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/if_vlan.h>
  20 #include <linux/kernel.h>
  21 #include <linux/netdevice.h>
  22 #include <linux/etherdevice.h>
  23 #include <linux/ethtool.h>
  24 #include <linux/skbuff.h>
  25
  26 #include <net/dst.h>
  27 #include <net/xfrm.h>
  28 #include <net/rtnetlink.h>
  29
  30 #include "datapath.h"
  31 #include "vport-internal_dev.h"
  32 #include "vport-netdev.h"
  33
  34 struct internal_dev {
  35         struct vport *vport;
  36 };
  37
  38 static struct vport_ops ovs_internal_vport_ops;
  39
  40 static struct internal_dev *internal_dev_priv(struct net_device *netdev)
  41 {
  42         return netdev_priv(netdev);
  43 }
  44
  45 /* Called with rcu_read_lock_bh. */
  46 static int internal_dev_xmit(struct sk_buff *skb, struct net_device *netdev)
  47 {
  48         int len, err;
  49
  50         len = skb->len;
  51         rcu_read_lock();
  52         err = ovs_vport_receive(internal_dev_priv(netdev)->vport, skb, NULL);
  53         rcu_read_unlock();
  54
  55         if (likely(!err)) {
  56                 struct pcpu_sw_netstats *tstats = this_cpu_ptr(netdev->tstats);
  57
  58                 u64_stats_update_begin(&tstats->syncp);
  59                 tstats->tx_bytes += len;
  60                 tstats->tx_packets++;
  61                 u64_stats_update_end(&tstats->syncp);
  62         } else {
  63                 netdev->stats.tx_errors++;
  64         }
  65         return 0;
  66 }
  67
  68 static int internal_dev_open(struct net_device *netdev)
  69 {
  70         netif_start_queue(netdev);
  71         return 0;
  72 }
  73
  74 static int internal_dev_stop(struct net_device *netdev)
  75 {
  76         netif_stop_queue(netdev);
  77         return 0;
  78 }
  79
  80 static void internal_dev_getinfo(struct net_device *netdev,
  81                                  struct ethtool_drvinfo *info)
  82 {
  83         strlcpy(info->driver, "openvswitch", sizeof(info->driver));
  84 }
  85
  86 static const struct ethtool_ops internal_dev_ethtool_ops = {
  87         .get_drvinfo    = internal_dev_getinfo,
  88         .get_link       = ethtool_op_get_link,
  89 };
  90
  91 static void internal_dev_destructor(struct net_device *dev)
  92 {
  93         struct vport *vport = ovs_internal_dev_get_vport(dev);
  94
  95         ovs_vport_free(vport);
  96 }
  97
  98 static void
  99 internal_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
 100 {
 101         int i;
 102
 103         memset(stats, 0, sizeof(*stats));
 104         stats->rx_errors  = dev->stats.rx_errors;
 105         stats->tx_errors  = dev->stats.tx_errors;
 106         stats->tx_dropped = dev->stats.tx_dropped;
 107         stats->rx_dropped = dev->stats.rx_dropped;
 108
 109         for_each_possible_cpu(i) {
 110                 const struct pcpu_sw_netstats *percpu_stats;
 111                 struct pcpu_sw_netstats local_stats;
 112                 unsigned int start;
 113
 114                 percpu_stats = per_cpu_ptr(dev->tstats, i);
 115
 116                 do {
 117                         start = u64_stats_fetch_begin_irq(&percpu_stats->syncp);
 118                         local_stats = *percpu_stats;
 119                 } while (u64_stats_fetch_retry_irq(&percpu_stats->syncp, start));
 120
 121                 stats->rx_bytes         += local_stats.rx_bytes;
 122                 stats->rx_packets       += local_stats.rx_packets;
 123                 stats->tx_bytes         += local_stats.tx_bytes;
 124                 stats->tx_packets       += local_stats.tx_packets;
 125         }
 126 }
 127
 128 static const struct net_device_ops internal_dev_netdev_ops = {
 129         .ndo_open = internal_dev_open,
 130         .ndo_stop = internal_dev_stop,
 131         .ndo_start_xmit = internal_dev_xmit,
 132         .ndo_set_mac_address = eth_mac_addr,
 133         .ndo_get_stats64 = internal_get_stats,
 134 };
 135
 136 static struct rtnl_link_ops internal_dev_link_ops __read_mostly = {
 137         .kind = "openvswitch",
 138 };
 139
 140 static void do_setup(struct net_device *netdev)
 141 {
 142         ether_setup(netdev);
 143
 144         netdev->max_mtu = ETH_MAX_MTU;
 145
 146         netdev->netdev_ops = &internal_dev_netdev_ops;
 147
 148         netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
 149         netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_OPENVSWITCH |
 150                               IFF_NO_QUEUE;
 151         netdev->needs_free_netdev = true;
 152         netdev->priv_destructor = internal_dev_destructor;
 153         netdev->ethtool_ops = &internal_dev_ethtool_ops;
 154         netdev->rtnl_link_ops = &internal_dev_link_ops;
 155
 156         netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
 157                            NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
 158                            NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
 159
 160         netdev->vlan_features = netdev->features;
 161         netdev->hw_enc_features = netdev->features;
 162         netdev->features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
 163         netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
 164
 165         eth_hw_addr_random(netdev);
 166 }
 167
 168 static struct vport *internal_dev_create(const struct vport_parms *parms)
 169 {
 170         struct vport *vport;
 171         struct internal_dev *internal_dev;
 172         struct net_device *dev;
 173         int err;
 174         bool free_vport = true;
 175
 176         vport = ovs_vport_alloc(0, &ovs_internal_vport_ops, parms);
 177         if (IS_ERR(vport)) {
 178                 err = PTR_ERR(vport);
 179                 goto error;
 180         }
 181
 182         dev = alloc_netdev(sizeof(struct internal_dev),
 183                            parms->name, NET_NAME_USER, do_setup);
 184         vport->dev = dev;
 185         if (!vport->dev) {
 186                 err = -ENOMEM;
 187                 goto error_free_vport;
 188         }
 189         vport->dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
 190         if (!vport->dev->tstats) {
 191                 err = -ENOMEM;
 192                 goto error_free_netdev;
 193         }
 194
 195         dev_net_set(vport->dev, ovs_dp_get_net(vport->dp));
 196         internal_dev = internal_dev_priv(vport->dev);
 197         internal_dev->vport = vport;
 198
 199         /* Restrict bridge port to current netns. */
 200         if (vport->port_no == OVSP_LOCAL)
 201                 vport->dev->features |= NETIF_F_NETNS_LOCAL;
 202
 203         rtnl_lock();
 204         err = register_netdevice(vport->dev);
 205         if (err) {
 206                 free_vport = false;
 207                 goto error_unlock;
 208         }
 209
 210         dev_set_promiscuity(vport->dev, 1);
 211         rtnl_unlock();
 212         netif_start_queue(vport->dev);
 213
 214         return vport;
 215
 216 error_unlock:
 217         rtnl_unlock();
 218         free_percpu(dev->tstats);
 219 error_free_netdev:
 220         free_netdev(dev);
 221 error_free_vport:
 222         if (free_vport)
 223                 ovs_vport_free(vport);
 224 error:
 225         return ERR_PTR(err);
 226 }
 227
 228 static void internal_dev_destroy(struct vport *vport)
 229 {
 230         netif_stop_queue(vport->dev);
 231         rtnl_lock();
 232         dev_set_promiscuity(vport->dev, -1);
 233
 234         /* unregister_netdevice() waits for an RCU grace period. */
 235         unregister_netdevice(vport->dev);
 236         free_percpu(vport->dev->tstats);
 237         rtnl_unlock();
 238 }
 239
 240 static netdev_tx_t internal_dev_recv(struct sk_buff *skb)
 241 {
 242         struct net_device *netdev = skb->dev;
 243         struct pcpu_sw_netstats *stats;
 244
 245         if (unlikely(!(netdev->flags & IFF_UP))) {
 246                 kfree_skb(skb);
 247                 netdev->stats.rx_dropped++;
 248                 return NETDEV_TX_OK;
 249         }
 250
 251         skb_dst_drop(skb);
 252         nf_reset(skb);
 253         secpath_reset(skb);
 254
 255         skb->pkt_type = PACKET_HOST;
 256         skb->protocol = eth_type_trans(skb, netdev);
 257         skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
 258
 259         stats = this_cpu_ptr(netdev->tstats);
 260         u64_stats_update_begin(&stats->syncp);
 261         stats->rx_packets++;
 262         stats->rx_bytes += skb->len;
 263         u64_stats_update_end(&stats->syncp);
 264
 265         netif_rx(skb);
 266         return NETDEV_TX_OK;
 267 }
 268
 269 static struct vport_ops ovs_internal_vport_ops = {
 270         .type           = OVS_VPORT_TYPE_INTERNAL,
 271         .create         = internal_dev_create,
 272         .destroy        = internal_dev_destroy,
 273         .send           = internal_dev_recv,
 274 };
 275
 276 int ovs_is_internal_dev(const struct net_device *netdev)
 277 {
 278         return netdev->netdev_ops == &internal_dev_netdev_ops;
 279 }
 280
 281 struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
 282 {
 283         if (!ovs_is_internal_dev(netdev))
 284                 return NULL;
 285
 286         return internal_dev_priv(netdev)->vport;
 287 }
 288
 289 int ovs_internal_dev_rtnl_link_register(void)
 290 {
 291         int err;
 292
 293         err = rtnl_link_register(&internal_dev_link_ops);
 294         if (err < 0)
 295                 return err;
 296
 297         err = ovs_vport_ops_register(&ovs_internal_vport_ops);
 298         if (err < 0)
 299                 rtnl_link_unregister(&internal_dev_link_ops);
 300
 301         return err;
 302 }
 303
 304 void ovs_internal_dev_rtnl_link_unregister(void)
 305 {
 306         ovs_vport_ops_unregister(&ovs_internal_vport_ops);
 307         rtnl_link_unregister(&internal_dev_link_ops);
 308 }