datapath/vport-internal_dev.c

   1 /*
   2  * Copyright (c) 2007-2012 Nicira, Inc.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of version 2 of the GNU General Public
   6  * License as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful, but
   9  * WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License
  14  * along with this program; if not, write to the Free Software
  15  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
  16  * 02110-1301, USA
  17  */
  18
  19 #include <linux/hardirq.h>
  20 #include <linux/if_vlan.h>
  21 #include <linux/kernel.h>
  22 #include <linux/netdevice.h>
  23 #include <linux/etherdevice.h>
  24 #include <linux/ethtool.h>
  25 #include <linux/netdev_features.h>
  26 #include <linux/skbuff.h>
  27 #include <linux/version.h>
  28
  29 #include <net/dst.h>
  30 #include <net/xfrm.h>
  31
  32 #include "datapath.h"
  33 #include "vlan.h"
  34 #include "vport-internal_dev.h"
  35 #include "vport-netdev.h"
  36
  37 struct internal_dev {
  38         struct vport *vport;
  39 };
  40
  41 static struct vport_ops ovs_internal_vport_ops;
  42
  43 static struct internal_dev *internal_dev_priv(struct net_device *netdev)
  44 {
  45         return netdev_priv(netdev);
  46 }
  47
  48 /* This function is only called by the kernel network layer.*/
  49 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
  50 static struct rtnl_link_stats64 *internal_dev_get_stats(struct net_device *netdev,
  51                                                         struct rtnl_link_stats64 *stats)
  52 {
  53 #else
  54 static struct net_device_stats *internal_dev_sys_stats(struct net_device *netdev)
  55 {
  56         struct net_device_stats *stats = &netdev->stats;
  57 #endif
  58         struct vport *vport = ovs_internal_dev_get_vport(netdev);
  59         struct ovs_vport_stats vport_stats;
  60
  61         ovs_vport_get_stats(vport, &vport_stats);
  62
  63         /* The tx and rx stats need to be swapped because the
  64          * switch and host OS have opposite perspectives.
  65          */
  66         stats->rx_packets       = vport_stats.tx_packets;
  67         stats->tx_packets       = vport_stats.rx_packets;
  68         stats->rx_bytes         = vport_stats.tx_bytes;
  69         stats->tx_bytes         = vport_stats.rx_bytes;
  70         stats->rx_errors        = vport_stats.tx_errors;
  71         stats->tx_errors        = vport_stats.rx_errors;
  72         stats->rx_dropped       = vport_stats.tx_dropped;
  73         stats->tx_dropped       = vport_stats.rx_dropped;
  74
  75         return stats;
  76 }
  77
  78 /* Called with rcu_read_lock_bh. */
  79 static int internal_dev_xmit(struct sk_buff *skb, struct net_device *netdev)
  80 {
  81         rcu_read_lock();
  82         ovs_vport_receive(internal_dev_priv(netdev)->vport, skb, NULL);
  83         rcu_read_unlock();
  84         return 0;
  85 }
  86
  87 static int internal_dev_open(struct net_device *netdev)
  88 {
  89         netif_start_queue(netdev);
  90         return 0;
  91 }
  92
  93 static int internal_dev_stop(struct net_device *netdev)
  94 {
  95         netif_stop_queue(netdev);
  96         return 0;
  97 }
  98
  99 static void internal_dev_getinfo(struct net_device *netdev,
 100                                  struct ethtool_drvinfo *info)
 101 {
 102         strlcpy(info->driver, "openvswitch", sizeof(info->driver));
 103 }
 104
 105 static const struct ethtool_ops internal_dev_ethtool_ops = {
 106         .get_drvinfo    = internal_dev_getinfo,
 107         .get_link       = ethtool_op_get_link,
 108 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,39)
 109         .get_sg         = ethtool_op_get_sg,
 110         .set_sg         = ethtool_op_set_sg,
 111         .get_tx_csum    = ethtool_op_get_tx_csum,
 112         .set_tx_csum    = ethtool_op_set_tx_hw_csum,
 113         .get_tso        = ethtool_op_get_tso,
 114         .set_tso        = ethtool_op_set_tso,
 115 #endif
 116 };
 117
 118 static int internal_dev_change_mtu(struct net_device *netdev, int new_mtu)
 119 {
 120         if (new_mtu < 68)
 121                 return -EINVAL;
 122
 123         netdev->mtu = new_mtu;
 124         return 0;
 125 }
 126
 127 static void internal_dev_destructor(struct net_device *dev)
 128 {
 129         struct vport *vport = ovs_internal_dev_get_vport(dev);
 130
 131         ovs_vport_free(vport);
 132         free_netdev(dev);
 133 }
 134
 135 static const struct net_device_ops internal_dev_netdev_ops = {
 136         .ndo_open = internal_dev_open,
 137         .ndo_stop = internal_dev_stop,
 138         .ndo_start_xmit = internal_dev_xmit,
 139         .ndo_set_mac_address = eth_mac_addr,
 140         .ndo_change_mtu = internal_dev_change_mtu,
 141 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
 142         .ndo_get_stats64 = internal_dev_get_stats,
 143 #else
 144         .ndo_get_stats = internal_dev_sys_stats,
 145 #endif
 146 };
 147
 148 static struct rtnl_link_ops internal_dev_link_ops __read_mostly = {
 149         .kind = "openvswitch",
 150 };
 151
 152 static void do_setup(struct net_device *netdev)
 153 {
 154         ether_setup(netdev);
 155
 156         netdev->netdev_ops = &internal_dev_netdev_ops;
 157
 158         netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
 159         netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
 160         netdev->destructor = internal_dev_destructor;
 161         netdev->ethtool_ops = &internal_dev_ethtool_ops;
 162         netdev->rtnl_link_ops = &internal_dev_link_ops;
 163         netdev->tx_queue_len = 0;
 164
 165         netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
 166                            NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
 167                            NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
 168
 169         netdev->vlan_features = netdev->features;
 170         netdev->features |= NETIF_F_HW_VLAN_CTAG_TX;
 171
 172 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
 173         netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
 174 #endif
 175
 176 #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0)
 177         netdev->hw_enc_features = netdev->features;
 178 #endif
 179
 180         eth_hw_addr_random(netdev);
 181 }
 182
 183 static struct vport *internal_dev_create(const struct vport_parms *parms)
 184 {
 185         struct vport *vport;
 186         struct netdev_vport *netdev_vport;
 187         struct internal_dev *internal_dev;
 188         int err;
 189
 190         vport = ovs_vport_alloc(sizeof(struct netdev_vport),
 191                                 &ovs_internal_vport_ops, parms);
 192         if (IS_ERR(vport)) {
 193                 err = PTR_ERR(vport);
 194                 goto error;
 195         }
 196
 197         netdev_vport = netdev_vport_priv(vport);
 198
 199         netdev_vport->dev = alloc_netdev(sizeof(struct internal_dev),
 200                                          parms->name, NET_NAME_UNKNOWN, do_setup);
 201         if (!netdev_vport->dev) {
 202                 err = -ENOMEM;
 203                 goto error_free_vport;
 204         }
 205
 206         dev_net_set(netdev_vport->dev, ovs_dp_get_net(vport->dp));
 207         internal_dev = internal_dev_priv(netdev_vport->dev);
 208         internal_dev->vport = vport;
 209
 210         /* Restrict bridge port to current netns. */
 211         if (vport->port_no == OVSP_LOCAL)
 212                 netdev_vport->dev->features |= NETIF_F_NETNS_LOCAL;
 213
 214         rtnl_lock();
 215         err = register_netdevice(netdev_vport->dev);
 216         if (err)
 217                 goto error_free_netdev;
 218
 219         dev_set_promiscuity(netdev_vport->dev, 1);
 220         rtnl_unlock();
 221         netif_start_queue(netdev_vport->dev);
 222
 223         return vport;
 224
 225 error_free_netdev:
 226         rtnl_unlock();
 227         free_netdev(netdev_vport->dev);
 228 error_free_vport:
 229         ovs_vport_free(vport);
 230 error:
 231         return ERR_PTR(err);
 232 }
 233
 234 static void internal_dev_destroy(struct vport *vport)
 235 {
 236         struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
 237
 238         netif_stop_queue(netdev_vport->dev);
 239         rtnl_lock();
 240         dev_set_promiscuity(netdev_vport->dev, -1);
 241
 242         /* unregister_netdevice() waits for an RCU grace period. */
 243         unregister_netdevice(netdev_vport->dev);
 244
 245         rtnl_unlock();
 246 }
 247
 248 static int internal_dev_recv(struct vport *vport, struct sk_buff *skb)
 249 {
 250         struct net_device *netdev = netdev_vport_priv(vport)->dev;
 251         int len;
 252
 253         if (unlikely(!(netdev->flags & IFF_UP))) {
 254                 kfree_skb(skb);
 255                 return 0;
 256         }
 257
 258 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,37)
 259         if (skb_vlan_tag_present(skb)) {
 260                 if (unlikely(!vlan_insert_tag_set_proto(skb,
 261                                                         skb->vlan_proto,
 262                                                         skb_vlan_tag_get(skb))))
 263                         return 0;
 264
 265                 if (skb->ip_summed == CHECKSUM_COMPLETE)
 266                         skb->csum = csum_add(skb->csum,
 267                                              csum_partial(skb->data + (2 * ETH_ALEN),
 268                                                           VLAN_HLEN, 0));
 269
 270                 vlan_set_tci(skb, 0);
 271         }
 272 #endif
 273
 274         len = skb->len;
 275
 276         skb_dst_drop(skb);
 277         nf_reset(skb);
 278         secpath_reset(skb);
 279
 280         skb->dev = netdev;
 281         skb->pkt_type = PACKET_HOST;
 282         skb->protocol = eth_type_trans(skb, netdev);
 283         skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
 284
 285         netif_rx(skb);
 286
 287         return len;
 288 }
 289
 290 static struct vport_ops ovs_internal_vport_ops = {
 291         .type           = OVS_VPORT_TYPE_INTERNAL,
 292         .create         = internal_dev_create,
 293         .destroy        = internal_dev_destroy,
 294         .get_name       = ovs_netdev_get_name,
 295         .send           = internal_dev_recv,
 296 };
 297
 298 int ovs_is_internal_dev(const struct net_device *netdev)
 299 {
 300         return netdev->netdev_ops == &internal_dev_netdev_ops;
 301 }
 302
 303 struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
 304 {
 305         if (!ovs_is_internal_dev(netdev))
 306                 return NULL;
 307
 308         return internal_dev_priv(netdev)->vport;
 309 }
 310
 311 int ovs_internal_dev_rtnl_link_register(void)
 312 {
 313         int err;
 314
 315         err = rtnl_link_register(&internal_dev_link_ops);
 316         if (err < 0)
 317                 return err;
 318
 319         err = ovs_vport_ops_register(&ovs_internal_vport_ops);
 320         if (err < 0)
 321                 rtnl_link_unregister(&internal_dev_link_ops);
 322
 323         return err;
 324 }
 325
 326 void ovs_internal_dev_rtnl_link_unregister(void)
 327 {
 328         ovs_vport_ops_unregister(&ovs_internal_vport_ops);
 329         rtnl_link_unregister(&internal_dev_link_ops);
 330 }