br_fdb.c 29.6 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-or-later
L
Linus Torvalds 已提交
2 3 4 5 6 7 8 9 10 11
/*
 *	Forwarding database
 *	Linux ethernet bridge
 *
 *	Authors:
 *	Lennert Buytenhek		<buytenh@gnu.org>
 */

#include <linux/kernel.h>
#include <linux/init.h>
12
#include <linux/rculist.h>
L
Linus Torvalds 已提交
13 14 15 16 17
#include <linux/spinlock.h>
#include <linux/times.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/jhash.h>
18
#include <linux/random.h>
19
#include <linux/slab.h>
A
Arun Sharma 已提交
20
#include <linux/atomic.h>
21
#include <asm/unaligned.h>
22
#include <linux/if_vlan.h>
23
#include <net/switchdev.h>
24
#include <trace/events/bridge.h>
L
Linus Torvalds 已提交
25 26
#include "br_private.h"

27 28 29 30 31 32 33
static const struct rhashtable_params br_fdb_rht_params = {
	.head_offset = offsetof(struct net_bridge_fdb_entry, rhnode),
	.key_offset = offsetof(struct net_bridge_fdb_entry, key),
	.key_len = sizeof(struct net_bridge_fdb_key),
	.automatic_shrinking = true,
};

34
static struct kmem_cache *br_fdb_cache __read_mostly;
L
Linus Torvalds 已提交
35
static int fdb_insert(struct net_bridge *br, struct net_bridge_port *source,
36
		      const unsigned char *addr, u16 vid);
37
static void fdb_notify(struct net_bridge *br,
38
		       const struct net_bridge_fdb_entry *, int, bool);
L
Linus Torvalds 已提交
39

40
int __init br_fdb_init(void)
L
Linus Torvalds 已提交
41 42 43 44
{
	br_fdb_cache = kmem_cache_create("bridge_fdb_cache",
					 sizeof(struct net_bridge_fdb_entry),
					 0,
45
					 SLAB_HWCACHE_ALIGN, NULL);
46 47 48 49
	if (!br_fdb_cache)
		return -ENOMEM;

	return 0;
L
Linus Torvalds 已提交
50 51
}

A
Andrew Morton 已提交
52
void br_fdb_fini(void)
L
Linus Torvalds 已提交
53 54 55 56
{
	kmem_cache_destroy(br_fdb_cache);
}

57 58 59 60 61 62 63 64 65
int br_fdb_hash_init(struct net_bridge *br)
{
	return rhashtable_init(&br->fdb_hash_tbl, &br_fdb_rht_params);
}

void br_fdb_hash_fini(struct net_bridge *br)
{
	rhashtable_destroy(&br->fdb_hash_tbl);
}
L
Linus Torvalds 已提交
66 67 68 69

/* if topology_changing then use forward_delay (default 15 sec)
 * otherwise keep longer (default 5 minutes)
 */
70
static inline unsigned long hold_time(const struct net_bridge *br)
L
Linus Torvalds 已提交
71 72 73 74
{
	return br->topology_change ? br->forward_delay : br->ageing_time;
}

75
static inline int has_expired(const struct net_bridge *br,
L
Linus Torvalds 已提交
76 77
				  const struct net_bridge_fdb_entry *fdb)
{
78 79 80
	return !test_bit(BR_FDB_STATIC, &fdb->flags) &&
	       !fdb->added_by_external_learn &&
	       time_before_eq(fdb->updated + hold_time(br), jiffies);
L
Linus Torvalds 已提交
81 82
}

83 84 85 86 87 88 89
static void fdb_rcu_free(struct rcu_head *head)
{
	struct net_bridge_fdb_entry *ent
		= container_of(head, struct net_bridge_fdb_entry, rcu);
	kmem_cache_free(br_fdb_cache, ent);
}

90
static struct net_bridge_fdb_entry *fdb_find_rcu(struct rhashtable *tbl,
91 92 93
						 const unsigned char *addr,
						 __u16 vid)
{
94
	struct net_bridge_fdb_key key;
95

96 97
	WARN_ON_ONCE(!rcu_read_lock_held());

98 99
	key.vlan_id = vid;
	memcpy(key.addr.addr, addr, sizeof(key.addr.addr));
100

101
	return rhashtable_lookup(tbl, &key, br_fdb_rht_params);
102 103 104 105 106 107 108 109 110
}

/* requires bridge hash_lock */
static struct net_bridge_fdb_entry *br_fdb_find(struct net_bridge *br,
						const unsigned char *addr,
						__u16 vid)
{
	struct net_bridge_fdb_entry *fdb;

111
	lockdep_assert_held_once(&br->hash_lock);
112

113
	rcu_read_lock();
114
	fdb = fdb_find_rcu(&br->fdb_hash_tbl, addr, vid);
115 116 117 118 119
	rcu_read_unlock();

	return fdb;
}

120 121 122 123 124 125 126 127 128 129 130 131 132 133
struct net_device *br_fdb_find_port(const struct net_device *br_dev,
				    const unsigned char *addr,
				    __u16 vid)
{
	struct net_bridge_fdb_entry *f;
	struct net_device *dev = NULL;
	struct net_bridge *br;

	ASSERT_RTNL();

	if (!netif_is_bridge_master(br_dev))
		return NULL;

	br = netdev_priv(br_dev);
134 135
	rcu_read_lock();
	f = br_fdb_find_rcu(br, addr, vid);
136 137
	if (f && f->dst)
		dev = f->dst->dev;
138
	rcu_read_unlock();
139 140 141 142 143

	return dev;
}
EXPORT_SYMBOL_GPL(br_fdb_find_port);

144 145 146 147
struct net_bridge_fdb_entry *br_fdb_find_rcu(struct net_bridge *br,
					     const unsigned char *addr,
					     __u16 vid)
{
148
	return fdb_find_rcu(&br->fdb_hash_tbl, addr, vid);
149 150
}

151 152 153 154 155
/* When a static FDB entry is added, the mac address from the entry is
 * added to the bridge private HW address list and all required ports
 * are then updated with the new information.
 * Called under RTNL.
 */
156
static void fdb_add_hw_addr(struct net_bridge *br, const unsigned char *addr)
157 158
{
	int err;
159
	struct net_bridge_port *p;
160 161 162 163 164 165 166 167 168 169 170 171 172

	ASSERT_RTNL();

	list_for_each_entry(p, &br->port_list, list) {
		if (!br_promisc_port(p)) {
			err = dev_uc_add(p->dev, addr);
			if (err)
				goto undo;
		}
	}

	return;
undo:
173 174 175
	list_for_each_entry_continue_reverse(p, &br->port_list, list) {
		if (!br_promisc_port(p))
			dev_uc_del(p->dev, addr);
176 177 178 179 180 181 182 183
	}
}

/* When a static FDB entry is deleted, the HW address from that entry is
 * also removed from the bridge private HW address list and updates all
 * the ports with needed information.
 * Called under RTNL.
 */
184
static void fdb_del_hw_addr(struct net_bridge *br, const unsigned char *addr)
185 186 187 188 189 190 191 192 193 194 195
{
	struct net_bridge_port *p;

	ASSERT_RTNL();

	list_for_each_entry(p, &br->port_list, list) {
		if (!br_promisc_port(p))
			dev_uc_del(p->dev, addr);
	}
}

196 197
static void fdb_delete(struct net_bridge *br, struct net_bridge_fdb_entry *f,
		       bool swdev_notify)
L
Linus Torvalds 已提交
198
{
199 200
	trace_fdb_delete(br, f);

201
	if (test_bit(BR_FDB_STATIC, &f->flags))
202
		fdb_del_hw_addr(br, f->key.addr.addr);
203

204 205 206
	hlist_del_init_rcu(&f->fdb_node);
	rhashtable_remove_fast(&br->fdb_hash_tbl, &f->rhnode,
			       br_fdb_rht_params);
207
	fdb_notify(br, f, RTM_DELNEIGH, swdev_notify);
208
	call_rcu(&f->rcu, fdb_rcu_free);
L
Linus Torvalds 已提交
209 210
}

211 212 213 214 215
/* Delete a local entry if no other port had the same address. */
static void fdb_delete_local(struct net_bridge *br,
			     const struct net_bridge_port *p,
			     struct net_bridge_fdb_entry *f)
{
216
	const unsigned char *addr = f->key.addr.addr;
217 218
	struct net_bridge_vlan_group *vg;
	const struct net_bridge_vlan *v;
219
	struct net_bridge_port *op;
220
	u16 vid = f->key.vlan_id;
221 222 223

	/* Maybe another port has same hw addr? */
	list_for_each_entry(op, &br->port_list, list) {
224
		vg = nbp_vlan_group(op);
225
		if (op != p && ether_addr_equal(op->dev->dev_addr, addr) &&
226
		    (!vid || br_vlan_find(vg, vid))) {
227
			f->dst = op;
228
			clear_bit(BR_FDB_ADDED_BY_USER, &f->flags);
229 230 231 232
			return;
		}
	}

233 234
	vg = br_vlan_group(br);
	v = br_vlan_find(vg, vid);
235 236
	/* Maybe bridge device has same hw addr? */
	if (p && ether_addr_equal(br->dev->dev_addr, addr) &&
237
	    (!vid || (v && br_vlan_should_use(v)))) {
238
		f->dst = NULL;
239
		clear_bit(BR_FDB_ADDED_BY_USER, &f->flags);
240 241 242
		return;
	}

243
	fdb_delete(br, f, true);
244 245
}

246 247 248 249 250 251 252
void br_fdb_find_delete_local(struct net_bridge *br,
			      const struct net_bridge_port *p,
			      const unsigned char *addr, u16 vid)
{
	struct net_bridge_fdb_entry *f;

	spin_lock_bh(&br->hash_lock);
253
	f = br_fdb_find(br, addr, vid);
254
	if (f && test_bit(BR_FDB_LOCAL, &f->flags) &&
255
	    !test_bit(BR_FDB_ADDED_BY_USER, &f->flags) && f->dst == p)
256 257 258 259
		fdb_delete_local(br, p, f);
	spin_unlock_bh(&br->hash_lock);
}

L
Linus Torvalds 已提交
260 261
void br_fdb_changeaddr(struct net_bridge_port *p, const unsigned char *newaddr)
{
262
	struct net_bridge_vlan_group *vg;
263
	struct net_bridge_fdb_entry *f;
L
Linus Torvalds 已提交
264
	struct net_bridge *br = p->br;
265
	struct net_bridge_vlan *v;
266

L
Linus Torvalds 已提交
267
	spin_lock_bh(&br->hash_lock);
268
	vg = nbp_vlan_group(p);
269
	hlist_for_each_entry(f, &br->fdb_list, fdb_node) {
270
		if (f->dst == p && test_bit(BR_FDB_LOCAL, &f->flags) &&
271
		    !test_bit(BR_FDB_ADDED_BY_USER, &f->flags)) {
272 273 274 275 276 277 278 279 280
			/* delete old one */
			fdb_delete_local(br, p, f);

			/* if this port has no vlan information
			 * configured, we can safely be done at
			 * this point.
			 */
			if (!vg || !vg->num_vlans)
				goto insert;
L
Linus Torvalds 已提交
281 282 283
		}
	}

284 285 286 287
insert:
	/* insert new address,  may fail if invalid address or dup. */
	fdb_insert(br, p, newaddr, 0);

288
	if (!vg || !vg->num_vlans)
289 290 291 292 293 294
		goto done;

	/* Now add entries for every VLAN configured on the port.
	 * This function runs under RTNL so the bitmap will not change
	 * from under us.
	 */
295 296
	list_for_each_entry(v, &vg->vlan_list, vlist)
		fdb_insert(br, p, newaddr, v->vid);
297

298
done:
L
Linus Torvalds 已提交
299 300 301
	spin_unlock_bh(&br->hash_lock);
}

302 303
void br_fdb_change_mac_address(struct net_bridge *br, const u8 *newaddr)
{
304
	struct net_bridge_vlan_group *vg;
305
	struct net_bridge_fdb_entry *f;
306
	struct net_bridge_vlan *v;
307

308 309
	spin_lock_bh(&br->hash_lock);

310
	/* If old entry was unassociated with any port, then delete it. */
311
	f = br_fdb_find(br, br->dev->dev_addr, 0);
312
	if (f && test_bit(BR_FDB_LOCAL, &f->flags) &&
313
	    !f->dst && !test_bit(BR_FDB_ADDED_BY_USER, &f->flags))
314
		fdb_delete_local(br, NULL, f);
315

316
	fdb_insert(br, NULL, newaddr, 0);
317 318 319
	vg = br_vlan_group(br);
	if (!vg || !vg->num_vlans)
		goto out;
320 321 322 323
	/* Now remove and add entries for every VLAN configured on the
	 * bridge.  This function runs under RTNL so the bitmap will not
	 * change from under us.
	 */
324
	list_for_each_entry(v, &vg->vlan_list, vlist) {
325 326
		if (!br_vlan_should_use(v))
			continue;
327
		f = br_fdb_find(br, br->dev->dev_addr, v->vid);
328
		if (f && test_bit(BR_FDB_LOCAL, &f->flags) &&
329
		    !f->dst && !test_bit(BR_FDB_ADDED_BY_USER, &f->flags))
330
			fdb_delete_local(br, NULL, f);
331
		fdb_insert(br, NULL, newaddr, v->vid);
332
	}
333 334
out:
	spin_unlock_bh(&br->hash_lock);
335 336
}

337
void br_fdb_cleanup(struct work_struct *work)
L
Linus Torvalds 已提交
338
{
339 340
	struct net_bridge *br = container_of(work, struct net_bridge,
					     gc_work.work);
341
	struct net_bridge_fdb_entry *f = NULL;
L
Linus Torvalds 已提交
342
	unsigned long delay = hold_time(br);
343 344
	unsigned long work_delay = delay;
	unsigned long now = jiffies;
L
Linus Torvalds 已提交
345

346 347 348 349 350 351 352
	/* this part is tricky, in order to avoid blocking learning and
	 * consequently forwarding, we rely on rcu to delete objects with
	 * delayed freeing allowing us to continue traversing
	 */
	rcu_read_lock();
	hlist_for_each_entry_rcu(f, &br->fdb_list, fdb_node) {
		unsigned long this_timer;
L
Linus Torvalds 已提交
353

354 355
		if (test_bit(BR_FDB_STATIC, &f->flags) ||
		    f->added_by_external_learn)
356
			continue;
357 358 359 360 361 362
		this_timer = f->updated + delay;
		if (time_after(this_timer, now)) {
			work_delay = min(work_delay, this_timer - now);
		} else {
			spin_lock_bh(&br->hash_lock);
			if (!hlist_unhashed(&f->fdb_node))
363
				fdb_delete(br, f, true);
364
			spin_unlock_bh(&br->hash_lock);
L
Linus Torvalds 已提交
365 366
		}
	}
367
	rcu_read_unlock();
L
Linus Torvalds 已提交
368

369 370 371
	/* Cleanup minimum 10 milliseconds apart */
	work_delay = max_t(unsigned long, work_delay, msecs_to_jiffies(10));
	mod_delayed_work(system_long_wq, &br->gc_work, work_delay);
L
Linus Torvalds 已提交
372 373
}

374 375 376
/* Completely flush all dynamic entries in forwarding database.*/
void br_fdb_flush(struct net_bridge *br)
{
377 378
	struct net_bridge_fdb_entry *f;
	struct hlist_node *tmp;
379 380

	spin_lock_bh(&br->hash_lock);
381
	hlist_for_each_entry_safe(f, tmp, &br->fdb_list, fdb_node) {
382
		if (!test_bit(BR_FDB_STATIC, &f->flags))
383
			fdb_delete(br, f, true);
384 385 386
	}
	spin_unlock_bh(&br->hash_lock);
}
387

L
Lucas De Marchi 已提交
388
/* Flush all entries referring to a specific port.
389
 * if do_all is set also flush static entries
390
 * if vid is set delete all entries that match the vlan_id
391
 */
392 393
void br_fdb_delete_by_port(struct net_bridge *br,
			   const struct net_bridge_port *p,
394
			   u16 vid,
395
			   int do_all)
L
Linus Torvalds 已提交
396
{
397 398
	struct net_bridge_fdb_entry *f;
	struct hlist_node *tmp;
L
Linus Torvalds 已提交
399 400

	spin_lock_bh(&br->hash_lock);
401 402 403
	hlist_for_each_entry_safe(f, tmp, &br->fdb_list, fdb_node) {
		if (f->dst != p)
			continue;
404

405
		if (!do_all)
406 407
			if (test_bit(BR_FDB_STATIC, &f->flags) ||
			    (vid && f->key.vlan_id != vid))
L
Linus Torvalds 已提交
408 409
				continue;

410
		if (test_bit(BR_FDB_LOCAL, &f->flags))
411 412
			fdb_delete_local(br, p, f);
		else
413
			fdb_delete(br, f, true);
L
Linus Torvalds 已提交
414 415 416 417
	}
	spin_unlock_bh(&br->hash_lock);
}

I
Igor Maravić 已提交
418
#if IS_ENABLED(CONFIG_ATM_LANE)
419 420 421
/* Interface used by ATM LANE hook to test
 * if an addr is on some other bridge port */
int br_fdb_test_addr(struct net_device *dev, unsigned char *addr)
L
Linus Torvalds 已提交
422 423
{
	struct net_bridge_fdb_entry *fdb;
424
	struct net_bridge_port *port;
425 426
	int ret;

L
Linus Torvalds 已提交
427
	rcu_read_lock();
428 429 430 431
	port = br_port_get_rcu(dev);
	if (!port)
		ret = 0;
	else {
432
		fdb = br_fdb_find_rcu(port->br, addr, 0);
433
		ret = fdb && fdb->dst && fdb->dst->dev != dev &&
434 435
			fdb->dst->state == BR_STATE_FORWARDING;
	}
L
Linus Torvalds 已提交
436 437
	rcu_read_unlock();

438
	return ret;
L
Linus Torvalds 已提交
439
}
440
#endif /* CONFIG_ATM_LANE */
L
Linus Torvalds 已提交
441 442

/*
443
 * Fill buffer with forwarding table records in
L
Linus Torvalds 已提交
444 445 446 447 448 449
 * the API format.
 */
int br_fdb_fillbuf(struct net_bridge *br, void *buf,
		   unsigned long maxnum, unsigned long skip)
{
	struct net_bridge_fdb_entry *f;
450 451
	struct __fdb_entry *fe = buf;
	int num = 0;
L
Linus Torvalds 已提交
452 453 454 455

	memset(buf, 0, maxnum*sizeof(struct __fdb_entry));

	rcu_read_lock();
456 457 458
	hlist_for_each_entry_rcu(f, &br->fdb_list, fdb_node) {
		if (num >= maxnum)
			break;
L
Linus Torvalds 已提交
459

460 461
		if (has_expired(br, f))
			continue;
L
Linus Torvalds 已提交
462

463 464 465
		/* ignore pseudo entry for local MAC address */
		if (!f->dst)
			continue;
466

467 468 469 470
		if (skip) {
			--skip;
			continue;
		}
L
Linus Torvalds 已提交
471

472 473
		/* convert from internal format to API */
		memcpy(fe->mac_addr, f->key.addr.addr, ETH_ALEN);
474

475 476 477
		/* due to ABI compat need to split into hi/lo */
		fe->port_no = f->dst->port_no;
		fe->port_hi = f->dst->port_no >> 8;
478

479
		fe->is_local = test_bit(BR_FDB_LOCAL, &f->flags);
480
		if (!test_bit(BR_FDB_STATIC, &f->flags))
481 482 483
			fe->ageing_timer_value = jiffies_delta_to_clock_t(jiffies - f->updated);
		++fe;
		++num;
L
Linus Torvalds 已提交
484 485 486 487 488 489
	}
	rcu_read_unlock();

	return num;
}

490
static struct net_bridge_fdb_entry *fdb_create(struct net_bridge *br,
L
Linus Torvalds 已提交
491
					       struct net_bridge_port *source,
492
					       const unsigned char *addr,
493 494 495
					       __u16 vid,
					       unsigned char is_local,
					       unsigned char is_static)
L
Linus Torvalds 已提交
496 497 498 499 500
{
	struct net_bridge_fdb_entry *fdb;

	fdb = kmem_cache_alloc(br_fdb_cache, GFP_ATOMIC);
	if (fdb) {
501
		memcpy(fdb->key.addr.addr, addr, ETH_ALEN);
L
Linus Torvalds 已提交
502
		fdb->dst = source;
503
		fdb->key.vlan_id = vid;
504 505 506
		fdb->flags = 0;
		if (is_local)
			set_bit(BR_FDB_LOCAL, &fdb->flags);
507 508
		if (is_static)
			set_bit(BR_FDB_STATIC, &fdb->flags);
509
		fdb->added_by_external_learn = 0;
510
		fdb->offloaded = 0;
511
		fdb->updated = fdb->used = jiffies;
512 513 514 515 516 517 518 519
		if (rhashtable_lookup_insert_fast(&br->fdb_hash_tbl,
						  &fdb->rhnode,
						  br_fdb_rht_params)) {
			kmem_cache_free(br_fdb_cache, fdb);
			fdb = NULL;
		} else {
			hlist_add_head_rcu(&fdb->fdb_node, &br->fdb_list);
		}
L
Linus Torvalds 已提交
520 521 522 523 524
	}
	return fdb;
}

static int fdb_insert(struct net_bridge *br, struct net_bridge_port *source,
525
		  const unsigned char *addr, u16 vid)
L
Linus Torvalds 已提交
526 527 528 529 530 531
{
	struct net_bridge_fdb_entry *fdb;

	if (!is_valid_ether_addr(addr))
		return -EINVAL;

532
	fdb = br_fdb_find(br, addr, vid);
L
Linus Torvalds 已提交
533
	if (fdb) {
534
		/* it is okay to have multiple ports with same
L
Linus Torvalds 已提交
535 536
		 * address, just use the first one.
		 */
537
		if (test_bit(BR_FDB_LOCAL, &fdb->flags))
L
Linus Torvalds 已提交
538
			return 0;
539 540
		br_warn(br, "adding interface %s with same address as a received packet (addr:%pM, vlan:%u)\n",
		       source ? source->dev->name : br->dev->name, addr, vid);
541
		fdb_delete(br, fdb, true);
542
	}
L
Linus Torvalds 已提交
543

544
	fdb = fdb_create(br, source, addr, vid, 1, 1);
545
	if (!fdb)
L
Linus Torvalds 已提交
546 547
		return -ENOMEM;

548
	fdb_add_hw_addr(br, addr);
549
	fdb_notify(br, fdb, RTM_NEWNEIGH, true);
L
Linus Torvalds 已提交
550 551 552
	return 0;
}

553
/* Add entry for local address of interface */
L
Linus Torvalds 已提交
554
int br_fdb_insert(struct net_bridge *br, struct net_bridge_port *source,
555
		  const unsigned char *addr, u16 vid)
L
Linus Torvalds 已提交
556 557 558 559
{
	int ret;

	spin_lock_bh(&br->hash_lock);
560
	ret = fdb_insert(br, source, addr, vid);
L
Linus Torvalds 已提交
561 562 563 564 565
	spin_unlock_bh(&br->hash_lock);
	return ret;
}

void br_fdb_update(struct net_bridge *br, struct net_bridge_port *source,
566
		   const unsigned char *addr, u16 vid, bool added_by_user)
L
Linus Torvalds 已提交
567 568
{
	struct net_bridge_fdb_entry *fdb;
569
	bool fdb_modified = false;
L
Linus Torvalds 已提交
570 571 572 573 574

	/* some users want to always flood. */
	if (hold_time(br) == 0)
		return;

575 576 577 578 579
	/* ignore packets unless we are using this port */
	if (!(source->state == BR_STATE_LEARNING ||
	      source->state == BR_STATE_FORWARDING))
		return;

580
	fdb = fdb_find_rcu(&br->fdb_hash_tbl, addr, vid);
L
Linus Torvalds 已提交
581 582
	if (likely(fdb)) {
		/* attempt to update an entry for a local interface */
583
		if (unlikely(test_bit(BR_FDB_LOCAL, &fdb->flags))) {
584
			if (net_ratelimit())
585 586
				br_warn(br, "received packet on %s with own address as source address (addr:%pM, vlan:%u)\n",
					source->dev->name, addr, vid);
L
Linus Torvalds 已提交
587
		} else {
588 589
			unsigned long now = jiffies;

L
Linus Torvalds 已提交
590
			/* fastpath: update of existing entry */
591 592
			if (unlikely(source != fdb->dst &&
				     !test_bit(BR_FDB_STICKY, &fdb->flags))) {
593 594
				fdb->dst = source;
				fdb_modified = true;
595 596 597
				/* Take over HW learned entry */
				if (unlikely(fdb->added_by_external_learn))
					fdb->added_by_external_learn = 0;
598
			}
599 600
			if (now != fdb->updated)
				fdb->updated = now;
601
			if (unlikely(added_by_user))
602
				set_bit(BR_FDB_ADDED_BY_USER, &fdb->flags);
603 604
			if (unlikely(fdb_modified)) {
				trace_br_fdb_update(br, source, addr, vid, added_by_user);
605
				fdb_notify(br, fdb, RTM_NEWNEIGH, true);
606
			}
L
Linus Torvalds 已提交
607 608
		}
	} else {
609
		spin_lock(&br->hash_lock);
610 611 612
		fdb = fdb_create(br, source, addr, vid, 0, 0);
		if (fdb) {
			if (unlikely(added_by_user))
613
				set_bit(BR_FDB_ADDED_BY_USER, &fdb->flags);
614 615
			trace_br_fdb_update(br, source, addr, vid,
					    added_by_user);
616
			fdb_notify(br, fdb, RTM_NEWNEIGH, true);
S
stephen hemminger 已提交
617
		}
L
Linus Torvalds 已提交
618 619 620
		/* else  we lose race and someone else inserts
		 * it first, don't bother updating
		 */
621
		spin_unlock(&br->hash_lock);
L
Linus Torvalds 已提交
622 623
	}
}
624

625 626
static int fdb_to_nud(const struct net_bridge *br,
		      const struct net_bridge_fdb_entry *fdb)
627
{
628
	if (test_bit(BR_FDB_LOCAL, &fdb->flags))
629
		return NUD_PERMANENT;
630
	else if (test_bit(BR_FDB_STATIC, &fdb->flags))
631
		return NUD_NOARP;
632
	else if (has_expired(br, fdb))
633 634 635 636 637
		return NUD_STALE;
	else
		return NUD_REACHABLE;
}

638
static int fdb_fill_info(struct sk_buff *skb, const struct net_bridge *br,
639
			 const struct net_bridge_fdb_entry *fdb,
640
			 u32 portid, u32 seq, int type, unsigned int flags)
641 642 643 644 645 646
{
	unsigned long now = jiffies;
	struct nda_cacheinfo ci;
	struct nlmsghdr *nlh;
	struct ndmsg *ndm;

647
	nlh = nlmsg_put(skb, portid, seq, type, sizeof(*ndm), flags);
648 649 650 651 652 653 654
	if (nlh == NULL)
		return -EMSGSIZE;

	ndm = nlmsg_data(nlh);
	ndm->ndm_family	 = AF_BRIDGE;
	ndm->ndm_pad1    = 0;
	ndm->ndm_pad2    = 0;
655
	ndm->ndm_flags	 = 0;
656
	ndm->ndm_type	 = 0;
657
	ndm->ndm_ifindex = fdb->dst ? fdb->dst->dev->ifindex : br->dev->ifindex;
658
	ndm->ndm_state   = fdb_to_nud(br, fdb);
659

660 661 662 663
	if (fdb->offloaded)
		ndm->ndm_flags |= NTF_OFFLOADED;
	if (fdb->added_by_external_learn)
		ndm->ndm_flags |= NTF_EXT_LEARNED;
664
	if (test_bit(BR_FDB_STICKY, &fdb->flags))
665
		ndm->ndm_flags |= NTF_STICKY;
666

667
	if (nla_put(skb, NDA_LLADDR, ETH_ALEN, &fdb->key.addr))
D
David S. Miller 已提交
668
		goto nla_put_failure;
669 670
	if (nla_put_u32(skb, NDA_MASTER, br->dev->ifindex))
		goto nla_put_failure;
671 672 673 674
	ci.ndm_used	 = jiffies_to_clock_t(now - fdb->used);
	ci.ndm_confirmed = 0;
	ci.ndm_updated	 = jiffies_to_clock_t(now - fdb->updated);
	ci.ndm_refcnt	 = 0;
D
David S. Miller 已提交
675 676
	if (nla_put(skb, NDA_CACHEINFO, sizeof(ci), &ci))
		goto nla_put_failure;
677

678 679
	if (fdb->key.vlan_id && nla_put(skb, NDA_VLAN, sizeof(u16),
					&fdb->key.vlan_id))
680 681
		goto nla_put_failure;

682 683
	nlmsg_end(skb, nlh);
	return 0;
684 685 686 687 688 689 690 691 692 693

nla_put_failure:
	nlmsg_cancel(skb, nlh);
	return -EMSGSIZE;
}

static inline size_t fdb_nlmsg_size(void)
{
	return NLMSG_ALIGN(sizeof(struct ndmsg))
		+ nla_total_size(ETH_ALEN) /* NDA_LLADDR */
694
		+ nla_total_size(sizeof(u32)) /* NDA_MASTER */
695
		+ nla_total_size(sizeof(u16)) /* NDA_VLAN */
696 697 698
		+ nla_total_size(sizeof(struct nda_cacheinfo));
}

699
static void fdb_notify(struct net_bridge *br,
700 701
		       const struct net_bridge_fdb_entry *fdb, int type,
		       bool swdev_notify)
702
{
703
	struct net *net = dev_net(br->dev);
704 705 706
	struct sk_buff *skb;
	int err = -ENOBUFS;

707 708
	if (swdev_notify)
		br_switchdev_fdb_notify(fdb, type);
709

710 711 712 713
	skb = nlmsg_new(fdb_nlmsg_size(), GFP_ATOMIC);
	if (skb == NULL)
		goto errout;

714
	err = fdb_fill_info(skb, br, fdb, 0, 0, type, 0);
715 716 717 718 719 720 721 722 723
	if (err < 0) {
		/* -EMSGSIZE implies BUG in fdb_nlmsg_size() */
		WARN_ON(err == -EMSGSIZE);
		kfree_skb(skb);
		goto errout;
	}
	rtnl_notify(skb, net, 0, RTNLGRP_NEIGH, NULL, GFP_ATOMIC);
	return;
errout:
724
	rtnl_set_sk_err(net, RTNLGRP_NEIGH, err);
725 726 727
}

/* Dump information about entries, in response to GETNEIGH */
728 729 730
int br_fdb_dump(struct sk_buff *skb,
		struct netlink_callback *cb,
		struct net_device *dev,
731
		struct net_device *filter_dev,
732
		int *idx)
733
{
734
	struct net_bridge *br = netdev_priv(dev);
735
	struct net_bridge_fdb_entry *f;
736
	int err = 0;
737

738
	if (!(dev->priv_flags & IFF_EBRIDGE))
739
		return err;
740

741 742 743
	if (!filter_dev) {
		err = ndo_dflt_fdb_dump(skb, cb, dev, NULL, idx);
		if (err < 0)
744
			return err;
745
	}
746

747 748 749 750 751 752
	rcu_read_lock();
	hlist_for_each_entry_rcu(f, &br->fdb_list, fdb_node) {
		if (*idx < cb->args[2])
			goto skip;
		if (filter_dev && (!f->dst || f->dst->dev != filter_dev)) {
			if (filter_dev != dev)
753
				goto skip;
754 755 756 757 758 759
			/* !f->dst is a special case for bridge
			 * It means the MAC belongs to the bridge
			 * Therefore need a little more filtering
			 * we only want to dump the !f->dst case
			 */
			if (f->dst)
760
				goto skip;
761
		}
762 763 764 765 766 767 768 769 770 771 772 773
		if (!filter_dev && f->dst)
			goto skip;

		err = fdb_fill_info(skb, br, f,
				    NETLINK_CB(cb->skb).portid,
				    cb->nlh->nlmsg_seq,
				    RTM_NEWNEIGH,
				    NLM_F_MULTI);
		if (err < 0)
			break;
skip:
		*idx += 1;
774
	}
775
	rcu_read_unlock();
776

777
	return err;
778
}
779

R
Roopa Prabhu 已提交
780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
int br_fdb_get(struct sk_buff *skb,
	       struct nlattr *tb[],
	       struct net_device *dev,
	       const unsigned char *addr,
	       u16 vid, u32 portid, u32 seq,
	       struct netlink_ext_ack *extack)
{
	struct net_bridge *br = netdev_priv(dev);
	struct net_bridge_fdb_entry *f;
	int err = 0;

	rcu_read_lock();
	f = br_fdb_find_rcu(br, addr, vid);
	if (!f) {
		NL_SET_ERR_MSG(extack, "Fdb entry not found");
		err = -ENOENT;
		goto errout;
	}

	err = fdb_fill_info(skb, br, f, portid, seq,
			    RTM_NEWNEIGH, 0);
errout:
	rcu_read_unlock();
	return err;
}

S
stephen hemminger 已提交
806
/* Update (create or replace) forwarding database entry */
807
static int fdb_add_entry(struct net_bridge *br, struct net_bridge_port *source,
808 809
			 const u8 *addr, u16 state, u16 flags, u16 vid,
			 u8 ndm_flags)
810
{
811
	bool is_sticky = !!(ndm_flags & NTF_STICKY);
812
	struct net_bridge_fdb_entry *fdb;
813
	bool modified = false;
814

815
	/* If the port cannot learn allow only local and static entries */
816
	if (source && !(state & NUD_PERMANENT) && !(state & NUD_NOARP) &&
817 818 819 820
	    !(source->state == BR_STATE_LEARNING ||
	      source->state == BR_STATE_FORWARDING))
		return -EPERM;

821 822 823 824 825 826
	if (!source && !(state & NUD_PERMANENT)) {
		pr_info("bridge: RTM_NEWNEIGH %s without NUD_PERMANENT\n",
			br->dev->name);
		return -EINVAL;
	}

827 828 829
	if (is_sticky && (state & NUD_PERMANENT))
		return -EINVAL;

830
	fdb = br_fdb_find(br, addr, vid);
831 832 833
	if (fdb == NULL) {
		if (!(flags & NLM_F_CREATE))
			return -ENOENT;
834

835
		fdb = fdb_create(br, source, addr, vid, 0, 0);
836 837
		if (!fdb)
			return -ENOMEM;
838 839

		modified = true;
840 841 842
	} else {
		if (flags & NLM_F_EXCL)
			return -EEXIST;
843 844 845 846 847

		if (fdb->dst != source) {
			fdb->dst = source;
			modified = true;
		}
S
stephen hemminger 已提交
848 849
	}

850
	if (fdb_to_nud(br, fdb) != state) {
851
		if (state & NUD_PERMANENT) {
852
			set_bit(BR_FDB_LOCAL, &fdb->flags);
853
			if (!test_and_set_bit(BR_FDB_STATIC, &fdb->flags))
854
				fdb_add_hw_addr(br, addr);
855
		} else if (state & NUD_NOARP) {
856
			clear_bit(BR_FDB_LOCAL, &fdb->flags);
857
			if (!test_and_set_bit(BR_FDB_STATIC, &fdb->flags))
858
				fdb_add_hw_addr(br, addr);
859
		} else {
860
			clear_bit(BR_FDB_LOCAL, &fdb->flags);
861
			if (test_and_clear_bit(BR_FDB_STATIC, &fdb->flags))
862
				fdb_del_hw_addr(br, addr);
863
		}
864

865 866
		modified = true;
	}
867

868 869
	if (is_sticky != test_bit(BR_FDB_STICKY, &fdb->flags)) {
		change_bit(BR_FDB_STICKY, &fdb->flags);
870 871 872
		modified = true;
	}

873
	set_bit(BR_FDB_ADDED_BY_USER, &fdb->flags);
874 875 876 877

	fdb->used = jiffies;
	if (modified) {
		fdb->updated = jiffies;
878
		fdb_notify(br, fdb, RTM_NEWNEIGH, true);
879
	}
880 881 882 883

	return 0;
}

884 885 886
static int __br_fdb_add(struct ndmsg *ndm, struct net_bridge *br,
			struct net_bridge_port *p, const unsigned char *addr,
			u16 nlh_flags, u16 vid)
887 888 889 890
{
	int err = 0;

	if (ndm->ndm_flags & NTF_USE) {
891 892 893 894 895
		if (!p) {
			pr_info("bridge: RTM_NEWNEIGH %s with NTF_USE is not supported\n",
				br->dev->name);
			return -EINVAL;
		}
896
		local_bh_disable();
897
		rcu_read_lock();
898
		br_fdb_update(br, p, addr, vid, true);
899
		rcu_read_unlock();
900
		local_bh_enable();
901
	} else if (ndm->ndm_flags & NTF_EXT_LEARNED) {
902
		err = br_fdb_external_learn_add(br, p, addr, vid, true);
903
	} else {
904 905
		spin_lock_bh(&br->hash_lock);
		err = fdb_add_entry(br, p, addr, ndm->ndm_state,
906
				    nlh_flags, vid, ndm->ndm_flags);
907
		spin_unlock_bh(&br->hash_lock);
908 909 910 911 912
	}

	return err;
}

913
/* Add new permanent fdb entry with RTM_NEWNEIGH */
914 915
int br_fdb_add(struct ndmsg *ndm, struct nlattr *tb[],
	       struct net_device *dev,
916 917
	       const unsigned char *addr, u16 vid, u16 nlh_flags,
	       struct netlink_ext_ack *extack)
918
{
919
	struct net_bridge_vlan_group *vg;
920
	struct net_bridge_port *p = NULL;
921
	struct net_bridge_vlan *v;
922
	struct net_bridge *br = NULL;
923
	int err = 0;
924

925 926
	trace_br_fdb_add(ndm, dev, addr, vid, nlh_flags);

S
stephen hemminger 已提交
927 928 929 930 931
	if (!(ndm->ndm_state & (NUD_PERMANENT|NUD_NOARP|NUD_REACHABLE))) {
		pr_info("bridge: RTM_NEWNEIGH with invalid state %#x\n", ndm->ndm_state);
		return -EINVAL;
	}

932 933 934 935 936
	if (is_zero_ether_addr(addr)) {
		pr_info("bridge: RTM_NEWNEIGH with invalid ether address\n");
		return -EINVAL;
	}

937 938 939 940 941 942 943 944 945 946
	if (dev->priv_flags & IFF_EBRIDGE) {
		br = netdev_priv(dev);
		vg = br_vlan_group(br);
	} else {
		p = br_port_get_rtnl(dev);
		if (!p) {
			pr_info("bridge: RTM_NEWNEIGH %s not a bridge port\n",
				dev->name);
			return -EINVAL;
		}
947
		br = p->br;
948
		vg = nbp_vlan_group(p);
949 950
	}

951
	if (vid) {
952
		v = br_vlan_find(vg, vid);
953 954
		if (!v || !br_vlan_should_use(v)) {
			pr_info("bridge: RTM_NEWNEIGH with unconfigured vlan %d on %s\n", vid, dev->name);
955 956 957 958
			return -EINVAL;
		}

		/* VID was specified, so use it. */
959
		err = __br_fdb_add(ndm, br, p, addr, nlh_flags, vid);
S
stephen hemminger 已提交
960
	} else {
961
		err = __br_fdb_add(ndm, br, p, addr, nlh_flags, 0);
962
		if (err || !vg || !vg->num_vlans)
963 964 965 966 967 968
			goto out;

		/* We have vlans configured on this port and user didn't
		 * specify a VLAN.  To be nice, add/update entry for every
		 * vlan on this port.
		 */
969
		list_for_each_entry(v, &vg->vlan_list, vlist) {
970 971
			if (!br_vlan_should_use(v))
				continue;
972
			err = __br_fdb_add(ndm, br, p, addr, nlh_flags, v->vid);
973 974 975
			if (err)
				goto out;
		}
S
stephen hemminger 已提交
976
	}
977

978
out:
979 980 981
	return err;
}

982 983
static int fdb_delete_by_addr_and_port(struct net_bridge *br,
				       const struct net_bridge_port *p,
984
				       const u8 *addr, u16 vlan)
985 986 987
{
	struct net_bridge_fdb_entry *fdb;

988
	fdb = br_fdb_find(br, addr, vlan);
989
	if (!fdb || fdb->dst != p)
990 991
		return -ENOENT;

992
	fdb_delete(br, fdb, true);
993

994 995 996
	return 0;
}

997 998
static int __br_fdb_delete(struct net_bridge *br,
			   const struct net_bridge_port *p,
999 1000 1001 1002
			   const unsigned char *addr, u16 vid)
{
	int err;

1003 1004 1005
	spin_lock_bh(&br->hash_lock);
	err = fdb_delete_by_addr_and_port(br, p, addr, vid);
	spin_unlock_bh(&br->hash_lock);
1006 1007 1008 1009

	return err;
}

1010
/* Remove neighbor entry with RTM_DELNEIGH */
1011 1012
int br_fdb_delete(struct ndmsg *ndm, struct nlattr *tb[],
		  struct net_device *dev,
1013
		  const unsigned char *addr, u16 vid)
1014
{
1015
	struct net_bridge_vlan_group *vg;
1016
	struct net_bridge_port *p = NULL;
1017
	struct net_bridge_vlan *v;
1018
	struct net_bridge *br;
1019
	int err;
1020

1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
	if (dev->priv_flags & IFF_EBRIDGE) {
		br = netdev_priv(dev);
		vg = br_vlan_group(br);
	} else {
		p = br_port_get_rtnl(dev);
		if (!p) {
			pr_info("bridge: RTM_DELNEIGH %s not a bridge port\n",
				dev->name);
			return -EINVAL;
		}
		vg = nbp_vlan_group(p);
1032
		br = p->br;
1033 1034
	}

1035
	if (vid) {
1036 1037
		v = br_vlan_find(vg, vid);
		if (!v) {
1038
			pr_info("bridge: RTM_DELNEIGH with unconfigured vlan %d on %s\n", vid, dev->name);
1039 1040
			return -EINVAL;
		}
1041

1042
		err = __br_fdb_delete(br, p, addr, vid);
1043
	} else {
1044
		err = -ENOENT;
1045
		err &= __br_fdb_delete(br, p, addr, 0);
1046
		if (!vg || !vg->num_vlans)
1047
			return err;
1048

1049 1050 1051
		list_for_each_entry(v, &vg->vlan_list, vlist) {
			if (!br_vlan_should_use(v))
				continue;
1052
			err &= __br_fdb_delete(br, p, addr, v->vid);
1053
		}
1054
	}
1055

1056 1057
	return err;
}
1058 1059 1060

int br_fdb_sync_static(struct net_bridge *br, struct net_bridge_port *p)
{
1061
	struct net_bridge_fdb_entry *f, *tmp;
1062
	int err = 0;
1063 1064 1065

	ASSERT_RTNL();

1066 1067 1068 1069
	/* the key here is that static entries change only under rtnl */
	rcu_read_lock();
	hlist_for_each_entry_rcu(f, &br->fdb_list, fdb_node) {
		/* We only care for static entries */
1070
		if (!test_bit(BR_FDB_STATIC, &f->flags))
1071 1072 1073 1074
			continue;
		err = dev_uc_add(p->dev, f->key.addr.addr);
		if (err)
			goto rollback;
1075
	}
1076 1077
done:
	rcu_read_unlock();
1078

1079
	return err;
1080

1081 1082 1083
rollback:
	hlist_for_each_entry_rcu(tmp, &br->fdb_list, fdb_node) {
		/* We only care for static entries */
1084
		if (!test_bit(BR_FDB_STATIC, &tmp->flags))
1085 1086 1087 1088
			continue;
		if (tmp == f)
			break;
		dev_uc_del(p->dev, tmp->key.addr.addr);
1089
	}
1090 1091

	goto done;
1092 1093 1094 1095
}

void br_fdb_unsync_static(struct net_bridge *br, struct net_bridge_port *p)
{
1096
	struct net_bridge_fdb_entry *f;
1097 1098 1099

	ASSERT_RTNL();

1100 1101 1102
	rcu_read_lock();
	hlist_for_each_entry_rcu(f, &br->fdb_list, fdb_node) {
		/* We only care for static entries */
1103
		if (!test_bit(BR_FDB_STATIC, &f->flags))
1104
			continue;
1105

1106
		dev_uc_del(p->dev, f->key.addr.addr);
1107
	}
1108
	rcu_read_unlock();
1109
}
1110

1111
int br_fdb_external_learn_add(struct net_bridge *br, struct net_bridge_port *p,
1112 1113
			      const unsigned char *addr, u16 vid,
			      bool swdev_notify)
1114 1115
{
	struct net_bridge_fdb_entry *fdb;
1116
	bool modified = false;
1117 1118
	int err = 0;

1119 1120
	trace_br_fdb_external_learn_add(br, p, addr, vid);

1121 1122
	spin_lock_bh(&br->hash_lock);

1123
	fdb = br_fdb_find(br, addr, vid);
1124
	if (!fdb) {
1125
		fdb = fdb_create(br, p, addr, vid, 0, 0);
1126 1127 1128 1129
		if (!fdb) {
			err = -ENOMEM;
			goto err_unlock;
		}
1130
		if (swdev_notify)
1131
			set_bit(BR_FDB_ADDED_BY_USER, &fdb->flags);
1132
		fdb->added_by_external_learn = 1;
1133
		fdb_notify(br, fdb, RTM_NEWNEIGH, swdev_notify);
1134
	} else {
1135
		fdb->updated = jiffies;
1136 1137 1138 1139 1140 1141 1142 1143 1144

		if (fdb->dst != p) {
			fdb->dst = p;
			modified = true;
		}

		if (fdb->added_by_external_learn) {
			/* Refresh entry */
			fdb->used = jiffies;
1145
		} else if (!test_bit(BR_FDB_ADDED_BY_USER, &fdb->flags)) {
1146 1147 1148 1149 1150
			/* Take over SW learned entry */
			fdb->added_by_external_learn = 1;
			modified = true;
		}

1151
		if (swdev_notify)
1152
			set_bit(BR_FDB_ADDED_BY_USER, &fdb->flags);
1153

1154
		if (modified)
1155
			fdb_notify(br, fdb, RTM_NEWNEIGH, swdev_notify);
1156 1157 1158 1159 1160 1161 1162 1163
	}

err_unlock:
	spin_unlock_bh(&br->hash_lock);

	return err;
}

1164
int br_fdb_external_learn_del(struct net_bridge *br, struct net_bridge_port *p,
1165 1166
			      const unsigned char *addr, u16 vid,
			      bool swdev_notify)
1167 1168 1169 1170 1171 1172
{
	struct net_bridge_fdb_entry *fdb;
	int err = 0;

	spin_lock_bh(&br->hash_lock);

1173
	fdb = br_fdb_find(br, addr, vid);
1174
	if (fdb && fdb->added_by_external_learn)
1175
		fdb_delete(br, fdb, swdev_notify);
1176 1177 1178 1179 1180 1181 1182
	else
		err = -ENOENT;

	spin_unlock_bh(&br->hash_lock);

	return err;
}
1183 1184

void br_fdb_offloaded_set(struct net_bridge *br, struct net_bridge_port *p,
1185
			  const unsigned char *addr, u16 vid, bool offloaded)
1186 1187 1188 1189 1190 1191 1192
{
	struct net_bridge_fdb_entry *fdb;

	spin_lock_bh(&br->hash_lock);

	fdb = br_fdb_find(br, addr, vid);
	if (fdb)
1193
		fdb->offloaded = offloaded;
1194 1195 1196

	spin_unlock_bh(&br->hash_lock);
}
P
Petr Machata 已提交
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216

void br_fdb_clear_offload(const struct net_device *dev, u16 vid)
{
	struct net_bridge_fdb_entry *f;
	struct net_bridge_port *p;

	ASSERT_RTNL();

	p = br_port_get_rtnl(dev);
	if (!p)
		return;

	spin_lock_bh(&p->br->hash_lock);
	hlist_for_each_entry(f, &p->br->fdb_list, fdb_node) {
		if (f->dst == p && f->key.vlan_id == vid)
			f->offloaded = 0;
	}
	spin_unlock_bh(&p->br->hash_lock);
}
EXPORT_SYMBOL_GPL(br_fdb_clear_offload);