Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
[pandora-kernel.git] / drivers / net / macvlan.c
index 2490aa3..93c3e6e 100644 (file)
@@ -29,6 +29,7 @@
 #include <linux/if_link.h>
 #include <linux/if_macvlan.h>
 #include <net/rtnetlink.h>
+#include <net/xfrm.h>
 
 #define MACVLAN_HASH_SIZE      (1 << BITS_PER_BYTE)
 
@@ -38,12 +39,28 @@ struct macvlan_port {
        struct list_head        vlans;
 };
 
+/**
+ *     struct macvlan_rx_stats - MACVLAN percpu rx stats
+ *     @rx_packets: number of received packets
+ *     @rx_bytes: number of received bytes
+ *     @multicast: number of received multicast packets
+ *     @rx_errors: number of errors
+ */
+struct macvlan_rx_stats {
+       unsigned long rx_packets;
+       unsigned long rx_bytes;
+       unsigned long multicast;
+       unsigned long rx_errors;
+};
+
 struct macvlan_dev {
        struct net_device       *dev;
        struct list_head        list;
        struct hlist_node       hlist;
        struct macvlan_port     *port;
        struct net_device       *lowerdev;
+       struct macvlan_rx_stats *rx_stats;
+       enum macvlan_mode       mode;
 };
 
 
@@ -101,41 +118,67 @@ static int macvlan_addr_busy(const struct macvlan_port *port,
        return 0;
 }
 
+static inline void macvlan_count_rx(const struct macvlan_dev *vlan,
+                                   unsigned int len, bool success,
+                                   bool multicast)
+{
+       struct macvlan_rx_stats *rx_stats;
+
+       rx_stats = per_cpu_ptr(vlan->rx_stats, smp_processor_id());
+       if (likely(success)) {
+               rx_stats->rx_packets++;;
+               rx_stats->rx_bytes += len;
+               if (multicast)
+                       rx_stats->multicast++;
+       } else {
+               rx_stats->rx_errors++;
+       }
+}
+
+static int macvlan_broadcast_one(struct sk_buff *skb, struct net_device *dev,
+                                const struct ethhdr *eth, bool local)
+{
+       if (!skb)
+               return NET_RX_DROP;
+
+       if (local)
+               return dev_forward_skb(dev, skb);
+
+       skb->dev = dev;
+       if (!compare_ether_addr_64bits(eth->h_dest,
+                                      dev->broadcast))
+               skb->pkt_type = PACKET_BROADCAST;
+       else
+               skb->pkt_type = PACKET_MULTICAST;
+
+       return netif_rx(skb);
+}
+
 static void macvlan_broadcast(struct sk_buff *skb,
-                             const struct macvlan_port *port)
+                             const struct macvlan_port *port,
+                             struct net_device *src,
+                             enum macvlan_mode mode)
 {
        const struct ethhdr *eth = eth_hdr(skb);
        const struct macvlan_dev *vlan;
        struct hlist_node *n;
-       struct net_device *dev;
        struct sk_buff *nskb;
        unsigned int i;
+       int err;
 
        if (skb->protocol == htons(ETH_P_PAUSE))
                return;
 
        for (i = 0; i < MACVLAN_HASH_SIZE; i++) {
                hlist_for_each_entry_rcu(vlan, n, &port->vlan_hash[i], hlist) {
-                       dev = vlan->dev;
-
-                       nskb = skb_clone(skb, GFP_ATOMIC);
-                       if (nskb == NULL) {
-                               dev->stats.rx_errors++;
-                               dev->stats.rx_dropped++;
+                       if (vlan->dev == src || !(vlan->mode & mode))
                                continue;
-                       }
 
-                       dev->stats.rx_bytes += skb->len + ETH_HLEN;
-                       dev->stats.rx_packets++;
-                       dev->stats.multicast++;
-
-                       nskb->dev = dev;
-                       if (!compare_ether_addr_64bits(eth->h_dest, dev->broadcast))
-                               nskb->pkt_type = PACKET_BROADCAST;
-                       else
-                               nskb->pkt_type = PACKET_MULTICAST;
-
-                       netif_rx(nskb);
+                       nskb = skb_clone(skb, GFP_ATOMIC);
+                       err = macvlan_broadcast_one(nskb, vlan->dev, eth,
+                                        mode == MACVLAN_MODE_BRIDGE);
+                       macvlan_count_rx(vlan, skb->len + ETH_HLEN,
+                                        err == NET_RX_SUCCESS, 1);
                }
        }
 }
@@ -146,14 +189,34 @@ static struct sk_buff *macvlan_handle_frame(struct sk_buff *skb)
        const struct ethhdr *eth = eth_hdr(skb);
        const struct macvlan_port *port;
        const struct macvlan_dev *vlan;
+       const struct macvlan_dev *src;
        struct net_device *dev;
+       unsigned int len;
 
        port = rcu_dereference(skb->dev->macvlan_port);
        if (port == NULL)
                return skb;
 
        if (is_multicast_ether_addr(eth->h_dest)) {
-               macvlan_broadcast(skb, port);
+               src = macvlan_hash_lookup(port, eth->h_source);
+               if (!src)
+                       /* frame comes from an external address */
+                       macvlan_broadcast(skb, port, NULL,
+                                         MACVLAN_MODE_PRIVATE |
+                                         MACVLAN_MODE_VEPA    |
+                                         MACVLAN_MODE_BRIDGE);
+               else if (src->mode == MACVLAN_MODE_VEPA)
+                       /* flood to everyone except source */
+                       macvlan_broadcast(skb, port, src->dev,
+                                         MACVLAN_MODE_VEPA |
+                                         MACVLAN_MODE_BRIDGE);
+               else if (src->mode == MACVLAN_MODE_BRIDGE)
+                       /*
+                        * flood only to VEPA ports, bridge ports
+                        * already saw the frame on the way out.
+                        */
+                       macvlan_broadcast(skb, port, src->dev,
+                                         MACVLAN_MODE_VEPA);
                return skb;
        }
 
@@ -166,16 +229,11 @@ static struct sk_buff *macvlan_handle_frame(struct sk_buff *skb)
                kfree_skb(skb);
                return NULL;
        }
-
+       len = skb->len + ETH_HLEN;
        skb = skb_share_check(skb, GFP_ATOMIC);
-       if (skb == NULL) {
-               dev->stats.rx_errors++;
-               dev->stats.rx_dropped++;
+       macvlan_count_rx(vlan, len, skb != NULL, 0);
+       if (!skb)
                return NULL;
-       }
-
-       dev->stats.rx_bytes += skb->len + ETH_HLEN;
-       dev->stats.rx_packets++;
 
        skb->dev = dev;
        skb->pkt_type = PACKET_HOST;
@@ -184,25 +242,53 @@ static struct sk_buff *macvlan_handle_frame(struct sk_buff *skb)
        return NULL;
 }
 
+static int macvlan_queue_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+       const struct macvlan_dev *vlan = netdev_priv(dev);
+       const struct macvlan_port *port = vlan->port;
+       const struct macvlan_dev *dest;
+
+       if (vlan->mode == MACVLAN_MODE_BRIDGE) {
+               const struct ethhdr *eth = (void *)skb->data;
+
+               /* send to other bridge ports directly */
+               if (is_multicast_ether_addr(eth->h_dest)) {
+                       macvlan_broadcast(skb, port, dev, MACVLAN_MODE_BRIDGE);
+                       goto xmit_world;
+               }
+
+               dest = macvlan_hash_lookup(port, eth->h_dest);
+               if (dest && dest->mode == MACVLAN_MODE_BRIDGE) {
+                       unsigned int length = skb->len + ETH_HLEN;
+                       int ret = dev_forward_skb(dest->dev, skb);
+                       macvlan_count_rx(dest, length,
+                                        ret == NET_RX_SUCCESS, 0);
+
+                       return NET_XMIT_SUCCESS;
+               }
+       }
+
+xmit_world:
+       skb->dev = vlan->lowerdev;
+       return dev_queue_xmit(skb);
+}
+
 static netdev_tx_t macvlan_start_xmit(struct sk_buff *skb,
                                      struct net_device *dev)
 {
        int i = skb_get_queue_mapping(skb);
        struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
-       const struct macvlan_dev *vlan = netdev_priv(dev);
        unsigned int len = skb->len;
        int ret;
 
-       skb->dev = vlan->lowerdev;
-       ret = dev_queue_xmit(skb);
-
+       ret = macvlan_queue_xmit(skb, dev);
        if (likely(ret == NET_XMIT_SUCCESS)) {
                txq->tx_packets++;
                txq->tx_bytes += len;
        } else
                txq->tx_dropped++;
 
-       return NETDEV_TX_OK;
+       return ret;
 }
 
 static int macvlan_hard_header(struct sk_buff *skb, struct net_device *dev,
@@ -366,9 +452,47 @@ static int macvlan_init(struct net_device *dev)
 
        macvlan_set_lockdep_class(dev);
 
+       vlan->rx_stats = alloc_percpu(struct macvlan_rx_stats);
+       if (!vlan->rx_stats)
+               return -ENOMEM;
+
        return 0;
 }
 
+static void macvlan_uninit(struct net_device *dev)
+{
+       struct macvlan_dev *vlan = netdev_priv(dev);
+
+       free_percpu(vlan->rx_stats);
+}
+
+static struct net_device_stats *macvlan_dev_get_stats(struct net_device *dev)
+{
+       struct net_device_stats *stats = &dev->stats;
+       struct macvlan_dev *vlan = netdev_priv(dev);
+
+       dev_txq_stats_fold(dev, stats);
+
+       if (vlan->rx_stats) {
+               struct macvlan_rx_stats *p, rx = {0};
+               int i;
+
+               for_each_possible_cpu(i) {
+                       p = per_cpu_ptr(vlan->rx_stats, i);
+                       rx.rx_packets += p->rx_packets;
+                       rx.rx_bytes   += p->rx_bytes;
+                       rx.rx_errors  += p->rx_errors;
+                       rx.multicast  += p->multicast;
+               }
+               stats->rx_packets = rx.rx_packets;
+               stats->rx_bytes   = rx.rx_bytes;
+               stats->rx_errors  = rx.rx_errors;
+               stats->rx_dropped = rx.rx_errors;
+               stats->multicast  = rx.multicast;
+       }
+       return stats;
+}
+
 static void macvlan_ethtool_get_drvinfo(struct net_device *dev,
                                        struct ethtool_drvinfo *drvinfo)
 {
@@ -405,6 +529,7 @@ static const struct ethtool_ops macvlan_ethtool_ops = {
 
 static const struct net_device_ops macvlan_netdev_ops = {
        .ndo_init               = macvlan_init,
+       .ndo_uninit             = macvlan_uninit,
        .ndo_open               = macvlan_open,
        .ndo_stop               = macvlan_stop,
        .ndo_start_xmit         = macvlan_start_xmit,
@@ -412,6 +537,7 @@ static const struct net_device_ops macvlan_netdev_ops = {
        .ndo_change_rx_flags    = macvlan_change_rx_flags,
        .ndo_set_mac_address    = macvlan_set_mac_address,
        .ndo_set_multicast_list = macvlan_set_multicast_list,
+       .ndo_get_stats          = macvlan_dev_get_stats,
        .ndo_validate_addr      = eth_validate_addr,
 };
 
@@ -483,6 +609,17 @@ static int macvlan_validate(struct nlattr *tb[], struct nlattr *data[])
                if (!is_valid_ether_addr(nla_data(tb[IFLA_ADDRESS])))
                        return -EADDRNOTAVAIL;
        }
+
+       if (data && data[IFLA_MACVLAN_MODE]) {
+               switch (nla_get_u32(data[IFLA_MACVLAN_MODE])) {
+               case MACVLAN_MODE_PRIVATE:
+               case MACVLAN_MODE_VEPA:
+               case MACVLAN_MODE_BRIDGE:
+                       break;
+               default:
+                       return -EINVAL;
+               }
+       }
        return 0;
 }
 
@@ -505,7 +642,7 @@ static int macvlan_get_tx_queues(struct net *net,
        return 0;
 }
 
-static int macvlan_newlink(struct net_device *dev,
+static int macvlan_newlink(struct net *src_net, struct net_device *dev,
                           struct nlattr *tb[], struct nlattr *data[])
 {
        struct macvlan_dev *vlan = netdev_priv(dev);
@@ -516,7 +653,7 @@ static int macvlan_newlink(struct net_device *dev,
        if (!tb[IFLA_LINK])
                return -EINVAL;
 
-       lowerdev = __dev_get_by_index(dev_net(dev), nla_get_u32(tb[IFLA_LINK]));
+       lowerdev = __dev_get_by_index(src_net, nla_get_u32(tb[IFLA_LINK]));
        if (lowerdev == NULL)
                return -ENODEV;
 
@@ -547,6 +684,10 @@ static int macvlan_newlink(struct net_device *dev,
        vlan->dev      = dev;
        vlan->port     = port;
 
+       vlan->mode     = MACVLAN_MODE_VEPA;
+       if (data && data[IFLA_MACVLAN_MODE])
+               vlan->mode = nla_get_u32(data[IFLA_MACVLAN_MODE]);
+
        err = register_netdevice(dev);
        if (err < 0)
                return err;
@@ -556,18 +697,48 @@ static int macvlan_newlink(struct net_device *dev,
        return 0;
 }
 
-static void macvlan_dellink(struct net_device *dev)
+static void macvlan_dellink(struct net_device *dev, struct list_head *head)
 {
        struct macvlan_dev *vlan = netdev_priv(dev);
        struct macvlan_port *port = vlan->port;
 
        list_del(&vlan->list);
-       unregister_netdevice(dev);
+       unregister_netdevice_queue(dev, head);
 
        if (list_empty(&port->vlans))
                macvlan_port_destroy(port->dev);
 }
 
+static int macvlan_changelink(struct net_device *dev,
+               struct nlattr *tb[], struct nlattr *data[])
+{
+       struct macvlan_dev *vlan = netdev_priv(dev);
+       if (data && data[IFLA_MACVLAN_MODE])
+               vlan->mode = nla_get_u32(data[IFLA_MACVLAN_MODE]);
+       return 0;
+}
+
+static size_t macvlan_get_size(const struct net_device *dev)
+{
+       return nla_total_size(4);
+}
+
+static int macvlan_fill_info(struct sk_buff *skb,
+                               const struct net_device *dev)
+{
+       struct macvlan_dev *vlan = netdev_priv(dev);
+
+       NLA_PUT_U32(skb, IFLA_MACVLAN_MODE, vlan->mode);
+       return 0;
+
+nla_put_failure:
+       return -EMSGSIZE;
+}
+
+static const struct nla_policy macvlan_policy[IFLA_MACVLAN_MAX + 1] = {
+       [IFLA_MACVLAN_MODE] = { .type = NLA_U32 },
+};
+
 static struct rtnl_link_ops macvlan_link_ops __read_mostly = {
        .kind           = "macvlan",
        .priv_size      = sizeof(struct macvlan_dev),
@@ -576,6 +747,11 @@ static struct rtnl_link_ops macvlan_link_ops __read_mostly = {
        .validate       = macvlan_validate,
        .newlink        = macvlan_newlink,
        .dellink        = macvlan_dellink,
+       .maxtype        = IFLA_MACVLAN_MAX,
+       .policy         = macvlan_policy,
+       .changelink     = macvlan_changelink,
+       .get_size       = macvlan_get_size,
+       .fill_info      = macvlan_fill_info,
 };
 
 static int macvlan_device_event(struct notifier_block *unused,
@@ -603,7 +779,7 @@ static int macvlan_device_event(struct notifier_block *unused,
                break;
        case NETDEV_UNREGISTER:
                list_for_each_entry_safe(vlan, next, &port->vlans, list)
-                       macvlan_dellink(vlan->dev);
+                       macvlan_dellink(vlan->dev, NULL);
                break;
        }
        return NOTIFY_DONE;