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 #include <net/dst_metadata.h>
  49
  50 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
  51 {
  52         void (*edemux)(struct sk_buff *skb);
  53
  54         /* if ingress device is enslaved to an L3 master device pass the
  55          * skb to its handler for processing
  56          */
  57         skb = l3mdev_ip6_rcv(skb);
  58         if (!skb)
  59                 return NET_RX_SUCCESS;
  60
  61         if (net->ipv4.sysctl_ip_early_demux && !skb_dst(skb) && skb->sk == NULL) {
  62                 const struct inet6_protocol *ipprot;
  63
  64                 ipprot = rcu_dereference(inet6_protos[ipv6_hdr(skb)->nexthdr]);
  65                 if (ipprot && (edemux = READ_ONCE(ipprot->early_demux)))
  66                         edemux(skb);
  67         }
  68         if (!skb_valid_dst(skb))
  69                 ip6_route_input(skb);
  70
  71         return dst_input(skb);
  72 }
  73
  74 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, struct net_device *orig_dev)
  75 {
  76         const struct ipv6hdr *hdr;
  77         u32 pkt_len;
  78         struct inet6_dev *idev;
  79         struct net *net = dev_net(skb->dev);
  80
  81         if (skb->pkt_type == PACKET_OTHERHOST) {
  82                 kfree_skb(skb);
  83                 return NET_RX_DROP;
  84         }
  85
  86         rcu_read_lock();
  87
  88         idev = __in6_dev_get(skb->dev);
  89
  90         __IP6_UPD_PO_STATS(net, idev, IPSTATS_MIB_IN, skb->len);
  91
  92         if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL ||
  93             !idev || unlikely(idev->cnf.disable_ipv6)) {
  94                 __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDISCARDS);
  95                 goto drop;
  96         }
  97
  98         memset(IP6CB(skb), 0, sizeof(struct inet6_skb_parm));
  99
 100         /*
 101          * Store incoming device index. When the packet will
 102          * be queued, we cannot refer to skb->dev anymore.
 103          *
 104          * BTW, when we send a packet for our own local address on a
 105          * non-loopback interface (e.g. ethX), it is being delivered
 106          * via the loopback interface (lo) here; skb->dev = loopback_dev.
 107          * It, however, should be considered as if it is being
 108          * arrived via the sending interface (ethX), because of the
 109          * nature of scoping architecture. --yoshfuji
 110          */
 111         IP6CB(skb)->iif = skb_valid_dst(skb) ? ip6_dst_idev(skb_dst(skb))->dev->ifindex : dev->ifindex;
 112
 113         if (unlikely(!pskb_may_pull(skb, sizeof(*hdr))))
 114                 goto err;
 115
 116         hdr = ipv6_hdr(skb);
 117
 118         if (hdr->version != 6)
 119                 goto err;
 120
 121         __IP6_ADD_STATS(net, idev,
 122                         IPSTATS_MIB_NOECTPKTS +
 123                                 (ipv6_get_dsfield(hdr) & INET_ECN_MASK),
 124                         max_t(unsigned short, 1, skb_shinfo(skb)->gso_segs));
 125         /*
 126          * RFC4291 2.5.3
 127          * The loopback address must not be used as the source address in IPv6
 128          * packets that are sent outside of a single node. [..]
 129          * A packet received on an interface with a destination address
 130          * of loopback must be dropped.
 131          */
 132         if ((ipv6_addr_loopback(&hdr->saddr) ||
 133              ipv6_addr_loopback(&hdr->daddr)) &&
 134              !(dev->flags & IFF_LOOPBACK))
 135                 goto err;
 136
 137         /* RFC4291 Errata ID: 3480
 138          * Interface-Local scope spans only a single interface on a
 139          * node and is useful only for loopback transmission of
 140          * multicast.  Packets with interface-local scope received
 141          * from another node must be discarded.
 142          */
 143         if (!(skb->pkt_type == PACKET_LOOPBACK ||
 144               dev->flags & IFF_LOOPBACK) &&
 145             ipv6_addr_is_multicast(&hdr->daddr) &&
 146             IPV6_ADDR_MC_SCOPE(&hdr->daddr) == 1)
 147                 goto err;
 148
 149         /* If enabled, drop unicast packets that were encapsulated in link-layer
 150          * multicast or broadcast to protected against the so-called "hole-196"
 151          * attack in 802.11 wireless.
 152          */
 153         if (!ipv6_addr_is_multicast(&hdr->daddr) &&
 154             (skb->pkt_type == PACKET_BROADCAST ||
 155              skb->pkt_type == PACKET_MULTICAST) &&
 156             idev->cnf.drop_unicast_in_l2_multicast)
 157                 goto err;
 158
 159         /* RFC4291 2.7
 160          * Nodes must not originate a packet to a multicast address whose scope
 161          * field contains the reserved value 0; if such a packet is received, it
 162          * must be silently dropped.
 163          */
 164         if (ipv6_addr_is_multicast(&hdr->daddr) &&
 165             IPV6_ADDR_MC_SCOPE(&hdr->daddr) == 0)
 166                 goto err;
 167
 168         /*
 169          * RFC4291 2.7
 170          * Multicast addresses must not be used as source addresses in IPv6
 171          * packets or appear in any Routing header.
 172          */
 173         if (ipv6_addr_is_multicast(&hdr->saddr))
 174                 goto err;
 175
 176         skb->transport_header = skb->network_header + sizeof(*hdr);
 177         IP6CB(skb)->nhoff = offsetof(struct ipv6hdr, nexthdr);
 178
 179         pkt_len = ntohs(hdr->payload_len);
 180
 181         /* pkt_len may be zero if Jumbo payload option is present */
 182         if (pkt_len || hdr->nexthdr != NEXTHDR_HOP) {
 183                 if (pkt_len + sizeof(struct ipv6hdr) > skb->len) {
 184                         __IP6_INC_STATS(net,
 185                                         idev, IPSTATS_MIB_INTRUNCATEDPKTS);
 186                         goto drop;
 187                 }
 188                 if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr))) {
 189                         __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
 190                         goto drop;
 191                 }
 192                 hdr = ipv6_hdr(skb);
 193         }
 194
 195         if (hdr->nexthdr == NEXTHDR_HOP) {
 196                 if (ipv6_parse_hopopts(skb) < 0) {
 197                         __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
 198                         rcu_read_unlock();
 199                         return NET_RX_DROP;
 200                 }
 201         }
 202
 203         rcu_read_unlock();
 204
 205         /* Must drop socket now because of tproxy. */
 206         skb_orphan(skb);
 207
 208         return NF_HOOK(NFPROTO_IPV6, NF_INET_PRE_ROUTING,
 209                        net, NULL, skb, dev, NULL,
 210                        ip6_rcv_finish);
 211 err:
 212         __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
 213 drop:
 214         rcu_read_unlock();
 215         kfree_skb(skb);
 216         return NET_RX_DROP;
 217 }
 218
 219 /*
 220  *      Deliver the packet to the host
 221  */
 222
 223
 224 static int ip6_input_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
 225 {
 226         const struct inet6_protocol *ipprot;
 227         struct inet6_dev *idev;
 228         unsigned int nhoff;
 229         int nexthdr;
 230         bool raw;
 231         bool have_final = false;
 232
 233         /*
 234          *      Parse extension headers
 235          */
 236
 237         rcu_read_lock();
 238 resubmit:
 239         idev = ip6_dst_idev(skb_dst(skb));
 240         if (!pskb_pull(skb, skb_transport_offset(skb)))
 241                 goto discard;
 242         nhoff = IP6CB(skb)->nhoff;
 243         nexthdr = skb_network_header(skb)[nhoff];
 244
 245 resubmit_final:
 246         raw = raw6_local_deliver(skb, nexthdr);
 247         ipprot = rcu_dereference(inet6_protos[nexthdr]);
 248         if (ipprot) {
 249                 int ret;
 250
 251                 if (have_final) {
 252                         if (!(ipprot->flags & INET6_PROTO_FINAL)) {
 253                                 /* Once we've seen a final protocol don't
 254                                  * allow encapsulation on any non-final
 255                                  * ones. This allows foo in UDP encapsulation
 256                                  * to work.
 257                                  */
 258                                 goto discard;
 259                         }
 260                 } else if (ipprot->flags & INET6_PROTO_FINAL) {
 261                         const struct ipv6hdr *hdr;
 262
 263                         /* Only do this once for first final protocol */
 264                         have_final = true;
 265
 266                         /* Free reference early: we don't need it any more,
 267                            and it may hold ip_conntrack module loaded
 268                            indefinitely. */
 269                         nf_reset(skb);
 270
 271                         skb_postpull_rcsum(skb, skb_network_header(skb),
 272                                            skb_network_header_len(skb));
 273                         hdr = ipv6_hdr(skb);
 274                         if (ipv6_addr_is_multicast(&hdr->daddr) &&
 275                             !ipv6_chk_mcast_addr(skb->dev, &hdr->daddr,
 276                             &hdr->saddr) &&
 277                             !ipv6_is_mld(skb, nexthdr, skb_network_header_len(skb)))
 278                                 goto discard;
 279                 }
 280                 if (!(ipprot->flags & INET6_PROTO_NOPOLICY) &&
 281                     !xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
 282                         goto discard;
 283
 284                 ret = ipprot->handler(skb);
 285                 if (ret > 0) {
 286                         if (ipprot->flags & INET6_PROTO_FINAL) {
 287                                 /* Not an extension header, most likely UDP
 288                                  * encapsulation. Use return value as nexthdr
 289                                  * protocol not nhoff (which presumably is
 290                                  * not set by handler).
 291                                  */
 292                                 nexthdr = ret;
 293                                 goto resubmit_final;
 294                         } else {
 295                                 goto resubmit;
 296                         }
 297                 } else if (ret == 0) {
 298                         __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDELIVERS);
 299                 }
 300         } else {
 301                 if (!raw) {
 302                         if (xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
 303                                 __IP6_INC_STATS(net, idev,
 304                                                 IPSTATS_MIB_INUNKNOWNPROTOS);
 305                                 icmpv6_send(skb, ICMPV6_PARAMPROB,
 306                                             ICMPV6_UNK_NEXTHDR, nhoff);
 307                         }
 308                         kfree_skb(skb);
 309                 } else {
 310                         __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDELIVERS);
 311                         consume_skb(skb);
 312                 }
 313         }
 314         rcu_read_unlock();
 315         return 0;
 316
 317 discard:
 318         __IP6_INC_STATS(net, idev, IPSTATS_MIB_INDISCARDS);
 319         rcu_read_unlock();
 320         kfree_skb(skb);
 321         return 0;
 322 }
 323
 324
 325 int ip6_input(struct sk_buff *skb)
 326 {
 327         return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_IN,
 328                        dev_net(skb->dev), NULL, skb, skb->dev, NULL,
 329                        ip6_input_finish);
 330 }
 331 EXPORT_SYMBOL_GPL(ip6_input);
 332
 333 int ip6_mc_input(struct sk_buff *skb)
 334 {
 335         const struct ipv6hdr *hdr;
 336         bool deliver;
 337
 338         __IP6_UPD_PO_STATS(dev_net(skb_dst(skb)->dev),
 339                          ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_INMCAST,
 340                          skb->len);
 341
 342         hdr = ipv6_hdr(skb);
 343         deliver = ipv6_chk_mcast_addr(skb->dev, &hdr->daddr, NULL);
 344
 345 #ifdef CONFIG_IPV6_MROUTE
 346         /*
 347          *      IPv6 multicast router mode is now supported ;)
 348          */
 349         if (dev_net(skb->dev)->ipv6.devconf_all->mc_forwarding &&
 350             !(ipv6_addr_type(&hdr->daddr) &
 351               (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)) &&
 352             likely(!(IP6CB(skb)->flags & IP6SKB_FORWARDED))) {
 353                 /*
 354                  * Okay, we try to forward - split and duplicate
 355                  * packets.
 356                  */
 357                 struct sk_buff *skb2;
 358                 struct inet6_skb_parm *opt = IP6CB(skb);
 359
 360                 /* Check for MLD */
 361                 if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
 362                         /* Check if this is a mld message */
 363                         u8 nexthdr = hdr->nexthdr;
 364                         __be16 frag_off;
 365                         int offset;
 366
 367                         /* Check if the value of Router Alert
 368                          * is for MLD (0x0000).
 369                          */
 370                         if (opt->ra == htons(IPV6_OPT_ROUTERALERT_MLD)) {
 371                                 deliver = false;
 372
 373                                 if (!ipv6_ext_hdr(nexthdr)) {
 374                                         /* BUG */
 375                                         goto out;
 376                                 }
 377                                 offset = ipv6_skip_exthdr(skb, sizeof(*hdr),
 378                                                           &nexthdr, &frag_off);
 379                                 if (offset < 0)
 380                                         goto out;
 381
 382                                 if (ipv6_is_mld(skb, nexthdr, offset))
 383                                         deliver = true;
 384
 385                                 goto out;
 386                         }
 387                         /* unknown RA - process it normally */
 388                 }
 389
 390                 if (deliver)
 391                         skb2 = skb_clone(skb, GFP_ATOMIC);
 392                 else {
 393                         skb2 = skb;
 394                         skb = NULL;
 395                 }
 396
 397                 if (skb2) {
 398                         ip6_mr_input(skb2);
 399                 }
 400         }
 401 out:
 402 #endif
 403         if (likely(deliver))
 404                 ip6_input(skb);
 405         else {
 406                 /* discard */
 407                 kfree_skb(skb);
 408         }
 409
 410         return 0;
 411 }