br_if.c 12.9 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
/*
 *	Userspace interface
 *	Linux ethernet bridge
 *
 *	Authors:
 *	Lennert Buytenhek		<buytenh@gnu.org>
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 */

#include <linux/kernel.h>
#include <linux/netdevice.h>
16
#include <linux/etherdevice.h>
17
#include <linux/netpoll.h>
Linus Torvalds's avatar
Linus Torvalds committed
18
19
20
21
22
#include <linux/ethtool.h>
#include <linux/if_arp.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/rtnetlink.h>
23
#include <linux/if_ether.h>
24
#include <linux/slab.h>
Linus Torvalds's avatar
Linus Torvalds committed
25
#include <net/sock.h>
26
#include <linux/if_vlan.h>
27
#include <net/switchdev.h>
Linus Torvalds's avatar
Linus Torvalds committed
28
29
30
31
32
33
34

#include "br_private.h"

/*
 * Determine initial path cost based on speed.
 * using recommendations from 802.1d standard
 *
35
 * Since driver might sleep need to not be holding any locks.
Linus Torvalds's avatar
Linus Torvalds committed
36
 */
37
static int port_cost(struct net_device *dev)
Linus Torvalds's avatar
Linus Torvalds committed
38
{
39
	struct ethtool_cmd ecmd;
40

41
	if (!__ethtool_get_settings(dev, &ecmd)) {
42
43
44
45
46
47
48
49
50
		switch (ethtool_cmd_speed(&ecmd)) {
		case SPEED_10000:
			return 2;
		case SPEED_1000:
			return 4;
		case SPEED_100:
			return 19;
		case SPEED_10:
			return 100;
Linus Torvalds's avatar
Linus Torvalds committed
51
52
53
54
55
56
57
58
59
60
61
62
63
		}
	}

	/* Old silly heuristics based on name */
	if (!strncmp(dev->name, "lec", 3))
		return 7;

	if (!strncmp(dev->name, "plip", 4))
		return 2500;

	return 100;	/* assume old 10Mbps */
}

64

tanxiaojun's avatar
tanxiaojun committed
65
/* Check for port carrier transitions. */
66
void br_port_carrier_check(struct net_bridge_port *p)
67
{
68
69
	struct net_device *dev = p->dev;
	struct net_bridge *br = p->br;
Stephen Hemminger's avatar
Stephen Hemminger committed
70

71
72
	if (!(p->flags & BR_ADMIN_COST) &&
	    netif_running(dev) && netif_oper_up(dev))
Stephen Hemminger's avatar
Stephen Hemminger committed
73
74
		p->path_cost = port_cost(dev);

75
76
77
78
	if (!netif_running(br->dev))
		return;

	spin_lock_bh(&br->lock);
79
	if (netif_running(dev) && netif_oper_up(dev)) {
80
81
82
83
84
		if (p->state == BR_STATE_DISABLED)
			br_stp_enable_port(p);
	} else {
		if (p->state != BR_STATE_DISABLED)
			br_stp_disable_port(p);
85
	}
86
	spin_unlock_bh(&br->lock);
87
88
}

89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
static void br_port_set_promisc(struct net_bridge_port *p)
{
	int err = 0;

	if (br_promisc_port(p))
		return;

	err = dev_set_promiscuity(p->dev, 1);
	if (err)
		return;

	br_fdb_unsync_static(p->br, p);
	p->flags |= BR_PROMISC;
}

static void br_port_clear_promisc(struct net_bridge_port *p)
{
	int err;

	/* Check if the port is already non-promisc or if it doesn't
	 * support UNICAST filtering.  Without unicast filtering support
	 * we'll end up re-enabling promisc mode anyway, so just check for
	 * it here.
	 */
	if (!br_promisc_port(p) || !(p->dev->priv_flags & IFF_UNICAST_FLT))
		return;

	/* Since we'll be clearing the promisc mode, program the port
	 * first so that we don't have interruption in traffic.
	 */
	err = br_fdb_sync_static(p->br, p);
	if (err)
		return;

	dev_set_promiscuity(p->dev, -1);
	p->flags &= ~BR_PROMISC;
}

/* When a port is added or removed or when certain port flags
 * change, this function is called to automatically manage
 * promiscuity setting of all the bridge ports.  We are always called
 * under RTNL so can skip using rcu primitives.
 */
void br_manage_promisc(struct net_bridge *br)
{
	struct net_bridge_port *p;
	bool set_all = false;

	/* If vlan filtering is disabled or bridge interface is placed
	 * into promiscuous mode, place all ports in promiscuous mode.
	 */
	if ((br->dev->flags & IFF_PROMISC) || !br_vlan_enabled(br))
		set_all = true;

	list_for_each_entry(p, &br->port_list, list) {
		if (set_all) {
			br_port_set_promisc(p);
		} else {
			/* If the number of auto-ports is <= 1, then all other
			 * ports will have their output configuration
			 * statically specified through fdbs.  Since ingress
			 * on the auto-port becomes forwarding/egress to other
			 * ports and egress configuration is statically known,
			 * we can say that ingress configuration of the
			 * auto-port is also statically known.
			 * This lets us disable promiscuous mode and write
			 * this config to hw.
			 */
157
158
			if (br->auto_cnt == 0 ||
			    (br->auto_cnt == 1 && br_auto_port(p)))
159
160
161
162
163
164
165
				br_port_clear_promisc(p);
			else
				br_port_set_promisc(p);
		}
	}
}

166
167
168
169
170
171
172
173
174
static void nbp_update_port_count(struct net_bridge *br)
{
	struct net_bridge_port *p;
	u32 cnt = 0;

	list_for_each_entry(p, &br->port_list, list) {
		if (br_auto_port(p))
			cnt++;
	}
175
176
177
178
179
180
181
182
	if (br->auto_cnt != cnt) {
		br->auto_cnt = cnt;
		br_manage_promisc(br);
	}
}

static void nbp_delete_promisc(struct net_bridge_port *p)
{
183
	/* If port is currently promiscuous, unset promiscuity.
184
185
186
187
188
189
190
191
	 * Otherwise, it is a static port so remove all addresses
	 * from it.
	 */
	dev_set_allmulti(p->dev, -1);
	if (br_promisc_port(p))
		dev_set_promiscuity(p->dev, -1);
	else
		br_fdb_unsync_static(p->br, p);
192
193
}

194
195
196
197
198
199
200
201
202
203
204
205
206
207
static void release_nbp(struct kobject *kobj)
{
	struct net_bridge_port *p
		= container_of(kobj, struct net_bridge_port, kobj);
	kfree(p);
}

static struct kobj_type brport_ktype = {
#ifdef CONFIG_SYSFS
	.sysfs_ops = &brport_sysfs_ops,
#endif
	.release = release_nbp,
};

Linus Torvalds's avatar
Linus Torvalds committed
208
209
210
211
212
213
214
215
static void destroy_nbp(struct net_bridge_port *p)
{
	struct net_device *dev = p->dev;

	p->br = NULL;
	p->dev = NULL;
	dev_put(dev);

216
	kobject_put(&p->kobj);
Linus Torvalds's avatar
Linus Torvalds committed
217
218
219
220
221
222
223
224
225
}

static void destroy_nbp_rcu(struct rcu_head *head)
{
	struct net_bridge_port *p =
			container_of(head, struct net_bridge_port, rcu);
	destroy_nbp(p);
}

226
227
228
229
230
231
232
233
234
/* Delete port(interface) from bridge is done in two steps.
 * via RCU. First step, marks device as down. That deletes
 * all the timers and stops new packets from flowing through.
 *
 * Final cleanup doesn't occur until after all CPU's finished
 * processing packets.
 *
 * Protected from multiple admin operations by RTNL mutex
 */
Linus Torvalds's avatar
Linus Torvalds committed
235
236
237
238
239
static void del_nbp(struct net_bridge_port *p)
{
	struct net_bridge *br = p->br;
	struct net_device *dev = p->dev;

240
	sysfs_remove_link(br->ifobj, p->dev->name);
241

242
	nbp_delete_promisc(p);
Linus Torvalds's avatar
Linus Torvalds committed
243
244
245
246
247

	spin_lock_bh(&br->lock);
	br_stp_disable_port(p);
	spin_unlock_bh(&br->lock);

248
249
	br_ifinfo_notify(RTM_DELLINK, p);

Linus Torvalds's avatar
Linus Torvalds committed
250
251
	list_del_rcu(&p->list);

252
	nbp_vlan_flush(p);
253
	br_fdb_delete_by_port(br, p, 0, 1);
254
255
	switchdev_deferred_process();

256
257
	nbp_update_port_count(br);

258
259
	netdev_upper_dev_unlink(dev, br->dev);

260
261
	dev->priv_flags &= ~IFF_BRIDGE_PORT;

262
	netdev_rx_handler_unregister(dev);
263

264
265
	br_multicast_del_port(p);

266
	kobject_uevent(&p->kobj, KOBJ_REMOVE);
267
268
	kobject_del(&p->kobj);

Herbert Xu's avatar
Herbert Xu committed
269
270
	br_netpoll_disable(p);

Linus Torvalds's avatar
Linus Torvalds committed
271
272
273
	call_rcu(&p->rcu, destroy_nbp_rcu);
}

274
275
/* Delete bridge device */
void br_dev_delete(struct net_device *dev, struct list_head *head)
Linus Torvalds's avatar
Linus Torvalds committed
276
{
277
	struct net_bridge *br = netdev_priv(dev);
Linus Torvalds's avatar
Linus Torvalds committed
278
279
280
281
282
283
	struct net_bridge_port *p, *n;

	list_for_each_entry_safe(p, n, &br->port_list, list) {
		del_nbp(p);
	}

284
	br_fdb_delete_by_port(br, NULL, 0, 1);
285

286
	br_vlan_flush(br);
287
	br_multicast_dev_del(br);
Linus Torvalds's avatar
Linus Torvalds committed
288
289
290
	del_timer_sync(&br->gc_timer);

	br_sysfs_delbr(br->dev);
291
	unregister_netdevice_queue(br->dev, head);
Linus Torvalds's avatar
Linus Torvalds committed
292
293
294
295
296
297
298
299
300
}

/* find an available port number */
static int find_portno(struct net_bridge *br)
{
	int index;
	struct net_bridge_port *p;
	unsigned long *inuse;

Stephen Hemminger's avatar
Stephen Hemminger committed
301
	inuse = kcalloc(BITS_TO_LONGS(BR_MAX_PORTS), sizeof(unsigned long),
Linus Torvalds's avatar
Linus Torvalds committed
302
303
304
305
306
307
308
309
310
311
312
313
314
315
			GFP_KERNEL);
	if (!inuse)
		return -ENOMEM;

	set_bit(0, inuse);	/* zero is reserved */
	list_for_each_entry(p, &br->port_list, list) {
		set_bit(p->port_no, inuse);
	}
	index = find_first_zero_bit(inuse, BR_MAX_PORTS);
	kfree(inuse);

	return (index >= BR_MAX_PORTS) ? -EXFULL : index;
}

316
/* called with RTNL but without bridge lock */
317
static struct net_bridge_port *new_nbp(struct net_bridge *br,
318
				       struct net_device *dev)
Linus Torvalds's avatar
Linus Torvalds committed
319
320
321
{
	int index;
	struct net_bridge_port *p;
322

Linus Torvalds's avatar
Linus Torvalds committed
323
324
325
326
	index = find_portno(br);
	if (index < 0)
		return ERR_PTR(index);

Stephen Hemminger's avatar
Stephen Hemminger committed
327
	p = kzalloc(sizeof(*p), GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
328
329
330
331
332
333
	if (p == NULL)
		return ERR_PTR(-ENOMEM);

	p->br = br;
	dev_hold(dev);
	p->dev = dev;
334
	p->path_cost = port_cost(dev);
335
	p->priority = 0x8000 >> BR_PORT_BITS;
Linus Torvalds's avatar
Linus Torvalds committed
336
	p->port_no = index;
337
	p->flags = BR_LEARNING | BR_FLOOD;
Linus Torvalds's avatar
Linus Torvalds committed
338
	br_init_port(p);
339
	br_set_state(p, BR_STATE_DISABLED);
340
	br_stp_port_timer_init(p);
341
	br_multicast_add_port(p);
Linus Torvalds's avatar
Linus Torvalds committed
342
343
344
345

	return p;
}

346
int br_add_bridge(struct net *net, const char *name)
Linus Torvalds's avatar
Linus Torvalds committed
347
348
{
	struct net_device *dev;
349
	int res;
Linus Torvalds's avatar
Linus Torvalds committed
350

351
	dev = alloc_netdev(sizeof(struct net_bridge), name, NET_NAME_UNKNOWN,
352
353
			   br_dev_setup);

354
	if (!dev)
Linus Torvalds's avatar
Linus Torvalds committed
355
356
		return -ENOMEM;

357
	dev_net_set(dev, net);
358
	dev->rtnl_link_ops = &br_link_ops;
359

360
361
362
363
	res = register_netdev(dev);
	if (res)
		free_netdev(dev);
	return res;
Linus Torvalds's avatar
Linus Torvalds committed
364
365
}

366
int br_del_bridge(struct net *net, const char *name)
Linus Torvalds's avatar
Linus Torvalds committed
367
368
369
370
371
{
	struct net_device *dev;
	int ret = 0;

	rtnl_lock();
372
	dev = __dev_get_by_name(net, name);
373
	if (dev == NULL)
Linus Torvalds's avatar
Linus Torvalds committed
374
375
376
377
378
379
380
381
382
383
		ret =  -ENXIO; 	/* Could not find device */

	else if (!(dev->priv_flags & IFF_EBRIDGE)) {
		/* Attempt to delete non bridge device! */
		ret = -EPERM;
	}

	else if (dev->flags & IFF_UP) {
		/* Not shutdown yet. */
		ret = -EBUSY;
384
	}
Linus Torvalds's avatar
Linus Torvalds committed
385

386
	else
387
		br_dev_delete(dev, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
388
389
390
391
392

	rtnl_unlock();
	return ret;
}

393
/* MTU of the bridge pseudo-device: ETH_DATA_LEN or the minimum of the ports */
Linus Torvalds's avatar
Linus Torvalds committed
394
395
396
397
398
399
400
401
int br_min_mtu(const struct net_bridge *br)
{
	const struct net_bridge_port *p;
	int mtu = 0;

	ASSERT_RTNL();

	if (list_empty(&br->port_list))
402
		mtu = ETH_DATA_LEN;
Linus Torvalds's avatar
Linus Torvalds committed
403
404
405
406
407
408
409
410
411
	else {
		list_for_each_entry(p, &br->port_list, list) {
			if (!mtu  || p->dev->mtu < mtu)
				mtu = p->dev->mtu;
		}
	}
	return mtu;
}

412
413
414
/*
 * Recomputes features using slave's features
 */
415
416
netdev_features_t br_features_recompute(struct net_bridge *br,
	netdev_features_t features)
417
418
{
	struct net_bridge_port *p;
419
	netdev_features_t mask;
420

421
	if (list_empty(&br->port_list))
422
		return features;
423

424
	mask = features;
425
	features &= ~NETIF_F_ONE_FOR_ALL;
426
427

	list_for_each_entry(p, &br->port_list, list) {
428
429
		features = netdev_increment_features(features,
						     p->dev->features, mask);
430
	}
431
	features = netdev_add_tso_features(features, mask);
432

433
	return features;
434
435
}

Linus Torvalds's avatar
Linus Torvalds committed
436
437
438
439
440
/* called with RTNL */
int br_add_if(struct net_bridge *br, struct net_device *dev)
{
	struct net_bridge_port *p;
	int err = 0;
441
	bool changed_addr;
Linus Torvalds's avatar
Linus Torvalds committed
442

443
444
445
446
447
448
	/* Don't allow bridging non-ethernet like devices, or DSA-enabled
	 * master network devices since the bridge layer rx_handler prevents
	 * the DSA fake ethertype handler to be invoked, so we do not strip off
	 * the DSA switch tag protocol header and the bridge layer just return
	 * RX_HANDLER_CONSUMED, stopping RX processing for these frames.
	 */
449
	if ((dev->flags & IFF_LOOPBACK) ||
450
	    dev->type != ARPHRD_ETHER || dev->addr_len != ETH_ALEN ||
451
452
	    !is_valid_ether_addr(dev->dev_addr) ||
	    netdev_uses_dsa(dev))
Linus Torvalds's avatar
Linus Torvalds committed
453
454
		return -EINVAL;

455
	/* No bridging of bridges */
456
	if (dev->netdev_ops->ndo_start_xmit == br_dev_xmit)
Linus Torvalds's avatar
Linus Torvalds committed
457
458
		return -ELOOP;

459
	/* Device is already being bridged */
460
	if (br_port_exists(dev))
Linus Torvalds's avatar
Linus Torvalds committed
461
462
		return -EBUSY;

463
464
465
466
	/* No bridging devices that dislike that (e.g. wireless) */
	if (dev->priv_flags & IFF_DONT_BRIDGE)
		return -EOPNOTSUPP;

467
468
	p = new_nbp(br, dev);
	if (IS_ERR(p))
Linus Torvalds's avatar
Linus Torvalds committed
469
470
		return PTR_ERR(p);

471
472
	call_netdevice_notifiers(NETDEV_JOIN, dev);

473
	err = dev_set_allmulti(dev, 1);
474
475
476
	if (err)
		goto put_back;

477
478
	err = kobject_init_and_add(&p->kobj, &brport_ktype, &(dev->dev.kobj),
				   SYSFS_BRIDGE_PORT_ATTR);
479
480
	if (err)
		goto err1;
Linus Torvalds's avatar
Linus Torvalds committed
481

482
483
484
	err = br_sysfs_addif(p);
	if (err)
		goto err2;
Linus Torvalds's avatar
Linus Torvalds committed
485

486
	err = br_netpoll_enable(p);
487
	if (err)
Herbert Xu's avatar
Herbert Xu committed
488
489
		goto err3;

490
	err = netdev_rx_handler_register(dev, br_handle_frame, p);
491
	if (err)
492
		goto err4;
493

494
495
496
	dev->priv_flags |= IFF_BRIDGE_PORT;

	err = netdev_master_upper_dev_link(dev, br->dev);
497
	if (err)
498
		goto err5;
499

500
	dev_disable_lro(dev);
501
502
503

	list_add_rcu(&p->list, &br->port_list);

504
505
	nbp_update_port_count(br);

506
507
	netdev_update_features(br->dev);

508
509
510
	if (br->dev->needed_headroom < dev->needed_headroom)
		br->dev->needed_headroom = dev->needed_headroom;

511
512
513
	if (br_fdb_insert(br, p, dev->dev_addr, 0))
		netdev_err(dev, "failed insert local address bridge forwarding table\n");

514
515
516
	if (nbp_vlan_init(p))
		netdev_err(dev, "failed to initialize vlan filtering on this port\n");

517
	spin_lock_bh(&br->lock);
518
	changed_addr = br_stp_recalculate_bridge_id(br);
519

520
	if (netif_running(dev) && netif_oper_up(dev) &&
521
522
	    (br->dev->flags & IFF_UP))
		br_stp_enable_port(p);
523
524
	spin_unlock_bh(&br->lock);

525
526
	br_ifinfo_notify(RTM_NEWLINK, p);

527
	if (changed_addr)
528
		call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);
529

530
	dev_set_mtu(br->dev, br_min_mtu(br));
531

532
	kobject_uevent(&p->kobj, KOBJ_ADD);
Linus Torvalds's avatar
Linus Torvalds committed
533

534
	return 0;
535

536
err5:
537
538
	dev->priv_flags &= ~IFF_BRIDGE_PORT;
	netdev_rx_handler_unregister(dev);
539
540
err4:
	br_netpoll_disable(p);
Herbert Xu's avatar
Herbert Xu committed
541
542
err3:
	sysfs_remove_link(br->ifobj, p->dev->name);
543
err2:
544
	kobject_put(&p->kobj);
545
	p = NULL; /* kobject_put frees */
546
err1:
547
	dev_set_allmulti(dev, -1);
548
549
put_back:
	dev_put(dev);
550
	kfree(p);
Linus Torvalds's avatar
Linus Torvalds committed
551
552
553
554
555
556
	return err;
}

/* called with RTNL */
int br_del_if(struct net_bridge *br, struct net_device *dev)
{
557
	struct net_bridge_port *p;
558
	bool changed_addr;
559

560
	p = br_port_get_rtnl(dev);
561
	if (!p || p->br != br)
Linus Torvalds's avatar
Linus Torvalds committed
562
563
		return -EINVAL;

564
565
566
567
	/* Since more than one interface can be attached to a bridge,
	 * there still maybe an alternate path for netconsole to use;
	 * therefore there is no reason for a NETDEV_RELEASE event.
	 */
Linus Torvalds's avatar
Linus Torvalds committed
568
569
	del_nbp(p);

570
571
	dev_set_mtu(br->dev, br_min_mtu(br));

Linus Torvalds's avatar
Linus Torvalds committed
572
	spin_lock_bh(&br->lock);
573
	changed_addr = br_stp_recalculate_bridge_id(br);
Linus Torvalds's avatar
Linus Torvalds committed
574
575
	spin_unlock_bh(&br->lock);

576
577
578
	if (changed_addr)
		call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);

579
580
	netdev_update_features(br->dev);

Linus Torvalds's avatar
Linus Torvalds committed
581
582
	return 0;
}
583
584
585
586
587
588
589
590

void br_port_flags_change(struct net_bridge_port *p, unsigned long mask)
{
	struct net_bridge *br = p->br;

	if (mask & BR_AUTO_MASK)
		nbp_update_port_count(br);
}