net/ipv6/ip6_input.c

   1 /*
   2  *      IPv6 input
   3  *      Linux INET6 implementation
   4  *
   5  *      Authors:
   6  *      Pedro Roque             <roque@di.fc.ul.pt>
   7  *      Ian P. Morris           <I.P.Morris@soton.ac.uk>
   8  *
   9  *      Based in linux/net/ipv4/ip_input.c
  10  *
  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.
  15  */
  16 /* Changes
  17  *
  18  *      Mitsuru KANDA @USAGI and
  19  *      YOSHIFUJI Hideaki @USAGI: Remove ipv6_parse_exthdrs().
  20  */
  21
  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
  33 #include <linux/netfilter.h>
  34 #include <linux/netfilter_ipv6.h>
  35
  36 #include <net/sock.h>
  37 #include <net/snmp.h>
  38
  39 #include <net/ipv6.h>
  40 #include <net/protocol.h>
  41 #include <net/transp_v6.h>
  42 #include <net/rawv6.h>
  43 #include <net/ndisc.h>
  44 #include <net/ip6_route.h>
  45 #include <net/addrconf.h>
  46 #include <net/xfrm.h>
  47 #include <net/inet_ecn.h>
  48
  49
  50 int ip6_rcv_finish(struct sk_buff *skb)
  51 {
  52         if (sysctl_ip_early_demux && !skb_dst(skb) && skb->sk == NULL) {
  53                 const struct inet6_protocol *ipprot;
  54
  55                 ipprot = rcu_dereference(inet6_protos[ipv6_hdr(skb)->nexthdr]);
  56                 if (ipprot && ipprot->early_demux)
  57                         ipprot->early_demux(skb);
  58         }
  59         if (!skb_dst(skb))
  60                 ip6_route_input(skb);
  61
  62         return dst_input(skb);
  63 }
  64
  65 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, struct net_device *orig_dev)
  66 {
  67         const struct ipv6hdr *hdr;
  68         u32 pkt_len;
  69         struct inet6_dev *idev;
  70         struct net *net = dev_net(skb->dev);
  71
  72         if (skb->pkt_type == PACKET_OTHERHOST) {
  73                 kfree_skb(skb);
  74                 return NET_RX_DROP;
  75         }
  76
  77         rcu_read_lock();
  78
  79         idev = __in6_dev_get(skb->dev);
  80
  81         IP6_UPD_PO_STATS_BH(net, idev, IPSTATS_MIB_IN, skb->len);
  82
  83         if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL ||
  84             !idev || unlikely(idev->cnf.disable_ipv6)) {
  85                 IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INDISCARDS);
  86                 goto drop;
  87         }
  88
  89         memset(IP6CB(skb), 0, sizeof(struct inet6_skb_parm));
  90
  91         /*
  92          * Store incoming device index. When the packet will
  93          * be queued, we cannot refer to skb->dev anymore.
  94          *
  95          * BTW, when we send a packet for our own local address on a
  96          * non-loopback interface (e.g. ethX), it is being delivered
  97          * via the loopback interface (lo) here; skb->dev = loopback_dev.
  98          * It, however, should be considered as if it is being
  99          * arrived via the sending interface (ethX), because of the
 100          * nature of scoping architecture. --yoshfuji
 101          */
 102         IP6CB(skb)->iif = skb_dst(skb) ? ip6_dst_idev(skb_dst(skb))->dev->ifindex : dev->ifindex;
 103
 104         if (unlikely(!pskb_may_pull(skb, sizeof(*hdr))))
 105                 goto err;
 106
 107         hdr = ipv6_hdr(skb);
 108
 109         if (hdr->version != 6)
 110                 goto err;
 111
 112         IP6_ADD_STATS_BH(dev_net(dev), idev,
 113                          IPSTATS_MIB_NOECTPKTS +
 114                                 (ipv6_get_dsfield(hdr) & INET_ECN_MASK),
 115                          max_t(unsigned short, 1, skb_shinfo(skb)->gso_segs));
 116         /*
 117          * RFC4291 2.5.3
 118          * A packet received on an interface with a destination address
 119          * of loopback must be dropped.
 120          */
 121         if (!(dev->flags & IFF_LOOPBACK) &&
 122             ipv6_addr_loopback(&hdr->daddr))
 123                 goto err;
 124
 125         /* RFC4291 Errata ID: 3480
 126          * Interface-Local scope spans only a single interface on a
 127          * node and is useful only for loopback transmission of
 128          * multicast.  Packets with interface-local scope received
 129          * from another node must be discarded.
 130          */
 131         if (!(skb->pkt_type == PACKET_LOOPBACK ||
 132               dev->flags & IFF_LOOPBACK) &&
 133             ipv6_addr_is_multicast(&hdr->daddr) &&
 134             IPV6_ADDR_MC_SCOPE(&hdr->daddr) == 1)
 135                 goto err;
 136
 137         /* RFC4291 2.7
 138          * Nodes must not originate a packet to a multicast address whose scope
 139          * field contains the reserved value 0; if such a packet is received, it
 140          * must be silently dropped.
 141          */
 142         if (ipv6_addr_is_multicast(&hdr->daddr) &&
 143             IPV6_ADDR_MC_SCOPE(&hdr->daddr) == 0)
 144                 goto err;
 145
 146         /*
 147          * RFC4291 2.7
 148          * Multicast addresses must not be used as source addresses in IPv6
 149          * packets or appear in any Routing header.
 150          */
 151         if (ipv6_addr_is_multicast(&hdr->saddr))
 152                 goto err;
 153
 154         skb->transport_header = skb->network_header + sizeof(*hdr);
 155         IP6CB(skb)->nhoff = offsetof(struct ipv6hdr, nexthdr);
 156
 157         pkt_len = ntohs(hdr->payload_len);
 158
 159         /* pkt_len may be zero if Jumbo payload option is present */
 160         if (pkt_len || hdr->nexthdr != NEXTHDR_HOP) {
 161                 if (pkt_len + sizeof(struct ipv6hdr) > skb->len) {
 162                         IP6_INC_STATS_BH(net,
 163                                          idev, IPSTATS_MIB_INTRUNCATEDPKTS);
 164                         goto drop;
 165                 }
 166                 if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr))) {
 167                         IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INHDRERRORS);
 168                         goto drop;
 169                 }
 170                 hdr = ipv6_hdr(skb);
 171         }
 172
 173         if (hdr->nexthdr == NEXTHDR_HOP) {
 174                 if (ipv6_parse_hopopts(skb) < 0) {
 175                         IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INHDRERRORS);
 176                         rcu_read_unlock();
 177                         return NET_RX_DROP;
 178                 }
 179         }
 180
 181         rcu_read_unlock();
 182
 183         /* Must drop socket now because of tproxy. */
 184         skb_orphan(skb);
 185
 186         return NF_HOOK(NFPROTO_IPV6, NF_INET_PRE_ROUTING, skb, dev, NULL,
 187                        ip6_rcv_finish);
 188 err:
 189         IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INHDRERRORS);
 190 drop:
 191         rcu_read_unlock();
 192         kfree_skb(skb);
 193         return NET_RX_DROP;
 194 }
 195
 196 /*
 197  *      Deliver the packet to the host
 198  */
 199
 200
 201 static int ip6_input_finish(struct sk_buff *skb)
 202 {
 203         struct net *net = dev_net(skb_dst(skb)->dev);
 204         const struct inet6_protocol *ipprot;
 205         struct inet6_dev *idev;
 206         unsigned int nhoff;
 207         int nexthdr;
 208         bool raw;
 209
 210         /*
 211          *      Parse extension headers
 212          */
 213
 214         rcu_read_lock();
 215 resubmit:
 216         idev = ip6_dst_idev(skb_dst(skb));
 217         if (!pskb_pull(skb, skb_transport_offset(skb)))
 218                 goto discard;
 219         nhoff = IP6CB(skb)->nhoff;
 220         nexthdr = skb_network_header(skb)[nhoff];
 221
 222         raw = raw6_local_deliver(skb, nexthdr);
 223         if ((ipprot = rcu_dereference(inet6_protos[nexthdr])) != NULL) {
 224                 int ret;
 225
 226                 if (ipprot->flags & INET6_PROTO_FINAL) {
 227                         const struct ipv6hdr *hdr;
 228
 229                         /* Free reference early: we don't need it any more,
 230                            and it may hold ip_conntrack module loaded
 231                            indefinitely. */
 232                         nf_reset(skb);
 233
 234                         skb_postpull_rcsum(skb, skb_network_header(skb),
 235                                            skb_network_header_len(skb));
 236                         hdr = ipv6_hdr(skb);
 237                         if (ipv6_addr_is_multicast(&hdr->daddr) &&
 238                             !ipv6_chk_mcast_addr(skb->dev, &hdr->daddr,
 239                             &hdr->saddr) &&
 240                             !ipv6_is_mld(skb, nexthdr, skb_network_header_len(skb)))
 241                                 goto discard;
 242                 }
 243                 if (!(ipprot->flags & INET6_PROTO_NOPOLICY) &&
 244                     !xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
 245                         goto discard;
 246
 247                 ret = ipprot->handler(skb);
 248                 if (ret > 0)
 249                         goto resubmit;
 250                 else if (ret == 0)
 251                         IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INDELIVERS);
 252         } else {
 253                 if (!raw) {
 254                         if (xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
 255                                 IP6_INC_STATS_BH(net, idev,
 256                                                  IPSTATS_MIB_INUNKNOWNPROTOS);
 257                                 icmpv6_send(skb, ICMPV6_PARAMPROB,
 258                                             ICMPV6_UNK_NEXTHDR, nhoff);
 259                         }
 260                         kfree_skb(skb);
 261                 } else {
 262                         IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INDELIVERS);
 263                         consume_skb(skb);
 264                 }
 265         }
 266         rcu_read_unlock();
 267         return 0;
 268
 269 discard:
 270         IP6_INC_STATS_BH(net, idev, IPSTATS_MIB_INDISCARDS);
 271         rcu_read_unlock();
 272         kfree_skb(skb);
 273         return 0;
 274 }
 275
 276
 277 int ip6_input(struct sk_buff *skb)
 278 {
 279         return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_IN, skb, skb->dev, NULL,
 280                        ip6_input_finish);
 281 }
 282
 283 int ip6_mc_input(struct sk_buff *skb)
 284 {
 285         const struct ipv6hdr *hdr;
 286         bool deliver;
 287
 288         IP6_UPD_PO_STATS_BH(dev_net(skb_dst(skb)->dev),
 289                          ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_INMCAST,
 290                          skb->len);
 291
 292         hdr = ipv6_hdr(skb);
 293         deliver = ipv6_chk_mcast_addr(skb->dev, &hdr->daddr, NULL);
 294
 295 #ifdef CONFIG_IPV6_MROUTE
 296         /*
 297          *      IPv6 multicast router mode is now supported ;)
 298          */
 299         if (dev_net(skb->dev)->ipv6.devconf_all->mc_forwarding &&
 300             !(ipv6_addr_type(&hdr->daddr) &
 301               (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)) &&
 302             likely(!(IP6CB(skb)->flags & IP6SKB_FORWARDED))) {
 303                 /*
 304                  * Okay, we try to forward - split and duplicate
 305                  * packets.
 306                  */
 307                 struct sk_buff *skb2;
 308                 struct inet6_skb_parm *opt = IP6CB(skb);
 309
 310                 /* Check for MLD */
 311                 if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
 312                         /* Check if this is a mld message */
 313                         u8 nexthdr = hdr->nexthdr;
 314                         __be16 frag_off;
 315                         int offset;
 316
 317                         /* Check if the value of Router Alert
 318                          * is for MLD (0x0000).
 319                          */
 320                         if (opt->ra == htons(IPV6_OPT_ROUTERALERT_MLD)) {
 321                                 deliver = false;
 322
 323                                 if (!ipv6_ext_hdr(nexthdr)) {
 324                                         /* BUG */
 325                                         goto out;
 326                                 }
 327                                 offset = ipv6_skip_exthdr(skb, sizeof(*hdr),
 328                                                           &nexthdr, &frag_off);
 329                                 if (offset < 0)
 330                                         goto out;
 331
 332                                 if (!ipv6_is_mld(skb, nexthdr, offset))
 333                                         goto out;
 334
 335                                 deliver = true;
 336                         }
 337                         /* unknown RA - process it normally */
 338                 }
 339
 340                 if (deliver)
 341                         skb2 = skb_clone(skb, GFP_ATOMIC);
 342                 else {
 343                         skb2 = skb;
 344                         skb = NULL;
 345                 }
 346
 347                 if (skb2) {
 348                         ip6_mr_input(skb2);
 349                 }
 350         }
 351 out:
 352 #endif
 353         if (likely(deliver))
 354                 ip6_input(skb);
 355         else {
 356                 /* discard */
 357                 kfree_skb(skb);
 358         }
 359
 360         return 0;
 361 }