]> git.proxmox.com Git - mirror_ovs.git/blob - datapath/vport-internal_dev.c
datapath: Enable tunnel GSO features.
[mirror_ovs.git] / 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 internal_dev *internal_dev_priv(struct net_device *netdev)
42 {
43 return netdev_priv(netdev);
44 }
45
46 /* This function is only called by the kernel network layer.*/
47 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
48 static struct rtnl_link_stats64 *internal_dev_get_stats(struct net_device *netdev,
49 struct rtnl_link_stats64 *stats)
50 {
51 #else
52 static struct net_device_stats *internal_dev_sys_stats(struct net_device *netdev)
53 {
54 struct net_device_stats *stats = &netdev->stats;
55 #endif
56 struct vport *vport = ovs_internal_dev_get_vport(netdev);
57 struct ovs_vport_stats vport_stats;
58
59 ovs_vport_get_stats(vport, &vport_stats);
60
61 /* The tx and rx stats need to be swapped because the
62 * switch and host OS have opposite perspectives. */
63 stats->rx_packets = vport_stats.tx_packets;
64 stats->tx_packets = vport_stats.rx_packets;
65 stats->rx_bytes = vport_stats.tx_bytes;
66 stats->tx_bytes = vport_stats.rx_bytes;
67 stats->rx_errors = vport_stats.tx_errors;
68 stats->tx_errors = vport_stats.rx_errors;
69 stats->rx_dropped = vport_stats.tx_dropped;
70 stats->tx_dropped = vport_stats.rx_dropped;
71
72 return stats;
73 }
74
75 /* Called with rcu_read_lock_bh. */
76 static int internal_dev_xmit(struct sk_buff *skb, struct net_device *netdev)
77 {
78 rcu_read_lock();
79 ovs_vport_receive(internal_dev_priv(netdev)->vport, skb, NULL);
80 rcu_read_unlock();
81 return 0;
82 }
83
84 static int internal_dev_open(struct net_device *netdev)
85 {
86 netif_start_queue(netdev);
87 return 0;
88 }
89
90 static int internal_dev_stop(struct net_device *netdev)
91 {
92 netif_stop_queue(netdev);
93 return 0;
94 }
95
96 static void internal_dev_getinfo(struct net_device *netdev,
97 struct ethtool_drvinfo *info)
98 {
99 strlcpy(info->driver, "openvswitch", sizeof(info->driver));
100 }
101
102 static const struct ethtool_ops internal_dev_ethtool_ops = {
103 .get_drvinfo = internal_dev_getinfo,
104 .get_link = ethtool_op_get_link,
105 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,39)
106 .get_sg = ethtool_op_get_sg,
107 .set_sg = ethtool_op_set_sg,
108 .get_tx_csum = ethtool_op_get_tx_csum,
109 .set_tx_csum = ethtool_op_set_tx_hw_csum,
110 .get_tso = ethtool_op_get_tso,
111 .set_tso = ethtool_op_set_tso,
112 #endif
113 };
114
115 static int internal_dev_change_mtu(struct net_device *netdev, int new_mtu)
116 {
117 if (new_mtu < 68)
118 return -EINVAL;
119
120 netdev->mtu = new_mtu;
121 return 0;
122 }
123
124 static void internal_dev_destructor(struct net_device *dev)
125 {
126 struct vport *vport = ovs_internal_dev_get_vport(dev);
127
128 ovs_vport_free(vport);
129 free_netdev(dev);
130 }
131
132 static const struct net_device_ops internal_dev_netdev_ops = {
133 .ndo_open = internal_dev_open,
134 .ndo_stop = internal_dev_stop,
135 .ndo_start_xmit = internal_dev_xmit,
136 .ndo_set_mac_address = eth_mac_addr,
137 .ndo_change_mtu = internal_dev_change_mtu,
138 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
139 .ndo_get_stats64 = internal_dev_get_stats,
140 #else
141 .ndo_get_stats = internal_dev_sys_stats,
142 #endif
143 };
144
145 static void do_setup(struct net_device *netdev)
146 {
147 ether_setup(netdev);
148
149 netdev->netdev_ops = &internal_dev_netdev_ops;
150
151 netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
152 netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
153 netdev->destructor = internal_dev_destructor;
154 SET_ETHTOOL_OPS(netdev, &internal_dev_ethtool_ops);
155 netdev->tx_queue_len = 0;
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->features |= NETIF_F_HW_VLAN_CTAG_TX;
163
164 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
165 netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
166 #endif
167
168 #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0)
169 netdev->hw_enc_features = netdev->features;
170 #endif
171
172 eth_hw_addr_random(netdev);
173 }
174
175 static struct vport *internal_dev_create(const struct vport_parms *parms)
176 {
177 struct vport *vport;
178 struct netdev_vport *netdev_vport;
179 struct internal_dev *internal_dev;
180 int err;
181
182 vport = ovs_vport_alloc(sizeof(struct netdev_vport),
183 &ovs_internal_vport_ops, parms);
184 if (IS_ERR(vport)) {
185 err = PTR_ERR(vport);
186 goto error;
187 }
188
189 netdev_vport = netdev_vport_priv(vport);
190
191 netdev_vport->dev = alloc_netdev(sizeof(struct internal_dev),
192 parms->name, do_setup);
193 if (!netdev_vport->dev) {
194 err = -ENOMEM;
195 goto error_free_vport;
196 }
197
198 dev_net_set(netdev_vport->dev, ovs_dp_get_net(vport->dp));
199 internal_dev = internal_dev_priv(netdev_vport->dev);
200 internal_dev->vport = vport;
201
202 /* Restrict bridge port to current netns. */
203 if (vport->port_no == OVSP_LOCAL)
204 netdev_vport->dev->features |= NETIF_F_NETNS_LOCAL;
205
206 rtnl_lock();
207 err = register_netdevice(netdev_vport->dev);
208 if (err)
209 goto error_free_netdev;
210
211 dev_set_promiscuity(netdev_vport->dev, 1);
212 rtnl_unlock();
213 netif_start_queue(netdev_vport->dev);
214
215 return vport;
216
217 error_free_netdev:
218 rtnl_unlock();
219 free_netdev(netdev_vport->dev);
220 error_free_vport:
221 ovs_vport_free(vport);
222 error:
223 return ERR_PTR(err);
224 }
225
226 static void internal_dev_destroy(struct vport *vport)
227 {
228 struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
229
230 netif_stop_queue(netdev_vport->dev);
231 rtnl_lock();
232 dev_set_promiscuity(netdev_vport->dev, -1);
233
234 /* unregister_netdevice() waits for an RCU grace period. */
235 unregister_netdevice(netdev_vport->dev);
236
237 rtnl_unlock();
238 }
239
240 static int internal_dev_recv(struct vport *vport, struct sk_buff *skb)
241 {
242 struct net_device *netdev = netdev_vport_priv(vport)->dev;
243 int len;
244
245 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,37)
246 if (vlan_tx_tag_present(skb)) {
247 if (unlikely(!__vlan_put_tag(skb,
248 skb->vlan_proto,
249 vlan_tx_tag_get(skb))))
250 return 0;
251
252 if (skb->ip_summed == CHECKSUM_COMPLETE)
253 skb->csum = csum_add(skb->csum,
254 csum_partial(skb->data + (2 * ETH_ALEN),
255 VLAN_HLEN, 0));
256
257 vlan_set_tci(skb, 0);
258 }
259 #endif
260
261 len = skb->len;
262
263 skb_dst_drop(skb);
264 nf_reset(skb);
265 secpath_reset(skb);
266
267 skb->dev = netdev;
268 skb->pkt_type = PACKET_HOST;
269 skb->protocol = eth_type_trans(skb, netdev);
270 skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
271
272 netif_rx(skb);
273
274 return len;
275 }
276
277 const struct vport_ops ovs_internal_vport_ops = {
278 .type = OVS_VPORT_TYPE_INTERNAL,
279 .create = internal_dev_create,
280 .destroy = internal_dev_destroy,
281 .get_name = ovs_netdev_get_name,
282 .send = internal_dev_recv,
283 };
284
285 int ovs_is_internal_dev(const struct net_device *netdev)
286 {
287 return netdev->netdev_ops == &internal_dev_netdev_ops;
288 }
289
290 struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
291 {
292 if (!ovs_is_internal_dev(netdev))
293 return NULL;
294
295 return internal_dev_priv(netdev)->vport;
296 }