1 // SPDX-License-Identifier: GPL-2.0-or-later
3 * IPv6 over IPv4 tunnel device - Simple Internet Transition (SIT)
4 * Linux INET6 implementation
7 * Pedro Roque <roque@di.fc.ul.pt>
8 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
11 * Roger Venning <r.venning@telstra.com>: 6to4 support
12 * Nate Thompson <nate@thebog.net>: 6to4 support
13 * Fred Templin <fred.l.templin@boeing.com>: isatap support
16 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
18 #include <linux/module.h>
19 #include <linux/capability.h>
20 #include <linux/errno.h>
21 #include <linux/types.h>
22 #include <linux/socket.h>
23 #include <linux/sockios.h>
24 #include <linux/net.h>
25 #include <linux/in6.h>
26 #include <linux/netdevice.h>
27 #include <linux/if_arp.h>
28 #include <linux/icmp.h>
29 #include <linux/slab.h>
30 #include <linux/uaccess.h>
31 #include <linux/init.h>
32 #include <linux/netfilter_ipv4.h>
33 #include <linux/if_ether.h>
39 #include <net/protocol.h>
40 #include <net/transp_v6.h>
41 #include <net/ip6_fib.h>
42 #include <net/ip6_route.h>
43 #include <net/ndisc.h>
44 #include <net/addrconf.h>
48 #include <net/ip_tunnels.h>
49 #include <net/inet_ecn.h>
51 #include <net/dsfield.h>
52 #include <net/net_namespace.h>
53 #include <net/netns/generic.h>
54 #include <net/inet_dscp.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
62 #define IP6_SIT_HASH_SIZE 16
63 #define HASH(addr) (((__force u32)addr^((__force u32)addr>>4))&0xF)
65 static bool log_ecn_error
= true;
66 module_param(log_ecn_error
, bool, 0644);
67 MODULE_PARM_DESC(log_ecn_error
, "Log packets received with corrupted ECN");
69 static int ipip6_tunnel_init(struct net_device
*dev
);
70 static void ipip6_tunnel_setup(struct net_device
*dev
);
71 static void ipip6_dev_free(struct net_device
*dev
);
72 static bool check_6rd(struct ip_tunnel
*tunnel
, const struct in6_addr
*v6dst
,
74 static struct rtnl_link_ops sit_link_ops __read_mostly
;
76 static unsigned int sit_net_id __read_mostly
;
78 struct ip_tunnel __rcu
*tunnels_r_l
[IP6_SIT_HASH_SIZE
];
79 struct ip_tunnel __rcu
*tunnels_r
[IP6_SIT_HASH_SIZE
];
80 struct ip_tunnel __rcu
*tunnels_l
[IP6_SIT_HASH_SIZE
];
81 struct ip_tunnel __rcu
*tunnels_wc
[1];
82 struct ip_tunnel __rcu
**tunnels
[4];
84 struct net_device
*fb_tunnel_dev
;
87 static inline struct sit_net
*dev_to_sit_net(struct net_device
*dev
)
89 struct ip_tunnel
*t
= netdev_priv(dev
);
91 return net_generic(t
->net
, sit_net_id
);
95 * Must be invoked with rcu_read_lock
97 static struct ip_tunnel
*ipip6_tunnel_lookup(struct net
*net
,
98 struct net_device
*dev
,
99 __be32 remote
, __be32 local
,
102 unsigned int h0
= HASH(remote
);
103 unsigned int h1
= HASH(local
);
105 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
106 int ifindex
= dev
? dev
->ifindex
: 0;
108 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_r_l
[h0
^ h1
]) {
109 if (local
== t
->parms
.iph
.saddr
&&
110 remote
== t
->parms
.iph
.daddr
&&
111 (!dev
|| !t
->parms
.link
|| ifindex
== t
->parms
.link
||
112 sifindex
== t
->parms
.link
) &&
113 (t
->dev
->flags
& IFF_UP
))
116 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_r
[h0
]) {
117 if (remote
== t
->parms
.iph
.daddr
&&
118 (!dev
|| !t
->parms
.link
|| ifindex
== t
->parms
.link
||
119 sifindex
== t
->parms
.link
) &&
120 (t
->dev
->flags
& IFF_UP
))
123 for_each_ip_tunnel_rcu(t
, sitn
->tunnels_l
[h1
]) {
124 if (local
== t
->parms
.iph
.saddr
&&
125 (!dev
|| !t
->parms
.link
|| ifindex
== t
->parms
.link
||
126 sifindex
== t
->parms
.link
) &&
127 (t
->dev
->flags
& IFF_UP
))
130 t
= rcu_dereference(sitn
->tunnels_wc
[0]);
131 if (t
&& (t
->dev
->flags
& IFF_UP
))
136 static struct ip_tunnel __rcu
**
137 __ipip6_bucket(struct sit_net
*sitn
, struct ip_tunnel_parm_kern
*parms
)
139 __be32 remote
= parms
->iph
.daddr
;
140 __be32 local
= parms
->iph
.saddr
;
152 return &sitn
->tunnels
[prio
][h
];
155 static inline struct ip_tunnel __rcu
**ipip6_bucket(struct sit_net
*sitn
,
158 return __ipip6_bucket(sitn
, &t
->parms
);
161 static void ipip6_tunnel_unlink(struct sit_net
*sitn
, struct ip_tunnel
*t
)
163 struct ip_tunnel __rcu
**tp
;
164 struct ip_tunnel
*iter
;
166 for (tp
= ipip6_bucket(sitn
, t
);
167 (iter
= rtnl_dereference(*tp
)) != NULL
;
170 rcu_assign_pointer(*tp
, t
->next
);
176 static void ipip6_tunnel_link(struct sit_net
*sitn
, struct ip_tunnel
*t
)
178 struct ip_tunnel __rcu
**tp
= ipip6_bucket(sitn
, t
);
180 rcu_assign_pointer(t
->next
, rtnl_dereference(*tp
));
181 rcu_assign_pointer(*tp
, t
);
184 static void ipip6_tunnel_clone_6rd(struct net_device
*dev
, struct sit_net
*sitn
)
186 #ifdef CONFIG_IPV6_SIT_6RD
187 struct ip_tunnel
*t
= netdev_priv(dev
);
189 if (dev
== sitn
->fb_tunnel_dev
|| !sitn
->fb_tunnel_dev
) {
190 ipv6_addr_set(&t
->ip6rd
.prefix
, htonl(0x20020000), 0, 0, 0);
191 t
->ip6rd
.relay_prefix
= 0;
192 t
->ip6rd
.prefixlen
= 16;
193 t
->ip6rd
.relay_prefixlen
= 0;
195 struct ip_tunnel
*t0
= netdev_priv(sitn
->fb_tunnel_dev
);
196 memcpy(&t
->ip6rd
, &t0
->ip6rd
, sizeof(t
->ip6rd
));
201 static int ipip6_tunnel_create(struct net_device
*dev
)
203 struct ip_tunnel
*t
= netdev_priv(dev
);
204 struct net
*net
= dev_net(dev
);
205 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
208 __dev_addr_set(dev
, &t
->parms
.iph
.saddr
, 4);
209 memcpy(dev
->broadcast
, &t
->parms
.iph
.daddr
, 4);
211 if (test_bit(IP_TUNNEL_SIT_ISATAP_BIT
, t
->parms
.i_flags
))
212 dev
->priv_flags
|= IFF_ISATAP
;
214 dev
->rtnl_link_ops
= &sit_link_ops
;
216 err
= register_netdevice(dev
);
220 ipip6_tunnel_clone_6rd(dev
, sitn
);
222 ipip6_tunnel_link(sitn
, t
);
229 static struct ip_tunnel
*ipip6_tunnel_locate(struct net
*net
,
230 struct ip_tunnel_parm_kern
*parms
,
233 __be32 remote
= parms
->iph
.daddr
;
234 __be32 local
= parms
->iph
.saddr
;
235 struct ip_tunnel
*t
, *nt
;
236 struct ip_tunnel __rcu
**tp
;
237 struct net_device
*dev
;
239 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
241 for (tp
= __ipip6_bucket(sitn
, parms
);
242 (t
= rtnl_dereference(*tp
)) != NULL
;
244 if (local
== t
->parms
.iph
.saddr
&&
245 remote
== t
->parms
.iph
.daddr
&&
246 parms
->link
== t
->parms
.link
) {
256 if (parms
->name
[0]) {
257 if (!dev_valid_name(parms
->name
))
259 strscpy(name
, parms
->name
, IFNAMSIZ
);
261 strcpy(name
, "sit%d");
263 dev
= alloc_netdev(sizeof(*t
), name
, NET_NAME_UNKNOWN
,
268 dev_net_set(dev
, net
);
270 nt
= netdev_priv(dev
);
273 if (ipip6_tunnel_create(dev
) < 0)
277 strcpy(parms
->name
, dev
->name
);
287 #define for_each_prl_rcu(start) \
288 for (prl = rcu_dereference(start); \
290 prl = rcu_dereference(prl->next))
292 static struct ip_tunnel_prl_entry
*
293 __ipip6_tunnel_locate_prl(struct ip_tunnel
*t
, __be32 addr
)
295 struct ip_tunnel_prl_entry
*prl
;
297 for_each_prl_rcu(t
->prl
)
298 if (prl
->addr
== addr
)
304 static int ipip6_tunnel_get_prl(struct net_device
*dev
, struct ip_tunnel_prl __user
*a
)
306 struct ip_tunnel
*t
= netdev_priv(dev
);
307 struct ip_tunnel_prl kprl
, *kp
;
308 struct ip_tunnel_prl_entry
*prl
;
309 unsigned int cmax
, c
= 0, ca
, len
;
312 if (dev
== dev_to_sit_net(dev
)->fb_tunnel_dev
)
315 if (copy_from_user(&kprl
, a
, sizeof(kprl
)))
317 cmax
= kprl
.datalen
/ sizeof(kprl
);
318 if (cmax
> 1 && kprl
.addr
!= htonl(INADDR_ANY
))
321 /* For simple GET or for root users,
322 * we try harder to allocate.
324 kp
= (cmax
<= 1 || capable(CAP_NET_ADMIN
)) ?
325 kcalloc(cmax
, sizeof(*kp
), GFP_KERNEL_ACCOUNT
| __GFP_NOWARN
) :
328 ca
= min(t
->prl_count
, cmax
);
331 /* We don't try hard to allocate much memory for
333 * For root users, retry allocating enough memory for
336 kp
= kcalloc(ca
, sizeof(*kp
), GFP_ATOMIC
| __GFP_ACCOUNT
|
345 for_each_prl_rcu(t
->prl
) {
348 if (kprl
.addr
!= htonl(INADDR_ANY
) && prl
->addr
!= kprl
.addr
)
350 kp
[c
].addr
= prl
->addr
;
351 kp
[c
].flags
= prl
->flags
;
353 if (kprl
.addr
!= htonl(INADDR_ANY
))
359 len
= sizeof(*kp
) * c
;
361 if ((len
&& copy_to_user(a
+ 1, kp
, len
)) || put_user(len
, &a
->datalen
))
370 ipip6_tunnel_add_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
, int chg
)
372 struct ip_tunnel_prl_entry
*p
;
375 if (a
->addr
== htonl(INADDR_ANY
))
380 for (p
= rtnl_dereference(t
->prl
); p
; p
= rtnl_dereference(p
->next
)) {
381 if (p
->addr
== a
->addr
) {
396 p
= kzalloc(sizeof(struct ip_tunnel_prl_entry
), GFP_KERNEL
);
406 rcu_assign_pointer(t
->prl
, p
);
411 static void prl_list_destroy_rcu(struct rcu_head
*head
)
413 struct ip_tunnel_prl_entry
*p
, *n
;
415 p
= container_of(head
, struct ip_tunnel_prl_entry
, rcu_head
);
417 n
= rcu_dereference_protected(p
->next
, 1);
424 ipip6_tunnel_del_prl(struct ip_tunnel
*t
, struct ip_tunnel_prl
*a
)
426 struct ip_tunnel_prl_entry
*x
;
427 struct ip_tunnel_prl_entry __rcu
**p
;
432 if (a
&& a
->addr
!= htonl(INADDR_ANY
)) {
434 (x
= rtnl_dereference(*p
)) != NULL
;
436 if (x
->addr
== a
->addr
) {
438 kfree_rcu(x
, rcu_head
);
445 x
= rtnl_dereference(t
->prl
);
448 call_rcu(&x
->rcu_head
, prl_list_destroy_rcu
);
456 static int ipip6_tunnel_prl_ctl(struct net_device
*dev
,
457 struct ip_tunnel_prl __user
*data
, int cmd
)
459 struct ip_tunnel
*t
= netdev_priv(dev
);
460 struct ip_tunnel_prl prl
;
463 if (!ns_capable(t
->net
->user_ns
, CAP_NET_ADMIN
))
465 if (dev
== dev_to_sit_net(dev
)->fb_tunnel_dev
)
468 if (copy_from_user(&prl
, data
, sizeof(prl
)))
473 err
= ipip6_tunnel_del_prl(t
, &prl
);
477 err
= ipip6_tunnel_add_prl(t
, &prl
, cmd
== SIOCCHGPRL
);
480 dst_cache_reset(&t
->dst_cache
);
481 netdev_state_change(dev
);
486 isatap_chksrc(struct sk_buff
*skb
, const struct iphdr
*iph
, struct ip_tunnel
*t
)
488 struct ip_tunnel_prl_entry
*p
;
492 p
= __ipip6_tunnel_locate_prl(t
, iph
->saddr
);
494 if (p
->flags
& PRL_DEFAULT
)
495 skb
->ndisc_nodetype
= NDISC_NODETYPE_DEFAULT
;
497 skb
->ndisc_nodetype
= NDISC_NODETYPE_NODEFAULT
;
499 const struct in6_addr
*addr6
= &ipv6_hdr(skb
)->saddr
;
501 if (ipv6_addr_is_isatap(addr6
) &&
502 (addr6
->s6_addr32
[3] == iph
->saddr
) &&
503 ipv6_chk_prefix(addr6
, t
->dev
))
504 skb
->ndisc_nodetype
= NDISC_NODETYPE_HOST
;
512 static void ipip6_tunnel_uninit(struct net_device
*dev
)
514 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
515 struct sit_net
*sitn
= net_generic(tunnel
->net
, sit_net_id
);
517 if (dev
== sitn
->fb_tunnel_dev
) {
518 RCU_INIT_POINTER(sitn
->tunnels_wc
[0], NULL
);
520 ipip6_tunnel_unlink(sitn
, tunnel
);
521 ipip6_tunnel_del_prl(tunnel
, NULL
);
523 dst_cache_reset(&tunnel
->dst_cache
);
524 netdev_put(dev
, &tunnel
->dev_tracker
);
527 static int ipip6_err(struct sk_buff
*skb
, u32 info
)
529 const struct iphdr
*iph
= (const struct iphdr
*)skb
->data
;
530 const int type
= icmp_hdr(skb
)->type
;
531 const int code
= icmp_hdr(skb
)->code
;
532 unsigned int data_len
= 0;
539 case ICMP_PARAMETERPROB
:
542 case ICMP_DEST_UNREACH
:
545 /* Impossible event. */
548 /* All others are translated to HOST_UNREACH.
549 rfc2003 contains "deep thoughts" about NET_UNREACH,
550 I believe they are just ether pollution. --ANK
555 case ICMP_TIME_EXCEEDED
:
556 if (code
!= ICMP_EXC_TTL
)
558 data_len
= icmp_hdr(skb
)->un
.reserved
[1] * 4; /* RFC 4884 4.1 */
566 sifindex
= netif_is_l3_master(skb
->dev
) ? IPCB(skb
)->iif
: 0;
567 t
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
568 iph
->daddr
, iph
->saddr
, sifindex
);
572 if (type
== ICMP_DEST_UNREACH
&& code
== ICMP_FRAG_NEEDED
) {
573 ipv4_update_pmtu(skb
, dev_net(skb
->dev
), info
,
574 t
->parms
.link
, iph
->protocol
);
578 if (type
== ICMP_REDIRECT
) {
579 ipv4_redirect(skb
, dev_net(skb
->dev
), t
->parms
.link
,
586 if (__in6_dev_get(skb
->dev
) &&
587 !ip6_err_gen_icmpv6_unreach(skb
, iph
->ihl
* 4, type
, data_len
))
590 if (t
->parms
.iph
.daddr
== 0)
593 if (t
->parms
.iph
.ttl
== 0 && type
== ICMP_TIME_EXCEEDED
)
596 if (time_before(jiffies
, t
->err_time
+ IPTUNNEL_ERR_TIMEO
))
600 t
->err_time
= jiffies
;
605 static inline bool is_spoofed_6rd(struct ip_tunnel
*tunnel
, const __be32 v4addr
,
606 const struct in6_addr
*v6addr
)
609 if (check_6rd(tunnel
, v6addr
, &v4embed
) && v4addr
!= v4embed
)
614 /* Checks if an address matches an address on the tunnel interface.
615 * Used to detect the NAT of proto 41 packets and let them pass spoofing test.
617 * This function is called after we considered the packet as spoofed
619 * We may have a router that is doing NAT for proto 41 packets
620 * for an internal station. Destination a.a.a.a/PREFIX:bbbb:bbbb
621 * will be translated to n.n.n.n/PREFIX:bbbb:bbbb. And is_spoofed_6rd
622 * function will return true, dropping the packet.
623 * But, we can still check if is spoofed against the IP
624 * addresses associated with the interface.
626 static bool only_dnatted(const struct ip_tunnel
*tunnel
,
627 const struct in6_addr
*v6dst
)
631 #ifdef CONFIG_IPV6_SIT_6RD
632 prefix_len
= tunnel
->ip6rd
.prefixlen
+ 32
633 - tunnel
->ip6rd
.relay_prefixlen
;
637 return ipv6_chk_custom_prefix(v6dst
, prefix_len
, tunnel
->dev
);
640 /* Returns true if a packet is spoofed */
641 static bool packet_is_spoofed(struct sk_buff
*skb
,
642 const struct iphdr
*iph
,
643 struct ip_tunnel
*tunnel
)
645 const struct ipv6hdr
*ipv6h
;
647 if (tunnel
->dev
->priv_flags
& IFF_ISATAP
) {
648 if (!isatap_chksrc(skb
, iph
, tunnel
))
654 if (tunnel
->dev
->flags
& IFF_POINTOPOINT
)
657 ipv6h
= ipv6_hdr(skb
);
659 if (unlikely(is_spoofed_6rd(tunnel
, iph
->saddr
, &ipv6h
->saddr
))) {
660 net_warn_ratelimited("Src spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
661 &iph
->saddr
, &ipv6h
->saddr
,
662 &iph
->daddr
, &ipv6h
->daddr
);
666 if (likely(!is_spoofed_6rd(tunnel
, iph
->daddr
, &ipv6h
->daddr
)))
669 if (only_dnatted(tunnel
, &ipv6h
->daddr
))
672 net_warn_ratelimited("Dst spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
673 &iph
->saddr
, &ipv6h
->saddr
,
674 &iph
->daddr
, &ipv6h
->daddr
);
678 static int ipip6_rcv(struct sk_buff
*skb
)
680 const struct iphdr
*iph
= ip_hdr(skb
);
681 struct ip_tunnel
*tunnel
;
685 sifindex
= netif_is_l3_master(skb
->dev
) ? IPCB(skb
)->iif
: 0;
686 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
687 iph
->saddr
, iph
->daddr
, sifindex
);
689 if (tunnel
->parms
.iph
.protocol
!= IPPROTO_IPV6
&&
690 tunnel
->parms
.iph
.protocol
!= 0)
693 skb
->mac_header
= skb
->network_header
;
694 skb_reset_network_header(skb
);
695 IPCB(skb
)->flags
= 0;
696 skb
->dev
= tunnel
->dev
;
698 if (packet_is_spoofed(skb
, iph
, tunnel
)) {
699 DEV_STATS_INC(tunnel
->dev
, rx_errors
);
703 if (iptunnel_pull_header(skb
, 0, htons(ETH_P_IPV6
),
704 !net_eq(tunnel
->net
, dev_net(tunnel
->dev
))))
707 /* skb can be uncloned in iptunnel_pull_header, so
708 * old iph is no longer valid
710 iph
= (const struct iphdr
*)skb_mac_header(skb
);
711 skb_reset_mac_header(skb
);
713 err
= IP_ECN_decapsulate(iph
, skb
);
716 net_info_ratelimited("non-ECT from %pI4 with TOS=%#x\n",
717 &iph
->saddr
, iph
->tos
);
719 DEV_STATS_INC(tunnel
->dev
, rx_frame_errors
);
720 DEV_STATS_INC(tunnel
->dev
, rx_errors
);
725 dev_sw_netstats_rx_add(tunnel
->dev
, skb
->len
);
732 /* no tunnel matched, let upstream know, ipsec may handle it */
739 static const struct tnl_ptk_info ipip_tpi
= {
740 /* no tunnel info required for ipip. */
741 .proto
= htons(ETH_P_IP
),
744 #if IS_ENABLED(CONFIG_MPLS)
745 static const struct tnl_ptk_info mplsip_tpi
= {
746 /* no tunnel info required for mplsip. */
747 .proto
= htons(ETH_P_MPLS_UC
),
751 static int sit_tunnel_rcv(struct sk_buff
*skb
, u8 ipproto
)
753 const struct iphdr
*iph
;
754 struct ip_tunnel
*tunnel
;
757 sifindex
= netif_is_l3_master(skb
->dev
) ? IPCB(skb
)->iif
: 0;
760 tunnel
= ipip6_tunnel_lookup(dev_net(skb
->dev
), skb
->dev
,
761 iph
->saddr
, iph
->daddr
, sifindex
);
763 const struct tnl_ptk_info
*tpi
;
765 if (tunnel
->parms
.iph
.protocol
!= ipproto
&&
766 tunnel
->parms
.iph
.protocol
!= 0)
769 if (!xfrm4_policy_check(NULL
, XFRM_POLICY_IN
, skb
))
771 #if IS_ENABLED(CONFIG_MPLS)
772 if (ipproto
== IPPROTO_MPLS
)
777 if (iptunnel_pull_header(skb
, 0, tpi
->proto
, false))
779 skb_reset_mac_header(skb
);
781 return ip_tunnel_rcv(tunnel
, skb
, tpi
, NULL
, log_ecn_error
);
791 static int ipip_rcv(struct sk_buff
*skb
)
793 return sit_tunnel_rcv(skb
, IPPROTO_IPIP
);
796 #if IS_ENABLED(CONFIG_MPLS)
797 static int mplsip_rcv(struct sk_buff
*skb
)
799 return sit_tunnel_rcv(skb
, IPPROTO_MPLS
);
804 * If the IPv6 address comes from 6rd / 6to4 (RFC 3056) addr space this function
805 * stores the embedded IPv4 address in v4dst and returns true.
807 static bool check_6rd(struct ip_tunnel
*tunnel
, const struct in6_addr
*v6dst
,
810 #ifdef CONFIG_IPV6_SIT_6RD
811 if (ipv6_prefix_equal(v6dst
, &tunnel
->ip6rd
.prefix
,
812 tunnel
->ip6rd
.prefixlen
)) {
813 unsigned int pbw0
, pbi0
;
817 pbw0
= tunnel
->ip6rd
.prefixlen
>> 5;
818 pbi0
= tunnel
->ip6rd
.prefixlen
& 0x1f;
820 d
= tunnel
->ip6rd
.relay_prefixlen
< 32 ?
821 (ntohl(v6dst
->s6_addr32
[pbw0
]) << pbi0
) >>
822 tunnel
->ip6rd
.relay_prefixlen
: 0;
824 pbi1
= pbi0
- tunnel
->ip6rd
.relay_prefixlen
;
826 d
|= ntohl(v6dst
->s6_addr32
[pbw0
+ 1]) >>
829 *v4dst
= tunnel
->ip6rd
.relay_prefix
| htonl(d
);
833 if (v6dst
->s6_addr16
[0] == htons(0x2002)) {
834 /* 6to4 v6 addr has 16 bits prefix, 32 v4addr, 16 SLA, ... */
835 memcpy(v4dst
, &v6dst
->s6_addr16
[1], 4);
842 static inline __be32
try_6rd(struct ip_tunnel
*tunnel
,
843 const struct in6_addr
*v6dst
)
846 check_6rd(tunnel
, v6dst
, &dst
);
851 * This function assumes it is being called from dev_queue_xmit()
852 * and that skb is filled properly by that function.
855 static netdev_tx_t
ipip6_tunnel_xmit(struct sk_buff
*skb
,
856 struct net_device
*dev
)
858 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
859 const struct iphdr
*tiph
= &tunnel
->parms
.iph
;
860 const struct ipv6hdr
*iph6
= ipv6_hdr(skb
);
861 u8 tos
= tunnel
->parms
.iph
.tos
;
862 __be16 df
= tiph
->frag_off
;
863 struct rtable
*rt
; /* Route to the other host */
864 struct net_device
*tdev
; /* Device to other host */
865 unsigned int max_headroom
; /* The extra header space needed */
866 __be32 dst
= tiph
->daddr
;
869 const struct in6_addr
*addr6
;
872 u8 protocol
= IPPROTO_IPV6
;
873 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
876 tos
= ipv6_get_dsfield(iph6
);
878 /* ISATAP (RFC4214) - must come before 6to4 */
879 if (dev
->priv_flags
& IFF_ISATAP
) {
880 struct neighbour
*neigh
= NULL
;
881 bool do_tx_error
= false;
884 neigh
= dst_neigh_lookup(skb_dst(skb
), &iph6
->daddr
);
887 net_dbg_ratelimited("nexthop == NULL\n");
891 addr6
= (const struct in6_addr
*)&neigh
->primary_key
;
892 addr_type
= ipv6_addr_type(addr6
);
894 if ((addr_type
& IPV6_ADDR_UNICAST
) &&
895 ipv6_addr_is_isatap(addr6
))
896 dst
= addr6
->s6_addr32
[3];
900 neigh_release(neigh
);
906 dst
= try_6rd(tunnel
, &iph6
->daddr
);
909 struct neighbour
*neigh
= NULL
;
910 bool do_tx_error
= false;
913 neigh
= dst_neigh_lookup(skb_dst(skb
), &iph6
->daddr
);
916 net_dbg_ratelimited("nexthop == NULL\n");
920 addr6
= (const struct in6_addr
*)&neigh
->primary_key
;
921 addr_type
= ipv6_addr_type(addr6
);
923 if (addr_type
== IPV6_ADDR_ANY
) {
924 addr6
= &ipv6_hdr(skb
)->daddr
;
925 addr_type
= ipv6_addr_type(addr6
);
928 if ((addr_type
& IPV6_ADDR_COMPATv4
) != 0)
929 dst
= addr6
->s6_addr32
[3];
933 neigh_release(neigh
);
938 flowi4_init_output(&fl4
, tunnel
->parms
.link
, tunnel
->fwmark
,
939 tos
& INET_DSCP_MASK
, RT_SCOPE_UNIVERSE
,
940 IPPROTO_IPV6
, 0, dst
, tiph
->saddr
, 0, 0,
941 sock_net_uid(tunnel
->net
, NULL
));
943 rt
= dst_cache_get_ip4(&tunnel
->dst_cache
, &fl4
.saddr
);
945 rt
= ip_route_output_flow(tunnel
->net
, &fl4
, NULL
);
947 DEV_STATS_INC(dev
, tx_carrier_errors
);
950 dst_cache_set_ip4(&tunnel
->dst_cache
, &rt
->dst
, fl4
.saddr
);
953 if (rt
->rt_type
!= RTN_UNICAST
&& rt
->rt_type
!= RTN_LOCAL
) {
955 DEV_STATS_INC(dev
, tx_carrier_errors
);
962 DEV_STATS_INC(dev
, collisions
);
966 if (iptunnel_handle_offloads(skb
, SKB_GSO_IPXIP4
)) {
972 mtu
= dst_mtu(&rt
->dst
) - t_hlen
;
974 if (mtu
< IPV4_MIN_MTU
) {
975 DEV_STATS_INC(dev
, collisions
);
980 if (mtu
< IPV6_MIN_MTU
) {
985 if (tunnel
->parms
.iph
.daddr
)
986 skb_dst_update_pmtu_no_confirm(skb
, mtu
);
988 if (skb
->len
> mtu
&& !skb_is_gso(skb
)) {
989 icmpv6_ndo_send(skb
, ICMPV6_PKT_TOOBIG
, 0, mtu
);
995 if (tunnel
->err_count
> 0) {
996 if (time_before(jiffies
,
997 tunnel
->err_time
+ IPTUNNEL_ERR_TIMEO
)) {
999 dst_link_failure(skb
);
1001 tunnel
->err_count
= 0;
1005 * Okay, now see if we can stuff it in the buffer as-is.
1007 max_headroom
= LL_RESERVED_SPACE(tdev
) + t_hlen
;
1009 if (skb_headroom(skb
) < max_headroom
|| skb_shared(skb
) ||
1010 (skb_cloned(skb
) && !skb_clone_writable(skb
, 0))) {
1011 struct sk_buff
*new_skb
= skb_realloc_headroom(skb
, max_headroom
);
1014 DEV_STATS_INC(dev
, tx_dropped
);
1016 return NETDEV_TX_OK
;
1019 skb_set_owner_w(new_skb
, skb
->sk
);
1022 iph6
= ipv6_hdr(skb
);
1026 ttl
= iph6
->hop_limit
;
1027 tos
= INET_ECN_encapsulate(tos
, ipv6_get_dsfield(iph6
));
1029 if (ip_tunnel_encap(skb
, &tunnel
->encap
, &protocol
, &fl4
) < 0) {
1034 skb_set_inner_ipproto(skb
, IPPROTO_IPV6
);
1036 iptunnel_xmit(NULL
, rt
, skb
, fl4
.saddr
, fl4
.daddr
, protocol
, tos
, ttl
,
1037 df
, !net_eq(tunnel
->net
, dev_net(dev
)));
1038 return NETDEV_TX_OK
;
1041 dst_link_failure(skb
);
1044 DEV_STATS_INC(dev
, tx_errors
);
1045 return NETDEV_TX_OK
;
1048 static netdev_tx_t
sit_tunnel_xmit__(struct sk_buff
*skb
,
1049 struct net_device
*dev
, u8 ipproto
)
1051 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1052 const struct iphdr
*tiph
= &tunnel
->parms
.iph
;
1054 if (iptunnel_handle_offloads(skb
, SKB_GSO_IPXIP4
))
1057 skb_set_inner_ipproto(skb
, ipproto
);
1059 ip_tunnel_xmit(skb
, dev
, tiph
, ipproto
);
1060 return NETDEV_TX_OK
;
1063 DEV_STATS_INC(dev
, tx_errors
);
1064 return NETDEV_TX_OK
;
1067 static netdev_tx_t
sit_tunnel_xmit(struct sk_buff
*skb
,
1068 struct net_device
*dev
)
1070 if (!pskb_inet_may_pull(skb
))
1073 switch (skb
->protocol
) {
1074 case htons(ETH_P_IP
):
1075 sit_tunnel_xmit__(skb
, dev
, IPPROTO_IPIP
);
1077 case htons(ETH_P_IPV6
):
1078 ipip6_tunnel_xmit(skb
, dev
);
1080 #if IS_ENABLED(CONFIG_MPLS)
1081 case htons(ETH_P_MPLS_UC
):
1082 sit_tunnel_xmit__(skb
, dev
, IPPROTO_MPLS
);
1089 return NETDEV_TX_OK
;
1092 DEV_STATS_INC(dev
, tx_errors
);
1094 return NETDEV_TX_OK
;
1098 static void ipip6_tunnel_bind_dev(struct net_device
*dev
)
1100 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1101 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1102 struct net_device
*tdev
= NULL
;
1103 int hlen
= LL_MAX_HEADER
;
1104 const struct iphdr
*iph
;
1107 iph
= &tunnel
->parms
.iph
;
1110 struct rtable
*rt
= ip_route_output_ports(tunnel
->net
, &fl4
,
1112 iph
->daddr
, iph
->saddr
,
1115 iph
->tos
& INET_DSCP_MASK
,
1116 tunnel
->parms
.link
);
1122 dev
->flags
|= IFF_POINTOPOINT
;
1125 if (!tdev
&& tunnel
->parms
.link
)
1126 tdev
= __dev_get_by_index(tunnel
->net
, tunnel
->parms
.link
);
1128 if (tdev
&& !netif_is_l3_master(tdev
)) {
1131 mtu
= tdev
->mtu
- t_hlen
;
1132 if (mtu
< IPV6_MIN_MTU
)
1134 WRITE_ONCE(dev
->mtu
, mtu
);
1135 hlen
= tdev
->hard_header_len
+ tdev
->needed_headroom
;
1137 dev
->needed_headroom
= t_hlen
+ hlen
;
1140 static void ipip6_tunnel_update(struct ip_tunnel
*t
,
1141 struct ip_tunnel_parm_kern
*p
,
1144 struct net
*net
= t
->net
;
1145 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1147 ipip6_tunnel_unlink(sitn
, t
);
1149 t
->parms
.iph
.saddr
= p
->iph
.saddr
;
1150 t
->parms
.iph
.daddr
= p
->iph
.daddr
;
1151 __dev_addr_set(t
->dev
, &p
->iph
.saddr
, 4);
1152 memcpy(t
->dev
->broadcast
, &p
->iph
.daddr
, 4);
1153 ipip6_tunnel_link(sitn
, t
);
1154 t
->parms
.iph
.ttl
= p
->iph
.ttl
;
1155 t
->parms
.iph
.tos
= p
->iph
.tos
;
1156 t
->parms
.iph
.frag_off
= p
->iph
.frag_off
;
1157 if (t
->parms
.link
!= p
->link
|| t
->fwmark
!= fwmark
) {
1158 t
->parms
.link
= p
->link
;
1160 ipip6_tunnel_bind_dev(t
->dev
);
1162 dst_cache_reset(&t
->dst_cache
);
1163 netdev_state_change(t
->dev
);
1166 #ifdef CONFIG_IPV6_SIT_6RD
1167 static int ipip6_tunnel_update_6rd(struct ip_tunnel
*t
,
1168 struct ip_tunnel_6rd
*ip6rd
)
1170 struct in6_addr prefix
;
1171 __be32 relay_prefix
;
1173 if (ip6rd
->relay_prefixlen
> 32 ||
1174 ip6rd
->prefixlen
+ (32 - ip6rd
->relay_prefixlen
) > 64)
1177 ipv6_addr_prefix(&prefix
, &ip6rd
->prefix
, ip6rd
->prefixlen
);
1178 if (!ipv6_addr_equal(&prefix
, &ip6rd
->prefix
))
1180 if (ip6rd
->relay_prefixlen
)
1181 relay_prefix
= ip6rd
->relay_prefix
&
1182 htonl(0xffffffffUL
<<
1183 (32 - ip6rd
->relay_prefixlen
));
1186 if (relay_prefix
!= ip6rd
->relay_prefix
)
1189 t
->ip6rd
.prefix
= prefix
;
1190 t
->ip6rd
.relay_prefix
= relay_prefix
;
1191 t
->ip6rd
.prefixlen
= ip6rd
->prefixlen
;
1192 t
->ip6rd
.relay_prefixlen
= ip6rd
->relay_prefixlen
;
1193 dst_cache_reset(&t
->dst_cache
);
1194 netdev_state_change(t
->dev
);
1199 ipip6_tunnel_get6rd(struct net_device
*dev
, struct ip_tunnel_parm __user
*data
)
1201 struct ip_tunnel
*t
= netdev_priv(dev
);
1202 struct ip_tunnel_parm_kern p
;
1203 struct ip_tunnel_6rd ip6rd
;
1205 if (dev
== dev_to_sit_net(dev
)->fb_tunnel_dev
) {
1206 if (!ip_tunnel_parm_from_user(&p
, data
))
1208 t
= ipip6_tunnel_locate(t
->net
, &p
, 0);
1211 t
= netdev_priv(dev
);
1213 ip6rd
.prefix
= t
->ip6rd
.prefix
;
1214 ip6rd
.relay_prefix
= t
->ip6rd
.relay_prefix
;
1215 ip6rd
.prefixlen
= t
->ip6rd
.prefixlen
;
1216 ip6rd
.relay_prefixlen
= t
->ip6rd
.relay_prefixlen
;
1217 if (copy_to_user(data
, &ip6rd
, sizeof(ip6rd
)))
1223 ipip6_tunnel_6rdctl(struct net_device
*dev
, struct ip_tunnel_6rd __user
*data
,
1226 struct ip_tunnel
*t
= netdev_priv(dev
);
1227 struct ip_tunnel_6rd ip6rd
;
1230 if (!ns_capable(t
->net
->user_ns
, CAP_NET_ADMIN
))
1232 if (copy_from_user(&ip6rd
, data
, sizeof(ip6rd
)))
1235 if (cmd
!= SIOCDEL6RD
) {
1236 err
= ipip6_tunnel_update_6rd(t
, &ip6rd
);
1240 ipip6_tunnel_clone_6rd(dev
, dev_to_sit_net(dev
));
1244 #endif /* CONFIG_IPV6_SIT_6RD */
1246 static bool ipip6_valid_ip_proto(u8 ipproto
)
1248 return ipproto
== IPPROTO_IPV6
||
1249 ipproto
== IPPROTO_IPIP
||
1250 #if IS_ENABLED(CONFIG_MPLS)
1251 ipproto
== IPPROTO_MPLS
||
1257 __ipip6_tunnel_ioctl_validate(struct net
*net
, struct ip_tunnel_parm_kern
*p
)
1259 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
1262 if (!ipip6_valid_ip_proto(p
->iph
.protocol
))
1264 if (p
->iph
.version
!= 4 ||
1265 p
->iph
.ihl
!= 5 || (p
->iph
.frag_off
& htons(~IP_DF
)))
1269 p
->iph
.frag_off
|= htons(IP_DF
);
1274 ipip6_tunnel_get(struct net_device
*dev
, struct ip_tunnel_parm_kern
*p
)
1276 struct ip_tunnel
*t
= netdev_priv(dev
);
1278 if (dev
== dev_to_sit_net(dev
)->fb_tunnel_dev
)
1279 t
= ipip6_tunnel_locate(t
->net
, p
, 0);
1281 t
= netdev_priv(dev
);
1282 memcpy(p
, &t
->parms
, sizeof(*p
));
1287 ipip6_tunnel_add(struct net_device
*dev
, struct ip_tunnel_parm_kern
*p
)
1289 struct ip_tunnel
*t
= netdev_priv(dev
);
1292 err
= __ipip6_tunnel_ioctl_validate(t
->net
, p
);
1296 t
= ipip6_tunnel_locate(t
->net
, p
, 1);
1303 ipip6_tunnel_change(struct net_device
*dev
, struct ip_tunnel_parm_kern
*p
)
1305 struct ip_tunnel
*t
= netdev_priv(dev
);
1308 err
= __ipip6_tunnel_ioctl_validate(t
->net
, p
);
1312 t
= ipip6_tunnel_locate(t
->net
, p
, 0);
1313 if (dev
== dev_to_sit_net(dev
)->fb_tunnel_dev
) {
1321 if (((dev
->flags
& IFF_POINTOPOINT
) && !p
->iph
.daddr
) ||
1322 (!(dev
->flags
& IFF_POINTOPOINT
) && p
->iph
.daddr
))
1324 t
= netdev_priv(dev
);
1327 ipip6_tunnel_update(t
, p
, t
->fwmark
);
1334 ipip6_tunnel_del(struct net_device
*dev
, struct ip_tunnel_parm_kern
*p
)
1336 struct ip_tunnel
*t
= netdev_priv(dev
);
1338 if (!ns_capable(t
->net
->user_ns
, CAP_NET_ADMIN
))
1341 if (dev
== dev_to_sit_net(dev
)->fb_tunnel_dev
) {
1342 t
= ipip6_tunnel_locate(t
->net
, p
, 0);
1345 if (t
== netdev_priv(dev_to_sit_net(dev
)->fb_tunnel_dev
))
1349 unregister_netdevice(dev
);
1354 ipip6_tunnel_ctl(struct net_device
*dev
, struct ip_tunnel_parm_kern
*p
,
1359 return ipip6_tunnel_get(dev
, p
);
1361 return ipip6_tunnel_add(dev
, p
);
1363 return ipip6_tunnel_change(dev
, p
);
1365 return ipip6_tunnel_del(dev
, p
);
1372 ipip6_tunnel_siocdevprivate(struct net_device
*dev
, struct ifreq
*ifr
,
1373 void __user
*data
, int cmd
)
1380 return ip_tunnel_siocdevprivate(dev
, ifr
, data
, cmd
);
1382 return ipip6_tunnel_get_prl(dev
, data
);
1386 return ipip6_tunnel_prl_ctl(dev
, data
, cmd
);
1387 #ifdef CONFIG_IPV6_SIT_6RD
1389 return ipip6_tunnel_get6rd(dev
, data
);
1393 return ipip6_tunnel_6rdctl(dev
, data
, cmd
);
1400 static const struct net_device_ops ipip6_netdev_ops
= {
1401 .ndo_init
= ipip6_tunnel_init
,
1402 .ndo_uninit
= ipip6_tunnel_uninit
,
1403 .ndo_start_xmit
= sit_tunnel_xmit
,
1404 .ndo_siocdevprivate
= ipip6_tunnel_siocdevprivate
,
1405 .ndo_get_iflink
= ip_tunnel_get_iflink
,
1406 .ndo_tunnel_ctl
= ipip6_tunnel_ctl
,
1409 static void ipip6_dev_free(struct net_device
*dev
)
1411 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1413 dst_cache_destroy(&tunnel
->dst_cache
);
1416 #define SIT_FEATURES (NETIF_F_SG | \
1417 NETIF_F_FRAGLIST | \
1419 NETIF_F_GSO_SOFTWARE | \
1422 static void ipip6_tunnel_setup(struct net_device
*dev
)
1424 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1425 int t_hlen
= tunnel
->hlen
+ sizeof(struct iphdr
);
1427 dev
->netdev_ops
= &ipip6_netdev_ops
;
1428 dev
->header_ops
= &ip_tunnel_header_ops
;
1429 dev
->needs_free_netdev
= true;
1430 dev
->priv_destructor
= ipip6_dev_free
;
1432 dev
->type
= ARPHRD_SIT
;
1433 dev
->mtu
= ETH_DATA_LEN
- t_hlen
;
1434 dev
->min_mtu
= IPV6_MIN_MTU
;
1435 dev
->max_mtu
= IP6_MAX_MTU
- t_hlen
;
1436 dev
->flags
= IFF_NOARP
;
1437 netif_keep_dst(dev
);
1440 dev
->features
|= SIT_FEATURES
;
1441 dev
->hw_features
|= SIT_FEATURES
;
1442 dev
->pcpu_stat_type
= NETDEV_PCPU_STAT_TSTATS
;
1446 static int ipip6_tunnel_init(struct net_device
*dev
)
1448 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1452 tunnel
->net
= dev_net(dev
);
1453 strcpy(tunnel
->parms
.name
, dev
->name
);
1455 ipip6_tunnel_bind_dev(dev
);
1457 err
= dst_cache_init(&tunnel
->dst_cache
, GFP_KERNEL
);
1461 netdev_hold(dev
, &tunnel
->dev_tracker
, GFP_KERNEL
);
1462 netdev_lockdep_set_classes(dev
);
1466 static void __net_init
ipip6_fb_tunnel_init(struct net_device
*dev
)
1468 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1469 struct iphdr
*iph
= &tunnel
->parms
.iph
;
1470 struct net
*net
= dev_net(dev
);
1471 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1474 iph
->protocol
= IPPROTO_IPV6
;
1478 rcu_assign_pointer(sitn
->tunnels_wc
[0], tunnel
);
1481 static int ipip6_validate(struct nlattr
*tb
[], struct nlattr
*data
[],
1482 struct netlink_ext_ack
*extack
)
1486 if (!data
|| !data
[IFLA_IPTUN_PROTO
])
1489 proto
= nla_get_u8(data
[IFLA_IPTUN_PROTO
]);
1490 if (!ipip6_valid_ip_proto(proto
))
1496 static void ipip6_netlink_parms(struct nlattr
*data
[],
1497 struct ip_tunnel_parm_kern
*parms
,
1500 memset(parms
, 0, sizeof(*parms
));
1502 parms
->iph
.version
= 4;
1503 parms
->iph
.protocol
= IPPROTO_IPV6
;
1505 parms
->iph
.ttl
= 64;
1510 ip_tunnel_netlink_parms(data
, parms
);
1512 if (data
[IFLA_IPTUN_FWMARK
])
1513 *fwmark
= nla_get_u32(data
[IFLA_IPTUN_FWMARK
]);
1516 #ifdef CONFIG_IPV6_SIT_6RD
1517 /* This function returns true when 6RD attributes are present in the nl msg */
1518 static bool ipip6_netlink_6rd_parms(struct nlattr
*data
[],
1519 struct ip_tunnel_6rd
*ip6rd
)
1522 memset(ip6rd
, 0, sizeof(*ip6rd
));
1527 if (data
[IFLA_IPTUN_6RD_PREFIX
]) {
1529 ip6rd
->prefix
= nla_get_in6_addr(data
[IFLA_IPTUN_6RD_PREFIX
]);
1532 if (data
[IFLA_IPTUN_6RD_RELAY_PREFIX
]) {
1534 ip6rd
->relay_prefix
=
1535 nla_get_be32(data
[IFLA_IPTUN_6RD_RELAY_PREFIX
]);
1538 if (data
[IFLA_IPTUN_6RD_PREFIXLEN
]) {
1540 ip6rd
->prefixlen
= nla_get_u16(data
[IFLA_IPTUN_6RD_PREFIXLEN
]);
1543 if (data
[IFLA_IPTUN_6RD_RELAY_PREFIXLEN
]) {
1545 ip6rd
->relay_prefixlen
=
1546 nla_get_u16(data
[IFLA_IPTUN_6RD_RELAY_PREFIXLEN
]);
1553 static int ipip6_newlink(struct net
*src_net
, struct net_device
*dev
,
1554 struct nlattr
*tb
[], struct nlattr
*data
[],
1555 struct netlink_ext_ack
*extack
)
1557 struct net
*net
= dev_net(dev
);
1558 struct ip_tunnel
*nt
;
1559 struct ip_tunnel_encap ipencap
;
1560 #ifdef CONFIG_IPV6_SIT_6RD
1561 struct ip_tunnel_6rd ip6rd
;
1565 nt
= netdev_priv(dev
);
1567 if (ip_tunnel_netlink_encap_parms(data
, &ipencap
)) {
1568 err
= ip_tunnel_encap_setup(nt
, &ipencap
);
1573 ipip6_netlink_parms(data
, &nt
->parms
, &nt
->fwmark
);
1575 if (ipip6_tunnel_locate(net
, &nt
->parms
, 0))
1578 err
= ipip6_tunnel_create(dev
);
1583 u32 mtu
= nla_get_u32(tb
[IFLA_MTU
]);
1585 if (mtu
>= IPV6_MIN_MTU
&&
1586 mtu
<= IP6_MAX_MTU
- dev
->hard_header_len
)
1590 #ifdef CONFIG_IPV6_SIT_6RD
1591 if (ipip6_netlink_6rd_parms(data
, &ip6rd
)) {
1592 err
= ipip6_tunnel_update_6rd(nt
, &ip6rd
);
1594 unregister_netdevice_queue(dev
, NULL
);
1601 static int ipip6_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
1602 struct nlattr
*data
[],
1603 struct netlink_ext_ack
*extack
)
1605 struct ip_tunnel
*t
= netdev_priv(dev
);
1606 struct ip_tunnel_encap ipencap
;
1607 struct ip_tunnel_parm_kern p
;
1608 struct net
*net
= t
->net
;
1609 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1610 #ifdef CONFIG_IPV6_SIT_6RD
1611 struct ip_tunnel_6rd ip6rd
;
1613 __u32 fwmark
= t
->fwmark
;
1616 if (dev
== sitn
->fb_tunnel_dev
)
1619 if (ip_tunnel_netlink_encap_parms(data
, &ipencap
)) {
1620 err
= ip_tunnel_encap_setup(t
, &ipencap
);
1625 ipip6_netlink_parms(data
, &p
, &fwmark
);
1627 if (((dev
->flags
& IFF_POINTOPOINT
) && !p
.iph
.daddr
) ||
1628 (!(dev
->flags
& IFF_POINTOPOINT
) && p
.iph
.daddr
))
1631 t
= ipip6_tunnel_locate(net
, &p
, 0);
1637 t
= netdev_priv(dev
);
1639 ipip6_tunnel_update(t
, &p
, fwmark
);
1641 #ifdef CONFIG_IPV6_SIT_6RD
1642 if (ipip6_netlink_6rd_parms(data
, &ip6rd
))
1643 return ipip6_tunnel_update_6rd(t
, &ip6rd
);
1649 static size_t ipip6_get_size(const struct net_device
*dev
)
1652 /* IFLA_IPTUN_LINK */
1654 /* IFLA_IPTUN_LOCAL */
1656 /* IFLA_IPTUN_REMOTE */
1658 /* IFLA_IPTUN_TTL */
1660 /* IFLA_IPTUN_TOS */
1662 /* IFLA_IPTUN_PMTUDISC */
1664 /* IFLA_IPTUN_FLAGS */
1666 /* IFLA_IPTUN_PROTO */
1668 #ifdef CONFIG_IPV6_SIT_6RD
1669 /* IFLA_IPTUN_6RD_PREFIX */
1670 nla_total_size(sizeof(struct in6_addr
)) +
1671 /* IFLA_IPTUN_6RD_RELAY_PREFIX */
1673 /* IFLA_IPTUN_6RD_PREFIXLEN */
1675 /* IFLA_IPTUN_6RD_RELAY_PREFIXLEN */
1678 /* IFLA_IPTUN_ENCAP_TYPE */
1680 /* IFLA_IPTUN_ENCAP_FLAGS */
1682 /* IFLA_IPTUN_ENCAP_SPORT */
1684 /* IFLA_IPTUN_ENCAP_DPORT */
1686 /* IFLA_IPTUN_FWMARK */
1691 static int ipip6_fill_info(struct sk_buff
*skb
, const struct net_device
*dev
)
1693 struct ip_tunnel
*tunnel
= netdev_priv(dev
);
1694 struct ip_tunnel_parm_kern
*parm
= &tunnel
->parms
;
1696 if (nla_put_u32(skb
, IFLA_IPTUN_LINK
, parm
->link
) ||
1697 nla_put_in_addr(skb
, IFLA_IPTUN_LOCAL
, parm
->iph
.saddr
) ||
1698 nla_put_in_addr(skb
, IFLA_IPTUN_REMOTE
, parm
->iph
.daddr
) ||
1699 nla_put_u8(skb
, IFLA_IPTUN_TTL
, parm
->iph
.ttl
) ||
1700 nla_put_u8(skb
, IFLA_IPTUN_TOS
, parm
->iph
.tos
) ||
1701 nla_put_u8(skb
, IFLA_IPTUN_PMTUDISC
,
1702 !!(parm
->iph
.frag_off
& htons(IP_DF
))) ||
1703 nla_put_u8(skb
, IFLA_IPTUN_PROTO
, parm
->iph
.protocol
) ||
1704 nla_put_be16(skb
, IFLA_IPTUN_FLAGS
,
1705 ip_tunnel_flags_to_be16(parm
->i_flags
)) ||
1706 nla_put_u32(skb
, IFLA_IPTUN_FWMARK
, tunnel
->fwmark
))
1707 goto nla_put_failure
;
1709 #ifdef CONFIG_IPV6_SIT_6RD
1710 if (nla_put_in6_addr(skb
, IFLA_IPTUN_6RD_PREFIX
,
1711 &tunnel
->ip6rd
.prefix
) ||
1712 nla_put_in_addr(skb
, IFLA_IPTUN_6RD_RELAY_PREFIX
,
1713 tunnel
->ip6rd
.relay_prefix
) ||
1714 nla_put_u16(skb
, IFLA_IPTUN_6RD_PREFIXLEN
,
1715 tunnel
->ip6rd
.prefixlen
) ||
1716 nla_put_u16(skb
, IFLA_IPTUN_6RD_RELAY_PREFIXLEN
,
1717 tunnel
->ip6rd
.relay_prefixlen
))
1718 goto nla_put_failure
;
1721 if (nla_put_u16(skb
, IFLA_IPTUN_ENCAP_TYPE
,
1722 tunnel
->encap
.type
) ||
1723 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_SPORT
,
1724 tunnel
->encap
.sport
) ||
1725 nla_put_be16(skb
, IFLA_IPTUN_ENCAP_DPORT
,
1726 tunnel
->encap
.dport
) ||
1727 nla_put_u16(skb
, IFLA_IPTUN_ENCAP_FLAGS
,
1728 tunnel
->encap
.flags
))
1729 goto nla_put_failure
;
1737 static const struct nla_policy ipip6_policy
[IFLA_IPTUN_MAX
+ 1] = {
1738 [IFLA_IPTUN_LINK
] = { .type
= NLA_U32
},
1739 [IFLA_IPTUN_LOCAL
] = { .type
= NLA_U32
},
1740 [IFLA_IPTUN_REMOTE
] = { .type
= NLA_U32
},
1741 [IFLA_IPTUN_TTL
] = { .type
= NLA_U8
},
1742 [IFLA_IPTUN_TOS
] = { .type
= NLA_U8
},
1743 [IFLA_IPTUN_PMTUDISC
] = { .type
= NLA_U8
},
1744 [IFLA_IPTUN_FLAGS
] = { .type
= NLA_U16
},
1745 [IFLA_IPTUN_PROTO
] = { .type
= NLA_U8
},
1746 #ifdef CONFIG_IPV6_SIT_6RD
1747 [IFLA_IPTUN_6RD_PREFIX
] = { .len
= sizeof(struct in6_addr
) },
1748 [IFLA_IPTUN_6RD_RELAY_PREFIX
] = { .type
= NLA_U32
},
1749 [IFLA_IPTUN_6RD_PREFIXLEN
] = { .type
= NLA_U16
},
1750 [IFLA_IPTUN_6RD_RELAY_PREFIXLEN
] = { .type
= NLA_U16
},
1752 [IFLA_IPTUN_ENCAP_TYPE
] = { .type
= NLA_U16
},
1753 [IFLA_IPTUN_ENCAP_FLAGS
] = { .type
= NLA_U16
},
1754 [IFLA_IPTUN_ENCAP_SPORT
] = { .type
= NLA_U16
},
1755 [IFLA_IPTUN_ENCAP_DPORT
] = { .type
= NLA_U16
},
1756 [IFLA_IPTUN_FWMARK
] = { .type
= NLA_U32
},
1759 static void ipip6_dellink(struct net_device
*dev
, struct list_head
*head
)
1761 struct net
*net
= dev_net(dev
);
1762 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1764 if (dev
!= sitn
->fb_tunnel_dev
)
1765 unregister_netdevice_queue(dev
, head
);
1768 static struct rtnl_link_ops sit_link_ops __read_mostly
= {
1770 .maxtype
= IFLA_IPTUN_MAX
,
1771 .policy
= ipip6_policy
,
1772 .priv_size
= sizeof(struct ip_tunnel
),
1773 .setup
= ipip6_tunnel_setup
,
1774 .validate
= ipip6_validate
,
1775 .newlink
= ipip6_newlink
,
1776 .changelink
= ipip6_changelink
,
1777 .get_size
= ipip6_get_size
,
1778 .fill_info
= ipip6_fill_info
,
1779 .dellink
= ipip6_dellink
,
1780 .get_link_net
= ip_tunnel_get_link_net
,
1783 static struct xfrm_tunnel sit_handler __read_mostly
= {
1784 .handler
= ipip6_rcv
,
1785 .err_handler
= ipip6_err
,
1789 static struct xfrm_tunnel ipip_handler __read_mostly
= {
1790 .handler
= ipip_rcv
,
1791 .err_handler
= ipip6_err
,
1795 #if IS_ENABLED(CONFIG_MPLS)
1796 static struct xfrm_tunnel mplsip_handler __read_mostly
= {
1797 .handler
= mplsip_rcv
,
1798 .err_handler
= ipip6_err
,
1803 static void __net_exit
sit_destroy_tunnels(struct net
*net
,
1804 struct list_head
*head
)
1806 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1807 struct net_device
*dev
, *aux
;
1810 for_each_netdev_safe(net
, dev
, aux
)
1811 if (dev
->rtnl_link_ops
== &sit_link_ops
)
1812 unregister_netdevice_queue(dev
, head
);
1814 for (prio
= 0; prio
< 4; prio
++) {
1816 for (h
= 0; h
< (prio
? IP6_SIT_HASH_SIZE
: 1); h
++) {
1817 struct ip_tunnel
*t
;
1819 t
= rtnl_dereference(sitn
->tunnels
[prio
][h
]);
1821 /* If dev is in the same netns, it has already
1822 * been added to the list by the previous loop.
1824 if (!net_eq(dev_net(t
->dev
), net
))
1825 unregister_netdevice_queue(t
->dev
,
1827 t
= rtnl_dereference(t
->next
);
1833 static int __net_init
sit_init_net(struct net
*net
)
1835 struct sit_net
*sitn
= net_generic(net
, sit_net_id
);
1836 struct ip_tunnel
*t
;
1839 sitn
->tunnels
[0] = sitn
->tunnels_wc
;
1840 sitn
->tunnels
[1] = sitn
->tunnels_l
;
1841 sitn
->tunnels
[2] = sitn
->tunnels_r
;
1842 sitn
->tunnels
[3] = sitn
->tunnels_r_l
;
1844 if (!net_has_fallback_tunnels(net
))
1847 sitn
->fb_tunnel_dev
= alloc_netdev(sizeof(struct ip_tunnel
), "sit0",
1849 ipip6_tunnel_setup
);
1850 if (!sitn
->fb_tunnel_dev
) {
1854 dev_net_set(sitn
->fb_tunnel_dev
, net
);
1855 sitn
->fb_tunnel_dev
->rtnl_link_ops
= &sit_link_ops
;
1856 /* FB netdevice is special: we have one, and only one per netns.
1857 * Allowing to move it to another netns is clearly unsafe.
1859 sitn
->fb_tunnel_dev
->netns_local
= true;
1861 err
= register_netdev(sitn
->fb_tunnel_dev
);
1865 ipip6_tunnel_clone_6rd(sitn
->fb_tunnel_dev
, sitn
);
1866 ipip6_fb_tunnel_init(sitn
->fb_tunnel_dev
);
1868 t
= netdev_priv(sitn
->fb_tunnel_dev
);
1870 strcpy(t
->parms
.name
, sitn
->fb_tunnel_dev
->name
);
1874 free_netdev(sitn
->fb_tunnel_dev
);
1879 static void __net_exit
sit_exit_batch_rtnl(struct list_head
*net_list
,
1880 struct list_head
*dev_to_kill
)
1885 list_for_each_entry(net
, net_list
, exit_list
)
1886 sit_destroy_tunnels(net
, dev_to_kill
);
1889 static struct pernet_operations sit_net_ops
= {
1890 .init
= sit_init_net
,
1891 .exit_batch_rtnl
= sit_exit_batch_rtnl
,
1893 .size
= sizeof(struct sit_net
),
1896 static void __exit
sit_cleanup(void)
1898 rtnl_link_unregister(&sit_link_ops
);
1899 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1900 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
1901 #if IS_ENABLED(CONFIG_MPLS)
1902 xfrm4_tunnel_deregister(&mplsip_handler
, AF_MPLS
);
1905 unregister_pernet_device(&sit_net_ops
);
1906 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1909 static int __init
sit_init(void)
1913 pr_info("IPv6, IPv4 and MPLS over IPv4 tunneling driver\n");
1915 err
= register_pernet_device(&sit_net_ops
);
1918 err
= xfrm4_tunnel_register(&sit_handler
, AF_INET6
);
1920 pr_info("%s: can't register ip6ip4\n", __func__
);
1921 goto xfrm_tunnel_failed
;
1923 err
= xfrm4_tunnel_register(&ipip_handler
, AF_INET
);
1925 pr_info("%s: can't register ip4ip4\n", __func__
);
1926 goto xfrm_tunnel4_failed
;
1928 #if IS_ENABLED(CONFIG_MPLS)
1929 err
= xfrm4_tunnel_register(&mplsip_handler
, AF_MPLS
);
1931 pr_info("%s: can't register mplsip\n", __func__
);
1932 goto xfrm_tunnel_mpls_failed
;
1935 err
= rtnl_link_register(&sit_link_ops
);
1937 goto rtnl_link_failed
;
1943 #if IS_ENABLED(CONFIG_MPLS)
1944 xfrm4_tunnel_deregister(&mplsip_handler
, AF_MPLS
);
1945 xfrm_tunnel_mpls_failed
:
1947 xfrm4_tunnel_deregister(&ipip_handler
, AF_INET
);
1948 xfrm_tunnel4_failed
:
1949 xfrm4_tunnel_deregister(&sit_handler
, AF_INET6
);
1951 unregister_pernet_device(&sit_net_ops
);
1955 module_init(sit_init
);
1956 module_exit(sit_cleanup
);
1957 MODULE_DESCRIPTION("IPv6-in-IPv4 tunnel SIT driver");
1958 MODULE_LICENSE("GPL");
1959 MODULE_ALIAS_RTNL_LINK("sit");
1960 MODULE_ALIAS_NETDEV("sit0");