2 * Copyright (c) 2007-2012 Nicira, Inc.
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.
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.
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
19 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
21 #include <linux/if_arp.h>
22 #include <linux/if_bridge.h>
23 #include <linux/if_vlan.h>
24 #include <linux/kernel.h>
25 #include <linux/llc.h>
26 #include <linux/rtnetlink.h>
27 #include <linux/skbuff.h>
34 #include "vport-internal_dev.h"
35 #include "vport-netdev.h"
37 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,37) && \
38 !defined(HAVE_VLAN_BUG_WORKAROUND)
39 #include <linux/module.h>
41 static int vlan_tso __read_mostly;
42 module_param(vlan_tso, int, 0644);
43 MODULE_PARM_DESC(vlan_tso, "Enable TSO for VLAN packets");
48 static void netdev_port_receive(struct vport *vport, struct sk_buff *skb);
50 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
51 /* Called with rcu_read_lock and bottom-halves disabled. */
52 static rx_handler_result_t netdev_frame_hook(struct sk_buff **pskb)
54 struct sk_buff *skb = *pskb;
57 if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
58 return RX_HANDLER_PASS;
60 vport = ovs_netdev_get_vport(skb->dev);
62 netdev_port_receive(vport, skb);
64 return RX_HANDLER_CONSUMED;
66 #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
67 /* Called with rcu_read_lock and bottom-halves disabled. */
68 static struct sk_buff *netdev_frame_hook(struct sk_buff *skb)
72 if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
75 vport = ovs_netdev_get_vport(skb->dev);
77 netdev_port_receive(vport, skb);
81 #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22)
83 * Used as br_handle_frame_hook. (Cannot run bridge at the same time, even on
84 * different set of devices!)
86 /* Called with rcu_read_lock and bottom-halves disabled. */
87 static struct sk_buff *netdev_frame_hook(struct net_bridge_port *p,
90 netdev_port_receive((struct vport *)p, skb);
93 #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)
95 * Used as br_handle_frame_hook. (Cannot run bridge at the same time, even on
96 * different set of devices!)
98 /* Called with rcu_read_lock and bottom-halves disabled. */
99 static int netdev_frame_hook(struct net_bridge_port *p, struct sk_buff **pskb)
101 netdev_port_receive((struct vport *)p, *pskb);
108 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
109 static int netdev_init(void) { return 0; }
110 static void netdev_exit(void) { }
112 static int netdev_init(void)
114 /* Hook into callback used by the bridge to intercept packets.
115 * Parasites we are. */
116 br_handle_frame_hook = netdev_frame_hook;
121 static void netdev_exit(void)
123 br_handle_frame_hook = NULL;
127 static struct vport *netdev_create(const struct vport_parms *parms)
130 struct netdev_vport *netdev_vport;
133 vport = ovs_vport_alloc(sizeof(struct netdev_vport),
134 &ovs_netdev_vport_ops, parms);
136 err = PTR_ERR(vport);
140 netdev_vport = netdev_vport_priv(vport);
142 netdev_vport->dev = dev_get_by_name(ovs_dp_get_net(vport->dp), parms->name);
143 if (!netdev_vport->dev) {
145 goto error_free_vport;
148 if (netdev_vport->dev->flags & IFF_LOOPBACK ||
149 netdev_vport->dev->type != ARPHRD_ETHER ||
150 ovs_is_internal_dev(netdev_vport->dev)) {
155 err = netdev_rx_handler_register(netdev_vport->dev, netdev_frame_hook,
160 dev_set_promiscuity(netdev_vport->dev, 1);
161 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
162 dev_disable_lro(netdev_vport->dev);
164 netdev_vport->dev->priv_flags |= IFF_OVS_DATAPATH;
169 dev_put(netdev_vport->dev);
171 ovs_vport_free(vport);
176 static void netdev_destroy(struct vport *vport)
178 struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
180 netdev_vport->dev->priv_flags &= ~IFF_OVS_DATAPATH;
181 netdev_rx_handler_unregister(netdev_vport->dev);
182 dev_set_promiscuity(netdev_vport->dev, -1);
186 dev_put(netdev_vport->dev);
187 ovs_vport_free(vport);
190 int ovs_netdev_set_addr(struct vport *vport, const unsigned char *addr)
192 struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
195 sa.sa_family = ARPHRD_ETHER;
196 memcpy(sa.sa_data, addr, ETH_ALEN);
198 return dev_set_mac_address(netdev_vport->dev, &sa);
201 const char *ovs_netdev_get_name(const struct vport *vport)
203 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
204 return netdev_vport->dev->name;
207 const unsigned char *ovs_netdev_get_addr(const struct vport *vport)
209 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
210 return netdev_vport->dev->dev_addr;
213 struct kobject *ovs_netdev_get_kobj(const struct vport *vport)
215 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
216 return &netdev_vport->dev->NETDEV_DEV_MEMBER.kobj;
219 unsigned ovs_netdev_get_dev_flags(const struct vport *vport)
221 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
222 return dev_get_flags(netdev_vport->dev);
225 int ovs_netdev_is_running(const struct vport *vport)
227 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
228 return netif_running(netdev_vport->dev);
231 unsigned char ovs_netdev_get_operstate(const struct vport *vport)
233 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
234 return netdev_vport->dev->operstate;
237 int ovs_netdev_get_ifindex(const struct vport *vport)
239 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
240 return netdev_vport->dev->ifindex;
243 int ovs_netdev_get_mtu(const struct vport *vport)
245 const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
246 return netdev_vport->dev->mtu;
249 /* Must be called with rcu_read_lock. */
250 static void netdev_port_receive(struct vport *vport, struct sk_buff *skb)
252 if (unlikely(!vport)) {
257 /* Make our own copy of the packet. Otherwise we will mangle the
258 * packet for anyone who came before us (e.g. tcpdump via AF_PACKET).
259 * (No one comes after us, since we tell handle_bridge() that we took
261 skb = skb_share_check(skb, GFP_ATOMIC);
265 skb_push(skb, ETH_HLEN);
267 if (unlikely(compute_ip_summed(skb, false))) {
271 vlan_copy_skb_tci(skb);
273 ovs_vport_receive(vport, skb);
276 static unsigned int packet_length(const struct sk_buff *skb)
278 unsigned int length = skb->len - ETH_HLEN;
280 if (skb->protocol == htons(ETH_P_8021Q))
286 static bool dev_supports_vlan_tx(struct net_device *dev)
288 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,37)
289 /* Software fallback means every device supports vlan_tci on TX. */
291 #elif defined(HAVE_VLAN_BUG_WORKAROUND)
292 return dev->features & NETIF_F_HW_VLAN_TX;
294 /* Assume that the driver is buggy. */
299 static int netdev_send(struct vport *vport, struct sk_buff *skb)
301 struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
302 int mtu = netdev_vport->dev->mtu;
305 if (unlikely(packet_length(skb) > mtu && !skb_is_gso(skb))) {
306 net_warn_ratelimited("%s: dropped over-mtu packet: %d > %d\n",
307 ovs_dp_name(vport->dp),
308 packet_length(skb), mtu);
312 if (unlikely(skb_warn_if_lro(skb)))
315 skb->dev = netdev_vport->dev;
316 forward_ip_summed(skb, true);
318 if (vlan_tx_tag_present(skb) && !dev_supports_vlan_tx(skb->dev)) {
321 features = netif_skb_features(skb);
324 features &= ~(NETIF_F_TSO | NETIF_F_TSO6 |
325 NETIF_F_UFO | NETIF_F_FSO);
327 if (netif_needs_gso(skb, features)) {
328 struct sk_buff *nskb;
330 nskb = skb_gso_segment(skb, features);
332 if (unlikely(skb_cloned(skb) &&
333 pskb_expand_head(skb, 0, 0, GFP_ATOMIC))) {
338 skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
354 skb = __vlan_put_tag(skb, vlan_tx_tag_get(skb));
357 vlan_set_tci(skb, 0);
368 skb = __vlan_put_tag(skb, vlan_tx_tag_get(skb));
371 vlan_set_tci(skb, 0);
381 ovs_vport_record_error(vport, VPORT_E_TX_DROPPED);
385 /* Returns null if this device is not attached to a datapath. */
386 struct vport *ovs_netdev_get_vport(struct net_device *dev)
388 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
389 #if IFF_OVS_DATAPATH != 0
390 if (likely(dev->priv_flags & IFF_OVS_DATAPATH))
392 if (likely(rcu_access_pointer(dev->rx_handler) == netdev_frame_hook))
394 return (struct vport *)rcu_dereference_rtnl(dev->rx_handler_data);
398 return (struct vport *)rcu_dereference_rtnl(dev->br_port);
402 const struct vport_ops ovs_netdev_vport_ops = {
403 .type = OVS_VPORT_TYPE_NETDEV,
404 .flags = VPORT_F_REQUIRED,
407 .create = netdev_create,
408 .destroy = netdev_destroy,
409 .set_addr = ovs_netdev_set_addr,
410 .get_name = ovs_netdev_get_name,
411 .get_addr = ovs_netdev_get_addr,
412 .get_kobj = ovs_netdev_get_kobj,
413 .get_dev_flags = ovs_netdev_get_dev_flags,
414 .is_running = ovs_netdev_is_running,
415 .get_operstate = ovs_netdev_get_operstate,
416 .get_ifindex = ovs_netdev_get_ifindex,
417 .get_mtu = ovs_netdev_get_mtu,
421 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,36)
423 * In kernels earlier than 2.6.36, Open vSwitch cannot safely coexist with the
424 * Linux bridge module, because there is only a single bridge hook function and
425 * only a single br_port member in struct net_device, so this prevents loading
426 * both bridge and openvswitch at the same time.
428 BRIDGE_MUTUAL_EXCLUSION;