3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
7 * Ian P. Morris <I.P.Morris@soton.ac.uk>
9 * Based in linux/net/ipv4/ip_input.c
11 * This program is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU General Public License
13 * as published by the Free Software Foundation; either version
14 * 2 of the License, or (at your option) any later version.
18 * Mitsuru KANDA @USAGI and
19 * YOSHIFUJI Hideaki @USAGI: Remove ipv6_parse_exthdrs().
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/netdevice.h>
28 #include <linux/in6.h>
29 #include <linux/icmpv6.h>
30 #include <linux/mroute6.h>
31 #include <linux/slab.h>
32 #include <linux/indirect_call_wrapper.h>
34 #include <linux/netfilter.h>
35 #include <linux/netfilter_ipv6.h>
41 #include <net/protocol.h>
42 #include <net/transp_v6.h>
43 #include <net/rawv6.h>
44 #include <net/ndisc.h>
45 #include <net/ip6_route.h>
46 #include <net/addrconf.h>
48 #include <net/inet_ecn.h>
49 #include <net/dst_metadata.h>
51 INDIRECT_CALLABLE_DECLARE(void udp_v6_early_demux(struct sk_buff *));
52 INDIRECT_CALLABLE_DECLARE(void tcp_v6_early_demux(struct sk_buff *));
53 static void ip6_rcv_finish_core(struct net *net, struct sock *sk,
56 void (*edemux)(struct sk_buff *skb);
58 if (net->ipv4.sysctl_ip_early_demux && !skb_dst(skb) && skb->sk == NULL) {
59 const struct inet6_protocol *ipprot;
61 ipprot = rcu_dereference(inet6_protos[ipv6_hdr(skb)->nexthdr]);
62 if (ipprot && (edemux = READ_ONCE(ipprot->early_demux)))
63 INDIRECT_CALL_2(edemux, tcp_v6_early_demux,
64 udp_v6_early_demux, skb);
66 if (!skb_valid_dst(skb))
70 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
72 /* if ingress device is enslaved to an L3 master device pass the
73 * skb to its handler for processing
75 skb = l3mdev_ip6_rcv(skb);
77 return NET_RX_SUCCESS;
78 ip6_rcv_finish_core(net, sk, skb);
80 return dst_input(skb);
83 static void ip6_sublist_rcv_finish(struct list_head *head)
85 struct sk_buff *skb, *next;
87 list_for_each_entry_safe(skb, next, head, list)
91 static void ip6_list_rcv_finish(struct net *net, struct sock *sk,
92 struct list_head *head)
94 struct dst_entry *curr_dst = NULL;
95 struct sk_buff *skb, *next;
96 struct list_head sublist;
98 INIT_LIST_HEAD(&sublist);
99 list_for_each_entry_safe(skb, next, head, list) {
100 struct dst_entry *dst;
102 skb_list_del_init(skb);
103 /* if ingress device is enslaved to an L3 master device pass the
104 * skb to its handler for processing
106 skb = l3mdev_ip6_rcv(skb);
109 ip6_rcv_finish_core(net, sk, skb);
111 if (curr_dst != dst) {
112 /* dispatch old sublist */
113 if (!list_empty(&sublist))
114 ip6_sublist_rcv_finish(&sublist);
115 /* start new sublist */
116 INIT_LIST_HEAD(&sublist);
119 list_add_tail(&skb->list, &sublist);
121 /* dispatch final sublist */
122 ip6_sublist_rcv_finish(&sublist);
125 static struct sk_buff *ip6_rcv_core(struct sk_buff *skb, struct net_device *dev,
128 const struct ipv6hdr *hdr;
130 struct inet6_dev *idev;
132 if (skb->pkt_type == PACKET_OTHERHOST) {
139 idev = __in6_dev_get(skb->dev);
141 __IP6_UPD_PO_STATS(net, idev, IPSTATS_MIB_IN, skb->len);
143 if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL ||
144 !idev || unlikely(idev->cnf.disable_ipv6)) {
145 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDISCARDS);
149 memset(IP6CB(skb), 0, sizeof(struct inet6_skb_parm));
152 * Store incoming device index. When the packet will
153 * be queued, we cannot refer to skb->dev anymore.
155 * BTW, when we send a packet for our own local address on a
156 * non-loopback interface (e.g. ethX), it is being delivered
157 * via the loopback interface (lo) here; skb->dev = loopback_dev.
158 * It, however, should be considered as if it is being
159 * arrived via the sending interface (ethX), because of the
160 * nature of scoping architecture. --yoshfuji
162 IP6CB(skb)->iif = skb_valid_dst(skb) ? ip6_dst_idev(skb_dst(skb))->dev->ifindex : dev->ifindex;
164 if (unlikely(!pskb_may_pull(skb, sizeof(*hdr))))
169 if (hdr->version != 6)
172 __IP6_ADD_STATS(net, idev,
173 IPSTATS_MIB_NOECTPKTS +
174 (ipv6_get_dsfield(hdr) & INET_ECN_MASK),
175 max_t(unsigned short, 1, skb_shinfo(skb)->gso_segs));
178 * The loopback address must not be used as the source address in IPv6
179 * packets that are sent outside of a single node. [..]
180 * A packet received on an interface with a destination address
181 * of loopback must be dropped.
183 if ((ipv6_addr_loopback(&hdr->saddr) ||
184 ipv6_addr_loopback(&hdr->daddr)) &&
185 !(dev->flags & IFF_LOOPBACK) &&
186 !netif_is_l3_master(dev))
189 /* RFC4291 Errata ID: 3480
190 * Interface-Local scope spans only a single interface on a
191 * node and is useful only for loopback transmission of
192 * multicast. Packets with interface-local scope received
193 * from another node must be discarded.
195 if (!(skb->pkt_type == PACKET_LOOPBACK ||
196 dev->flags & IFF_LOOPBACK) &&
197 ipv6_addr_is_multicast(&hdr->daddr) &&
198 IPV6_ADDR_MC_SCOPE(&hdr->daddr) == 1)
201 /* If enabled, drop unicast packets that were encapsulated in link-layer
202 * multicast or broadcast to protected against the so-called "hole-196"
203 * attack in 802.11 wireless.
205 if (!ipv6_addr_is_multicast(&hdr->daddr) &&
206 (skb->pkt_type == PACKET_BROADCAST ||
207 skb->pkt_type == PACKET_MULTICAST) &&
208 idev->cnf.drop_unicast_in_l2_multicast)
212 * Nodes must not originate a packet to a multicast address whose scope
213 * field contains the reserved value 0; if such a packet is received, it
214 * must be silently dropped.
216 if (ipv6_addr_is_multicast(&hdr->daddr) &&
217 IPV6_ADDR_MC_SCOPE(&hdr->daddr) == 0)
222 * Multicast addresses must not be used as source addresses in IPv6
223 * packets or appear in any Routing header.
225 if (ipv6_addr_is_multicast(&hdr->saddr))
228 skb->transport_header = skb->network_header + sizeof(*hdr);
229 IP6CB(skb)->nhoff = offsetof(struct ipv6hdr, nexthdr);
231 pkt_len = ntohs(hdr->payload_len);
233 /* pkt_len may be zero if Jumbo payload option is present */
234 if (pkt_len || hdr->nexthdr != NEXTHDR_HOP) {
235 if (pkt_len + sizeof(struct ipv6hdr) > skb->len) {
237 idev, IPSTATS_MIB_INTRUNCATEDPKTS);
240 if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr))) {
241 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
247 if (hdr->nexthdr == NEXTHDR_HOP) {
248 if (ipv6_parse_hopopts(skb) < 0) {
249 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
257 /* Must drop socket now because of tproxy. */
262 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
269 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, struct net_device *orig_dev)
271 struct net *net = dev_net(skb->dev);
273 skb = ip6_rcv_core(skb, dev, net);
276 return NF_HOOK(NFPROTO_IPV6, NF_INET_PRE_ROUTING,
277 net, NULL, skb, dev, NULL,
281 static void ip6_sublist_rcv(struct list_head *head, struct net_device *dev,
284 NF_HOOK_LIST(NFPROTO_IPV6, NF_INET_PRE_ROUTING, net, NULL,
285 head, dev, NULL, ip6_rcv_finish);
286 ip6_list_rcv_finish(net, NULL, head);
289 /* Receive a list of IPv6 packets */
290 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
291 struct net_device *orig_dev)
293 struct net_device *curr_dev = NULL;
294 struct net *curr_net = NULL;
295 struct sk_buff *skb, *next;
296 struct list_head sublist;
298 INIT_LIST_HEAD(&sublist);
299 list_for_each_entry_safe(skb, next, head, list) {
300 struct net_device *dev = skb->dev;
301 struct net *net = dev_net(dev);
303 skb_list_del_init(skb);
304 skb = ip6_rcv_core(skb, dev, net);
308 if (curr_dev != dev || curr_net != net) {
309 /* dispatch old sublist */
310 if (!list_empty(&sublist))
311 ip6_sublist_rcv(&sublist, curr_dev, curr_net);
312 /* start new sublist */
313 INIT_LIST_HEAD(&sublist);
317 list_add_tail(&skb->list, &sublist);
319 /* dispatch final sublist */
320 ip6_sublist_rcv(&sublist, curr_dev, curr_net);
323 INDIRECT_CALLABLE_DECLARE(int udpv6_rcv(struct sk_buff *));
324 INDIRECT_CALLABLE_DECLARE(int tcp_v6_rcv(struct sk_buff *));
327 * Deliver the packet to the host
329 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr,
332 const struct inet6_protocol *ipprot;
333 struct inet6_dev *idev;
338 * Parse extension headers
342 idev = ip6_dst_idev(skb_dst(skb));
343 nhoff = IP6CB(skb)->nhoff;
345 if (!pskb_pull(skb, skb_transport_offset(skb)))
347 nexthdr = skb_network_header(skb)[nhoff];
351 raw = raw6_local_deliver(skb, nexthdr);
352 ipprot = rcu_dereference(inet6_protos[nexthdr]);
357 if (!(ipprot->flags & INET6_PROTO_FINAL)) {
358 /* Once we've seen a final protocol don't
359 * allow encapsulation on any non-final
360 * ones. This allows foo in UDP encapsulation
365 } else if (ipprot->flags & INET6_PROTO_FINAL) {
366 const struct ipv6hdr *hdr;
367 int sdif = inet6_sdif(skb);
368 struct net_device *dev;
370 /* Only do this once for first final protocol */
373 /* Free reference early: we don't need it any more,
374 and it may hold ip_conntrack module loaded
378 skb_postpull_rcsum(skb, skb_network_header(skb),
379 skb_network_header_len(skb));
382 /* skb->dev passed may be master dev for vrfs. */
384 dev = dev_get_by_index_rcu(net, sdif);
391 if (ipv6_addr_is_multicast(&hdr->daddr) &&
392 !ipv6_chk_mcast_addr(dev, &hdr->daddr,
394 !ipv6_is_mld(skb, nexthdr, skb_network_header_len(skb)))
397 if (!(ipprot->flags & INET6_PROTO_NOPOLICY) &&
398 !xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
401 ret = INDIRECT_CALL_2(ipprot->handler, tcp_v6_rcv, udpv6_rcv,
404 if (ipprot->flags & INET6_PROTO_FINAL) {
405 /* Not an extension header, most likely UDP
406 * encapsulation. Use return value as nexthdr
407 * protocol not nhoff (which presumably is
408 * not set by handler).
415 } else if (ret == 0) {
416 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDELIVERS);
420 if (xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
421 __IP6_INC_STATS(net, idev,
422 IPSTATS_MIB_INUNKNOWNPROTOS);
423 icmpv6_send(skb, ICMPV6_PARAMPROB,
424 ICMPV6_UNK_NEXTHDR, nhoff);
428 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDELIVERS);
435 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDISCARDS);
439 static int ip6_input_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
442 ip6_protocol_deliver_rcu(net, skb, 0, false);
449 int ip6_input(struct sk_buff *skb)
451 return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_IN,
452 dev_net(skb->dev), NULL, skb, skb->dev, NULL,
455 EXPORT_SYMBOL_GPL(ip6_input);
457 int ip6_mc_input(struct sk_buff *skb)
459 int sdif = inet6_sdif(skb);
460 const struct ipv6hdr *hdr;
461 struct net_device *dev;
464 __IP6_UPD_PO_STATS(dev_net(skb_dst(skb)->dev),
465 __in6_dev_get_safely(skb->dev), IPSTATS_MIB_INMCAST,
468 /* skb->dev passed may be master dev for vrfs. */
471 dev = dev_get_by_index_rcu(dev_net(skb->dev), sdif);
482 deliver = ipv6_chk_mcast_addr(dev, &hdr->daddr, NULL);
486 #ifdef CONFIG_IPV6_MROUTE
488 * IPv6 multicast router mode is now supported ;)
490 if (dev_net(skb->dev)->ipv6.devconf_all->mc_forwarding &&
491 !(ipv6_addr_type(&hdr->daddr) &
492 (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)) &&
493 likely(!(IP6CB(skb)->flags & IP6SKB_FORWARDED))) {
495 * Okay, we try to forward - split and duplicate
498 struct sk_buff *skb2;
499 struct inet6_skb_parm *opt = IP6CB(skb);
502 if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
503 /* Check if this is a mld message */
504 u8 nexthdr = hdr->nexthdr;
508 /* Check if the value of Router Alert
509 * is for MLD (0x0000).
511 if (opt->ra == htons(IPV6_OPT_ROUTERALERT_MLD)) {
514 if (!ipv6_ext_hdr(nexthdr)) {
518 offset = ipv6_skip_exthdr(skb, sizeof(*hdr),
519 &nexthdr, &frag_off);
523 if (ipv6_is_mld(skb, nexthdr, offset))
528 /* unknown RA - process it normally */
532 skb2 = skb_clone(skb, GFP_ATOMIC);