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 netdev_tx_t
  47 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 NETDEV_TX_OK;
  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 void internal_dev_destructor(struct net_device *dev)
  93 {
  94         struct vport *vport = ovs_internal_dev_get_vport(dev);
  95
  96         ovs_vport_free(vport);
  97 }
  98
  99 static void
 100 internal_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
 101 {
 102         int i;
 103
 104         memset(stats, 0, sizeof(*stats));
 105         stats->rx_errors  = dev->stats.rx_errors;
 106         stats->tx_errors  = dev->stats.tx_errors;
 107         stats->tx_dropped = dev->stats.tx_dropped;
 108         stats->rx_dropped = dev->stats.rx_dropped;
 109
 110         for_each_possible_cpu(i) {
 111                 const struct pcpu_sw_netstats *percpu_stats;
 112                 struct pcpu_sw_netstats local_stats;
 113                 unsigned int start;
 114
 115                 percpu_stats = per_cpu_ptr(dev->tstats, i);
 116
 117                 do {
 118                         start = u64_stats_fetch_begin_irq(&percpu_stats->syncp);
 119                         local_stats = *percpu_stats;
 120                 } while (u64_stats_fetch_retry_irq(&percpu_stats->syncp, start));
 121
 122                 stats->rx_bytes         += local_stats.rx_bytes;
 123                 stats->rx_packets       += local_stats.rx_packets;
 124                 stats->tx_bytes         += local_stats.tx_bytes;
 125                 stats->tx_packets       += local_stats.tx_packets;
 126         }
 127 }
 128
 129 static const struct net_device_ops internal_dev_netdev_ops = {
 130         .ndo_open = internal_dev_open,
 131         .ndo_stop = internal_dev_stop,
 132         .ndo_start_xmit = internal_dev_xmit,
 133         .ndo_set_mac_address = eth_mac_addr,
 134         .ndo_get_stats64 = internal_get_stats,
 135 };
 136
 137 static struct rtnl_link_ops internal_dev_link_ops __read_mostly = {
 138         .kind = "openvswitch",
 139 };
 140
 141 static void do_setup(struct net_device *netdev)
 142 {
 143         ether_setup(netdev);
 144
 145         netdev->max_mtu = ETH_MAX_MTU;
 146
 147         netdev->netdev_ops = &internal_dev_netdev_ops;
 148
 149         netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
 150         netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_OPENVSWITCH |
 151                               IFF_NO_QUEUE;
 152         netdev->needs_free_netdev = true;
 153         netdev->priv_destructor = internal_dev_destructor;
 154         netdev->ethtool_ops = &internal_dev_ethtool_ops;
 155         netdev->rtnl_link_ops = &internal_dev_link_ops;
 156
 157         netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
 158                            NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
 159                            NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
 160
 161         netdev->vlan_features = netdev->features;
 162         netdev->hw_enc_features = netdev->features;
 163         netdev->features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
 164         netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
 165
 166         eth_hw_addr_random(netdev);
 167 }
 168
 169 static struct vport *internal_dev_create(const struct vport_parms *parms)
 170 {
 171         struct vport *vport;
 172         struct internal_dev *internal_dev;
 173         struct net_device *dev;
 174         int err;
 175         bool free_vport = true;
 176
 177         vport = ovs_vport_alloc(0, &ovs_internal_vport_ops, parms);
 178         if (IS_ERR(vport)) {
 179                 err = PTR_ERR(vport);
 180                 goto error;
 181         }
 182
 183         dev = alloc_netdev(sizeof(struct internal_dev),
 184                            parms->name, NET_NAME_USER, do_setup);
 185         vport->dev = dev;
 186         if (!vport->dev) {
 187                 err = -ENOMEM;
 188                 goto error_free_vport;
 189         }
 190         vport->dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
 191         if (!vport->dev->tstats) {
 192                 err = -ENOMEM;
 193                 goto error_free_netdev;
 194         }
 195
 196         dev_net_set(vport->dev, ovs_dp_get_net(vport->dp));
 197         internal_dev = internal_dev_priv(vport->dev);
 198         internal_dev->vport = vport;
 199
 200         /* Restrict bridge port to current netns. */
 201         if (vport->port_no == OVSP_LOCAL)
 202                 vport->dev->features |= NETIF_F_NETNS_LOCAL;
 203
 204         rtnl_lock();
 205         err = register_netdevice(vport->dev);
 206         if (err) {
 207                 free_vport = false;
 208                 goto error_unlock;
 209         }
 210
 211         dev_set_promiscuity(vport->dev, 1);
 212         rtnl_unlock();
 213         netif_start_queue(vport->dev);
 214
 215         return vport;
 216
 217 error_unlock:
 218         rtnl_unlock();
 219         free_percpu(dev->tstats);
 220 error_free_netdev:
 221         free_netdev(dev);
 222 error_free_vport:
 223         if (free_vport)
 224                 ovs_vport_free(vport);
 225 error:
 226         return ERR_PTR(err);
 227 }
 228
 229 static void internal_dev_destroy(struct vport *vport)
 230 {
 231         netif_stop_queue(vport->dev);
 232         rtnl_lock();
 233         dev_set_promiscuity(vport->dev, -1);
 234
 235         /* unregister_netdevice() waits for an RCU grace period. */
 236         unregister_netdevice(vport->dev);
 237         free_percpu(vport->dev->tstats);
 238         rtnl_unlock();
 239 }
 240
 241 static netdev_tx_t internal_dev_recv(struct sk_buff *skb)
 242 {
 243         struct net_device *netdev = skb->dev;
 244         struct pcpu_sw_netstats *stats;
 245
 246         if (unlikely(!(netdev->flags & IFF_UP))) {
 247                 kfree_skb(skb);
 248                 netdev->stats.rx_dropped++;
 249                 return NETDEV_TX_OK;
 250         }
 251
 252         skb_dst_drop(skb);
 253         nf_reset(skb);
 254         secpath_reset(skb);
 255
 256         skb->pkt_type = PACKET_HOST;
 257         skb->protocol = eth_type_trans(skb, netdev);
 258         skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
 259
 260         stats = this_cpu_ptr(netdev->tstats);
 261         u64_stats_update_begin(&stats->syncp);
 262         stats->rx_packets++;
 263         stats->rx_bytes += skb->len;
 264         u64_stats_update_end(&stats->syncp);
 265
 266         netif_rx(skb);
 267         return NETDEV_TX_OK;
 268 }
 269
 270 static struct vport_ops ovs_internal_vport_ops = {
 271         .type           = OVS_VPORT_TYPE_INTERNAL,
 272         .create         = internal_dev_create,
 273         .destroy        = internal_dev_destroy,
 274         .send           = internal_dev_recv,
 275 };
 276
 277 int ovs_is_internal_dev(const struct net_device *netdev)
 278 {
 279         return netdev->netdev_ops == &internal_dev_netdev_ops;
 280 }
 281
 282 struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
 283 {
 284         if (!ovs_is_internal_dev(netdev))
 285                 return NULL;
 286
 287         return internal_dev_priv(netdev)->vport;
 288 }
 289
 290 int ovs_internal_dev_rtnl_link_register(void)
 291 {
 292         int err;
 293
 294         err = rtnl_link_register(&internal_dev_link_ops);
 295         if (err < 0)
 296                 return err;
 297
 298         err = ovs_vport_ops_register(&ovs_internal_vport_ops);
 299         if (err < 0)
 300                 rtnl_link_unregister(&internal_dev_link_ops);
 301
 302         return err;
 303 }
 304
 305 void ovs_internal_dev_rtnl_link_unregister(void)
 306 {
 307         ovs_vport_ops_unregister(&ovs_internal_vport_ops);
 308         rtnl_link_unregister(&internal_dev_link_ops);
 309 }