(struct net_device
*dev
)
1372 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1373 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1375 dev
->netdev_ops
= &ipip6_netdev_ops
;
1376 dev
->needs_free_netdev
= true;
1377 dev
->priv_destructor
= ipip6_dev_free
;
1379 dev
->type
= ARPHRD_SIT
;
1380 dev
->hard_header_len
= LL_MAX_HEADER
+ t_hlen
;
1381 dev
->mtu
= ETH_DATA_LEN
- t_hlen
;
1382 dev
->min_mtu
= IPV6_MIN_MTU
;
1383 dev
->max_mtu
= IP6_MAX_MTU
- t_hlen
;
1384 dev
->flags
= IFF_NOARP
;
1385 netif_keep_dst(dev
);
1387 dev
->features
|= NETIF_F_LLTX
;
1388 dev
->features
|= SIT_FEATURES
;
1389 dev
->hw_features
|= SIT_FEATURES
;
1392 static int ipip6_tunnel_init(struct net_device
*dev
)
1394 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1398 tunnel
->net
= dev_net(dev
);
1399 strcpy(tunnel
->parms
.name
, dev
->name
);
1401 ipip6_tunnel_bind_dev(dev
);
1402 dev
->tstats
= netdev_alloc_pcpu_stats(struct pcpu_sw_netstats
);
1406 err
= dst_cache_init(&tunnel
->dst_cache
, GFP_KERNEL
);
1408 free_percpu(dev
->tstats
);
1416 static void __net_init
ipip6_fb_tunnel_init(struct net_device
*dev
)
1418 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1419 struct iphdr
*iph
= &tunnel
->parms
.iph
;
1420 struct net
*net
= dev_net(dev
);
1421 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1424 iph
->protocol
= IPPROTO_IPV6
;
1429 rcu_assign_pointer(sitn
->tunnels_wc
[0], tunnel
);
1432 static int ipip6_validate(struct nlattr
*tb
[], struct nlattr
*data
[],
1433 struct netlink_ext_ack
*extack
)
1437 if (!data
|| !data
[IFLA_IPTUN_PROTO
])
1440 proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1441 if (!ipip6_valid_ip_proto(proto
))
1447 static void ipip6_netlink_parms(struct nlattr
*data
[],
1448 struct ip_tunnel_parm
*parms
,
1451 memset(parms
, 0, sizeof(*parms
));
1453 parms
->iph
.version
= 4;
1454 parms
->iph
.protocol
= IPPROTO_IPV6
;
1456 parms
->iph
.ttl
= 64;
1461 if (data
[IFLA_IPTUN_LINK
])
1462 parms
->link
= nla_get_u32(data
[IFLA_IPTUN_LINK
]);
1464 if (data
[IFLA_IPTUN_LOCAL
])
1465 parms
->iph
.saddr
= nla_get_be32(data
[IFLA_IPTUN_LOCAL
]);
1467 if (data
[IFLA_IPTUN_REMOTE
])
1468 parms
->iph
.daddr
= nla_get_be32(data
[IFLA_IPTUN_REMOTE
]);
1470 if (data
[IFLA_IPTUN_TTL
]) {
1471 parms
->iph
.ttl
= nla_get_u8(data
[IFLA_IPTUN_TTL
]);
1473 parms
->iph
.frag_off
= htons(IP_DF
);
1476 if (data
[IFLA_IPTUN_TOS
])
1477 parms
->iph
.tos
= nla_get_u8(data
[IFLA_IPTUN_TOS
]);
1479 if (!data
[IFLA_IPTUN_PMTUDISC
] || nla_get_u8(data
[IFLA_IPTUN_PMTUDISC
]))
1480 parms
->iph
.frag_off
= htons(IP_DF
);
1482 if (data
[IFLA_IPTUN_FLAGS
])
1483 parms
->i_flags
= nla_get_be16(data
[IFLA_IPTUN_FLAGS
]);
1485 if (data
[IFLA_IPTUN_PROTO
])
1486 parms
->iph
.protocol
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1488 if (data
[IFLA_IPTUN_FWMARK
])
1489 *fwmark
= nla_get_u32(data
[IFLA_IPTUN_FWMARK
]);
1492 /* This function returns true when ENCAP attributes are present in the nl msg */
1493 static bool ipip6_netlink_encap_parms(struct nlattr
*data
[],
1494 struct ip_tunnel_encap
*ipencap
)
1498 memset(ipencap
, 0, sizeof(*ipencap
));
1503 if (data
[IFLA_IPTUN_ENCAP_TYPE
]) {
1505 ipencap
->type
= nla_get_u16(data
[IFLA_IPTUN_ENCAP_TYPE
]);
1508 if (data
[IFLA_IPTUN_ENCAP_FLAGS
]) {
1510 ipencap
->flags
= nla_get_u16(data
[IFLA_IPTUN_ENCAP_FLAGS
]);
1513 if (data
[IFLA_IPTUN_ENCAP_SPORT
]) {
1515 ipencap
->sport
= nla_get_be16(data
[IFLA_IPTUN_ENCAP_SPORT
]);
1518 if (data
[IFLA_IPTUN_ENCAP_DPORT
]) {
1520 ipencap
->dport
= nla_get_be16(data
[IFLA_IPTUN_ENCAP_DPORT
]);
1526 #ifdef CONFIG_IPV6_SIT_6RD
1527 /* This function returns true when 6RD attributes are present in the nl msg */
1528 static bool ipip6_netlink_6rd_parms(struct nlattr
*data
[],
1529 struct ip_tunnel_6rd
*ip6rd
)
1532 memset(ip6rd
, 0, sizeof(*ip6rd
));
1537 if (data
[IFLA_IPTUN_6RD_PREFIX
]) {
1539 ip6rd
->prefix
= nla_get_in6_addr(data
[IFLA_IPTUN_6RD_PREFIX
]);
1542 if (data
[IFLA_IPTUN_6RD_RELAY_PREFIX
]) {
1544 ip6rd
->relay_prefix
=
1545 nla_get_be32(data
[IFLA_IPTUN_6RD_RELAY_PREFIX
]);
1548 if (data
[IFLA_IPTUN_6RD_PREFIXLEN
]) {
1550 ip6rd
->prefixlen
= nla_get_u16(data
[IFLA_IPTUN_6RD_PREFIXLEN
]);
1553 if (data
[IFLA_IPTUN_6RD_RELAY_PREFIXLEN
]) {
1555 ip6rd
->relay_prefixlen
=
1556 nla_get_u16(data
[IFLA_IPTUN_6RD_RELAY_PREFIXLEN
]);
1563 static int ipip6_newlink(struct net
*src_net
, struct net_device
*dev
,
1564 struct nlattr
*tb
[], struct nlattr
*data
[],
1565 struct netlink_ext_ack
*extack
)
1567 struct net
*net
= dev_net(dev
);
1568 struct ip_tunnel
*nt
;
1569 struct ip_tunnel_encap ipencap
;
1570 #ifdef CONFIG_IPV6_SIT_6RD
1571 struct ip_tunnel_6rd ip6rd
;
1575 nt
= netdev_priv(dev
);
1577 if (ipip6_netlink_encap_parms(data
, &ipencap
)) {
1578 err
= ip_tunnel_encap_setup(nt
, &ipencap
);
1583 ipip6_netlink_parms(data
, &nt
->parms
, &nt
->fwmark
);
1585 if (ipip6_tunnel_locate(net
, &nt
->parms
, 0))
1588 err
= ipip6_tunnel_create(dev
);
1593 u32 mtu
= nla_get_u32(tb
[IFLA_MTU
]);
1595 if (mtu
>= IPV6_MIN_MTU
&&
1596 mtu
<= IP6_MAX_MTU
- dev
->hard_header_len
)
1600 #ifdef CONFIG_IPV6_SIT_6RD
1601 if (ipip6_netlink_6rd_parms(data
, &ip6rd
))
1602 err
= ipip6_tunnel_update_6rd(nt
, &ip6rd
);
1608 static int ipip6_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
1609 struct nlattr
*data
[],
1610 struct netlink_ext_ack
*extack
)
1612 struct ip_tunnel
*t
= netdev_priv(dev
);
1613 struct ip_tunnel_parm p
;
1614 struct ip_tunnel_encap ipencap
;
1615 struct net
*net
= t
->net
;
1616 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1617 #ifdef CONFIG_IPV6_SIT_6RD
1618 struct ip_tunnel_6rd ip6rd
;
1620 __u32 fwmark
= t
->fwmark
;
1623 if (dev
== sitn
->fb_tunnel_dev
)
1626 if (ipip6_netlink_encap_parms(data
, &ipencap
)) {
1627 err
= ip_tunnel_encap_setup(t
, &ipencap
);
1632 ipip6_netlink_parms(data
, &p
, &fwmark
);
1634 if (((dev
->flags
& IFF_POINTOPOINT
) && !p
.iph
.daddr
) ||
1635 (!(dev
->flags
& IFF_POINTOPOINT
) && p
.iph
.daddr
))
1638 t
= ipip6_tunnel_locate(net
, &p
, 0);
1644 t
= netdev_priv(dev
);
1646 ipip6_tunnel_update(t
, &p
, fwmark
);
1648 #ifdef CONFIG_IPV6_SIT_6RD
1649 if (ipip6_netlink_6rd_parms(data
, &ip6rd
))
1650 return ipip6_tunnel_update_6rd(t
, &ip6rd
);
1656 static size_t ipip6_get_size(const struct net_device
*dev
)
1659 /* IFLA_IPTUN_LINK */
1661 /* IFLA_IPTUN_LOCAL */
1663 /* IFLA_IPTUN_REMOTE */
1665 /* IFLA_IPTUN_TTL */
1667 /* IFLA_IPTUN_TOS */
1669 /* IFLA_IPTUN_PMTUDISC */
1671 /* IFLA_IPTUN_FLAGS */
1673 /* IFLA_IPTUN_PROTO */
1675 #ifdef CONFIG_IPV6_SIT_6RD
1676 /* IFLA_IPTUN_6RD_PREFIX */
1677 nla_total_size(sizeof(struct in6_addr
)) +
1678 /* IFLA_IPTUN_6RD_RELAY_PREFIX */
1680 /* IFLA_IPTUN_6RD_PREFIXLEN */
1682 /* IFLA_IPTUN_6RD_RELAY_PREFIXLEN */
1685 /* IFLA_IPTUN_ENCAP_TYPE */
1687 /* IFLA_IPTUN_ENCAP_FLAGS */
1689 /* IFLA_IPTUN_ENCAP_SPORT */
1691 /* IFLA_IPTUN_ENCAP_DPORT */
1693 /* IFLA_IPTUN_FWMARK */
1698 static int ipip6_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
1700 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1701 struct ip_tunnel_parm
*parm
= &tunnel
->parms
;
1703 if (nla_put_u32(skb
, IFLA_IPTUN_LINK
, parm
->link
) ||
1704 nla_put_in_addr(skb
, IFLA_IPTUN_LOCAL
, parm
->iph
.saddr
) ||
1705 nla_put_in_addr(skb
, IFLA_IPTUN_REMOTE
, parm
->iph
.daddr
) ||
1706 nla_put_u8(skb
, IFLA_IPTUN_TTL
, parm
->iph
.ttl
) ||
1707 nla_put_u8(skb
, IFLA_IPTUN_TOS
, parm
->iph
.tos
) ||
1708 nla_put_u8(skb
, IFLA_IPTUN_PMTUDISC
,
1709 !!(parm
->iph
.frag_off
& htons(IP_DF
))) ||
1710 nla_put_u8(skb
, IFLA_IPTUN_PROTO
, parm
->iph
.protocol
) ||
1711 nla_put_be16(skb
, IFLA_IPTUN_FLAGS
, parm
->i_flags
) ||
1712 nla_put_u32(skb
, IFLA_IPTUN_FWMARK
, tunnel
->fwmark
))
1713 goto nla_put_failure
;
1715 #ifdef CONFIG_IPV6_SIT_6RD
1716 if (nla_put_in6_addr(skb
, IFLA_IPTUN_6RD_PREFIX
,
1717 &tunnel
->ip6rd
.prefix
) ||
1718 nla_put_in_addr(skb
, IFLA_IPTUN_6RD_RELAY_PREFIX
,
1719 tunnel
->ip6rd
.relay_prefix
) ||
1720 nla_put_u16(skb
, IFLA_IPTUN_6RD_PREFIXLEN
,
1721 tunnel
->ip6rd
.prefixlen
) ||
1722 nla_put_u16(skb
, IFLA_IPTUN_6RD_RELAY_PREFIXLEN
,
1723 tunnel
->ip6rd
.relay_prefixlen
))
1724 goto nla_put_failure
;
1727 if (nla_put_u16(skb
, IFLA_IPTUN_ENCAP_TYPE
,
1728 tunnel
->encap
.type
) ||
1729 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_SPORT
,
1730 tunnel
->encap
.sport
) ||
1731 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_DPORT
,
1732 tunnel
->encap
.dport
) ||
1733 nla_put_u16(skb
, IFLA_IPTUN_ENCAP_FLAGS
,
1734 tunnel
->encap
.flags
))
1735 goto nla_put_failure
;
1743 static const struct nla_policy ipip6_policy
[IFLA_IPTUN_MAX
+ 1] = {
1744 [IFLA_IPTUN_LINK
] = { .type
= NLA_U32
},
1745 [IFLA_IPTUN_LOCAL
] = { .type
= NLA_U32
},
1746 [IFLA_IPTUN_REMOTE
] = { .type
= NLA_U32
},
1747 [IFLA_IPTUN_TTL
] = { .type
= NLA_U8
},
1748 [IFLA_IPTUN_TOS
] = { .type
= NLA_U8
},
1749 [IFLA_IPTUN_PMTUDISC
] = { .type
= NLA_U8
},
1750 [IFLA_IPTUN_FLAGS
] = { .type
= NLA_U16
},
1751 [IFLA_IPTUN_PROTO
] = { .type
= NLA_U8
},
1752 #ifdef CONFIG_IPV6_SIT_6RD
1753 [IFLA_IPTUN_6RD_PREFIX
] = { .len
= sizeof(struct in6_addr
) },
1754 [IFLA_IPTUN_6RD_RELAY_PREFIX
] = { .type
= NLA_U32
},
1755 [IFLA_IPTUN_6RD_PREFIXLEN
] = { .type
= NLA_U16
},
1756 [IFLA_IPTUN_6RD_RELAY_PREFIXLEN
] = { .type
= NLA_U16
},
1758 [IFLA_IPTUN_ENCAP_TYPE
] = { .type
= NLA_U16
},
1759 [IFLA_IPTUN_ENCAP_FLAGS
] = { .type
= NLA_U16
},
1760 [IFLA_IPTUN_ENCAP_SPORT
] = { .type
= NLA_U16
},
1761 [IFLA_IPTUN_ENCAP_DPORT
] = { .type
= NLA_U16
},
1762 [IFLA_IPTUN_FWMARK
] = { .type
= NLA_U32
},
1765 static void ipip6_dellink(struct net_device
*dev
, struct list_head
*head
)
1767 struct net
*net
= dev_net(dev
);
1768 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1770 if (dev
!= sitn
->fb_tunnel_dev
)
1771 unregister_netdevice_queue(dev
, head
);
1774 static struct rtnl_link_ops sit_link_ops __read_mostly
= {
1776 .maxtype
= IFLA_IPTUN_MAX
,
1777 .policy
= ipip6_policy
,
1778 .priv_size
= sizeof(struct ip_tunnel
),
1779 .setup
= ipip6_tunnel_setup
,
1780 .validate
= ipip6_validate
,
1781 .newlink
= ipip6_newlink
,
1782 .changelink
= ipip6_changelink
,
1783 .get_size
= ipip6_get_size
,
1784 .fill_info
= ipip6_fill_info
,
1785 .dellink
= ipip6_dellink
,
1786 .get_link_net
= ip_tunnel_get_link_net
,
1789 static struct xfrm_tunnel sit_handler __read_mostly
= {
1790 .handler
= ipip6_rcv
,
1791 .err_handler
= ipip6_err
,
1795 static struct xfrm_tunnel ipip_handler __read_mostly
= {
1796 .handler
= ipip_rcv
,
1797 .err_handler
= ipip6_err
,
1801 #if IS_ENABLED(CONFIG_MPLS)
1802 static struct xfrm_tunnel mplsip_handler __read_mostly
= {
1803 .handler
= mplsip_rcv
,
1804 .err_handler
= ipip6_err
,
1809 static void __net_exit
sit_destroy_tunnels(struct net
*net
,
1810 struct list_head
*head
)
1812 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1813 struct net_device
*dev
, *aux
;
1816 for_each_netdev_safe(net
, dev
, aux
)
1817 if (dev
->rtnl_link_ops
== &sit_link_ops
)
1818 unregister_netdevice_queue(dev
, head
);
1820 for (prio
= 1; prio
< 4; prio
++) {
1822 for (h
= 0; h
< IP6_SIT_HASH_SIZE
; h
++) {
1823 struct ip_tunnel
*t
;
1825 t
= rtnl_dereference(sitn
->tunnels
[prio
][h
]);
1827 /* If dev is in the same netns, it has already
1828 * been added to the list by the previous loop.
1830 if (!net_eq(dev_net(t
->dev
), net
))
1831 unregister_netdevice_queue(t
->dev
,
1833 t
= rtnl_dereference(t
->next
);
1839 static int __net_init
sit_init_net(struct net
*net
)
1841 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1842 struct ip_tunnel
*t
;
1845 sitn
->tunnels
[0] = sitn
->tunnels_wc
;
1846 sitn
->tunnels
[1] = sitn
->tunnels_l
;
1847 sitn
->tunnels
[2] = sitn
->tunnels_r
;
1848 sitn
->tunnels
[3] = sitn
->tunnels_r_l
;
1850 if (!net_has_fallback_tunnels(net
))
1853 sitn
->fb_tunnel_dev
= alloc_netdev(sizeof(struct ip_tunnel
), "sit0",
1855 ipip6_tunnel_setup
);
1856 if (!sitn
->fb_tunnel_dev
) {
1860 dev_net_set(sitn
->fb_tunnel_dev
, net
);
1861 sitn
->fb_tunnel_dev
->rtnl_link_ops
= &sit_link_ops
;
1862 /* FB netdevice is special: we have one, and only one per netns.
1863 * Allowing to move it to another netns is clearly unsafe.
1865 sitn
->fb_tunnel_dev
->features
|= NETIF_F_NETNS_LOCAL
;
1867 err
= register_netdev(sitn
->fb_tunnel_dev
);
1871 ipip6_tunnel_clone_6rd(sitn
->fb_tunnel_dev
, sitn
);
1872 ipip6_fb_tunnel_init(sitn
->fb_tunnel_dev
);
1874 t
= netdev_priv(sitn
->fb_tunnel_dev
);
1876 strcpy(t
->parms
.name
, sitn
->fb_tunnel_dev
->name
);
1880 ipip6_dev_free(sitn
->fb_tunnel_dev
);
1881 free_netdev(sitn
->fb_tunnel_dev
);
1886 static void __net_exit
sit_exit_batch_net(struct list_head
*net_list
)
1892 list_for_each_entry(net
, net_list
, exit_list
)
1893 sit_destroy_tunnels(net
, &list
);
1895 unregister_netdevice_many(&list
);
1899 static struct pernet_operations sit_net_ops
= {
1900 .init
= sit_init_net
,
1901 .exit_batch
= sit_exit_batch_net
,
1903 .size
= sizeof(struct sit_net
),
1906 static void __exit
sit_cleanup(void)
1908 rtnl_link_unregister(&sit_link_ops
);
1909 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1910 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
1911 #if IS_ENABLED(CONFIG_MPLS)
1912 xfrm4_tunnel_deregister(&mplsip_handler
, AF_MPLS
);
1915 unregister_pernet_device(&sit_net_ops
);
1916 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1919 static int __init
sit_init(void)
1923 pr_info("IPv6, IPv4 and MPLS over IPv4 tunneling driver\n");
1925 err
= register_pernet_device(&sit_net_ops
);
1928 err
= xfrm4_tunnel_register(&sit_handler
, AF_INET6
);
1930 pr_info("%s: can't register ip6ip4\n", __func__
);
1931 goto xfrm_tunnel_failed
;
1933 err
= xfrm4_tunnel_register(&ipip_handler
, AF_INET
);
1935 pr_info("%s: can't register ip4ip4\n", __func__
);
1936 goto xfrm_tunnel4_failed
;
1938 #if IS_ENABLED(CONFIG_MPLS)
1939 err
= xfrm4_tunnel_register(&mplsip_handler
, AF_MPLS
);
1941 pr_info("%s: can't register mplsip\n", __func__
);
1942 goto xfrm_tunnel_mpls_failed
;
1945 err
= rtnl_link_register(&sit_link_ops
);
1947 goto rtnl_link_failed
;
1953 #if IS_ENABLED(CONFIG_MPLS)
1954 xfrm4_tunnel_deregister(&mplsip_handler
, AF_MPLS
);
1955 xfrm_tunnel_mpls_failed
:
1957 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
1958 xfrm_tunnel4_failed
:
1959 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1961 unregister_pernet_device(&sit_net_ops
);
1965 module_init(sit_init
);
1966 module_exit(sit_cleanup
);
1967 MODULE_LICENSE("GPL");
1968 MODULE_ALIAS_RTNL_LINK("sit");
1969 MODULE_ALIAS_NETDEV("sit0");