bridge: Add core IGMP snooping support
authorHerbert Xu <herbert@gondor.apana.org.au>
Sat, 27 Feb 2010 19:41:45 +0000 (19:41 +0000)
committerDavid S. Miller <davem@davemloft.net>
Sun, 28 Feb 2010 08:48:45 +0000 (00:48 -0800)
This patch adds the core functionality of IGMP snooping support
without actually hooking it up.  So this patch should be a no-op
as far as the bridge's external behaviour is concerned.

All the new code and data is controlled by the Kconfig option
BRIDGE_IGMP_SNOOPING.  A run-time toggle is also available.

The multicast switching is done using an hash table that is
lockless on the read-side through RCU.  On the write-side the
new multicast_lock is used for all operations.  The hash table
supports dynamic growth/rehashing.

The hash table will be rehashed if any chain length exceeds a
preset limit.  If rehashing does not reduce the maximum chain
length then snooping will be disabled.

These features may be added in future (in no particular order):

* IGMPv3 source support
* Non-querier router detection
* IPv6

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/bridge/Kconfig
net/bridge/Makefile
net/bridge/br_multicast.c [new file with mode: 0644]
net/bridge/br_private.h

index e143ca678881748ab645fff4ae2e9f55ed2317f5..78dd5497210a543cbe4836222530c66036dbf325 100644 (file)
@@ -31,3 +31,15 @@ config BRIDGE
          will be called bridge.
 
          If unsure, say N.
+
+config BRIDGE_IGMP_SNOOPING
+       bool "IGMP snooping"
+       default y
+       ---help---
+         If you say Y here, then the Ethernet bridge will be able selectively
+         forward multicast traffic based on IGMP traffic received from each
+         port.
+
+         Say N to exclude this support and reduce the binary size.
+
+         If unsure, say Y.
index f444c12cde5a66d0cf864ef2238de8307d8b2707..d0359ea8ee7901472adee37ff9f22e62d51ce585 100644 (file)
@@ -12,4 +12,6 @@ bridge-$(CONFIG_SYSFS) += br_sysfs_if.o br_sysfs_br.o
 
 bridge-$(CONFIG_BRIDGE_NETFILTER) += br_netfilter.o
 
+bridge-$(CONFIG_BRIDGE_IGMP_SNOOPING) += br_multicast.o
+
 obj-$(CONFIG_BRIDGE_NF_EBTABLES) += netfilter/
diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c
new file mode 100644 (file)
index 0000000..746b5a6
--- /dev/null
@@ -0,0 +1,1135 @@
+/*
+ * Bridge multicast support.
+ *
+ * Copyright (c) 2010 Herbert Xu <herbert@gondor.apana.org.au>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation; either version 2 of the License, or (at your option)
+ * any later version.
+ *
+ */
+
+#include <linux/err.h>
+#include <linux/if_ether.h>
+#include <linux/igmp.h>
+#include <linux/jhash.h>
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/netfilter_bridge.h>
+#include <linux/random.h>
+#include <linux/rculist.h>
+#include <linux/skbuff.h>
+#include <linux/slab.h>
+#include <linux/timer.h>
+#include <net/ip.h>
+
+#include "br_private.h"
+
+static inline int br_ip_hash(struct net_bridge_mdb_htable *mdb, __be32 ip)
+{
+       return jhash_1word(mdb->secret, (u32)ip) & (mdb->max - 1);
+}
+
+static struct net_bridge_mdb_entry *__br_mdb_ip_get(
+       struct net_bridge_mdb_htable *mdb, __be32 dst, int hash)
+{
+       struct net_bridge_mdb_entry *mp;
+       struct hlist_node *p;
+
+       hlist_for_each_entry(mp, p, &mdb->mhash[hash], hlist[mdb->ver]) {
+               if (dst == mp->addr)
+                       return mp;
+       }
+
+       return NULL;
+}
+
+static struct net_bridge_mdb_entry *br_mdb_ip_get(
+       struct net_bridge_mdb_htable *mdb, __be32 dst)
+{
+       return __br_mdb_ip_get(mdb, dst, br_ip_hash(mdb, dst));
+}
+
+struct net_bridge_mdb_entry *br_mdb_get(struct net_bridge *br,
+                                       struct sk_buff *skb)
+{
+       struct net_bridge_mdb_htable *mdb = br->mdb;
+
+       if (!mdb || br->multicast_disabled)
+               return NULL;
+
+       switch (skb->protocol) {
+       case htons(ETH_P_IP):
+               if (BR_INPUT_SKB_CB(skb)->igmp)
+                       break;
+               return br_mdb_ip_get(mdb, ip_hdr(skb)->daddr);
+       }
+
+       return NULL;
+}
+
+static void br_mdb_free(struct rcu_head *head)
+{
+       struct net_bridge_mdb_htable *mdb =
+               container_of(head, struct net_bridge_mdb_htable, rcu);
+       struct net_bridge_mdb_htable *old = mdb->old;
+
+       mdb->old = NULL;
+       kfree(old->mhash);
+       kfree(old);
+}
+
+static int br_mdb_copy(struct net_bridge_mdb_htable *new,
+                      struct net_bridge_mdb_htable *old,
+                      int elasticity)
+{
+       struct net_bridge_mdb_entry *mp;
+       struct hlist_node *p;
+       int maxlen;
+       int len;
+       int i;
+
+       for (i = 0; i < old->max; i++)
+               hlist_for_each_entry(mp, p, &old->mhash[i], hlist[old->ver])
+                       hlist_add_head(&mp->hlist[new->ver],
+                                      &new->mhash[br_ip_hash(new, mp->addr)]);
+
+       if (!elasticity)
+               return 0;
+
+       maxlen = 0;
+       for (i = 0; i < new->max; i++) {
+               len = 0;
+               hlist_for_each_entry(mp, p, &new->mhash[i], hlist[new->ver])
+                       len++;
+               if (len > maxlen)
+                       maxlen = len;
+       }
+
+       return maxlen > elasticity ? -EINVAL : 0;
+}
+
+static void br_multicast_free_pg(struct rcu_head *head)
+{
+       struct net_bridge_port_group *p =
+               container_of(head, struct net_bridge_port_group, rcu);
+
+       kfree(p);
+}
+
+static void br_multicast_free_group(struct rcu_head *head)
+{
+       struct net_bridge_mdb_entry *mp =
+               container_of(head, struct net_bridge_mdb_entry, rcu);
+
+       kfree(mp);
+}
+
+static void br_multicast_group_expired(unsigned long data)
+{
+       struct net_bridge_mdb_entry *mp = (void *)data;
+       struct net_bridge *br = mp->br;
+       struct net_bridge_mdb_htable *mdb;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) || timer_pending(&mp->timer))
+               goto out;
+
+       if (!hlist_unhashed(&mp->mglist))
+               hlist_del_init(&mp->mglist);
+
+       if (mp->ports)
+               goto out;
+
+       mdb = br->mdb;
+       hlist_del_rcu(&mp->hlist[mdb->ver]);
+       mdb->size--;
+
+       del_timer(&mp->query_timer);
+       call_rcu_bh(&mp->rcu, br_multicast_free_group);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+static void br_multicast_del_pg(struct net_bridge *br,
+                               struct net_bridge_port_group *pg)
+{
+       struct net_bridge_mdb_htable *mdb = br->mdb;
+       struct net_bridge_mdb_entry *mp;
+       struct net_bridge_port_group *p;
+       struct net_bridge_port_group **pp;
+
+       mp = br_mdb_ip_get(mdb, pg->addr);
+       if (WARN_ON(!mp))
+               return;
+
+       for (pp = &mp->ports; (p = *pp); pp = &p->next) {
+               if (p != pg)
+                       continue;
+
+               *pp = p->next;
+               hlist_del_init(&p->mglist);
+               del_timer(&p->timer);
+               del_timer(&p->query_timer);
+               call_rcu_bh(&p->rcu, br_multicast_free_pg);
+
+               if (!mp->ports && hlist_unhashed(&mp->mglist) &&
+                   netif_running(br->dev))
+                       mod_timer(&mp->timer, jiffies);
+
+               return;
+       }
+
+       WARN_ON(1);
+}
+
+static void br_multicast_port_group_expired(unsigned long data)
+{
+       struct net_bridge_port_group *pg = (void *)data;
+       struct net_bridge *br = pg->port->br;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) || timer_pending(&pg->timer) ||
+           hlist_unhashed(&pg->mglist))
+               goto out;
+
+       br_multicast_del_pg(br, pg);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+static int br_mdb_rehash(struct net_bridge_mdb_htable **mdbp, int max,
+                        int elasticity)
+{
+       struct net_bridge_mdb_htable *old = *mdbp;
+       struct net_bridge_mdb_htable *mdb;
+       int err;
+
+       mdb = kmalloc(sizeof(*mdb), GFP_ATOMIC);
+       if (!mdb)
+               return -ENOMEM;
+
+       mdb->max = max;
+       mdb->old = old;
+
+       mdb->mhash = kzalloc(max * sizeof(*mdb->mhash), GFP_ATOMIC);
+       if (!mdb->mhash) {
+               kfree(mdb);
+               return -ENOMEM;
+       }
+
+       mdb->size = old ? old->size : 0;
+       mdb->ver = old ? old->ver ^ 1 : 0;
+
+       if (!old || elasticity)
+               get_random_bytes(&mdb->secret, sizeof(mdb->secret));
+       else
+               mdb->secret = old->secret;
+
+       if (!old)
+               goto out;
+
+       err = br_mdb_copy(mdb, old, elasticity);
+       if (err) {
+               kfree(mdb->mhash);
+               kfree(mdb);
+               return err;
+       }
+
+       call_rcu_bh(&mdb->rcu, br_mdb_free);
+
+out:
+       rcu_assign_pointer(*mdbp, mdb);
+
+       return 0;
+}
+
+static struct sk_buff *br_multicast_alloc_query(struct net_bridge *br,
+                                               __be32 group)
+{
+       struct sk_buff *skb;
+       struct igmphdr *ih;
+       struct ethhdr *eth;
+       struct iphdr *iph;
+
+       skb = netdev_alloc_skb_ip_align(br->dev, sizeof(*eth) + sizeof(*iph) +
+                                                sizeof(*ih) + 4);
+       if (!skb)
+               goto out;
+
+       skb->protocol = htons(ETH_P_IP);
+
+       skb_reset_mac_header(skb);
+       eth = eth_hdr(skb);
+
+       memcpy(eth->h_source, br->dev->dev_addr, 6);
+       eth->h_dest[0] = 1;
+       eth->h_dest[1] = 0;
+       eth->h_dest[2] = 0x5e;
+       eth->h_dest[3] = 0;
+       eth->h_dest[4] = 0;
+       eth->h_dest[5] = 1;
+       eth->h_proto = htons(ETH_P_IP);
+       skb_put(skb, sizeof(*eth));
+
+       skb_set_network_header(skb, skb->len);
+       iph = ip_hdr(skb);
+
+       iph->version = 4;
+       iph->ihl = 6;
+       iph->tos = 0xc0;
+       iph->tot_len = htons(sizeof(*iph) + sizeof(*ih) + 4);
+       iph->id = 0;
+       iph->frag_off = htons(IP_DF);
+       iph->ttl = 1;
+       iph->protocol = IPPROTO_IGMP;
+       iph->saddr = 0;
+       iph->daddr = htonl(INADDR_ALLHOSTS_GROUP);
+       ((u8 *)&iph[1])[0] = IPOPT_RA;
+       ((u8 *)&iph[1])[1] = 4;
+       ((u8 *)&iph[1])[2] = 0;
+       ((u8 *)&iph[1])[3] = 0;
+       ip_send_check(iph);
+       skb_put(skb, 24);
+
+       skb_set_transport_header(skb, skb->len);
+       ih = igmp_hdr(skb);
+       ih->type = IGMP_HOST_MEMBERSHIP_QUERY;
+       ih->code = (group ? br->multicast_last_member_interval :
+                           br->multicast_query_response_interval) /
+                  (HZ / IGMP_TIMER_SCALE);
+       ih->group = group;
+       ih->csum = 0;
+       ih->csum = ip_compute_csum((void *)ih, sizeof(struct igmphdr));
+       skb_put(skb, sizeof(*ih));
+
+       __skb_pull(skb, sizeof(*eth));
+
+out:
+       return skb;
+}
+
+static void br_multicast_send_group_query(struct net_bridge_mdb_entry *mp)
+{
+       struct net_bridge *br = mp->br;
+       struct sk_buff *skb;
+
+       skb = br_multicast_alloc_query(br, mp->addr);
+       if (!skb)
+               goto timer;
+
+       netif_rx(skb);
+
+timer:
+       if (++mp->queries_sent < br->multicast_last_member_count)
+               mod_timer(&mp->query_timer,
+                         jiffies + br->multicast_last_member_interval);
+}
+
+static void br_multicast_group_query_expired(unsigned long data)
+{
+       struct net_bridge_mdb_entry *mp = (void *)data;
+       struct net_bridge *br = mp->br;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) || hlist_unhashed(&mp->mglist) ||
+           mp->queries_sent >= br->multicast_last_member_count)
+               goto out;
+
+       br_multicast_send_group_query(mp);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+static void br_multicast_send_port_group_query(struct net_bridge_port_group *pg)
+{
+       struct net_bridge_port *port = pg->port;
+       struct net_bridge *br = port->br;
+       struct sk_buff *skb;
+
+       skb = br_multicast_alloc_query(br, pg->addr);
+       if (!skb)
+               goto timer;
+
+       br_deliver(port, skb);
+
+timer:
+       if (++pg->queries_sent < br->multicast_last_member_count)
+               mod_timer(&pg->query_timer,
+                         jiffies + br->multicast_last_member_interval);
+}
+
+static void br_multicast_port_group_query_expired(unsigned long data)
+{
+       struct net_bridge_port_group *pg = (void *)data;
+       struct net_bridge_port *port = pg->port;
+       struct net_bridge *br = port->br;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) || hlist_unhashed(&pg->mglist) ||
+           pg->queries_sent >= br->multicast_last_member_count)
+               goto out;
+
+       br_multicast_send_port_group_query(pg);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+static struct net_bridge_mdb_entry *br_multicast_get_group(
+       struct net_bridge *br, struct net_bridge_port *port, __be32 group,
+       int hash)
+{
+       struct net_bridge_mdb_htable *mdb = br->mdb;
+       struct net_bridge_mdb_entry *mp;
+       struct hlist_node *p;
+       unsigned count = 0;
+       unsigned max;
+       int elasticity;
+       int err;
+
+       hlist_for_each_entry(mp, p, &mdb->mhash[hash], hlist[mdb->ver]) {
+               count++;
+               if (unlikely(group == mp->addr)) {
+                       return mp;
+               }
+       }
+
+       elasticity = 0;
+       max = mdb->max;
+
+       if (unlikely(count > br->hash_elasticity && count)) {
+               if (net_ratelimit())
+                       printk(KERN_INFO "%s: Multicast hash table "
+                              "chain limit reached: %s\n",
+                              br->dev->name, port ? port->dev->name :
+                                                    br->dev->name);
+
+               elasticity = br->hash_elasticity;
+       }
+
+       if (mdb->size >= max) {
+               max *= 2;
+               if (unlikely(max >= br->hash_max)) {
+                       printk(KERN_WARNING "%s: Multicast hash table maximum "
+                              "reached, disabling snooping: %s, %d\n",
+                              br->dev->name, port ? port->dev->name :
+                                                    br->dev->name,
+                              max);
+                       err = -E2BIG;
+disable:
+                       br->multicast_disabled = 1;
+                       goto err;
+               }
+       }
+
+       if (max > mdb->max || elasticity) {
+               if (mdb->old) {
+                       if (net_ratelimit())
+                               printk(KERN_INFO "%s: Multicast hash table "
+                                      "on fire: %s\n",
+                                      br->dev->name, port ? port->dev->name :
+                                                            br->dev->name);
+                       err = -EEXIST;
+                       goto err;
+               }
+
+               err = br_mdb_rehash(&br->mdb, max, elasticity);
+               if (err) {
+                       printk(KERN_WARNING "%s: Cannot rehash multicast "
+                              "hash table, disabling snooping: "
+                              "%s, %d, %d\n",
+                              br->dev->name, port ? port->dev->name :
+                                                    br->dev->name,
+                              mdb->size, err);
+                       goto disable;
+               }
+
+               err = -EAGAIN;
+               goto err;
+       }
+
+       return NULL;
+
+err:
+       mp = ERR_PTR(err);
+       return mp;
+}
+
+static struct net_bridge_mdb_entry *br_multicast_new_group(
+       struct net_bridge *br, struct net_bridge_port *port, __be32 group)
+{
+       struct net_bridge_mdb_htable *mdb = br->mdb;
+       struct net_bridge_mdb_entry *mp;
+       int hash;
+
+       if (!mdb) {
+               if (br_mdb_rehash(&br->mdb, BR_HASH_SIZE, 0))
+                       return NULL;
+               goto rehash;
+       }
+
+       hash = br_ip_hash(mdb, group);
+       mp = br_multicast_get_group(br, port, group, hash);
+       switch (PTR_ERR(mp)) {
+       case 0:
+               break;
+
+       case -EAGAIN:
+rehash:
+               mdb = br->mdb;
+               hash = br_ip_hash(mdb, group);
+               break;
+
+       default:
+               goto out;
+       }
+
+       mp = kzalloc(sizeof(*mp), GFP_ATOMIC);
+       if (unlikely(!mp))
+               goto out;
+
+       mp->br = br;
+       mp->addr = group;
+       setup_timer(&mp->timer, br_multicast_group_expired,
+                   (unsigned long)mp);
+       setup_timer(&mp->query_timer, br_multicast_group_query_expired,
+                   (unsigned long)mp);
+
+       hlist_add_head_rcu(&mp->hlist[mdb->ver], &mdb->mhash[hash]);
+       mdb->size++;
+
+out:
+       return mp;
+}
+
+static int br_multicast_add_group(struct net_bridge *br,
+                                 struct net_bridge_port *port, __be32 group)
+{
+       struct net_bridge_mdb_entry *mp;
+       struct net_bridge_port_group *p;
+       struct net_bridge_port_group **pp;
+       unsigned long now = jiffies;
+       int err;
+
+       if (ipv4_is_local_multicast(group))
+               return 0;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) ||
+           (port && port->state == BR_STATE_DISABLED))
+               goto out;
+
+       mp = br_multicast_new_group(br, port, group);
+       err = PTR_ERR(mp);
+       if (unlikely(IS_ERR(mp) || !mp))
+               goto err;
+
+       if (!port) {
+               hlist_add_head(&mp->mglist, &br->mglist);
+               mod_timer(&mp->timer, now + br->multicast_membership_interval);
+               goto out;
+       }
+
+       for (pp = &mp->ports; (p = *pp); pp = &p->next) {
+               if (p->port == port)
+                       goto found;
+               if ((unsigned long)p->port < (unsigned long)port)
+                       break;
+       }
+
+       p = kzalloc(sizeof(*p), GFP_ATOMIC);
+       err = -ENOMEM;
+       if (unlikely(!p))
+               goto err;
+
+       p->addr = group;
+       p->port = port;
+       p->next = *pp;
+       hlist_add_head(&p->mglist, &port->mglist);
+       setup_timer(&p->timer, br_multicast_port_group_expired,
+                   (unsigned long)p);
+       setup_timer(&p->query_timer, br_multicast_port_group_query_expired,
+                   (unsigned long)p);
+
+       rcu_assign_pointer(*pp, p);
+
+found:
+       mod_timer(&p->timer, now + br->multicast_membership_interval);
+out:
+       err = 0;
+
+err:
+       spin_unlock(&br->multicast_lock);
+       return err;
+}
+
+static void br_multicast_router_expired(unsigned long data)
+{
+       struct net_bridge_port *port = (void *)data;
+       struct net_bridge *br = port->br;
+
+       spin_lock(&br->multicast_lock);
+       if (port->multicast_router != 1 ||
+           timer_pending(&port->multicast_router_timer) ||
+           hlist_unhashed(&port->rlist))
+               goto out;
+
+       hlist_del_init_rcu(&port->rlist);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+static void br_multicast_local_router_expired(unsigned long data)
+{
+}
+
+static void br_multicast_send_query(struct net_bridge *br,
+                                   struct net_bridge_port *port, u32 sent)
+{
+       unsigned long time;
+       struct sk_buff *skb;
+
+       if (!netif_running(br->dev) || br->multicast_disabled ||
+           timer_pending(&br->multicast_querier_timer))
+               return;
+
+       skb = br_multicast_alloc_query(br, 0);
+       if (!skb)
+               goto timer;
+
+       if (port) {
+               __skb_push(skb, sizeof(struct ethhdr));
+               skb->dev = port->dev;
+               NF_HOOK(PF_BRIDGE, NF_BR_LOCAL_OUT, skb, NULL, skb->dev,
+                       dev_queue_xmit);
+       } else
+               netif_rx(skb);
+
+timer:
+       time = jiffies;
+       time += sent < br->multicast_startup_query_count ?
+               br->multicast_startup_query_interval :
+               br->multicast_query_interval;
+       mod_timer(port ? &port->multicast_query_timer :
+                        &br->multicast_query_timer, time);
+}
+
+static void br_multicast_port_query_expired(unsigned long data)
+{
+       struct net_bridge_port *port = (void *)data;
+       struct net_bridge *br = port->br;
+
+       spin_lock(&br->multicast_lock);
+       if (port && (port->state == BR_STATE_DISABLED ||
+                    port->state == BR_STATE_BLOCKING))
+               goto out;
+
+       if (port->multicast_startup_queries_sent <
+           br->multicast_startup_query_count)
+               port->multicast_startup_queries_sent++;
+
+       br_multicast_send_query(port->br, port,
+                               port->multicast_startup_queries_sent);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+void br_multicast_add_port(struct net_bridge_port *port)
+{
+       port->multicast_router = 1;
+
+       setup_timer(&port->multicast_router_timer, br_multicast_router_expired,
+                   (unsigned long)port);
+       setup_timer(&port->multicast_query_timer,
+                   br_multicast_port_query_expired, (unsigned long)port);
+}
+
+void br_multicast_del_port(struct net_bridge_port *port)
+{
+       del_timer_sync(&port->multicast_router_timer);
+}
+
+void br_multicast_enable_port(struct net_bridge_port *port)
+{
+       struct net_bridge *br = port->br;
+
+       spin_lock(&br->multicast_lock);
+       if (br->multicast_disabled || !netif_running(br->dev))
+               goto out;
+
+       port->multicast_startup_queries_sent = 0;
+
+       if (try_to_del_timer_sync(&port->multicast_query_timer) >= 0 ||
+           del_timer(&port->multicast_query_timer))
+               mod_timer(&port->multicast_query_timer, jiffies);
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+void br_multicast_disable_port(struct net_bridge_port *port)
+{
+       struct net_bridge *br = port->br;
+       struct net_bridge_port_group *pg;
+       struct hlist_node *p, *n;
+
+       spin_lock(&br->multicast_lock);
+       hlist_for_each_entry_safe(pg, p, n, &port->mglist, mglist)
+               br_multicast_del_pg(br, pg);
+
+       if (!hlist_unhashed(&port->rlist))
+               hlist_del_init_rcu(&port->rlist);
+       del_timer(&port->multicast_router_timer);
+       del_timer(&port->multicast_query_timer);
+       spin_unlock(&br->multicast_lock);
+}
+
+static int br_multicast_igmp3_report(struct net_bridge *br,
+                                    struct net_bridge_port *port,
+                                    struct sk_buff *skb)
+{
+       struct igmpv3_report *ih;
+       struct igmpv3_grec *grec;
+       int i;
+       int len;
+       int num;
+       int type;
+       int err = 0;
+       __be32 group;
+
+       if (!pskb_may_pull(skb, sizeof(*ih)))
+               return -EINVAL;
+
+       ih = igmpv3_report_hdr(skb);
+       num = ntohs(ih->ngrec);
+       len = sizeof(*ih);
+
+       for (i = 0; i < num; i++) {
+               len += sizeof(*grec);
+               if (!pskb_may_pull(skb, len))
+                       return -EINVAL;
+
+               grec = (void *)(skb->data + len);
+               group = grec->grec_mca;
+               type = grec->grec_type;
+
+               len += grec->grec_nsrcs * 4;
+               if (!pskb_may_pull(skb, len))
+                       return -EINVAL;
+
+               /* We treat this as an IGMPv2 report for now. */
+               switch (type) {
+               case IGMPV3_MODE_IS_INCLUDE:
+               case IGMPV3_MODE_IS_EXCLUDE:
+               case IGMPV3_CHANGE_TO_INCLUDE:
+               case IGMPV3_CHANGE_TO_EXCLUDE:
+               case IGMPV3_ALLOW_NEW_SOURCES:
+               case IGMPV3_BLOCK_OLD_SOURCES:
+                       break;
+
+               default:
+                       continue;
+               }
+
+               err = br_multicast_add_group(br, port, group);
+               if (err)
+                       break;
+       }
+
+       return err;
+}
+
+static void br_multicast_mark_router(struct net_bridge *br,
+                                    struct net_bridge_port *port)
+{
+       unsigned long now = jiffies;
+       struct hlist_node *p;
+       struct hlist_node **h;
+
+       if (!port) {
+               if (br->multicast_router == 1)
+                       mod_timer(&br->multicast_router_timer,
+                                 now + br->multicast_querier_interval);
+               return;
+       }
+
+       if (port->multicast_router != 1)
+               return;
+
+       if (!hlist_unhashed(&port->rlist))
+               goto timer;
+
+       for (h = &br->router_list.first;
+            (p = *h) &&
+            (unsigned long)container_of(p, struct net_bridge_port, rlist) >
+            (unsigned long)port;
+            h = &p->next)
+               ;
+
+       port->rlist.pprev = h;
+       port->rlist.next = p;
+       rcu_assign_pointer(*h, &port->rlist);
+       if (p)
+               p->pprev = &port->rlist.next;
+
+timer:
+       mod_timer(&port->multicast_router_timer,
+                 now + br->multicast_querier_interval);
+}
+
+static void br_multicast_query_received(struct net_bridge *br,
+                                       struct net_bridge_port *port,
+                                       __be32 saddr)
+{
+       if (saddr)
+               mod_timer(&br->multicast_querier_timer,
+                         jiffies + br->multicast_querier_interval);
+       else if (timer_pending(&br->multicast_querier_timer))
+               return;
+
+       br_multicast_mark_router(br, port);
+}
+
+static int br_multicast_query(struct net_bridge *br,
+                             struct net_bridge_port *port,
+                             struct sk_buff *skb)
+{
+       struct iphdr *iph = ip_hdr(skb);
+       struct igmphdr *ih = igmp_hdr(skb);
+       struct net_bridge_mdb_entry *mp;
+       struct igmpv3_query *ih3;
+       struct net_bridge_port_group *p;
+       struct net_bridge_port_group **pp;
+       unsigned long max_delay;
+       unsigned long now = jiffies;
+       __be32 group;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) ||
+           (port && port->state == BR_STATE_DISABLED))
+               goto out;
+
+       br_multicast_query_received(br, port, iph->saddr);
+
+       group = ih->group;
+
+       if (skb->len == sizeof(*ih)) {
+               max_delay = ih->code * (HZ / IGMP_TIMER_SCALE);
+
+               if (!max_delay) {
+                       max_delay = 10 * HZ;
+                       group = 0;
+               }
+       } else {
+               if (!pskb_may_pull(skb, sizeof(struct igmpv3_query)))
+                       return -EINVAL;
+
+               ih3 = igmpv3_query_hdr(skb);
+               if (ih3->nsrcs)
+                       return 0;
+
+               max_delay = ih3->code ? 1 :
+                           IGMPV3_MRC(ih3->code) * (HZ / IGMP_TIMER_SCALE);
+       }
+
+       if (!group)
+               goto out;
+
+       mp = br_mdb_ip_get(br->mdb, group);
+       if (!mp)
+               goto out;
+
+       max_delay *= br->multicast_last_member_count;
+
+       if (!hlist_unhashed(&mp->mglist) &&
+           (timer_pending(&mp->timer) ?
+            time_after(mp->timer.expires, now + max_delay) :
+            try_to_del_timer_sync(&mp->timer) >= 0))
+               mod_timer(&mp->timer, now + max_delay);
+
+       for (pp = &mp->ports; (p = *pp); pp = &p->next) {
+               if (timer_pending(&p->timer) ?
+                   time_after(p->timer.expires, now + max_delay) :
+                   try_to_del_timer_sync(&p->timer) >= 0)
+                       mod_timer(&mp->timer, now + max_delay);
+       }
+
+out:
+       spin_unlock(&br->multicast_lock);
+       return 0;
+}
+
+static void br_multicast_leave_group(struct net_bridge *br,
+                                    struct net_bridge_port *port,
+                                    __be32 group)
+{
+       struct net_bridge_mdb_htable *mdb;
+       struct net_bridge_mdb_entry *mp;
+       struct net_bridge_port_group *p;
+       unsigned long now;
+       unsigned long time;
+
+       if (ipv4_is_local_multicast(group))
+               return;
+
+       spin_lock(&br->multicast_lock);
+       if (!netif_running(br->dev) ||
+           (port && port->state == BR_STATE_DISABLED) ||
+           timer_pending(&br->multicast_querier_timer))
+               goto out;
+
+       mdb = br->mdb;
+       mp = br_mdb_ip_get(mdb, group);
+       if (!mp)
+               goto out;
+
+       now = jiffies;
+       time = now + br->multicast_last_member_count *
+                    br->multicast_last_member_interval;
+
+       if (!port) {
+               if (!hlist_unhashed(&mp->mglist) &&
+                   (timer_pending(&mp->timer) ?
+                    time_after(mp->timer.expires, time) :
+                    try_to_del_timer_sync(&mp->timer) >= 0)) {
+                       mod_timer(&mp->timer, time);
+
+                       mp->queries_sent = 0;
+                       mod_timer(&mp->query_timer, now);
+               }
+
+               goto out;
+       }
+
+       for (p = mp->ports; p; p = p->next) {
+               if (p->port != port)
+                       continue;
+
+               if (!hlist_unhashed(&p->mglist) &&
+                   (timer_pending(&p->timer) ?
+                    time_after(p->timer.expires, time) :
+                    try_to_del_timer_sync(&p->timer) >= 0)) {
+                       mod_timer(&p->timer, time);
+
+                       p->queries_sent = 0;
+                       mod_timer(&p->query_timer, now);
+               }
+
+               break;
+       }
+
+out:
+       spin_unlock(&br->multicast_lock);
+}
+
+static int br_multicast_ipv4_rcv(struct net_bridge *br,
+                                struct net_bridge_port *port,
+                                struct sk_buff *skb)
+{
+       struct sk_buff *skb2 = skb;
+       struct iphdr *iph;
+       struct igmphdr *ih;
+       unsigned len;
+       unsigned offset;
+       int err;
+
+       BR_INPUT_SKB_CB(skb)->igmp = 0;
+       BR_INPUT_SKB_CB(skb)->mrouters_only = 0;
+
+       /* We treat OOM as packet loss for now. */
+       if (!pskb_may_pull(skb, sizeof(*iph)))
+               return -EINVAL;
+
+       iph = ip_hdr(skb);
+
+       if (iph->ihl < 5 || iph->version != 4)
+               return -EINVAL;
+
+       if (!pskb_may_pull(skb, ip_hdrlen(skb)))
+               return -EINVAL;
+
+       iph = ip_hdr(skb);
+
+       if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl)))
+               return -EINVAL;
+
+       if (iph->protocol != IPPROTO_IGMP)
+               return 0;
+
+       len = ntohs(iph->tot_len);
+       if (skb->len < len || len < ip_hdrlen(skb))
+               return -EINVAL;
+
+       if (skb->len > len) {
+               skb2 = skb_clone(skb, GFP_ATOMIC);
+               if (!skb2)
+                       return -ENOMEM;
+
+               err = pskb_trim_rcsum(skb2, len);
+               if (err)
+                       return err;
+       }
+
+       len -= ip_hdrlen(skb2);
+       offset = skb_network_offset(skb2) + ip_hdrlen(skb2);
+       __skb_pull(skb2, offset);
+       skb_reset_transport_header(skb2);
+
+       err = -EINVAL;
+       if (!pskb_may_pull(skb2, sizeof(*ih)))
+               goto out;
+
+       iph = ip_hdr(skb2);
+
+       switch (skb2->ip_summed) {
+       case CHECKSUM_COMPLETE:
+               if (!csum_fold(skb2->csum))
+                       break;
+               /* fall through */
+       case CHECKSUM_NONE:
+               skb2->csum = 0;
+               if (skb_checksum_complete(skb2))
+                       return -EINVAL;
+       }
+
+       err = 0;
+
+       BR_INPUT_SKB_CB(skb)->igmp = 1;
+       ih = igmp_hdr(skb2);
+
+       switch (ih->type) {
+       case IGMP_HOST_MEMBERSHIP_REPORT:
+       case IGMPV2_HOST_MEMBERSHIP_REPORT:
+               BR_INPUT_SKB_CB(skb2)->mrouters_only = 1;
+               err = br_multicast_add_group(br, port, ih->group);
+               break;
+       case IGMPV3_HOST_MEMBERSHIP_REPORT:
+               err = br_multicast_igmp3_report(br, port, skb2);
+               break;
+       case IGMP_HOST_MEMBERSHIP_QUERY:
+               err = br_multicast_query(br, port, skb2);
+               break;
+       case IGMP_HOST_LEAVE_MESSAGE:
+               br_multicast_leave_group(br, port, ih->group);
+               break;
+       }
+
+out:
+       __skb_push(skb2, offset);
+       if (skb2 != skb)
+               kfree_skb(skb2);
+       return err;
+}
+
+int br_multicast_rcv(struct net_bridge *br, struct net_bridge_port *port,
+                    struct sk_buff *skb)
+{
+       if (br->multicast_disabled)
+               return 0;
+
+       switch (skb->protocol) {
+       case htons(ETH_P_IP):
+               return br_multicast_ipv4_rcv(br, port, skb);
+       }
+
+       return 0;
+}
+
+static void br_multicast_query_expired(unsigned long data)
+{
+       struct net_bridge *br = (void *)data;
+
+       spin_lock(&br->multicast_lock);
+       if (br->multicast_startup_queries_sent <
+           br->multicast_startup_query_count)
+               br->multicast_startup_queries_sent++;
+
+       br_multicast_send_query(br, NULL, br->multicast_startup_queries_sent);
+
+       spin_unlock(&br->multicast_lock);
+}
+
+void br_multicast_init(struct net_bridge *br)
+{
+       br->hash_elasticity = 4;
+       br->hash_max = 512;
+
+       br->multicast_router = 1;
+       br->multicast_last_member_count = 2;
+       br->multicast_startup_query_count = 2;
+
+       br->multicast_last_member_interval = HZ;
+       br->multicast_query_response_interval = 10 * HZ;
+       br->multicast_startup_query_interval = 125 * HZ / 4;
+       br->multicast_query_interval = 125 * HZ;
+       br->multicast_querier_interval = 255 * HZ;
+       br->multicast_membership_interval = 260 * HZ;
+
+       spin_lock_init(&br->multicast_lock);
+       setup_timer(&br->multicast_router_timer,
+                   br_multicast_local_router_expired, 0);
+       setup_timer(&br->multicast_querier_timer,
+                   br_multicast_local_router_expired, 0);
+       setup_timer(&br->multicast_query_timer, br_multicast_query_expired,
+                   (unsigned long)br);
+}
+
+void br_multicast_open(struct net_bridge *br)
+{
+       br->multicast_startup_queries_sent = 0;
+
+       if (br->multicast_disabled)
+               return;
+
+       mod_timer(&br->multicast_query_timer, jiffies);
+}
+
+void br_multicast_stop(struct net_bridge *br)
+{
+       struct net_bridge_mdb_htable *mdb;
+       struct net_bridge_mdb_entry *mp;
+       struct hlist_node *p, *n;
+       u32 ver;
+       int i;
+
+       del_timer_sync(&br->multicast_router_timer);
+       del_timer_sync(&br->multicast_querier_timer);
+       del_timer_sync(&br->multicast_query_timer);
+
+       spin_lock_bh(&br->multicast_lock);
+       mdb = br->mdb;
+       if (!mdb)
+               goto out;
+
+       br->mdb = NULL;
+
+       ver = mdb->ver;
+       for (i = 0; i < mdb->max; i++) {
+               hlist_for_each_entry_safe(mp, p, n, &mdb->mhash[i],
+                                         hlist[ver]) {
+                       del_timer(&mp->timer);
+                       del_timer(&mp->query_timer);
+                       call_rcu_bh(&mp->rcu, br_multicast_free_group);
+               }
+       }
+
+       if (mdb->old) {
+               spin_unlock_bh(&br->multicast_lock);
+               synchronize_rcu_bh();
+               spin_lock_bh(&br->multicast_lock);
+               WARN_ON(mdb->old);
+       }
+
+       mdb->old = mdb;
+       call_rcu_bh(&mdb->rcu, br_mdb_free);
+
+out:
+       spin_unlock_bh(&br->multicast_lock);
+}
index fad5a2669d3460e76df10908d4ff096c100848ae..44345c9afdd3a15775595c07d27402b013501788 100644 (file)
@@ -57,6 +57,41 @@ struct net_bridge_fdb_entry
        unsigned char                   is_static;
 };
 
+struct net_bridge_port_group {
+       struct net_bridge_port          *port;
+       struct net_bridge_port_group    *next;
+       struct hlist_node               mglist;
+       struct rcu_head                 rcu;
+       struct timer_list               timer;
+       struct timer_list               query_timer;
+       __be32                          addr;
+       u32                             queries_sent;
+};
+
+struct net_bridge_mdb_entry
+{
+       struct hlist_node               hlist[2];
+       struct hlist_node               mglist;
+       struct net_bridge               *br;
+       struct net_bridge_port_group    *ports;
+       struct rcu_head                 rcu;
+       struct timer_list               timer;
+       struct timer_list               query_timer;
+       __be32                          addr;
+       u32                             queries_sent;
+};
+
+struct net_bridge_mdb_htable
+{
+       struct hlist_head               *mhash;
+       struct rcu_head                 rcu;
+       struct net_bridge_mdb_htable    *old;
+       u32                             size;
+       u32                             max;
+       u32                             secret;
+       u32                             ver;
+};
+
 struct net_bridge_port
 {
        struct net_bridge               *br;
@@ -84,6 +119,15 @@ struct net_bridge_port
 
        unsigned long                   flags;
 #define BR_HAIRPIN_MODE                0x00000001
+
+#ifdef CONFIG_BRIDGE_IGMP_SNOOPING
+       u32                             multicast_startup_queries_sent;
+       unsigned char                   multicast_router;
+       struct timer_list               multicast_router_timer;
+       struct timer_list               multicast_query_timer;
+       struct hlist_head               mglist;
+       struct hlist_node               rlist;
+#endif
 };
 
 struct net_bridge
@@ -124,6 +168,35 @@ struct net_bridge
        unsigned char                   topology_change;
        unsigned char                   topology_change_detected;
 
+#ifdef CONFIG_BRIDGE_IGMP_SNOOPING
+       unsigned char                   multicast_router;
+
+       u8                              multicast_disabled:1;
+
+       u32                             hash_elasticity;
+       u32                             hash_max;
+
+       u32                             multicast_last_member_count;
+       u32                             multicast_startup_queries_sent;
+       u32                             multicast_startup_query_count;
+
+       unsigned long                   multicast_last_member_interval;
+       unsigned long                   multicast_membership_interval;
+       unsigned long                   multicast_querier_interval;
+       unsigned long                   multicast_query_interval;
+       unsigned long                   multicast_query_response_interval;
+       unsigned long                   multicast_startup_query_interval;
+
+       spinlock_t                      multicast_lock;
+       struct net_bridge_mdb_htable    *mdb;
+       struct hlist_head               router_list;
+       struct hlist_head               mglist;
+
+       struct timer_list               multicast_router_timer;
+       struct timer_list               multicast_querier_timer;
+       struct timer_list               multicast_query_timer;
+#endif
+
        struct timer_list               hello_timer;
        struct timer_list               tcn_timer;
        struct timer_list               topology_change_timer;
@@ -133,6 +206,8 @@ struct net_bridge
 
 struct br_input_skb_cb {
        struct net_device *brdev;
+       int igmp;
+       int mrouters_only;
 };
 
 #define BR_INPUT_SKB_CB(__skb) ((struct br_input_skb_cb *)(__skb)->cb)
@@ -204,6 +279,70 @@ extern struct sk_buff *br_handle_frame(struct net_bridge_port *p,
 extern int br_dev_ioctl(struct net_device *dev, struct ifreq *rq, int cmd);
 extern int br_ioctl_deviceless_stub(struct net *net, unsigned int cmd, void __user *arg);
 
+/* br_multicast.c */
+#ifdef CONFIG_BRIDGE_IGMP_SNOOPING
+extern int br_multicast_rcv(struct net_bridge *br,
+                           struct net_bridge_port *port,
+                           struct sk_buff *skb);
+extern struct net_bridge_mdb_entry *br_mdb_get(struct net_bridge *br,
+                                              struct sk_buff *skb);
+extern void br_multicast_add_port(struct net_bridge_port *port);
+extern void br_multicast_del_port(struct net_bridge_port *port);
+extern void br_multicast_enable_port(struct net_bridge_port *port);
+extern void br_multicast_disable_port(struct net_bridge_port *port);
+extern void br_multicast_init(struct net_bridge *br);
+extern void br_multicast_open(struct net_bridge *br);
+extern void br_multicast_stop(struct net_bridge *br);
+#else
+static inline int br_multicast_rcv(struct net_bridge *br,
+                                  struct net_bridge_port *port,
+                                  struct sk_buff *skb)
+{
+       return 0;
+}
+
+static inline struct net_bridge_mdb_entry *br_mdb_get(struct net_bridge *br,
+                                                     struct sk_buff *skb)
+{
+       return NULL;
+}
+
+static inline void br_multicast_add_port(struct net_bridge_port *port)
+{
+}
+
+static inline void br_multicast_del_port(struct net_bridge_port *port)
+{
+}
+
+static inline void br_multicast_enable_port(struct net_bridge_port *port)
+{
+}
+
+static inline void br_multicast_disable_port(struct net_bridge_port *port)
+{
+}
+
+static inline void br_multicast_init(struct net_bridge *br)
+{
+}
+
+static inline void br_multicast_open(struct net_bridge *br)
+{
+}
+
+static inline void br_multicast_stop(struct net_bridge *br)
+{
+}
+#endif
+
+static inline bool br_multicast_is_router(struct net_bridge *br)
+{
+       return br->multicast_router == 2 ||
+              (br->multicast_router == 1 &&
+               timer_pending(&br->multicast_router_timer));
+}
+
 /* br_netfilter.c */
 #ifdef CONFIG_BRIDGE_NETFILTER
 extern int br_netfilter_init(void);