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
35struct internal_dev {
36	struct vport *vport;
37};
38
39static struct vport_ops ovs_internal_vport_ops;
40
41static 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. */
47static 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
69static int internal_dev_open(struct net_device *netdev)
70{
71	netif_start_queue(netdev);
72	return 0;
73}
74
75static int internal_dev_stop(struct net_device *netdev)
76{
77	netif_stop_queue(netdev);
78	return 0;
79}
80
81static void internal_dev_getinfo(struct net_device *netdev,
82				 struct ethtool_drvinfo *info)
83{
84	strlcpy(info->driver, "openvswitch", sizeof(info->driver));
85}
86
87static const struct ethtool_ops internal_dev_ethtool_ops = {
88	.get_drvinfo	= internal_dev_getinfo,
89	.get_link	= ethtool_op_get_link,
90};
91
92static 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
101static 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
109static struct rtnl_link_stats64 *
110internal_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
141static const struct net_device_ops internal_dev_netdev_ops = {
142	.ndo_open = internal_dev_open,
143	.ndo_stop = internal_dev_stop,
144	.ndo_start_xmit = internal_dev_xmit,
145	.ndo_set_mac_address = eth_mac_addr,
146	.ndo_change_mtu = internal_dev_change_mtu,
147	.ndo_get_stats64 = internal_get_stats,
148};
149
150static struct rtnl_link_ops internal_dev_link_ops __read_mostly = {
151	.kind = "openvswitch",
152};
153
154static void do_setup(struct net_device *netdev)
155{
156	ether_setup(netdev);
157
158	netdev->netdev_ops = &internal_dev_netdev_ops;
159
160	netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
161	netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_OPENVSWITCH;
162	netdev->destructor = internal_dev_destructor;
163	netdev->ethtool_ops = &internal_dev_ethtool_ops;
164	netdev->rtnl_link_ops = &internal_dev_link_ops;
165	netdev->tx_queue_len = 0;
166
167	netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
168			   NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
169			   NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
170
171	netdev->vlan_features = netdev->features;
172	netdev->hw_enc_features = netdev->features;
173	netdev->features |= NETIF_F_HW_VLAN_CTAG_TX;
174	netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
175
176	eth_hw_addr_random(netdev);
177}
178
179static struct vport *internal_dev_create(const struct vport_parms *parms)
180{
181	struct vport *vport;
182	struct internal_dev *internal_dev;
183	int err;
184
185	vport = ovs_vport_alloc(0, &ovs_internal_vport_ops, parms);
186	if (IS_ERR(vport)) {
187		err = PTR_ERR(vport);
188		goto error;
189	}
190
191	vport->dev = alloc_netdev(sizeof(struct internal_dev),
192				  parms->name, NET_NAME_UNKNOWN, do_setup);
193	if (!vport->dev) {
194		err = -ENOMEM;
195		goto error_free_vport;
196	}
197	vport->dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
198	if (!vport->dev->tstats) {
199		err = -ENOMEM;
200		goto error_free_netdev;
201	}
202
203	dev_net_set(vport->dev, ovs_dp_get_net(vport->dp));
204	internal_dev = internal_dev_priv(vport->dev);
205	internal_dev->vport = vport;
206
207	/* Restrict bridge port to current netns. */
208	if (vport->port_no == OVSP_LOCAL)
209		vport->dev->features |= NETIF_F_NETNS_LOCAL;
210
211	rtnl_lock();
212	err = register_netdevice(vport->dev);
213	if (err)
214		goto error_unlock;
215
216	dev_set_promiscuity(vport->dev, 1);
217	rtnl_unlock();
218	netif_start_queue(vport->dev);
219
220	return vport;
221
222error_unlock:
223	rtnl_unlock();
224	free_percpu(vport->dev->tstats);
225error_free_netdev:
226	free_netdev(vport->dev);
227error_free_vport:
228	ovs_vport_free(vport);
229error:
230	return ERR_PTR(err);
231}
232
233static void internal_dev_destroy(struct vport *vport)
234{
235	netif_stop_queue(vport->dev);
236	rtnl_lock();
237	dev_set_promiscuity(vport->dev, -1);
238
239	/* unregister_netdevice() waits for an RCU grace period. */
240	unregister_netdevice(vport->dev);
241	free_percpu(vport->dev->tstats);
242	rtnl_unlock();
243}
244
245static netdev_tx_t internal_dev_recv(struct sk_buff *skb)
246{
247	struct net_device *netdev = skb->dev;
248	struct pcpu_sw_netstats *stats;
249
250	if (unlikely(!(netdev->flags & IFF_UP))) {
251		kfree_skb(skb);
252		netdev->stats.rx_dropped++;
253		return NETDEV_TX_OK;
254	}
255
256	skb_dst_drop(skb);
257	nf_reset(skb);
258	secpath_reset(skb);
259
260	skb->pkt_type = PACKET_HOST;
261	skb->protocol = eth_type_trans(skb, netdev);
262	skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
263
264	stats = this_cpu_ptr(netdev->tstats);
265	u64_stats_update_begin(&stats->syncp);
266	stats->rx_packets++;
267	stats->rx_bytes += skb->len;
268	u64_stats_update_end(&stats->syncp);
269
270	netif_rx(skb);
271	return NETDEV_TX_OK;
272}
273
274static struct vport_ops ovs_internal_vport_ops = {
275	.type		= OVS_VPORT_TYPE_INTERNAL,
276	.create		= internal_dev_create,
277	.destroy	= internal_dev_destroy,
278	.send		= internal_dev_recv,
279};
280
281int ovs_is_internal_dev(const struct net_device *netdev)
282{
283	return netdev->netdev_ops == &internal_dev_netdev_ops;
284}
285
286struct vport *ovs_internal_dev_get_vport(struct net_device *netdev)
287{
288	if (!ovs_is_internal_dev(netdev))
289		return NULL;
290
291	return internal_dev_priv(netdev)->vport;
292}
293
294int ovs_internal_dev_rtnl_link_register(void)
295{
296	int err;
297
298	err = rtnl_link_register(&internal_dev_link_ops);
299	if (err < 0)
300		return err;
301
302	err = ovs_vport_ops_register(&ovs_internal_vport_ops);
303	if (err < 0)
304		rtnl_link_unregister(&internal_dev_link_ops);
305
306	return err;
307}
308
309void ovs_internal_dev_rtnl_link_unregister(void)
310{
311	ovs_vport_ops_unregister(&ovs_internal_vport_ops);
312	rtnl_link_unregister(&internal_dev_link_ops);
313}
314