Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / net / openvswitch / vport-internal_dev.c
diff --git a/kernel/net/openvswitch/vport-internal_dev.c b/kernel/net/openvswitch/vport-internal_dev.c
new file mode 100644 (file)
index 0000000..6a55f71
--- /dev/null
@@ -0,0 +1,288 @@
+/*
+ * Copyright (c) 2007-2012 Nicira, Inc.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of version 2 of the GNU General Public
+ * License as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA
+ */
+
+#include <linux/hardirq.h>
+#include <linux/if_vlan.h>
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/ethtool.h>
+#include <linux/skbuff.h>
+
+#include <net/dst.h>
+#include <net/xfrm.h>
+#include <net/rtnetlink.h>
+
+#include "datapath.h"
+#include "vport-internal_dev.h"
+#include "vport-netdev.h"
+
+struct internal_dev {
+       struct vport *vport;
+};
+
+static struct vport_ops ovs_internal_vport_ops;
+
+static struct internal_dev *internal_dev_priv(struct net_device *netdev)
+{
+       return netdev_priv(netdev);
+}
+
+/* This function is only called by the kernel network layer.*/
+static struct rtnl_link_stats64 *internal_dev_get_stats(struct net_device *netdev,
+                                                       struct rtnl_link_stats64 *stats)
+{
+       struct vport *vport = ovs_internal_dev_get_vport(netdev);
+       struct ovs_vport_stats vport_stats;
+
+       ovs_vport_get_stats(vport, &vport_stats);
+
+       /* The tx and rx stats need to be swapped because the
+        * switch and host OS have opposite perspectives. */
+       stats->rx_packets       = vport_stats.tx_packets;
+       stats->tx_packets       = vport_stats.rx_packets;
+       stats->rx_bytes         = vport_stats.tx_bytes;
+       stats->tx_bytes         = vport_stats.rx_bytes;
+       stats->rx_errors        = vport_stats.tx_errors;
+       stats->tx_errors        = vport_stats.rx_errors;
+       stats->rx_dropped       = vport_stats.tx_dropped;
+       stats->tx_dropped       = vport_stats.rx_dropped;
+
+       return stats;
+}
+
+/* Called with rcu_read_lock_bh. */
+static int internal_dev_xmit(struct sk_buff *skb, struct net_device *netdev)
+{
+       rcu_read_lock();
+       ovs_vport_receive(internal_dev_priv(netdev)->vport, skb, NULL);
+       rcu_read_unlock();
+       return 0;
+}
+
+static int internal_dev_open(struct net_device *netdev)
+{
+       netif_start_queue(netdev);
+       return 0;
+}
+
+static int internal_dev_stop(struct net_device *netdev)
+{
+       netif_stop_queue(netdev);
+       return 0;
+}
+
+static void internal_dev_getinfo(struct net_device *netdev,
+                                struct ethtool_drvinfo *info)
+{
+       strlcpy(info->driver, "openvswitch", sizeof(info->driver));
+}
+
+static const struct ethtool_ops internal_dev_ethtool_ops = {
+       .get_drvinfo    = internal_dev_getinfo,
+       .get_link       = ethtool_op_get_link,
+};
+
+static int internal_dev_change_mtu(struct net_device *netdev, int new_mtu)
+{
+       if (new_mtu < 68)
+               return -EINVAL;
+
+       netdev->mtu = new_mtu;
+       return 0;
+}
+
+static void internal_dev_destructor(struct net_device *dev)
+{
+       struct vport *vport = ovs_internal_dev_get_vport(dev);
+
+       ovs_vport_free(vport);
+       free_netdev(dev);
+}
+
+static const struct net_device_ops internal_dev_netdev_ops = {
+       .ndo_open = internal_dev_open,
+       .ndo_stop = internal_dev_stop,
+       .ndo_start_xmit = internal_dev_xmit,
+       .ndo_set_mac_address = eth_mac_addr,
+       .ndo_change_mtu = internal_dev_change_mtu,
+       .ndo_get_stats64 = internal_dev_get_stats,
+};
+
+static struct rtnl_link_ops internal_dev_link_ops __read_mostly = {
+       .kind = "openvswitch",
+};
+
+static void do_setup(struct net_device *netdev)
+{
+       ether_setup(netdev);
+
+       netdev->netdev_ops = &internal_dev_netdev_ops;
+
+       netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
+       netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
+       netdev->destructor = internal_dev_destructor;
+       netdev->ethtool_ops = &internal_dev_ethtool_ops;
+       netdev->rtnl_link_ops = &internal_dev_link_ops;
+       netdev->tx_queue_len = 0;
+
+       netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
+                          NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
+                          NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
+
+       netdev->vlan_features = netdev->features;
+       netdev->hw_enc_features = netdev->features;
+       netdev->features |= NETIF_F_HW_VLAN_CTAG_TX;
+       netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
+
+       eth_hw_addr_random(netdev);
+}
+
+static struct vport *internal_dev_create(const struct vport_parms *parms)
+{
+       struct vport *vport;
+       struct netdev_vport *netdev_vport;
+       struct internal_dev *internal_dev;
+       int err;
+
+       vport = ovs_vport_alloc(sizeof(struct netdev_vport),
+                               &ovs_internal_vport_ops, parms);
+       if (IS_ERR(vport)) {
+               err = PTR_ERR(vport);
+               goto error;
+       }
+
+       netdev_vport = netdev_vport_priv(vport);
+
+       netdev_vport->dev = alloc_netdev(sizeof(struct internal_dev),
+                                        parms->name, NET_NAME_UNKNOWN,
+                                        do_setup);
+       if (!netdev_vport->dev) {
+               err = -ENOMEM;
+               goto error_free_vport;
+       }
+
+       dev_net_set(netdev_vport->dev, ovs_dp_get_net(vport->dp));
+       internal_dev = internal_dev_priv(netdev_vport->dev);
+       internal_dev->vport = vport;
+
+       /* Restrict bridge port to current netns. */
+       if (vport->port_no == OVSP_LOCAL)
+               netdev_vport->dev->features |= NETIF_F_NETNS_LOCAL;
+
+       rtnl_lock();
+       err = register_netdevice(netdev_vport->dev);
+       if (err)
+               goto error_free_netdev;
+
+       dev_set_promiscuity(netdev_vport->dev, 1);
+       rtnl_unlock();
+       netif_start_queue(netdev_vport->dev);
+
+       return vport;
+
+error_free_netdev:
+       rtnl_unlock();
+       free_netdev(netdev_vport->dev);
+error_free_vport:
+       ovs_vport_free(vport);
+error:
+       return ERR_PTR(err);
+}
+
+static void internal_dev_destroy(struct vport *vport)
+{
+       struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
+
+       netif_stop_queue(netdev_vport->dev);
+       rtnl_lock();
+       dev_set_promiscuity(netdev_vport->dev, -1);
+
+       /* unregister_netdevice() waits for an RCU grace period. */
+       unregister_netdevice(netdev_vport->dev);
+
+       rtnl_unlock();
+}
+
+static int internal_dev_recv(struct vport *vport, struct sk_buff *skb)
+{
+       struct net_device *netdev = netdev_vport_priv(vport)->dev;
+       int len;
+
+       if (unlikely(!(netdev->flags & IFF_UP))) {
+               kfree_skb(skb);
+               return 0;
+       }
+
+       len = skb->len;
+
+       skb_dst_drop(skb);
+       nf_reset(skb);
+       secpath_reset(skb);
+
+       skb->dev = netdev;
+       skb->pkt_type = PACKET_HOST;
+       skb->protocol = eth_type_trans(skb, netdev);
+       skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
+
+       netif_rx(skb);
+
+       return len;
+}
+
+static struct vport_ops ovs_internal_vport_ops = {
+       .type           = OVS_VPORT_TYPE_INTERNAL,
+       .create         = internal_dev_create,
+       .destroy        = internal_dev_destroy,
+       .get_name       = ovs_netdev_get_name,
+       .send           = internal_dev_recv,
+};
+
+int ovs_is_internal_dev(const struct net_device *netdev)
+{
+       return netdev->netdev_ops == &internal_dev_netdev_ops;
+}
+
+struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
+{
+       if (!ovs_is_internal_dev(netdev))
+               return NULL;
+
+       return internal_dev_priv(netdev)->vport;
+}
+
+int ovs_internal_dev_rtnl_link_register(void)
+{
+       int err;
+
+       err = rtnl_link_register(&internal_dev_link_ops);
+       if (err < 0)
+               return err;
+
+       err = ovs_vport_ops_register(&ovs_internal_vport_ops);
+       if (err < 0)
+               rtnl_link_unregister(&internal_dev_link_ops);
+
+       return err;
+}
+
+void ovs_internal_dev_rtnl_link_unregister(void)
+{
+       ovs_vport_ops_unregister(&ovs_internal_vport_ops);
+       rtnl_link_unregister(&internal_dev_link_ops);
+}