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 #include <linux/etherdevice.h>
21 #include <linux/if_vlan.h>
22 #include <linux/jhash.h>
23 #include <linux/kernel.h>
24 #include <linux/list.h>
25 #include <linux/mutex.h>
26 #include <linux/percpu.h>
27 #include <linux/rcupdate.h>
28 #include <linux/rtnetlink.h>
29 #include <linux/compat.h>
30 #include <linux/version.h>
31 #include <net/net_namespace.h>
35 #include "vport-internal_dev.h"
37 /* List of statically compiled vport implementations. Don't forget to also
38 * add yours to the list at the bottom of vport.h. */
39 static const struct vport_ops *base_vport_ops_list[] = {
40 &ovs_netdev_vport_ops,
41 &ovs_internal_vport_ops,
44 &ovs_gre_ft_vport_ops,
46 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,26)
47 &ovs_capwap_vport_ops,
52 static const struct vport_ops **vport_ops_list;
53 static int n_vport_types;
55 /* Protected by RCU read lock for reading, RTNL lock for writing. */
56 static struct hlist_head *dev_table;
57 #define VPORT_HASH_BUCKETS 1024
60 * ovs_vport_init - initialize vport subsystem
62 * Called at module load time to initialize the vport subsystem and any
63 * compiled in vport types.
65 int ovs_vport_init(void)
70 dev_table = kzalloc(VPORT_HASH_BUCKETS * sizeof(struct hlist_head),
77 vport_ops_list = kmalloc(ARRAY_SIZE(base_vport_ops_list) *
78 sizeof(struct vport_ops *), GFP_KERNEL);
79 if (!vport_ops_list) {
84 for (i = 0; i < ARRAY_SIZE(base_vport_ops_list); i++) {
85 const struct vport_ops *new_ops = base_vport_ops_list[i];
88 err = new_ops->init();
93 vport_ops_list[n_vport_types++] = new_ops;
94 else if (new_ops->flags & VPORT_F_REQUIRED) {
109 * ovs_vport_exit - shutdown vport subsystem
111 * Called at module exit time to shutdown the vport subsystem and any
112 * initialized vport types.
114 void ovs_vport_exit(void)
118 for (i = 0; i < n_vport_types; i++) {
119 if (vport_ops_list[i]->exit)
120 vport_ops_list[i]->exit();
123 kfree(vport_ops_list);
127 static struct hlist_head *hash_bucket(struct net *net, const char *name)
129 unsigned int hash = jhash(name, strlen(name), (unsigned long) net);
130 return &dev_table[hash & (VPORT_HASH_BUCKETS - 1)];
134 * ovs_vport_locate - find a port that has already been created
136 * @name: name of port to find
138 * Must be called with RTNL or RCU read lock.
140 struct vport *ovs_vport_locate(struct net *net, const char *name)
142 struct hlist_head *bucket = hash_bucket(net, name);
144 struct hlist_node *node;
146 hlist_for_each_entry_rcu(vport, node, bucket, hash_node)
147 if (!strcmp(name, vport->ops->get_name(vport)) &&
148 net_eq(ovs_dp_get_net(vport->dp), net))
155 * ovs_vport_alloc - allocate and initialize new vport
157 * @priv_size: Size of private data area to allocate.
158 * @ops: vport device ops
160 * Allocate and initialize a new vport defined by @ops. The vport will contain
161 * a private data area of size @priv_size that can be accessed using
162 * vport_priv(). vports that are no longer needed should be released with
165 struct vport *ovs_vport_alloc(int priv_size, const struct vport_ops *ops,
166 const struct vport_parms *parms)
171 alloc_size = sizeof(struct vport);
173 alloc_size = ALIGN(alloc_size, VPORT_ALIGN);
174 alloc_size += priv_size;
177 vport = kzalloc(alloc_size, GFP_KERNEL);
179 return ERR_PTR(-ENOMEM);
181 vport->dp = parms->dp;
182 vport->port_no = parms->port_no;
183 vport->upcall_portid = parms->upcall_portid;
185 INIT_HLIST_NODE(&vport->dp_hash_node);
187 vport->percpu_stats = alloc_percpu(struct vport_percpu_stats);
188 if (!vport->percpu_stats) {
190 return ERR_PTR(-ENOMEM);
193 spin_lock_init(&vport->stats_lock);
199 * ovs_vport_free - uninitialize and free vport
201 * @vport: vport to free
203 * Frees a vport allocated with ovs_vport_alloc() when it is no longer needed.
205 * The caller must ensure that an RCU grace period has passed since the last
206 * time @vport was in a datapath.
208 void ovs_vport_free(struct vport *vport)
210 free_percpu(vport->percpu_stats);
215 * ovs_vport_add - add vport device (for kernel callers)
217 * @parms: Information about new vport.
219 * Creates a new vport with the specified configuration (which is dependent on
220 * device type). RTNL lock must be held.
222 struct vport *ovs_vport_add(const struct vport_parms *parms)
230 for (i = 0; i < n_vport_types; i++) {
231 if (vport_ops_list[i]->type == parms->type) {
232 struct hlist_head *bucket;
234 vport = vport_ops_list[i]->create(parms);
236 err = PTR_ERR(vport);
240 bucket = hash_bucket(ovs_dp_get_net(vport->dp),
241 vport->ops->get_name(vport));
242 hlist_add_head_rcu(&vport->hash_node, bucket);
254 * ovs_vport_set_options - modify existing vport device (for kernel callers)
256 * @vport: vport to modify.
257 * @port: New configuration.
259 * Modifies an existing device with the specified configuration (which is
260 * dependent on device type). RTNL lock must be held.
262 int ovs_vport_set_options(struct vport *vport, struct nlattr *options)
266 if (!vport->ops->set_options)
268 return vport->ops->set_options(vport, options);
272 * ovs_vport_del - delete existing vport device
274 * @vport: vport to delete.
276 * Detaches @vport from its datapath and destroys it. It is possible to fail
277 * for reasons such as lack of memory. RTNL lock must be held.
279 void ovs_vport_del(struct vport *vport)
283 hlist_del_rcu(&vport->hash_node);
285 vport->ops->destroy(vport);
289 * ovs_vport_set_addr - set device Ethernet address (for kernel callers)
291 * @vport: vport on which to set Ethernet address.
292 * @addr: New address.
294 * Sets the Ethernet address of the given device. Some devices may not support
295 * setting the Ethernet address, in which case the result will always be
296 * -EOPNOTSUPP. RTNL lock must be held.
298 int ovs_vport_set_addr(struct vport *vport, const unsigned char *addr)
302 if (!is_valid_ether_addr(addr))
303 return -EADDRNOTAVAIL;
305 if (vport->ops->set_addr)
306 return vport->ops->set_addr(vport, addr);
312 * ovs_vport_set_stats - sets offset device stats
314 * @vport: vport on which to set stats
315 * @stats: stats to set
317 * Provides a set of transmit, receive, and error stats to be added as an
318 * offset to the collect data when stats are retreived. Some devices may not
319 * support setting the stats, in which case the result will always be
322 * Must be called with RTNL lock.
324 void ovs_vport_set_stats(struct vport *vport, struct ovs_vport_stats *stats)
328 spin_lock_bh(&vport->stats_lock);
329 vport->offset_stats = *stats;
330 spin_unlock_bh(&vport->stats_lock);
334 * ovs_vport_get_stats - retrieve device stats
336 * @vport: vport from which to retrieve the stats
337 * @stats: location to store stats
339 * Retrieves transmit, receive, and error stats for the given device.
341 * Must be called with RTNL lock or rcu_read_lock.
343 void ovs_vport_get_stats(struct vport *vport, struct ovs_vport_stats *stats)
347 /* We potentially have 3 sources of stats that need to be
348 * combined: those we have collected (split into err_stats and
349 * percpu_stats), offset_stats from set_stats(), and device
350 * error stats from netdev->get_stats() (for errors that happen
351 * downstream and therefore aren't reported through our
352 * vport_record_error() function).
353 * Stats from first two sources are merged and reported by ovs over
354 * OVS_VPORT_ATTR_STATS.
355 * netdev-stats can be directly read over netlink-ioctl.
358 spin_lock_bh(&vport->stats_lock);
360 *stats = vport->offset_stats;
362 stats->rx_errors += vport->err_stats.rx_errors;
363 stats->tx_errors += vport->err_stats.tx_errors;
364 stats->tx_dropped += vport->err_stats.tx_dropped;
365 stats->rx_dropped += vport->err_stats.rx_dropped;
367 spin_unlock_bh(&vport->stats_lock);
369 for_each_possible_cpu(i) {
370 const struct vport_percpu_stats *percpu_stats;
371 struct vport_percpu_stats local_stats;
374 percpu_stats = per_cpu_ptr(vport->percpu_stats, i);
377 start = u64_stats_fetch_begin_bh(&percpu_stats->sync);
378 local_stats = *percpu_stats;
379 } while (u64_stats_fetch_retry_bh(&percpu_stats->sync, start));
381 stats->rx_bytes += local_stats.rx_bytes;
382 stats->rx_packets += local_stats.rx_packets;
383 stats->tx_bytes += local_stats.tx_bytes;
384 stats->tx_packets += local_stats.tx_packets;
389 * ovs_vport_get_options - retrieve device options
391 * @vport: vport from which to retrieve the options.
392 * @skb: sk_buff where options should be appended.
394 * Retrieves the configuration of the given device, appending an
395 * %OVS_VPORT_ATTR_OPTIONS attribute that in turn contains nested
396 * vport-specific attributes to @skb.
398 * Returns 0 if successful, -EMSGSIZE if @skb has insufficient room, or another
399 * negative error code if a real error occurred. If an error occurs, @skb is
402 * Must be called with RTNL lock or rcu_read_lock.
404 int ovs_vport_get_options(const struct vport *vport, struct sk_buff *skb)
408 nla = nla_nest_start(skb, OVS_VPORT_ATTR_OPTIONS);
412 if (vport->ops->get_options) {
413 int err = vport->ops->get_options(vport, skb);
415 nla_nest_cancel(skb, nla);
420 nla_nest_end(skb, nla);
425 * ovs_vport_receive - pass up received packet to the datapath for processing
427 * @vport: vport that received the packet
428 * @skb: skb that was received
430 * Must be called with rcu_read_lock. The packet cannot be shared and
431 * skb->data should point to the Ethernet header. The caller must have already
432 * called compute_ip_summed() to initialize the checksumming fields.
434 void ovs_vport_receive(struct vport *vport, struct sk_buff *skb)
436 struct vport_percpu_stats *stats;
438 stats = this_cpu_ptr(vport->percpu_stats);
439 u64_stats_update_begin(&stats->sync);
441 stats->rx_bytes += skb->len;
442 u64_stats_update_end(&stats->sync);
444 if (!(vport->ops->flags & VPORT_F_FLOW))
445 OVS_CB(skb)->flow = NULL;
447 if (!(vport->ops->flags & VPORT_F_TUN_ID))
448 OVS_CB(skb)->tun_key = NULL;
450 ovs_dp_process_received_packet(vport, skb);
454 * ovs_vport_send - send a packet on a device
456 * @vport: vport on which to send the packet
459 * Sends the given packet and returns the length of data sent. Either RTNL
460 * lock or rcu_read_lock must be held.
462 int ovs_vport_send(struct vport *vport, struct sk_buff *skb)
464 int sent = vport->ops->send(vport, skb);
467 struct vport_percpu_stats *stats;
469 stats = this_cpu_ptr(vport->percpu_stats);
471 u64_stats_update_begin(&stats->sync);
473 stats->tx_bytes += sent;
474 u64_stats_update_end(&stats->sync);
480 * ovs_vport_record_error - indicate device error to generic stats layer
482 * @vport: vport that encountered the error
483 * @err_type: one of enum vport_err_type types to indicate the error type
485 * If using the vport generic stats layer indicate that an error of the given
488 void ovs_vport_record_error(struct vport *vport, enum vport_err_type err_type)
490 spin_lock(&vport->stats_lock);
493 case VPORT_E_RX_DROPPED:
494 vport->err_stats.rx_dropped++;
497 case VPORT_E_RX_ERROR:
498 vport->err_stats.rx_errors++;
501 case VPORT_E_TX_DROPPED:
502 vport->err_stats.tx_dropped++;
505 case VPORT_E_TX_ERROR:
506 vport->err_stats.tx_errors++;
510 spin_unlock(&vport->stats_lock);