SAFE public projects git trees. - safe/jmp/linux-2.6/blob - net/bridge/br_multicast.c

   1 /*
   2  * Bridge multicast support.
   3  *
   4  * Copyright (c) 2010 Herbert Xu <herbert@gondor.apana.org.au>
   5  *
   6  * This program is free software; you can redistribute it and/or modify it
   7  * under the terms of the GNU General Public License as published by the Free
   8  * Software Foundation; either version 2 of the License, or (at your option)
   9  * any later version.
  10  *
  11  */
  12
  13 #include <linux/err.h>
  14 #include <linux/if_ether.h>
  15 #include <linux/igmp.h>
  16 #include <linux/jhash.h>
  17 #include <linux/kernel.h>
  18 #include <linux/log2.h>
  19 #include <linux/netdevice.h>
  20 #include <linux/netfilter_bridge.h>
  21 #include <linux/random.h>
  22 #include <linux/rculist.h>
  23 #include <linux/skbuff.h>
  24 #include <linux/slab.h>
  25 #include <linux/timer.h>
  26 #include <net/ip.h>
  27 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
  28 #include <net/ipv6.h>
  29 #include <net/mld.h>
  30 #include <net/addrconf.h>
  31 #endif
  32
  33 #include "br_private.h"
  34
  35 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
  36 static inline int ipv6_is_local_multicast(const struct in6_addr *addr)
  37 {
  38         if (ipv6_addr_is_multicast(addr) &&
  39             IPV6_ADDR_MC_SCOPE(addr) <= IPV6_ADDR_SCOPE_LINKLOCAL)
  40                 return 1;
  41         return 0;
  42 }
  43 #endif
  44
  45 static inline int br_ip_equal(const struct br_ip *a, const struct br_ip *b)
  46 {
  47         if (a->proto != b->proto)
  48                 return 0;
  49         switch (a->proto) {
  50         case htons(ETH_P_IP):
  51                 return a->u.ip4 == b->u.ip4;
  52 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
  53         case htons(ETH_P_IPV6):
  54                 return ipv6_addr_equal(&a->u.ip6, &b->u.ip6);
  55 #endif
  56         }
  57         return 0;
  58 }
  59
  60 static inline int __br_ip4_hash(struct net_bridge_mdb_htable *mdb, __be32 ip)
  61 {
  62         return jhash_1word(mdb->secret, (__force u32)ip) & (mdb->max - 1);
  63 }
  64
  65 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
  66 static inline int __br_ip6_hash(struct net_bridge_mdb_htable *mdb,
  67                                 const struct in6_addr *ip)
  68 {
  69         return jhash2((__force u32 *)ip->s6_addr32, 4, mdb->secret) & (mdb->max - 1);
  70 }
  71 #endif
  72
  73 static inline int br_ip_hash(struct net_bridge_mdb_htable *mdb,
  74                              struct br_ip *ip)
  75 {
  76         switch (ip->proto) {
  77         case htons(ETH_P_IP):
  78                 return __br_ip4_hash(mdb, ip->u.ip4);
  79 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
  80         case htons(ETH_P_IPV6):
  81                 return __br_ip6_hash(mdb, &ip->u.ip6);
  82 #endif
  83         }
  84         return 0;
  85 }
  86
  87 static struct net_bridge_mdb_entry *__br_mdb_ip_get(
  88         struct net_bridge_mdb_htable *mdb, struct br_ip *dst, int hash)
  89 {
  90         struct net_bridge_mdb_entry *mp;
  91         struct hlist_node *p;
  92
  93         hlist_for_each_entry_rcu(mp, p, &mdb->mhash[hash], hlist[mdb->ver]) {
  94                 if (br_ip_equal(&mp->addr, dst))
  95                         return mp;
  96         }
  97
  98         return NULL;
  99 }
 100
 101 static struct net_bridge_mdb_entry *br_mdb_ip4_get(
 102         struct net_bridge_mdb_htable *mdb, __be32 dst)
 103 {
 104         struct br_ip br_dst;
 105
 106         br_dst.u.ip4 = dst;
 107         br_dst.proto = htons(ETH_P_IP);
 108
 109         return __br_mdb_ip_get(mdb, &br_dst, __br_ip4_hash(mdb, dst));
 110 }
 111
 112 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 113 static struct net_bridge_mdb_entry *br_mdb_ip6_get(
 114         struct net_bridge_mdb_htable *mdb, const struct in6_addr *dst)
 115 {
 116         struct br_ip br_dst;
 117
 118         ipv6_addr_copy(&br_dst.u.ip6, dst);
 119         br_dst.proto = htons(ETH_P_IPV6);
 120
 121         return __br_mdb_ip_get(mdb, &br_dst, __br_ip6_hash(mdb, dst));
 122 }
 123 #endif
 124
 125 static struct net_bridge_mdb_entry *br_mdb_ip_get(
 126         struct net_bridge_mdb_htable *mdb, struct br_ip *dst)
 127 {
 128         return __br_mdb_ip_get(mdb, dst, br_ip_hash(mdb, dst));
 129 }
 130
 131 struct net_bridge_mdb_entry *br_mdb_get(struct net_bridge *br,
 132                                         struct sk_buff *skb)
 133 {
 134         struct net_bridge_mdb_htable *mdb = br->mdb;
 135         struct br_ip ip;
 136
 137         if (!mdb || br->multicast_disabled)
 138                 return NULL;
 139
 140         if (BR_INPUT_SKB_CB(skb)->igmp)
 141                 return NULL;
 142
 143         ip.proto = skb->protocol;
 144
 145         switch (skb->protocol) {
 146         case htons(ETH_P_IP):
 147                 ip.u.ip4 = ip_hdr(skb)->daddr;
 148                 break;
 149 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 150         case htons(ETH_P_IPV6):
 151                 ipv6_addr_copy(&ip.u.ip6, &ipv6_hdr(skb)->daddr);
 152                 break;
 153 #endif
 154         default:
 155                 return NULL;
 156         }
 157
 158         return br_mdb_ip_get(mdb, &ip);
 159 }
 160
 161 static void br_mdb_free(struct rcu_head *head)
 162 {
 163         struct net_bridge_mdb_htable *mdb =
 164                 container_of(head, struct net_bridge_mdb_htable, rcu);
 165         struct net_bridge_mdb_htable *old = mdb->old;
 166
 167         mdb->old = NULL;
 168         kfree(old->mhash);
 169         kfree(old);
 170 }
 171
 172 static int br_mdb_copy(struct net_bridge_mdb_htable *new,
 173                        struct net_bridge_mdb_htable *old,
 174                        int elasticity)
 175 {
 176         struct net_bridge_mdb_entry *mp;
 177         struct hlist_node *p;
 178         int maxlen;
 179         int len;
 180         int i;
 181
 182         for (i = 0; i < old->max; i++)
 183                 hlist_for_each_entry(mp, p, &old->mhash[i], hlist[old->ver])
 184                         hlist_add_head(&mp->hlist[new->ver],
 185                                        &new->mhash[br_ip_hash(new, &mp->addr)]);
 186
 187         if (!elasticity)
 188                 return 0;
 189
 190         maxlen = 0;
 191         for (i = 0; i < new->max; i++) {
 192                 len = 0;
 193                 hlist_for_each_entry(mp, p, &new->mhash[i], hlist[new->ver])
 194                         len++;
 195                 if (len > maxlen)
 196                         maxlen = len;
 197         }
 198
 199         return maxlen > elasticity ? -EINVAL : 0;
 200 }
 201
 202 static void br_multicast_free_pg(struct rcu_head *head)
 203 {
 204         struct net_bridge_port_group *p =
 205                 container_of(head, struct net_bridge_port_group, rcu);
 206
 207         kfree(p);
 208 }
 209
 210 static void br_multicast_free_group(struct rcu_head *head)
 211 {
 212         struct net_bridge_mdb_entry *mp =
 213                 container_of(head, struct net_bridge_mdb_entry, rcu);
 214
 215         kfree(mp);
 216 }
 217
 218 static void br_multicast_group_expired(unsigned long data)
 219 {
 220         struct net_bridge_mdb_entry *mp = (void *)data;
 221         struct net_bridge *br = mp->br;
 222         struct net_bridge_mdb_htable *mdb;
 223
 224         spin_lock(&br->multicast_lock);
 225         if (!netif_running(br->dev) || timer_pending(&mp->timer))
 226                 goto out;
 227
 228         if (!hlist_unhashed(&mp->mglist))
 229                 hlist_del_init(&mp->mglist);
 230
 231         if (mp->ports)
 232                 goto out;
 233
 234         mdb = br->mdb;
 235         hlist_del_rcu(&mp->hlist[mdb->ver]);
 236         mdb->size--;
 237
 238         del_timer(&mp->query_timer);
 239         call_rcu_bh(&mp->rcu, br_multicast_free_group);
 240
 241 out:
 242         spin_unlock(&br->multicast_lock);
 243 }
 244
 245 static void br_multicast_del_pg(struct net_bridge *br,
 246                                 struct net_bridge_port_group *pg)
 247 {
 248         struct net_bridge_mdb_htable *mdb = br->mdb;
 249         struct net_bridge_mdb_entry *mp;
 250         struct net_bridge_port_group *p;
 251         struct net_bridge_port_group **pp;
 252
 253         mp = br_mdb_ip_get(mdb, &pg->addr);
 254         if (WARN_ON(!mp))
 255                 return;
 256
 257         for (pp = &mp->ports; (p = *pp); pp = &p->next) {
 258                 if (p != pg)
 259                         continue;
 260
 261                 *pp = p->next;
 262                 hlist_del_init(&p->mglist);
 263                 del_timer(&p->timer);
 264                 del_timer(&p->query_timer);
 265                 call_rcu_bh(&p->rcu, br_multicast_free_pg);
 266
 267                 if (!mp->ports && hlist_unhashed(&mp->mglist) &&
 268                     netif_running(br->dev))
 269                         mod_timer(&mp->timer, jiffies);
 270
 271                 return;
 272         }
 273
 274         WARN_ON(1);
 275 }
 276
 277 static void br_multicast_port_group_expired(unsigned long data)
 278 {
 279         struct net_bridge_port_group *pg = (void *)data;
 280         struct net_bridge *br = pg->port->br;
 281
 282         spin_lock(&br->multicast_lock);
 283         if (!netif_running(br->dev) || timer_pending(&pg->timer) ||
 284             hlist_unhashed(&pg->mglist))
 285                 goto out;
 286
 287         br_multicast_del_pg(br, pg);
 288
 289 out:
 290         spin_unlock(&br->multicast_lock);
 291 }
 292
 293 static int br_mdb_rehash(struct net_bridge_mdb_htable **mdbp, int max,
 294                          int elasticity)
 295 {
 296         struct net_bridge_mdb_htable *old = *mdbp;
 297         struct net_bridge_mdb_htable *mdb;
 298         int err;
 299
 300         mdb = kmalloc(sizeof(*mdb), GFP_ATOMIC);
 301         if (!mdb)
 302                 return -ENOMEM;
 303
 304         mdb->max = max;
 305         mdb->old = old;
 306
 307         mdb->mhash = kzalloc(max * sizeof(*mdb->mhash), GFP_ATOMIC);
 308         if (!mdb->mhash) {
 309                 kfree(mdb);
 310                 return -ENOMEM;
 311         }
 312
 313         mdb->size = old ? old->size : 0;
 314         mdb->ver = old ? old->ver ^ 1 : 0;
 315
 316         if (!old || elasticity)
 317                 get_random_bytes(&mdb->secret, sizeof(mdb->secret));
 318         else
 319                 mdb->secret = old->secret;
 320
 321         if (!old)
 322                 goto out;
 323
 324         err = br_mdb_copy(mdb, old, elasticity);
 325         if (err) {
 326                 kfree(mdb->mhash);
 327                 kfree(mdb);
 328                 return err;
 329         }
 330
 331         call_rcu_bh(&mdb->rcu, br_mdb_free);
 332
 333 out:
 334         rcu_assign_pointer(*mdbp, mdb);
 335
 336         return 0;
 337 }
 338
 339 static struct sk_buff *br_ip4_multicast_alloc_query(struct net_bridge *br,
 340                                                     __be32 group)
 341 {
 342         struct sk_buff *skb;
 343         struct igmphdr *ih;
 344         struct ethhdr *eth;
 345         struct iphdr *iph;
 346
 347         skb = netdev_alloc_skb_ip_align(br->dev, sizeof(*eth) + sizeof(*iph) +
 348                                                  sizeof(*ih) + 4);
 349         if (!skb)
 350                 goto out;
 351
 352         skb->protocol = htons(ETH_P_IP);
 353
 354         skb_reset_mac_header(skb);
 355         eth = eth_hdr(skb);
 356
 357         memcpy(eth->h_source, br->dev->dev_addr, 6);
 358         eth->h_dest[0] = 1;
 359         eth->h_dest[1] = 0;
 360         eth->h_dest[2] = 0x5e;
 361         eth->h_dest[3] = 0;
 362         eth->h_dest[4] = 0;
 363         eth->h_dest[5] = 1;
 364         eth->h_proto = htons(ETH_P_IP);
 365         skb_put(skb, sizeof(*eth));
 366
 367         skb_set_network_header(skb, skb->len);
 368         iph = ip_hdr(skb);
 369
 370         iph->version = 4;
 371         iph->ihl = 6;
 372         iph->tos = 0xc0;
 373         iph->tot_len = htons(sizeof(*iph) + sizeof(*ih) + 4);
 374         iph->id = 0;
 375         iph->frag_off = htons(IP_DF);
 376         iph->ttl = 1;
 377         iph->protocol = IPPROTO_IGMP;
 378         iph->saddr = 0;
 379         iph->daddr = htonl(INADDR_ALLHOSTS_GROUP);
 380         ((u8 *)&iph[1])[0] = IPOPT_RA;
 381         ((u8 *)&iph[1])[1] = 4;
 382         ((u8 *)&iph[1])[2] = 0;
 383         ((u8 *)&iph[1])[3] = 0;
 384         ip_send_check(iph);
 385         skb_put(skb, 24);
 386
 387         skb_set_transport_header(skb, skb->len);
 388         ih = igmp_hdr(skb);
 389         ih->type = IGMP_HOST_MEMBERSHIP_QUERY;
 390         ih->code = (group ? br->multicast_last_member_interval :
 391                             br->multicast_query_response_interval) /
 392                    (HZ / IGMP_TIMER_SCALE);
 393         ih->group = group;
 394         ih->csum = 0;
 395         ih->csum = ip_compute_csum((void *)ih, sizeof(struct igmphdr));
 396         skb_put(skb, sizeof(*ih));
 397
 398         __skb_pull(skb, sizeof(*eth));
 399
 400 out:
 401         return skb;
 402 }
 403
 404 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 405 static struct sk_buff *br_ip6_multicast_alloc_query(struct net_bridge *br,
 406                                                     struct in6_addr *group)
 407 {
 408         struct sk_buff *skb;
 409         struct ipv6hdr *ip6h;
 410         struct mld_msg *mldq;
 411         struct ethhdr *eth;
 412         u8 *hopopt;
 413         unsigned long interval;
 414
 415         skb = netdev_alloc_skb_ip_align(br->dev, sizeof(*eth) + sizeof(*ip6h) +
 416                                                  8 + sizeof(*mldq));
 417         if (!skb)
 418                 goto out;
 419
 420         skb->protocol = htons(ETH_P_IPV6);
 421
 422         /* Ethernet header */
 423         skb_reset_mac_header(skb);
 424         eth = eth_hdr(skb);
 425
 426         memcpy(eth->h_source, br->dev->dev_addr, 6);
 427         ipv6_eth_mc_map(group, eth->h_dest);
 428         eth->h_proto = htons(ETH_P_IPV6);
 429         skb_put(skb, sizeof(*eth));
 430
 431         /* IPv6 header + HbH option */
 432         skb_set_network_header(skb, skb->len);
 433         ip6h = ipv6_hdr(skb);
 434
 435         *(__force __be32 *)ip6h = htonl(0x60000000);
 436         ip6h->payload_len = 8 + sizeof(*mldq);
 437         ip6h->nexthdr = IPPROTO_HOPOPTS;
 438         ip6h->hop_limit = 1;
 439         ipv6_addr_set(&ip6h->saddr, 0, 0, 0, 0);
 440         ipv6_addr_set(&ip6h->daddr, htonl(0xff020000), 0, 0, htonl(1));
 441
 442         hopopt = (u8 *)(ip6h + 1);
 443         hopopt[0] = IPPROTO_ICMPV6;             /* next hdr */
 444         hopopt[1] = 0;                          /* length of HbH */
 445         hopopt[2] = IPV6_TLV_ROUTERALERT;       /* Router Alert */
 446         hopopt[3] = 2;                          /* Length of RA Option */
 447         hopopt[4] = 0;                          /* Type = 0x0000 (MLD) */
 448         hopopt[5] = 0;
 449         hopopt[6] = IPV6_TLV_PAD0;              /* Pad0 */
 450         hopopt[7] = IPV6_TLV_PAD0;              /* Pad0 */
 451
 452         skb_put(skb, sizeof(*ip6h) + 8);
 453
 454         /* ICMPv6 */
 455         skb_set_transport_header(skb, skb->len);
 456         mldq = (struct mld_msg *) icmp6_hdr(skb);
 457
 458         interval = ipv6_addr_any(group) ? br->multicast_last_member_interval :
 459                                           br->multicast_query_response_interval;
 460
 461         mldq->mld_type = ICMPV6_MGM_QUERY;
 462         mldq->mld_code = 0;
 463         mldq->mld_cksum = 0;
 464         mldq->mld_maxdelay = htons((u16)jiffies_to_msecs(interval));
 465         mldq->mld_reserved = 0;
 466         ipv6_addr_copy(&mldq->mld_mca, group);
 467
 468         /* checksum */
 469         mldq->mld_cksum = csum_ipv6_magic(&ip6h->saddr, &ip6h->daddr,
 470                                           sizeof(*mldq), IPPROTO_ICMPV6,
 471                                           csum_partial(mldq,
 472                                                        sizeof(*mldq), 0));
 473         skb_put(skb, sizeof(*mldq));
 474
 475         __skb_pull(skb, sizeof(*eth));
 476
 477 out:
 478         return skb;
 479 }
 480 #endif
 481
 482 static struct sk_buff *br_multicast_alloc_query(struct net_bridge *br,
 483                                                 struct br_ip *addr)
 484 {
 485         switch (addr->proto) {
 486         case htons(ETH_P_IP):
 487                 return br_ip4_multicast_alloc_query(br, addr->u.ip4);
 488 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 489         case htons(ETH_P_IPV6):
 490                 return br_ip6_multicast_alloc_query(br, &addr->u.ip6);
 491 #endif
 492         }
 493         return NULL;
 494 }
 495
 496 static void br_multicast_send_group_query(struct net_bridge_mdb_entry *mp)
 497 {
 498         struct net_bridge *br = mp->br;
 499         struct sk_buff *skb;
 500
 501         skb = br_multicast_alloc_query(br, &mp->addr);
 502         if (!skb)
 503                 goto timer;
 504
 505         netif_rx(skb);
 506
 507 timer:
 508         if (++mp->queries_sent < br->multicast_last_member_count)
 509                 mod_timer(&mp->query_timer,
 510                           jiffies + br->multicast_last_member_interval);
 511 }
 512
 513 static void br_multicast_group_query_expired(unsigned long data)
 514 {
 515         struct net_bridge_mdb_entry *mp = (void *)data;
 516         struct net_bridge *br = mp->br;
 517
 518         spin_lock(&br->multicast_lock);
 519         if (!netif_running(br->dev) || hlist_unhashed(&mp->mglist) ||
 520             mp->queries_sent >= br->multicast_last_member_count)
 521                 goto out;
 522
 523         br_multicast_send_group_query(mp);
 524
 525 out:
 526         spin_unlock(&br->multicast_lock);
 527 }
 528
 529 static void br_multicast_send_port_group_query(struct net_bridge_port_group *pg)
 530 {
 531         struct net_bridge_port *port = pg->port;
 532         struct net_bridge *br = port->br;
 533         struct sk_buff *skb;
 534
 535         skb = br_multicast_alloc_query(br, &pg->addr);
 536         if (!skb)
 537                 goto timer;
 538
 539         br_deliver(port, skb);
 540
 541 timer:
 542         if (++pg->queries_sent < br->multicast_last_member_count)
 543                 mod_timer(&pg->query_timer,
 544                           jiffies + br->multicast_last_member_interval);
 545 }
 546
 547 static void br_multicast_port_group_query_expired(unsigned long data)
 548 {
 549         struct net_bridge_port_group *pg = (void *)data;
 550         struct net_bridge_port *port = pg->port;
 551         struct net_bridge *br = port->br;
 552
 553         spin_lock(&br->multicast_lock);
 554         if (!netif_running(br->dev) || hlist_unhashed(&pg->mglist) ||
 555             pg->queries_sent >= br->multicast_last_member_count)
 556                 goto out;
 557
 558         br_multicast_send_port_group_query(pg);
 559
 560 out:
 561         spin_unlock(&br->multicast_lock);
 562 }
 563
 564 static struct net_bridge_mdb_entry *br_multicast_get_group(
 565         struct net_bridge *br, struct net_bridge_port *port,
 566         struct br_ip *group, int hash)
 567 {
 568         struct net_bridge_mdb_htable *mdb = br->mdb;
 569         struct net_bridge_mdb_entry *mp;
 570         struct hlist_node *p;
 571         unsigned count = 0;
 572         unsigned max;
 573         int elasticity;
 574         int err;
 575
 576         hlist_for_each_entry(mp, p, &mdb->mhash[hash], hlist[mdb->ver]) {
 577                 count++;
 578                 if (unlikely(br_ip_equal(group, &mp->addr)))
 579                         return mp;
 580         }
 581
 582         elasticity = 0;
 583         max = mdb->max;
 584
 585         if (unlikely(count > br->hash_elasticity && count)) {
 586                 if (net_ratelimit())
 587                         printk(KERN_INFO "%s: Multicast hash table "
 588                                "chain limit reached: %s\n",
 589                                br->dev->name, port ? port->dev->name :
 590                                                      br->dev->name);
 591
 592                 elasticity = br->hash_elasticity;
 593         }
 594
 595         if (mdb->size >= max) {
 596                 max *= 2;
 597                 if (unlikely(max >= br->hash_max)) {
 598                         printk(KERN_WARNING "%s: Multicast hash table maximum "
 599                                "reached, disabling snooping: %s, %d\n",
 600                                br->dev->name, port ? port->dev->name :
 601                                                      br->dev->name,
 602                                max);
 603                         err = -E2BIG;
 604 disable:
 605                         br->multicast_disabled = 1;
 606                         goto err;
 607                 }
 608         }
 609
 610         if (max > mdb->max || elasticity) {
 611                 if (mdb->old) {
 612                         if (net_ratelimit())
 613                                 printk(KERN_INFO "%s: Multicast hash table "
 614                                        "on fire: %s\n",
 615                                        br->dev->name, port ? port->dev->name :
 616                                                              br->dev->name);
 617                         err = -EEXIST;
 618                         goto err;
 619                 }
 620
 621                 err = br_mdb_rehash(&br->mdb, max, elasticity);
 622                 if (err) {
 623                         printk(KERN_WARNING "%s: Cannot rehash multicast "
 624                                "hash table, disabling snooping: "
 625                                "%s, %d, %d\n",
 626                                br->dev->name, port ? port->dev->name :
 627                                                      br->dev->name,
 628                                mdb->size, err);
 629                         goto disable;
 630                 }
 631
 632                 err = -EAGAIN;
 633                 goto err;
 634         }
 635
 636         return NULL;
 637
 638 err:
 639         mp = ERR_PTR(err);
 640         return mp;
 641 }
 642
 643 static struct net_bridge_mdb_entry *br_multicast_new_group(
 644         struct net_bridge *br, struct net_bridge_port *port,
 645         struct br_ip *group)
 646 {
 647         struct net_bridge_mdb_htable *mdb = br->mdb;
 648         struct net_bridge_mdb_entry *mp;
 649         int hash;
 650
 651         if (!mdb) {
 652                 if (br_mdb_rehash(&br->mdb, BR_HASH_SIZE, 0))
 653                         return NULL;
 654                 goto rehash;
 655         }
 656
 657         hash = br_ip_hash(mdb, group);
 658         mp = br_multicast_get_group(br, port, group, hash);
 659         switch (PTR_ERR(mp)) {
 660         case 0:
 661                 break;
 662
 663         case -EAGAIN:
 664 rehash:
 665                 mdb = br->mdb;
 666                 hash = br_ip_hash(mdb, group);
 667                 break;
 668
 669         default:
 670                 goto out;
 671         }
 672
 673         mp = kzalloc(sizeof(*mp), GFP_ATOMIC);
 674         if (unlikely(!mp))
 675                 goto out;
 676
 677         mp->br = br;
 678         mp->addr = *group;
 679         setup_timer(&mp->timer, br_multicast_group_expired,
 680                     (unsigned long)mp);
 681         setup_timer(&mp->query_timer, br_multicast_group_query_expired,
 682                     (unsigned long)mp);
 683
 684         hlist_add_head_rcu(&mp->hlist[mdb->ver], &mdb->mhash[hash]);
 685         mdb->size++;
 686
 687 out:
 688         return mp;
 689 }
 690
 691 static int br_multicast_add_group(struct net_bridge *br,
 692                                   struct net_bridge_port *port,
 693                                   struct br_ip *group)
 694 {
 695         struct net_bridge_mdb_entry *mp;
 696         struct net_bridge_port_group *p;
 697         struct net_bridge_port_group **pp;
 698         unsigned long now = jiffies;
 699         int err;
 700
 701         spin_lock(&br->multicast_lock);
 702         if (!netif_running(br->dev) ||
 703             (port && port->state == BR_STATE_DISABLED))
 704                 goto out;
 705
 706         mp = br_multicast_new_group(br, port, group);
 707         err = PTR_ERR(mp);
 708         if (unlikely(IS_ERR(mp) || !mp))
 709                 goto err;
 710
 711         if (!port) {
 712                 hlist_add_head(&mp->mglist, &br->mglist);
 713                 mod_timer(&mp->timer, now + br->multicast_membership_interval);
 714                 goto out;
 715         }
 716
 717         for (pp = &mp->ports; (p = *pp); pp = &p->next) {
 718                 if (p->port == port)
 719                         goto found;
 720                 if ((unsigned long)p->port < (unsigned long)port)
 721                         break;
 722         }
 723
 724         p = kzalloc(sizeof(*p), GFP_ATOMIC);
 725         err = -ENOMEM;
 726         if (unlikely(!p))
 727                 goto err;
 728
 729         p->addr = *group;
 730         p->port = port;
 731         p->next = *pp;
 732         hlist_add_head(&p->mglist, &port->mglist);
 733         setup_timer(&p->timer, br_multicast_port_group_expired,
 734                     (unsigned long)p);
 735         setup_timer(&p->query_timer, br_multicast_port_group_query_expired,
 736                     (unsigned long)p);
 737
 738         rcu_assign_pointer(*pp, p);
 739
 740 found:
 741         mod_timer(&p->timer, now + br->multicast_membership_interval);
 742 out:
 743         err = 0;
 744
 745 err:
 746         spin_unlock(&br->multicast_lock);
 747         return err;
 748 }
 749
 750 static int br_ip4_multicast_add_group(struct net_bridge *br,
 751                                       struct net_bridge_port *port,
 752                                       __be32 group)
 753 {
 754         struct br_ip br_group;
 755
 756         if (ipv4_is_local_multicast(group))
 757                 return 0;
 758
 759         br_group.u.ip4 = group;
 760         br_group.proto = htons(ETH_P_IP);
 761
 762         return br_multicast_add_group(br, port, &br_group);
 763 }
 764
 765 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 766 static int br_ip6_multicast_add_group(struct net_bridge *br,
 767                                       struct net_bridge_port *port,
 768                                       const struct in6_addr *group)
 769 {
 770         struct br_ip br_group;
 771
 772         if (ipv6_is_local_multicast(group))
 773                 return 0;
 774
 775         ipv6_addr_copy(&br_group.u.ip6, group);
 776         br_group.proto = htons(ETH_P_IP);
 777
 778         return br_multicast_add_group(br, port, &br_group);
 779 }
 780 #endif
 781
 782 static void br_multicast_router_expired(unsigned long data)
 783 {
 784         struct net_bridge_port *port = (void *)data;
 785         struct net_bridge *br = port->br;
 786
 787         spin_lock(&br->multicast_lock);
 788         if (port->multicast_router != 1 ||
 789             timer_pending(&port->multicast_router_timer) ||
 790             hlist_unhashed(&port->rlist))
 791                 goto out;
 792
 793         hlist_del_init_rcu(&port->rlist);
 794
 795 out:
 796         spin_unlock(&br->multicast_lock);
 797 }
 798
 799 static void br_multicast_local_router_expired(unsigned long data)
 800 {
 801 }
 802
 803 static void __br_multicast_send_query(struct net_bridge *br,
 804                                       struct net_bridge_port *port,
 805                                       struct br_ip *ip)
 806 {
 807         struct sk_buff *skb;
 808
 809         skb = br_multicast_alloc_query(br, ip);
 810         if (!skb)
 811                 return;
 812
 813         if (port) {
 814                 __skb_push(skb, sizeof(struct ethhdr));
 815                 skb->dev = port->dev;
 816                 NF_HOOK(PF_BRIDGE, NF_BR_LOCAL_OUT, skb, NULL, skb->dev,
 817                         dev_queue_xmit);
 818         } else
 819                 netif_rx(skb);
 820 }
 821
 822 static void br_multicast_send_query(struct net_bridge *br,
 823                                     struct net_bridge_port *port, u32 sent)
 824 {
 825         unsigned long time;
 826         struct br_ip br_group;
 827
 828         if (!netif_running(br->dev) || br->multicast_disabled ||
 829             timer_pending(&br->multicast_querier_timer))
 830                 return;
 831
 832         memset(&br_group.u, 0, sizeof(br_group.u));
 833
 834         br_group.proto = htons(ETH_P_IP);
 835         __br_multicast_send_query(br, port, &br_group);
 836
 837 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 838         br_group.proto = htons(ETH_P_IPV6);
 839         __br_multicast_send_query(br, port, &br_group);
 840 #endif
 841
 842         time = jiffies;
 843         time += sent < br->multicast_startup_query_count ?
 844                 br->multicast_startup_query_interval :
 845                 br->multicast_query_interval;
 846         mod_timer(port ? &port->multicast_query_timer :
 847                          &br->multicast_query_timer, time);
 848 }
 849
 850 static void br_multicast_port_query_expired(unsigned long data)
 851 {
 852         struct net_bridge_port *port = (void *)data;
 853         struct net_bridge *br = port->br;
 854
 855         spin_lock(&br->multicast_lock);
 856         if (port->state == BR_STATE_DISABLED ||
 857             port->state == BR_STATE_BLOCKING)
 858                 goto out;
 859
 860         if (port->multicast_startup_queries_sent <
 861             br->multicast_startup_query_count)
 862                 port->multicast_startup_queries_sent++;
 863
 864         br_multicast_send_query(port->br, port,
 865                                 port->multicast_startup_queries_sent);
 866
 867 out:
 868         spin_unlock(&br->multicast_lock);
 869 }
 870
 871 void br_multicast_add_port(struct net_bridge_port *port)
 872 {
 873         port->multicast_router = 1;
 874
 875         setup_timer(&port->multicast_router_timer, br_multicast_router_expired,
 876                     (unsigned long)port);
 877         setup_timer(&port->multicast_query_timer,
 878                     br_multicast_port_query_expired, (unsigned long)port);
 879 }
 880
 881 void br_multicast_del_port(struct net_bridge_port *port)
 882 {
 883         del_timer_sync(&port->multicast_router_timer);
 884 }
 885
 886 static void __br_multicast_enable_port(struct net_bridge_port *port)
 887 {
 888         port->multicast_startup_queries_sent = 0;
 889
 890         if (try_to_del_timer_sync(&port->multicast_query_timer) >= 0 ||
 891             del_timer(&port->multicast_query_timer))
 892                 mod_timer(&port->multicast_query_timer, jiffies);
 893 }
 894
 895 void br_multicast_enable_port(struct net_bridge_port *port)
 896 {
 897         struct net_bridge *br = port->br;
 898
 899         spin_lock(&br->multicast_lock);
 900         if (br->multicast_disabled || !netif_running(br->dev))
 901                 goto out;
 902
 903         __br_multicast_enable_port(port);
 904
 905 out:
 906         spin_unlock(&br->multicast_lock);
 907 }
 908
 909 void br_multicast_disable_port(struct net_bridge_port *port)
 910 {
 911         struct net_bridge *br = port->br;
 912         struct net_bridge_port_group *pg;
 913         struct hlist_node *p, *n;
 914
 915         spin_lock(&br->multicast_lock);
 916         hlist_for_each_entry_safe(pg, p, n, &port->mglist, mglist)
 917                 br_multicast_del_pg(br, pg);
 918
 919         if (!hlist_unhashed(&port->rlist))
 920                 hlist_del_init_rcu(&port->rlist);
 921         del_timer(&port->multicast_router_timer);
 922         del_timer(&port->multicast_query_timer);
 923         spin_unlock(&br->multicast_lock);
 924 }
 925
 926 static int br_ip4_multicast_igmp3_report(struct net_bridge *br,
 927                                          struct net_bridge_port *port,
 928                                          struct sk_buff *skb)
 929 {
 930         struct igmpv3_report *ih;
 931         struct igmpv3_grec *grec;
 932         int i;
 933         int len;
 934         int num;
 935         int type;
 936         int err = 0;
 937         __be32 group;
 938
 939         if (!pskb_may_pull(skb, sizeof(*ih)))
 940                 return -EINVAL;
 941
 942         ih = igmpv3_report_hdr(skb);
 943         num = ntohs(ih->ngrec);
 944         len = sizeof(*ih);
 945
 946         for (i = 0; i < num; i++) {
 947                 len += sizeof(*grec);
 948                 if (!pskb_may_pull(skb, len))
 949                         return -EINVAL;
 950
 951                 grec = (void *)(skb->data + len - sizeof(*grec));
 952                 group = grec->grec_mca;
 953                 type = grec->grec_type;
 954
 955                 len += ntohs(grec->grec_nsrcs) * 4;
 956                 if (!pskb_may_pull(skb, len))
 957                         return -EINVAL;
 958
 959                 /* We treat this as an IGMPv2 report for now. */
 960                 switch (type) {
 961                 case IGMPV3_MODE_IS_INCLUDE:
 962                 case IGMPV3_MODE_IS_EXCLUDE:
 963                 case IGMPV3_CHANGE_TO_INCLUDE:
 964                 case IGMPV3_CHANGE_TO_EXCLUDE:
 965                 case IGMPV3_ALLOW_NEW_SOURCES:
 966                 case IGMPV3_BLOCK_OLD_SOURCES:
 967                         break;
 968
 969                 default:
 970                         continue;
 971                 }
 972
 973                 err = br_ip4_multicast_add_group(br, port, group);
 974                 if (err)
 975                         break;
 976         }
 977
 978         return err;
 979 }
 980
 981 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 982 static int br_ip6_multicast_mld2_report(struct net_bridge *br,
 983                                         struct net_bridge_port *port,
 984                                         struct sk_buff *skb)
 985 {
 986         struct icmp6hdr *icmp6h;
 987         struct mld2_grec *grec;
 988         int i;
 989         int len;
 990         int num;
 991         int err = 0;
 992
 993         if (!pskb_may_pull(skb, sizeof(*icmp6h)))
 994                 return -EINVAL;
 995
 996         icmp6h = icmp6_hdr(skb);
 997         num = ntohs(icmp6h->icmp6_dataun.un_data16[1]);
 998         len = sizeof(*icmp6h);
 999
1000         for (i = 0; i < num; i++) {
1001                 __be16 *nsrcs, _nsrcs;
1002
1003                 nsrcs = skb_header_pointer(skb,
1004                                            len + offsetof(struct mld2_grec,
1005                                                           grec_mca),
1006                                            sizeof(_nsrcs), &_nsrcs);
1007                 if (!nsrcs)
1008                         return -EINVAL;
1009
1010                 if (!pskb_may_pull(skb,
1011                                    len + sizeof(*grec) +
1012                                    sizeof(struct in6_addr) * (*nsrcs)))
1013                         return -EINVAL;
1014
1015                 grec = (struct mld2_grec *)(skb->data + len);
1016                 len += sizeof(*grec) + sizeof(struct in6_addr) * (*nsrcs);
1017
1018                 /* We treat these as MLDv1 reports for now. */
1019                 switch (grec->grec_type) {
1020                 case MLD2_MODE_IS_INCLUDE:
1021                 case MLD2_MODE_IS_EXCLUDE:
1022                 case MLD2_CHANGE_TO_INCLUDE:
1023                 case MLD2_CHANGE_TO_EXCLUDE:
1024                 case MLD2_ALLOW_NEW_SOURCES:
1025                 case MLD2_BLOCK_OLD_SOURCES:
1026                         break;
1027
1028                 default:
1029                         continue;
1030                 }
1031
1032                 err = br_ip6_multicast_add_group(br, port, &grec->grec_mca);
1033                 if (!err)
1034                         break;
1035         }
1036
1037         return err;
1038 }
1039 #endif
1040
1041 static void br_multicast_add_router(struct net_bridge *br,
1042                                     struct net_bridge_port *port)
1043 {
1044         struct hlist_node *p;
1045         struct hlist_node **h;
1046
1047         for (h = &br->router_list.first;
1048              (p = *h) &&
1049              (unsigned long)container_of(p, struct net_bridge_port, rlist) >
1050              (unsigned long)port;
1051              h = &p->next)
1052                 ;
1053
1054         port->rlist.pprev = h;
1055         port->rlist.next = p;
1056         rcu_assign_pointer(*h, &port->rlist);
1057         if (p)
1058                 p->pprev = &port->rlist.next;
1059 }
1060
1061 static void br_multicast_mark_router(struct net_bridge *br,
1062                                      struct net_bridge_port *port)
1063 {
1064         unsigned long now = jiffies;
1065
1066         if (!port) {
1067                 if (br->multicast_router == 1)
1068                         mod_timer(&br->multicast_router_timer,
1069                                   now + br->multicast_querier_interval);
1070                 return;
1071         }
1072
1073         if (port->multicast_router != 1)
1074                 return;
1075
1076         if (!hlist_unhashed(&port->rlist))
1077                 goto timer;
1078
1079         br_multicast_add_router(br, port);
1080
1081 timer:
1082         mod_timer(&port->multicast_router_timer,
1083                   now + br->multicast_querier_interval);
1084 }
1085
1086 static void br_multicast_query_received(struct net_bridge *br,
1087                                         struct net_bridge_port *port,
1088                                         int saddr)
1089 {
1090         if (saddr)
1091                 mod_timer(&br->multicast_querier_timer,
1092                           jiffies + br->multicast_querier_interval);
1093         else if (timer_pending(&br->multicast_querier_timer))
1094                 return;
1095
1096         br_multicast_mark_router(br, port);
1097 }
1098
1099 static int br_ip4_multicast_query(struct net_bridge *br,
1100                                   struct net_bridge_port *port,
1101                                   struct sk_buff *skb)
1102 {
1103         struct iphdr *iph = ip_hdr(skb);
1104         struct igmphdr *ih = igmp_hdr(skb);
1105         struct net_bridge_mdb_entry *mp;
1106         struct igmpv3_query *ih3;
1107         struct net_bridge_port_group *p;
1108         struct net_bridge_port_group **pp;
1109         unsigned long max_delay;
1110         unsigned long now = jiffies;
1111         __be32 group;
1112         int err = 0;
1113
1114         spin_lock(&br->multicast_lock);
1115         if (!netif_running(br->dev) ||
1116             (port && port->state == BR_STATE_DISABLED))
1117                 goto out;
1118
1119         br_multicast_query_received(br, port, !!iph->saddr);
1120
1121         group = ih->group;
1122
1123         if (skb->len == sizeof(*ih)) {
1124                 max_delay = ih->code * (HZ / IGMP_TIMER_SCALE);
1125
1126                 if (!max_delay) {
1127                         max_delay = 10 * HZ;
1128                         group = 0;
1129                 }
1130         } else {
1131                 if (!pskb_may_pull(skb, sizeof(struct igmpv3_query))) {
1132                         err = -EINVAL;
1133                         goto out;
1134                 }
1135
1136                 ih3 = igmpv3_query_hdr(skb);
1137                 if (ih3->nsrcs)
1138                         goto out;
1139
1140                 max_delay = ih3->code ?
1141                             IGMPV3_MRC(ih3->code) * (HZ / IGMP_TIMER_SCALE) : 1;
1142         }
1143
1144         if (!group)
1145                 goto out;
1146
1147         mp = br_mdb_ip4_get(br->mdb, group);
1148         if (!mp)
1149                 goto out;
1150
1151         max_delay *= br->multicast_last_member_count;
1152
1153         if (!hlist_unhashed(&mp->mglist) &&
1154             (timer_pending(&mp->timer) ?
1155              time_after(mp->timer.expires, now + max_delay) :
1156              try_to_del_timer_sync(&mp->timer) >= 0))
1157                 mod_timer(&mp->timer, now + max_delay);
1158
1159         for (pp = &mp->ports; (p = *pp); pp = &p->next) {
1160                 if (timer_pending(&p->timer) ?
1161                     time_after(p->timer.expires, now + max_delay) :
1162                     try_to_del_timer_sync(&p->timer) >= 0)
1163                         mod_timer(&mp->timer, now + max_delay);
1164         }
1165
1166 out:
1167         spin_unlock(&br->multicast_lock);
1168         return err;
1169 }
1170
1171 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
1172 static int br_ip6_multicast_query(struct net_bridge *br,
1173                                   struct net_bridge_port *port,
1174                                   struct sk_buff *skb)
1175 {
1176         struct ipv6hdr *ip6h = ipv6_hdr(skb);
1177         struct mld_msg *mld = (struct mld_msg *) icmp6_hdr(skb);
1178         struct net_bridge_mdb_entry *mp;
1179         struct mld2_query *mld2q;
1180         struct net_bridge_port_group *p, **pp;
1181         unsigned long max_delay;
1182         unsigned long now = jiffies;
1183         struct in6_addr *group = NULL;
1184         int err = 0;
1185
1186         spin_lock(&br->multicast_lock);
1187         if (!netif_running(br->dev) ||
1188             (port && port->state == BR_STATE_DISABLED))
1189                 goto out;
1190
1191         br_multicast_query_received(br, port, !ipv6_addr_any(&ip6h->saddr));
1192
1193         if (skb->len == sizeof(*mld)) {
1194                 if (!pskb_may_pull(skb, sizeof(*mld))) {
1195                         err = -EINVAL;
1196                         goto out;
1197                 }
1198                 mld = (struct mld_msg *) icmp6_hdr(skb);
1199                 max_delay = msecs_to_jiffies(htons(mld->mld_maxdelay));
1200                 if (max_delay)
1201                         group = &mld->mld_mca;
1202         } else if (skb->len >= sizeof(*mld2q)) {
1203                 if (!pskb_may_pull(skb, sizeof(*mld2q))) {
1204                         err = -EINVAL;
1205                         goto out;
1206                 }
1207                 mld2q = (struct mld2_query *)icmp6_hdr(skb);
1208                 if (!mld2q->mld2q_nsrcs)
1209                         group = &mld2q->mld2q_mca;
1210                 max_delay = mld2q->mld2q_mrc ? MLDV2_MRC(mld2q->mld2q_mrc) : 1;
1211         }
1212
1213         if (!group)
1214                 goto out;
1215
1216         mp = br_mdb_ip6_get(br->mdb, group);
1217         if (!mp)
1218                 goto out;
1219
1220         max_delay *= br->multicast_last_member_count;
1221         if (!hlist_unhashed(&mp->mglist) &&
1222             (timer_pending(&mp->timer) ?
1223              time_after(mp->timer.expires, now + max_delay) :
1224              try_to_del_timer_sync(&mp->timer) >= 0))
1225                 mod_timer(&mp->timer, now + max_delay);
1226
1227         for (pp = &mp->ports; (p = *pp); pp = &p->next) {
1228                 if (timer_pending(&p->timer) ?
1229                     time_after(p->timer.expires, now + max_delay) :
1230                     try_to_del_timer_sync(&p->timer) >= 0)
1231                         mod_timer(&mp->timer, now + max_delay);
1232         }
1233
1234 out:
1235         spin_unlock(&br->multicast_lock);
1236         return err;
1237 }
1238 #endif
1239
1240 static void br_multicast_leave_group(struct net_bridge *br,
1241                                      struct net_bridge_port *port,
1242                                      struct br_ip *group)
1243 {
1244         struct net_bridge_mdb_htable *mdb;
1245         struct net_bridge_mdb_entry *mp;
1246         struct net_bridge_port_group *p;
1247         unsigned long now;
1248         unsigned long time;
1249
1250         spin_lock(&br->multicast_lock);
1251         if (!netif_running(br->dev) ||
1252             (port && port->state == BR_STATE_DISABLED) ||
1253             timer_pending(&br->multicast_querier_timer))
1254                 goto out;
1255
1256         mdb = br->mdb;
1257         mp = br_mdb_ip_get(mdb, group);
1258         if (!mp)
1259                 goto out;
1260
1261         now = jiffies;
1262         time = now + br->multicast_last_member_count *
1263                      br->multicast_last_member_interval;
1264
1265         if (!port) {
1266                 if (!hlist_unhashed(&mp->mglist) &&
1267                     (timer_pending(&mp->timer) ?
1268                      time_after(mp->timer.expires, time) :
1269                      try_to_del_timer_sync(&mp->timer) >= 0)) {
1270                         mod_timer(&mp->timer, time);
1271
1272                         mp->queries_sent = 0;
1273                         mod_timer(&mp->query_timer, now);
1274                 }
1275
1276                 goto out;
1277         }
1278
1279         for (p = mp->ports; p; p = p->next) {
1280                 if (p->port != port)
1281                         continue;
1282
1283                 if (!hlist_unhashed(&p->mglist) &&
1284                     (timer_pending(&p->timer) ?
1285                      time_after(p->timer.expires, time) :
1286                      try_to_del_timer_sync(&p->timer) >= 0)) {
1287                         mod_timer(&p->timer, time);
1288
1289                         p->queries_sent = 0;
1290                         mod_timer(&p->query_timer, now);
1291                 }
1292
1293                 break;
1294         }
1295
1296 out:
1297         spin_unlock(&br->multicast_lock);
1298 }
1299
1300 static void br_ip4_multicast_leave_group(struct net_bridge *br,
1301                                          struct net_bridge_port *port,
1302                                          __be32 group)
1303 {
1304         struct br_ip br_group;
1305
1306         if (ipv4_is_local_multicast(group))
1307                 return;
1308
1309         br_group.u.ip4 = group;
1310         br_group.proto = htons(ETH_P_IP);
1311
1312         br_multicast_leave_group(br, port, &br_group);
1313 }
1314
1315 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
1316 static void br_ip6_multicast_leave_group(struct net_bridge *br,
1317                                          struct net_bridge_port *port,
1318                                          const struct in6_addr *group)
1319 {
1320         struct br_ip br_group;
1321
1322         if (ipv6_is_local_multicast(group))
1323                 return;
1324
1325         ipv6_addr_copy(&br_group.u.ip6, group);
1326         br_group.proto = htons(ETH_P_IPV6);
1327
1328         br_multicast_leave_group(br, port, &br_group);
1329 }
1330 #endif
1331
1332 static int br_multicast_ipv4_rcv(struct net_bridge *br,
1333                                  struct net_bridge_port *port,
1334                                  struct sk_buff *skb)
1335 {
1336         struct sk_buff *skb2 = skb;
1337         struct iphdr *iph;
1338         struct igmphdr *ih;
1339         unsigned len;
1340         unsigned offset;
1341         int err;
1342
1343         /* We treat OOM as packet loss for now. */
1344         if (!pskb_may_pull(skb, sizeof(*iph)))
1345                 return -EINVAL;
1346
1347         iph = ip_hdr(skb);
1348
1349         if (iph->ihl < 5 || iph->version != 4)
1350                 return -EINVAL;
1351
1352         if (!pskb_may_pull(skb, ip_hdrlen(skb)))
1353                 return -EINVAL;
1354
1355         iph = ip_hdr(skb);
1356
1357         if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl)))
1358                 return -EINVAL;
1359
1360         if (iph->protocol != IPPROTO_IGMP)
1361                 return 0;
1362
1363         len = ntohs(iph->tot_len);
1364         if (skb->len < len || len < ip_hdrlen(skb))
1365                 return -EINVAL;
1366
1367         if (skb->len > len) {
1368                 skb2 = skb_clone(skb, GFP_ATOMIC);
1369                 if (!skb2)
1370                         return -ENOMEM;
1371
1372                 err = pskb_trim_rcsum(skb2, len);
1373                 if (err)
1374                         goto err_out;
1375         }
1376
1377         len -= ip_hdrlen(skb2);
1378         offset = skb_network_offset(skb2) + ip_hdrlen(skb2);
1379         __skb_pull(skb2, offset);
1380         skb_reset_transport_header(skb2);
1381
1382         err = -EINVAL;
1383         if (!pskb_may_pull(skb2, sizeof(*ih)))
1384                 goto out;
1385
1386         switch (skb2->ip_summed) {
1387         case CHECKSUM_COMPLETE:
1388                 if (!csum_fold(skb2->csum))
1389                         break;
1390                 /* fall through */
1391         case CHECKSUM_NONE:
1392                 skb2->csum = 0;
1393                 if (skb_checksum_complete(skb2))
1394                         goto out;
1395         }
1396
1397         err = 0;
1398
1399         BR_INPUT_SKB_CB(skb)->igmp = 1;
1400         ih = igmp_hdr(skb2);
1401
1402         switch (ih->type) {
1403         case IGMP_HOST_MEMBERSHIP_REPORT:
1404         case IGMPV2_HOST_MEMBERSHIP_REPORT:
1405                 BR_INPUT_SKB_CB(skb2)->mrouters_only = 1;
1406                 err = br_ip4_multicast_add_group(br, port, ih->group);
1407                 break;
1408         case IGMPV3_HOST_MEMBERSHIP_REPORT:
1409                 err = br_ip4_multicast_igmp3_report(br, port, skb2);
1410                 break;
1411         case IGMP_HOST_MEMBERSHIP_QUERY:
1412                 err = br_ip4_multicast_query(br, port, skb2);
1413                 break;
1414         case IGMP_HOST_LEAVE_MESSAGE:
1415                 br_ip4_multicast_leave_group(br, port, ih->group);
1416                 break;
1417         }
1418
1419 out:
1420         __skb_push(skb2, offset);
1421 err_out:
1422         if (skb2 != skb)
1423                 kfree_skb(skb2);
1424         return err;
1425 }
1426
1427 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
1428 static int br_multicast_ipv6_rcv(struct net_bridge *br,
1429                                  struct net_bridge_port *port,
1430                                  struct sk_buff *skb)
1431 {
1432         struct sk_buff *skb2 = skb;
1433         struct ipv6hdr *ip6h;
1434         struct icmp6hdr *icmp6h;
1435         u8 nexthdr;
1436         unsigned len;
1437         unsigned offset;
1438         int err;
1439
1440         if (!pskb_may_pull(skb, sizeof(*ip6h)))
1441                 return -EINVAL;
1442
1443         ip6h = ipv6_hdr(skb);
1444
1445         /*
1446          * We're interested in MLD messages only.
1447          *  - Version is 6
1448          *  - MLD has always Router Alert hop-by-hop option
1449          *  - But we do not support jumbrograms.
1450          */
1451         if (ip6h->version != 6 ||
1452             ip6h->nexthdr != IPPROTO_HOPOPTS ||
1453             ip6h->payload_len == 0)
1454                 return 0;
1455
1456         len = ntohs(ip6h->payload_len);
1457         if (skb->len < len)
1458                 return -EINVAL;
1459
1460         nexthdr = ip6h->nexthdr;
1461         offset = ipv6_skip_exthdr(skb, sizeof(*ip6h), &nexthdr);
1462
1463         if (offset < 0 || nexthdr != IPPROTO_ICMPV6)
1464                 return 0;
1465
1466         /* Okay, we found ICMPv6 header */
1467         skb2 = skb_clone(skb, GFP_ATOMIC);
1468         if (!skb2)
1469                 return -ENOMEM;
1470
1471         len -= offset - skb_network_offset(skb2);
1472
1473         __skb_pull(skb2, offset);
1474         skb_reset_transport_header(skb2);
1475
1476         err = -EINVAL;
1477         if (!pskb_may_pull(skb2, sizeof(*icmp6h)))
1478                 goto out;
1479
1480         icmp6h = icmp6_hdr(skb2);
1481
1482         switch (icmp6h->icmp6_type) {
1483         case ICMPV6_MGM_QUERY:
1484         case ICMPV6_MGM_REPORT:
1485         case ICMPV6_MGM_REDUCTION:
1486         case ICMPV6_MLD2_REPORT:
1487                 break;
1488         default:
1489                 err = 0;
1490                 goto out;
1491         }
1492
1493         /* Okay, we found MLD message. Check further. */
1494         if (skb2->len > len) {
1495                 err = pskb_trim_rcsum(skb2, len);
1496                 if (err)
1497                         goto out;
1498         }
1499
1500         switch (skb2->ip_summed) {
1501         case CHECKSUM_COMPLETE:
1502                 if (!csum_fold(skb2->csum))
1503                         break;
1504                 /*FALLTHROUGH*/
1505         case CHECKSUM_NONE:
1506                 skb2->csum = 0;
1507                 if (skb_checksum_complete(skb2))
1508                         goto out;
1509         }
1510
1511         err = 0;
1512
1513         BR_INPUT_SKB_CB(skb)->igmp = 1;
1514
1515         switch (icmp6h->icmp6_type) {
1516         case ICMPV6_MGM_REPORT:
1517             {
1518                 struct mld_msg *mld = (struct mld_msg *)icmp6h;
1519                 BR_INPUT_SKB_CB(skb2)->mrouters_only = 1;
1520                 err = br_ip6_multicast_add_group(br, port, &mld->mld_mca);
1521                 break;
1522             }
1523         case ICMPV6_MLD2_REPORT:
1524                 err = br_ip6_multicast_mld2_report(br, port, skb2);
1525                 break;
1526         case ICMPV6_MGM_QUERY:
1527                 err = br_ip6_multicast_query(br, port, skb2);
1528                 break;
1529         case ICMPV6_MGM_REDUCTION:
1530             {
1531                 struct mld_msg *mld = (struct mld_msg *)icmp6h;
1532                 br_ip6_multicast_leave_group(br, port, &mld->mld_mca);
1533             }
1534         }
1535
1536 out:
1537         __skb_push(skb2, offset);
1538         if (skb2 != skb)
1539                 kfree_skb(skb2);
1540         return err;
1541 }
1542 #endif
1543
1544 int br_multicast_rcv(struct net_bridge *br, struct net_bridge_port *port,
1545                      struct sk_buff *skb)
1546 {
1547         BR_INPUT_SKB_CB(skb)->igmp = 0;
1548         BR_INPUT_SKB_CB(skb)->mrouters_only = 0;
1549
1550         if (br->multicast_disabled)
1551                 return 0;
1552
1553         switch (skb->protocol) {
1554         case htons(ETH_P_IP):
1555                 return br_multicast_ipv4_rcv(br, port, skb);
1556 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
1557         case htons(ETH_P_IPV6):
1558                 return br_multicast_ipv6_rcv(br, port, skb);
1559 #endif
1560         }
1561
1562         return 0;
1563 }
1564
1565 static void br_multicast_query_expired(unsigned long data)
1566 {
1567         struct net_bridge *br = (void *)data;
1568
1569         spin_lock(&br->multicast_lock);
1570         if (br->multicast_startup_queries_sent <
1571             br->multicast_startup_query_count)
1572                 br->multicast_startup_queries_sent++;
1573
1574         br_multicast_send_query(br, NULL, br->multicast_startup_queries_sent);
1575
1576         spin_unlock(&br->multicast_lock);
1577 }
1578
1579 void br_multicast_init(struct net_bridge *br)
1580 {
1581         br->hash_elasticity = 4;
1582         br->hash_max = 512;
1583
1584         br->multicast_router = 1;
1585         br->multicast_last_member_count = 2;
1586         br->multicast_startup_query_count = 2;
1587
1588         br->multicast_last_member_interval = HZ;
1589         br->multicast_query_response_interval = 10 * HZ;
1590         br->multicast_startup_query_interval = 125 * HZ / 4;
1591         br->multicast_query_interval = 125 * HZ;
1592         br->multicast_querier_interval = 255 * HZ;
1593         br->multicast_membership_interval = 260 * HZ;
1594
1595         spin_lock_init(&br->multicast_lock);
1596         setup_timer(&br->multicast_router_timer,
1597                     br_multicast_local_router_expired, 0);
1598         setup_timer(&br->multicast_querier_timer,
1599                     br_multicast_local_router_expired, 0);
1600         setup_timer(&br->multicast_query_timer, br_multicast_query_expired,
1601                     (unsigned long)br);
1602 }
1603
1604 void br_multicast_open(struct net_bridge *br)
1605 {
1606         br->multicast_startup_queries_sent = 0;
1607
1608         if (br->multicast_disabled)
1609                 return;
1610
1611         mod_timer(&br->multicast_query_timer, jiffies);
1612 }
1613
1614 void br_multicast_stop(struct net_bridge *br)
1615 {
1616         struct net_bridge_mdb_htable *mdb;
1617         struct net_bridge_mdb_entry *mp;
1618         struct hlist_node *p, *n;
1619         u32 ver;
1620         int i;
1621
1622         del_timer_sync(&br->multicast_router_timer);
1623         del_timer_sync(&br->multicast_querier_timer);
1624         del_timer_sync(&br->multicast_query_timer);
1625
1626         spin_lock_bh(&br->multicast_lock);
1627         mdb = br->mdb;
1628         if (!mdb)
1629                 goto out;
1630
1631         br->mdb = NULL;
1632
1633         ver = mdb->ver;
1634         for (i = 0; i < mdb->max; i++) {
1635                 hlist_for_each_entry_safe(mp, p, n, &mdb->mhash[i],
1636                                           hlist[ver]) {
1637                         del_timer(&mp->timer);
1638                         del_timer(&mp->query_timer);
1639                         call_rcu_bh(&mp->rcu, br_multicast_free_group);
1640                 }
1641         }
1642
1643         if (mdb->old) {
1644                 spin_unlock_bh(&br->multicast_lock);
1645                 rcu_barrier_bh();
1646                 spin_lock_bh(&br->multicast_lock);
1647                 WARN_ON(mdb->old);
1648         }
1649
1650         mdb->old = mdb;
1651         call_rcu_bh(&mdb->rcu, br_mdb_free);
1652
1653 out:
1654         spin_unlock_bh(&br->multicast_lock);
1655 }
1656
1657 int br_multicast_set_router(struct net_bridge *br, unsigned long val)
1658 {
1659         int err = -ENOENT;
1660
1661         spin_lock_bh(&br->multicast_lock);
1662         if (!netif_running(br->dev))
1663                 goto unlock;
1664
1665         switch (val) {
1666         case 0:
1667         case 2:
1668                 del_timer(&br->multicast_router_timer);
1669                 /* fall through */
1670         case 1:
1671                 br->multicast_router = val;
1672                 err = 0;
1673                 break;
1674
1675         default:
1676                 err = -EINVAL;
1677                 break;
1678         }
1679
1680 unlock:
1681         spin_unlock_bh(&br->multicast_lock);
1682
1683         return err;
1684 }
1685
1686 int br_multicast_set_port_router(struct net_bridge_port *p, unsigned long val)
1687 {
1688         struct net_bridge *br = p->br;
1689         int err = -ENOENT;
1690
1691         spin_lock(&br->multicast_lock);
1692         if (!netif_running(br->dev) || p->state == BR_STATE_DISABLED)
1693                 goto unlock;
1694
1695         switch (val) {
1696         case 0:
1697         case 1:
1698         case 2:
1699                 p->multicast_router = val;
1700                 err = 0;
1701
1702                 if (val < 2 && !hlist_unhashed(&p->rlist))
1703                         hlist_del_init_rcu(&p->rlist);
1704
1705                 if (val == 1)
1706                         break;
1707
1708                 del_timer(&p->multicast_router_timer);
1709
1710                 if (val == 0)
1711                         break;
1712
1713                 br_multicast_add_router(br, p);
1714                 break;
1715
1716         default:
1717                 err = -EINVAL;
1718                 break;
1719         }
1720
1721 unlock:
1722         spin_unlock(&br->multicast_lock);
1723
1724         return err;
1725 }
1726
1727 int br_multicast_toggle(struct net_bridge *br, unsigned long val)
1728 {
1729         struct net_bridge_port *port;
1730         int err = -ENOENT;
1731
1732         spin_lock(&br->multicast_lock);
1733         if (!netif_running(br->dev))
1734                 goto unlock;
1735
1736         err = 0;
1737         if (br->multicast_disabled == !val)
1738                 goto unlock;
1739
1740         br->multicast_disabled = !val;
1741         if (br->multicast_disabled)
1742                 goto unlock;
1743
1744         if (br->mdb) {
1745                 if (br->mdb->old) {
1746                         err = -EEXIST;
1747 rollback:
1748                         br->multicast_disabled = !!val;
1749                         goto unlock;
1750                 }
1751
1752                 err = br_mdb_rehash(&br->mdb, br->mdb->max,
1753                                     br->hash_elasticity);
1754                 if (err)
1755                         goto rollback;
1756         }
1757
1758         br_multicast_open(br);
1759         list_for_each_entry(port, &br->port_list, list) {
1760                 if (port->state == BR_STATE_DISABLED ||
1761                     port->state == BR_STATE_BLOCKING)
1762                         continue;
1763
1764                 __br_multicast_enable_port(port);
1765         }
1766
1767 unlock:
1768         spin_unlock(&br->multicast_lock);
1769
1770         return err;
1771 }
1772
1773 int br_multicast_set_hash_max(struct net_bridge *br, unsigned long val)
1774 {
1775         int err = -ENOENT;
1776         u32 old;
1777
1778         spin_lock(&br->multicast_lock);
1779         if (!netif_running(br->dev))
1780                 goto unlock;
1781
1782         err = -EINVAL;
1783         if (!is_power_of_2(val))
1784                 goto unlock;
1785         if (br->mdb && val < br->mdb->size)
1786                 goto unlock;
1787
1788         err = 0;
1789
1790         old = br->hash_max;
1791         br->hash_max = val;
1792
1793         if (br->mdb) {
1794                 if (br->mdb->old) {
1795                         err = -EEXIST;
1796 rollback:
1797                         br->hash_max = old;
1798                         goto unlock;
1799                 }
1800
1801                 err = br_mdb_rehash(&br->mdb, br->hash_max,
1802                                     br->hash_elasticity);
1803                 if (err)
1804                         goto rollback;
1805         }
1806
1807 unlock:
1808         spin_unlock(&br->multicast_lock);
1809
1810         return err;
1811 }