net/openvswitch/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/skbuff.h>
  26
  27 #include <net/dst.h>
  28 #include <net/xfrm.h>
  29 #include <net/rtnetlink.h>
  30
  31 #include "datapath.h"
  32 #include "vport-internal_dev.h"
  33 #include "vport-netdev.h"
  34
  35 struct internal_dev {
  36         struct vport *vport;
  37 };
  38
  39 static struct vport_ops ovs_internal_vport_ops;
  40
  41 static struct internal_dev *internal_dev_priv(struct net_device *netdev)
  42 {
  43         return netdev_priv(netdev);
  44 }
  45
  46 /* Called with rcu_read_lock_bh. */
  47 static int internal_dev_xmit(struct sk_buff *skb, struct net_device *netdev)
  48 {
  49         int len, err;
  50
  51         len = skb->len;
  52         rcu_read_lock();
  53         err = ovs_vport_receive(internal_dev_priv(netdev)->vport, skb, NULL);
  54         rcu_read_unlock();
  55
  56         if (likely(!err)) {
  57                 struct pcpu_sw_netstats *tstats = this_cpu_ptr(netdev->tstats);
  58
  59                 u64_stats_update_begin(&tstats->syncp);
  60                 tstats->tx_bytes += len;
  61                 tstats->tx_packets++;
  62                 u64_stats_update_end(&tstats->syncp);
  63         } else {
  64                 netdev->stats.tx_errors++;
  65         }
  66         return 0;
  67 }
  68
  69 static int internal_dev_open(struct net_device *netdev)
  70 {
  71         netif_start_queue(netdev);
  72         return 0;
  73 }
  74
  75 static int internal_dev_stop(struct net_device *netdev)
  76 {
  77         netif_stop_queue(netdev);
  78         return 0;
  79 }
  80
  81 static void internal_dev_getinfo(struct net_device *netdev,
  82                                  struct ethtool_drvinfo *info)
  83 {
  84         strlcpy(info->driver, "openvswitch", sizeof(info->driver));
  85 }
  86
  87 static const struct ethtool_ops internal_dev_ethtool_ops = {
  88         .get_drvinfo    = internal_dev_getinfo,
  89         .get_link       = ethtool_op_get_link,
  90 };
  91
  92 static int internal_dev_change_mtu(struct net_device *netdev, int new_mtu)
  93 {
  94         if (new_mtu < 68)
  95                 return -EINVAL;
  96
  97         netdev->mtu = new_mtu;
  98         return 0;
  99 }
 100
 101 static void internal_dev_destructor(struct net_device *dev)
 102 {
 103         struct vport *vport = ovs_internal_dev_get_vport(dev);
 104
 105         ovs_vport_free(vport);
 106         free_netdev(dev);
 107 }
 108
 109 static struct rtnl_link_stats64 *
 110 internal_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
 111 {
 112         int i;
 113
 114         memset(stats, 0, sizeof(*stats));
 115         stats->rx_errors  = dev->stats.rx_errors;
 116         stats->tx_errors  = dev->stats.tx_errors;
 117         stats->tx_dropped = dev->stats.tx_dropped;
 118         stats->rx_dropped = dev->stats.rx_dropped;
 119
 120         for_each_possible_cpu(i) {
 121                 const struct pcpu_sw_netstats *percpu_stats;
 122                 struct pcpu_sw_netstats local_stats;
 123                 unsigned int start;
 124
 125                 percpu_stats = per_cpu_ptr(dev->tstats, i);
 126
 127                 do {
 128                         start = u64_stats_fetch_begin_irq(&percpu_stats->syncp);
 129                         local_stats = *percpu_stats;
 130                 } while (u64_stats_fetch_retry_irq(&percpu_stats->syncp, start));
 131
 132                 stats->rx_bytes         += local_stats.rx_bytes;
 133                 stats->rx_packets       += local_stats.rx_packets;
 134                 stats->tx_bytes         += local_stats.tx_bytes;
 135                 stats->tx_packets       += local_stats.tx_packets;
 136         }
 137
 138         return stats;
 139 }
 140
 141 static void internal_set_rx_headroom(struct net_device *dev, int new_hr)
 142 {
 143         dev->needed_headroom = new_hr;
 144 }
 145
 146 static const struct net_device_ops internal_dev_netdev_ops = {
 147         .ndo_open = internal_dev_open,
 148         .ndo_stop = internal_dev_stop,
 149         .ndo_start_xmit = internal_dev_xmit,
 150         .ndo_set_mac_address = eth_mac_addr,
 151         .ndo_change_mtu = internal_dev_change_mtu,
 152         .ndo_get_stats64 = internal_get_stats,
 153         .ndo_set_rx_headroom = internal_set_rx_headroom,
 154 };
 155
 156 static struct rtnl_link_ops internal_dev_link_ops __read_mostly = {
 157         .kind = "openvswitch",
 158 };
 159
 160 static void do_setup(struct net_device *netdev)
 161 {
 162         ether_setup(netdev);
 163
 164         netdev->netdev_ops = &internal_dev_netdev_ops;
 165
 166         netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
 167         netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_OPENVSWITCH |
 168                               IFF_PHONY_HEADROOM | IFF_NO_QUEUE;
 169         netdev->destructor = internal_dev_destructor;
 170         netdev->ethtool_ops = &internal_dev_ethtool_ops;
 171         netdev->rtnl_link_ops = &internal_dev_link_ops;
 172
 173         netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
 174                            NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
 175                            NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
 176
 177         netdev->vlan_features = netdev->features;
 178         netdev->hw_enc_features = netdev->features;
 179         netdev->features |= NETIF_F_HW_VLAN_CTAG_TX;
 180         netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
 181
 182         eth_hw_addr_random(netdev);
 183 }
 184
 185 static struct vport *internal_dev_create(const struct vport_parms *parms)
 186 {
 187         struct vport *vport;
 188         struct internal_dev *internal_dev;
 189         int err;
 190
 191         vport = ovs_vport_alloc(0, &ovs_internal_vport_ops, parms);
 192         if (IS_ERR(vport)) {
 193                 err = PTR_ERR(vport);
 194                 goto error;
 195         }
 196
 197         vport->dev = alloc_netdev(sizeof(struct internal_dev),
 198                                   parms->name, NET_NAME_USER, do_setup);
 199         if (!vport->dev) {
 200                 err = -ENOMEM;
 201                 goto error_free_vport;
 202         }
 203         vport->dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
 204         if (!vport->dev->tstats) {
 205                 err = -ENOMEM;
 206                 goto error_free_netdev;
 207         }
 208         vport->dev->needed_headroom = vport->dp->max_headroom;
 209
 210         dev_net_set(vport->dev, ovs_dp_get_net(vport->dp));
 211         internal_dev = internal_dev_priv(vport->dev);
 212         internal_dev->vport = vport;
 213
 214         /* Restrict bridge port to current netns. */
 215         if (vport->port_no == OVSP_LOCAL)
 216                 vport->dev->features |= NETIF_F_NETNS_LOCAL;
 217
 218         rtnl_lock();
 219         err = register_netdevice(vport->dev);
 220         if (err)
 221                 goto error_unlock;
 222
 223         dev_set_promiscuity(vport->dev, 1);
 224         rtnl_unlock();
 225         netif_start_queue(vport->dev);
 226
 227         return vport;
 228
 229 error_unlock:
 230         rtnl_unlock();
 231         free_percpu(vport->dev->tstats);
 232 error_free_netdev:
 233         free_netdev(vport->dev);
 234 error_free_vport:
 235         ovs_vport_free(vport);
 236 error:
 237         return ERR_PTR(err);
 238 }
 239
 240 static void internal_dev_destroy(struct vport *vport)
 241 {
 242         netif_stop_queue(vport->dev);
 243         rtnl_lock();
 244         dev_set_promiscuity(vport->dev, -1);
 245
 246         /* unregister_netdevice() waits for an RCU grace period. */
 247         unregister_netdevice(vport->dev);
 248         free_percpu(vport->dev->tstats);
 249         rtnl_unlock();
 250 }
 251
 252 static netdev_tx_t internal_dev_recv(struct sk_buff *skb)
 253 {
 254         struct net_device *netdev = skb->dev;
 255         struct pcpu_sw_netstats *stats;
 256
 257         if (unlikely(!(netdev->flags & IFF_UP))) {
 258                 kfree_skb(skb);
 259                 netdev->stats.rx_dropped++;
 260                 return NETDEV_TX_OK;
 261         }
 262
 263         skb_dst_drop(skb);
 264         nf_reset(skb);
 265         secpath_reset(skb);
 266
 267         skb->pkt_type = PACKET_HOST;
 268         skb->protocol = eth_type_trans(skb, netdev);
 269         skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
 270
 271         stats = this_cpu_ptr(netdev->tstats);
 272         u64_stats_update_begin(&stats->syncp);
 273         stats->rx_packets++;
 274         stats->rx_bytes += skb->len;
 275         u64_stats_update_end(&stats->syncp);
 276
 277         netif_rx(skb);
 278         return NETDEV_TX_OK;
 279 }
 280
 281 static struct vport_ops ovs_internal_vport_ops = {
 282         .type           = OVS_VPORT_TYPE_INTERNAL,
 283         .create         = internal_dev_create,
 284         .destroy        = internal_dev_destroy,
 285         .send           = internal_dev_recv,
 286 };
 287
 288 int ovs_is_internal_dev(const struct net_device *netdev)
 289 {
 290         return netdev->netdev_ops == &internal_dev_netdev_ops;
 291 }
 292
 293 struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
 294 {
 295         if (!ovs_is_internal_dev(netdev))
 296                 return NULL;
 297
 298         return internal_dev_priv(netdev)->vport;
 299 }
 300
 301 int ovs_internal_dev_rtnl_link_register(void)
 302 {
 303         int err;
 304
 305         err = rtnl_link_register(&internal_dev_link_ops);
 306         if (err < 0)
 307                 return err;
 308
 309         err = ovs_vport_ops_register(&ovs_internal_vport_ops);
 310         if (err < 0)
 311                 rtnl_link_unregister(&internal_dev_link_ops);
 312
 313         return err;
 314 }
 315
 316 void ovs_internal_dev_rtnl_link_unregister(void)
 317 {
 318         ovs_vport_ops_unregister(&ovs_internal_vport_ops);
 319         rtnl_link_unregister(&internal_dev_link_ops);
 320 }