bond_main.c 130.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
/*
 * originally based on the dummy device.
 *
 * Copyright 1999, Thomas Davis, tadavis@lbl.gov.
 * Licensed under the GPL. Based on dummy.c, and eql.c devices.
 *
 * bonding.c: an Ethernet Bonding driver
 *
 * This is useful to talk to a Cisco EtherChannel compatible equipment:
 *	Cisco 5500
 *	Sun Trunking (Solaris)
 *	Alteon AceDirector Trunks
 *	Linux Bonding
 *	and probably many L2 switches ...
 *
 * How it works:
 *    ifconfig bond0 ipaddress netmask up
 *      will setup a network device, with an ip address.  No mac address
 *	will be assigned at this time.  The hw mac address will come from
 *	the first slave bonded to the channel.  All slaves will then use
 *	this hw mac address.
 *
 *    ifconfig bond0 down
 *         will release all slaves, marking them as down.
 *
 *    ifenslave bond0 eth0
 *	will attach eth0 to bond0 as a slave.  eth0 hw mac address will either
 *	a: be used as initial mac address
 *	b: if a hw mac address already is there, eth0's hw mac address
 *	   will then be set from bond0.
 *
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/types.h>
#include <linux/fcntl.h>
#include <linux/interrupt.h>
#include <linux/ptrace.h>
#include <linux/ioport.h>
#include <linux/in.h>
42
#include <net/ip.h>
L
Linus Torvalds 已提交
43
#include <linux/ip.h>
44 45
#include <linux/tcp.h>
#include <linux/udp.h>
L
Linus Torvalds 已提交
46 47 48 49 50 51 52 53
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/init.h>
#include <linux/timer.h>
#include <linux/socket.h>
#include <linux/ctype.h>
#include <linux/inet.h>
#include <linux/bitops.h>
S
Stephen Hemminger 已提交
54
#include <linux/io.h>
L
Linus Torvalds 已提交
55
#include <asm/dma.h>
S
Stephen Hemminger 已提交
56
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
57 58 59
#include <linux/errno.h>
#include <linux/netdevice.h>
#include <linux/inetdevice.h>
60
#include <linux/igmp.h>
L
Linus Torvalds 已提交
61 62 63 64 65 66 67 68 69 70 71
#include <linux/etherdevice.h>
#include <linux/skbuff.h>
#include <net/sock.h>
#include <linux/rtnetlink.h>
#include <linux/smp.h>
#include <linux/if_ether.h>
#include <net/arp.h>
#include <linux/mii.h>
#include <linux/ethtool.h>
#include <linux/if_vlan.h>
#include <linux/if_bonding.h>
D
David Sterba 已提交
72
#include <linux/jiffies.h>
73
#include <linux/preempt.h>
J
Jay Vosburgh 已提交
74
#include <net/route.h>
75
#include <net/net_namespace.h>
76
#include <net/netns/generic.h>
77
#include <net/pkt_sched.h>
78
#include <linux/rculist.h>
79
#include <net/flow_keys.h>
80
#include <net/switchdev.h>
81 82 83
#include <net/bonding.h>
#include <net/bond_3ad.h>
#include <net/bond_alb.h>
L
Linus Torvalds 已提交
84

85 86
#include "bonding_priv.h"

L
Linus Torvalds 已提交
87 88 89 90 91
/*---------------------------- Module parameters ----------------------------*/

/* monitor all links that often (in milliseconds). <=0 disables monitoring */

static int max_bonds	= BOND_DEFAULT_MAX_BONDS;
92
static int tx_queues	= BOND_DEFAULT_TX_QUEUES;
93
static int num_peer_notif = 1;
94
static int miimon;
S
Stephen Hemminger 已提交
95 96
static int updelay;
static int downdelay;
L
Linus Torvalds 已提交
97
static int use_carrier	= 1;
S
Stephen Hemminger 已提交
98 99
static char *mode;
static char *primary;
100
static char *primary_reselect;
S
Stephen Hemminger 已提交
101
static char *lacp_rate;
102
static int min_links;
S
Stephen Hemminger 已提交
103 104
static char *ad_select;
static char *xmit_hash_policy;
105
static int arp_interval;
S
Stephen Hemminger 已提交
106 107
static char *arp_ip_target[BOND_MAX_ARP_TARGETS];
static char *arp_validate;
108
static char *arp_all_targets;
S
Stephen Hemminger 已提交
109
static char *fail_over_mac;
110
static int all_slaves_active;
111
static struct bond_params bonding_defaults;
112
static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
113
static int packets_per_slave = 1;
114
static int lp_interval = BOND_ALB_DEFAULT_LP_INTERVAL;
L
Linus Torvalds 已提交
115 116 117

module_param(max_bonds, int, 0);
MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
118 119
module_param(tx_queues, int, 0);
MODULE_PARM_DESC(tx_queues, "Max number of transmit queues (default = 16)");
120
module_param_named(num_grat_arp, num_peer_notif, int, 0644);
121 122
MODULE_PARM_DESC(num_grat_arp, "Number of peer notifications to send on "
			       "failover event (alias of num_unsol_na)");
123
module_param_named(num_unsol_na, num_peer_notif, int, 0644);
124 125
MODULE_PARM_DESC(num_unsol_na, "Number of peer notifications to send on "
			       "failover event (alias of num_grat_arp)");
L
Linus Torvalds 已提交
126 127 128 129 130
module_param(miimon, int, 0);
MODULE_PARM_DESC(miimon, "Link check interval in milliseconds");
module_param(updelay, int, 0);
MODULE_PARM_DESC(updelay, "Delay before considering link up, in milliseconds");
module_param(downdelay, int, 0);
131 132
MODULE_PARM_DESC(downdelay, "Delay before considering link down, "
			    "in milliseconds");
L
Linus Torvalds 已提交
133
module_param(use_carrier, int, 0);
134 135
MODULE_PARM_DESC(use_carrier, "Use netif_carrier_ok (vs MII ioctls) in miimon; "
			      "0 for off, 1 for on (default)");
L
Linus Torvalds 已提交
136
module_param(mode, charp, 0);
137
MODULE_PARM_DESC(mode, "Mode of operation; 0 for balance-rr, "
138 139 140
		       "1 for active-backup, 2 for balance-xor, "
		       "3 for broadcast, 4 for 802.3ad, 5 for balance-tlb, "
		       "6 for balance-alb");
L
Linus Torvalds 已提交
141 142
module_param(primary, charp, 0);
MODULE_PARM_DESC(primary, "Primary network device to use");
143 144 145 146 147 148 149 150
module_param(primary_reselect, charp, 0);
MODULE_PARM_DESC(primary_reselect, "Reselect primary slave "
				   "once it comes up; "
				   "0 for always (default), "
				   "1 for only if speed of primary is "
				   "better, "
				   "2 for only on active slave "
				   "failure");
L
Linus Torvalds 已提交
151
module_param(lacp_rate, charp, 0);
152 153
MODULE_PARM_DESC(lacp_rate, "LACPDU tx rate to request from 802.3ad partner; "
			    "0 for slow, 1 for fast");
154
module_param(ad_select, charp, 0);
155 156 157
MODULE_PARM_DESC(ad_select, "803.ad aggregation selection logic; "
			    "0 for stable (default), 1 for bandwidth, "
			    "2 for count");
158 159 160
module_param(min_links, int, 0);
MODULE_PARM_DESC(min_links, "Minimum number of available links before turning on carrier");

161
module_param(xmit_hash_policy, charp, 0);
162 163
MODULE_PARM_DESC(xmit_hash_policy, "balance-xor and 802.3ad hashing method; "
				   "0 for layer 2 (default), 1 for layer 3+4, "
164 165
				   "2 for layer 2+3, 3 for encap layer 2+3, "
				   "4 for encap layer 3+4");
L
Linus Torvalds 已提交
166 167 168 169
module_param(arp_interval, int, 0);
MODULE_PARM_DESC(arp_interval, "arp interval in milliseconds");
module_param_array(arp_ip_target, charp, NULL, 0);
MODULE_PARM_DESC(arp_ip_target, "arp targets in n.n.n.n form");
170
module_param(arp_validate, charp, 0);
171 172 173
MODULE_PARM_DESC(arp_validate, "validate src/dst of ARP probes; "
			       "0 for none (default), 1 for active, "
			       "2 for backup, 3 for all");
174 175
module_param(arp_all_targets, charp, 0);
MODULE_PARM_DESC(arp_all_targets, "fail on any/all arp targets timeout; 0 for any (default), 1 for all");
176
module_param(fail_over_mac, charp, 0);
177 178 179
MODULE_PARM_DESC(fail_over_mac, "For active-backup, do not set all slaves to "
				"the same MAC; 0 for none (default), "
				"1 for active, 2 for follow");
180
module_param(all_slaves_active, int, 0);
181
MODULE_PARM_DESC(all_slaves_active, "Keep all frames received on an interface "
182
				     "by setting active flag for all slaves; "
183
				     "0 for never (default), 1 for always.");
184
module_param(resend_igmp, int, 0);
185 186
MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on "
			      "link failure");
187 188 189 190
module_param(packets_per_slave, int, 0);
MODULE_PARM_DESC(packets_per_slave, "Packets to send per slave in balance-rr "
				    "mode; 0 for a random slave, 1 packet per "
				    "slave (default), >1 packets per slave.");
191 192 193 194
module_param(lp_interval, uint, 0);
MODULE_PARM_DESC(lp_interval, "The number of seconds between instances where "
			      "the bonding driver sends learning packets to "
			      "each slaves peer switch. The default is 1.");
L
Linus Torvalds 已提交
195 196 197

/*----------------------------- Global variables ----------------------------*/

198
#ifdef CONFIG_NET_POLL_CONTROLLER
199
atomic_t netpoll_block_tx = ATOMIC_INIT(0);
200 201
#endif

202
int bond_net_id __read_mostly;
L
Linus Torvalds 已提交
203

S
Stephen Hemminger 已提交
204 205
static __be32 arp_target[BOND_MAX_ARP_TARGETS];
static int arp_ip_count;
L
Linus Torvalds 已提交
206
static int bond_mode	= BOND_MODE_ROUNDROBIN;
S
Stephen Hemminger 已提交
207 208
static int xmit_hashtype = BOND_XMIT_POLICY_LAYER2;
static int lacp_fast;
L
Linus Torvalds 已提交
209 210 211

/*-------------------------- Forward declarations ---------------------------*/

212
static int bond_init(struct net_device *bond_dev);
213
static void bond_uninit(struct net_device *bond_dev);
214 215
static struct rtnl_link_stats64 *bond_get_stats(struct net_device *bond_dev,
						struct rtnl_link_stats64 *stats);
216
static void bond_slave_arr_handler(struct work_struct *work);
L
Linus Torvalds 已提交
217 218 219

/*---------------------------- General routines -----------------------------*/

220
const char *bond_mode_name(int mode)
L
Linus Torvalds 已提交
221
{
222 223 224 225 226
	static const char *names[] = {
		[BOND_MODE_ROUNDROBIN] = "load balancing (round-robin)",
		[BOND_MODE_ACTIVEBACKUP] = "fault-tolerance (active-backup)",
		[BOND_MODE_XOR] = "load balancing (xor)",
		[BOND_MODE_BROADCAST] = "fault-tolerance (broadcast)",
S
Stephen Hemminger 已提交
227
		[BOND_MODE_8023AD] = "IEEE 802.3ad Dynamic link aggregation",
228 229 230 231
		[BOND_MODE_TLB] = "transmit load balancing",
		[BOND_MODE_ALB] = "adaptive load balancing",
	};

232
	if (mode < BOND_MODE_ROUNDROBIN || mode > BOND_MODE_ALB)
L
Linus Torvalds 已提交
233
		return "unknown";
234 235

	return names[mode];
L
Linus Torvalds 已提交
236 237 238 239 240 241
}

/*---------------------------------- VLAN -----------------------------------*/

/**
 * bond_dev_queue_xmit - Prepare skb for xmit.
S
Stephen Hemminger 已提交
242
 *
L
Linus Torvalds 已提交
243 244 245 246
 * @bond: bond device that got this skb for tx.
 * @skb: hw accel VLAN tagged skb to transmit
 * @slave_dev: slave that is supposed to xmit this skbuff
 */
247
void bond_dev_queue_xmit(struct bonding *bond, struct sk_buff *skb,
S
Stephen Hemminger 已提交
248
			struct net_device *slave_dev)
L
Linus Torvalds 已提交
249
{
250
	skb->dev = slave_dev;
251

252
	BUILD_BUG_ON(sizeof(skb->queue_mapping) !=
253 254
		     sizeof(qdisc_skb_cb(skb)->slave_dev_queue_mapping));
	skb->queue_mapping = qdisc_skb_cb(skb)->slave_dev_queue_mapping;
255

256
	if (unlikely(netpoll_tx_running(bond->dev)))
257
		bond_netpoll_send_skb(bond_get_slave_by_dev(bond, slave_dev), skb);
258
	else
259
		dev_queue_xmit(skb);
L
Linus Torvalds 已提交
260 261
}

262
/* In the following 2 functions, bond_vlan_rx_add_vid and bond_vlan_rx_kill_vid,
J
Jiri Pirko 已提交
263
 * We don't protect the slave list iteration with a lock because:
L
Linus Torvalds 已提交
264 265 266 267
 * a. This operation is performed in IOCTL context,
 * b. The operation is protected by the RTNL semaphore in the 8021q code,
 * c. Holding a lock with BH disabled while directly calling a base driver
 *    entry point is generally a BAD idea.
S
Stephen Hemminger 已提交
268
 *
L
Linus Torvalds 已提交
269 270 271 272 273 274 275 276 277 278 279 280 281 282
 * The design of synchronization/protection for this operation in the 8021q
 * module is good for one or more VLAN devices over a single physical device
 * and cannot be extended for a teaming solution like bonding, so there is a
 * potential race condition here where a net device from the vlan group might
 * be referenced (either by a base driver or the 8021q code) while it is being
 * removed from the system. However, it turns out we're not making matters
 * worse, and if it works for regular VLAN usage it will work here too.
*/

/**
 * bond_vlan_rx_add_vid - Propagates adding an id to slaves
 * @bond_dev: bonding net device that got called
 * @vid: vlan id being added
 */
283 284
static int bond_vlan_rx_add_vid(struct net_device *bond_dev,
				__be16 proto, u16 vid)
L
Linus Torvalds 已提交
285
{
286
	struct bonding *bond = netdev_priv(bond_dev);
287
	struct slave *slave, *rollback_slave;
288
	struct list_head *iter;
289
	int res;
L
Linus Torvalds 已提交
290

291
	bond_for_each_slave(bond, slave, iter) {
292
		res = vlan_vid_add(slave->dev, proto, vid);
293 294
		if (res)
			goto unwind;
L
Linus Torvalds 已提交
295 296
	}

297
	return 0;
298 299

unwind:
300
	/* unwind to the slave that failed */
301
	bond_for_each_slave(bond, rollback_slave, iter) {
302 303 304 305 306
		if (rollback_slave == slave)
			break;

		vlan_vid_del(rollback_slave->dev, proto, vid);
	}
307 308

	return res;
L
Linus Torvalds 已提交
309 310 311 312 313 314 315
}

/**
 * bond_vlan_rx_kill_vid - Propagates deleting an id to slaves
 * @bond_dev: bonding net device that got called
 * @vid: vlan id being removed
 */
316 317
static int bond_vlan_rx_kill_vid(struct net_device *bond_dev,
				 __be16 proto, u16 vid)
L
Linus Torvalds 已提交
318
{
319
	struct bonding *bond = netdev_priv(bond_dev);
320
	struct list_head *iter;
L
Linus Torvalds 已提交
321 322
	struct slave *slave;

323
	bond_for_each_slave(bond, slave, iter)
324
		vlan_vid_del(slave->dev, proto, vid);
L
Linus Torvalds 已提交
325

326 327
	if (bond_is_lb(bond))
		bond_alb_clear_vlan(bond, vid);
328 329

	return 0;
L
Linus Torvalds 已提交
330 331 332 333
}

/*------------------------------- Link status -------------------------------*/

334
/* Set the carrier state for the master according to the state of its
335 336 337 338 339
 * slaves.  If any slaves are up, the master is up.  In 802.3ad mode,
 * do special 802.3ad magic.
 *
 * Returns zero if carrier state does not change, nonzero if it does.
 */
340
int bond_set_carrier(struct bonding *bond)
341
{
342
	struct list_head *iter;
343 344
	struct slave *slave;

345
	if (!bond_has_slaves(bond))
346 347
		goto down;

348
	if (BOND_MODE(bond) == BOND_MODE_8023AD)
349 350
		return bond_3ad_set_carrier(bond);

351
	bond_for_each_slave(bond, slave, iter) {
352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368
		if (slave->link == BOND_LINK_UP) {
			if (!netif_carrier_ok(bond->dev)) {
				netif_carrier_on(bond->dev);
				return 1;
			}
			return 0;
		}
	}

down:
	if (netif_carrier_ok(bond->dev)) {
		netif_carrier_off(bond->dev);
		return 1;
	}
	return 0;
}

369
/* Get link speed and duplex from the slave's base driver
L
Linus Torvalds 已提交
370
 * using ethtool. If for some reason the call fails or the
371
 * values are invalid, set speed and duplex to -1,
372
 * and return.
L
Linus Torvalds 已提交
373
 */
374
static void bond_update_speed_duplex(struct slave *slave)
L
Linus Torvalds 已提交
375 376
{
	struct net_device *slave_dev = slave->dev;
377
	struct ethtool_cmd ecmd;
378
	u32 slave_speed;
379
	int res;
L
Linus Torvalds 已提交
380

381 382
	slave->speed = SPEED_UNKNOWN;
	slave->duplex = DUPLEX_UNKNOWN;
L
Linus Torvalds 已提交
383

384
	res = __ethtool_get_settings(slave_dev, &ecmd);
385
	if (res < 0)
386
		return;
L
Linus Torvalds 已提交
387

388
	slave_speed = ethtool_cmd_speed(&ecmd);
J
Jiri Pirko 已提交
389
	if (slave_speed == 0 || slave_speed == ((__u32) -1))
390
		return;
L
Linus Torvalds 已提交
391

392
	switch (ecmd.duplex) {
L
Linus Torvalds 已提交
393 394 395 396
	case DUPLEX_FULL:
	case DUPLEX_HALF:
		break;
	default:
397
		return;
L
Linus Torvalds 已提交
398 399
	}

400
	slave->speed = slave_speed;
401
	slave->duplex = ecmd.duplex;
L
Linus Torvalds 已提交
402

403
	return;
L
Linus Torvalds 已提交
404 405
}

406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
const char *bond_slave_link_status(s8 link)
{
	switch (link) {
	case BOND_LINK_UP:
		return "up";
	case BOND_LINK_FAIL:
		return "going down";
	case BOND_LINK_DOWN:
		return "down";
	case BOND_LINK_BACK:
		return "going back";
	default:
		return "unknown";
	}
}

422
/* if <dev> supports MII link status reporting, check its link status.
L
Linus Torvalds 已提交
423 424
 *
 * We either do MII/ETHTOOL ioctls, or check netif_carrier_ok(),
S
Stephen Hemminger 已提交
425
 * depending upon the setting of the use_carrier parameter.
L
Linus Torvalds 已提交
426 427 428 429 430 431 432 433 434 435 436
 *
 * Return either BMSR_LSTATUS, meaning that the link is up (or we
 * can't tell and just pretend it is), or 0, meaning that the link is
 * down.
 *
 * If reporting is non-zero, instead of faking link up, return -1 if
 * both ETHTOOL and MII ioctls fail (meaning the device does not
 * support them).  If use_carrier is set, return whatever it says.
 * It'd be nice if there was a good way to tell if a driver supports
 * netif_carrier, but there really isn't.
 */
S
Stephen Hemminger 已提交
437 438
static int bond_check_dev_link(struct bonding *bond,
			       struct net_device *slave_dev, int reporting)
L
Linus Torvalds 已提交
439
{
440
	const struct net_device_ops *slave_ops = slave_dev->netdev_ops;
441
	int (*ioctl)(struct net_device *, struct ifreq *, int);
L
Linus Torvalds 已提交
442 443 444
	struct ifreq ifr;
	struct mii_ioctl_data *mii;

445 446 447
	if (!reporting && !netif_running(slave_dev))
		return 0;

448
	if (bond->params.use_carrier)
L
Linus Torvalds 已提交
449 450
		return netif_carrier_ok(slave_dev) ? BMSR_LSTATUS : 0;

451
	/* Try to get link status using Ethtool first. */
452 453 454
	if (slave_dev->ethtool_ops->get_link)
		return slave_dev->ethtool_ops->get_link(slave_dev) ?
			BMSR_LSTATUS : 0;
455

S
Stephen Hemminger 已提交
456
	/* Ethtool can't be used, fallback to MII ioctls. */
457
	ioctl = slave_ops->ndo_do_ioctl;
L
Linus Torvalds 已提交
458
	if (ioctl) {
459 460 461 462 463 464 465 466
		/* TODO: set pointer to correct ioctl on a per team member
		 *       bases to make this more efficient. that is, once
		 *       we determine the correct ioctl, we will always
		 *       call it and not the others for that team
		 *       member.
		 */

		/* We cannot assume that SIOCGMIIPHY will also read a
L
Linus Torvalds 已提交
467 468 469 470 471 472 473 474 475
		 * register; not all network drivers (e.g., e100)
		 * support that.
		 */

		/* Yes, the mii is overlaid on the ifreq.ifr_ifru */
		strncpy(ifr.ifr_name, slave_dev->name, IFNAMSIZ);
		mii = if_mii(&ifr);
		if (IOCTL(slave_dev, &ifr, SIOCGMIIPHY) == 0) {
			mii->reg_num = MII_BMSR;
S
Stephen Hemminger 已提交
476 477
			if (IOCTL(slave_dev, &ifr, SIOCGMIIREG) == 0)
				return mii->val_out & BMSR_LSTATUS;
L
Linus Torvalds 已提交
478 479 480
		}
	}

481
	/* If reporting, report that either there's no dev->do_ioctl,
482
	 * or both SIOCGMIIREG and get_link failed (meaning that we
L
Linus Torvalds 已提交
483 484 485
	 * cannot report link status).  If not reporting, pretend
	 * we're ok.
	 */
S
Stephen Hemminger 已提交
486
	return reporting ? -1 : BMSR_LSTATUS;
L
Linus Torvalds 已提交
487 488 489 490
}

/*----------------------------- Multicast list ------------------------------*/

491
/* Push the promiscuity flag down to appropriate slaves */
492
static int bond_set_promiscuity(struct bonding *bond, int inc)
L
Linus Torvalds 已提交
493
{
494
	struct list_head *iter;
495
	int err = 0;
496

497
	if (bond_uses_primary(bond)) {
498
		struct slave *curr_active = rtnl_dereference(bond->curr_active_slave);
499 500 501

		if (curr_active)
			err = dev_set_promiscuity(curr_active->dev, inc);
L
Linus Torvalds 已提交
502 503
	} else {
		struct slave *slave;
504

505
		bond_for_each_slave(bond, slave, iter) {
506 507 508
			err = dev_set_promiscuity(slave->dev, inc);
			if (err)
				return err;
L
Linus Torvalds 已提交
509 510
		}
	}
511
	return err;
L
Linus Torvalds 已提交
512 513
}

514
/* Push the allmulti flag down to all slaves */
515
static int bond_set_allmulti(struct bonding *bond, int inc)
L
Linus Torvalds 已提交
516
{
517
	struct list_head *iter;
518
	int err = 0;
519

520
	if (bond_uses_primary(bond)) {
521
		struct slave *curr_active = rtnl_dereference(bond->curr_active_slave);
522 523 524

		if (curr_active)
			err = dev_set_allmulti(curr_active->dev, inc);
L
Linus Torvalds 已提交
525 526
	} else {
		struct slave *slave;
527

528
		bond_for_each_slave(bond, slave, iter) {
529 530 531
			err = dev_set_allmulti(slave->dev, inc);
			if (err)
				return err;
L
Linus Torvalds 已提交
532 533
		}
	}
534
	return err;
L
Linus Torvalds 已提交
535 536
}

537
/* Retrieve the list of registered multicast addresses for the bonding
538 539 540
 * device and retransmit an IGMP JOIN request to the current active
 * slave.
 */
541
static void bond_resend_igmp_join_requests_delayed(struct work_struct *work)
542
{
543 544 545
	struct bonding *bond = container_of(work, struct bonding,
					    mcast_work.work);

546
	if (!rtnl_trylock()) {
547
		queue_delayed_work(bond->wq, &bond->mcast_work, 1);
548
		return;
549
	}
550
	call_netdevice_notifiers(NETDEV_RESEND_IGMP, bond->dev);
551

552 553
	if (bond->igmp_retrans > 1) {
		bond->igmp_retrans--;
554
		queue_delayed_work(bond->wq, &bond->mcast_work, HZ/5);
555
	}
556
	rtnl_unlock();
557 558
}

559
/* Flush bond's hardware addresses from slave */
560
static void bond_hw_addr_flush(struct net_device *bond_dev,
S
Stephen Hemminger 已提交
561
			       struct net_device *slave_dev)
L
Linus Torvalds 已提交
562
{
563
	struct bonding *bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
564

565 566
	dev_uc_unsync(slave_dev, bond_dev);
	dev_mc_unsync(slave_dev, bond_dev);
L
Linus Torvalds 已提交
567

568
	if (BOND_MODE(bond) == BOND_MODE_8023AD) {
L
Linus Torvalds 已提交
569 570 571
		/* del lacpdu mc addr from mc list */
		u8 lacpdu_multicast[ETH_ALEN] = MULTICAST_LACPDU_ADDR;

572
		dev_mc_del(slave_dev, lacpdu_multicast);
L
Linus Torvalds 已提交
573 574 575 576 577
	}
}

/*--------------------------- Active slave change ---------------------------*/

578
/* Update the hardware address list and promisc/allmulti for the new and
579 580
 * old active slaves (if any).  Modes that are not using primary keep all
 * slaves up date at all times; only the modes that use primary need to call
581
 * this function to swap these settings during a failover.
L
Linus Torvalds 已提交
582
 */
583 584
static void bond_hw_addr_swap(struct bonding *bond, struct slave *new_active,
			      struct slave *old_active)
L
Linus Torvalds 已提交
585 586
{
	if (old_active) {
S
Stephen Hemminger 已提交
587
		if (bond->dev->flags & IFF_PROMISC)
L
Linus Torvalds 已提交
588 589
			dev_set_promiscuity(old_active->dev, -1);

S
Stephen Hemminger 已提交
590
		if (bond->dev->flags & IFF_ALLMULTI)
L
Linus Torvalds 已提交
591 592
			dev_set_allmulti(old_active->dev, -1);

593
		bond_hw_addr_flush(bond->dev, old_active->dev);
L
Linus Torvalds 已提交
594 595 596
	}

	if (new_active) {
597
		/* FIXME: Signal errors upstream. */
S
Stephen Hemminger 已提交
598
		if (bond->dev->flags & IFF_PROMISC)
L
Linus Torvalds 已提交
599 600
			dev_set_promiscuity(new_active->dev, 1);

S
Stephen Hemminger 已提交
601
		if (bond->dev->flags & IFF_ALLMULTI)
L
Linus Torvalds 已提交
602 603
			dev_set_allmulti(new_active->dev, 1);

604
		netif_addr_lock_bh(bond->dev);
605 606
		dev_uc_sync(new_active->dev, bond->dev);
		dev_mc_sync(new_active->dev, bond->dev);
607
		netif_addr_unlock_bh(bond->dev);
L
Linus Torvalds 已提交
608 609 610
	}
}

611 612 613 614 615 616 617 618 619 620
/**
 * bond_set_dev_addr - clone slave's address to bond
 * @bond_dev: bond net device
 * @slave_dev: slave net device
 *
 * Should be called with RTNL held.
 */
static void bond_set_dev_addr(struct net_device *bond_dev,
			      struct net_device *slave_dev)
{
621 622
	netdev_dbg(bond_dev, "bond_dev=%p slave_dev=%p slave_dev->addr_len=%d\n",
		   bond_dev, slave_dev, slave_dev->addr_len);
623 624 625 626 627
	memcpy(bond_dev->dev_addr, slave_dev->dev_addr, slave_dev->addr_len);
	bond_dev->addr_assign_type = NET_ADDR_STOLEN;
	call_netdevice_notifiers(NETDEV_CHANGEADDR, bond_dev);
}

628
/* bond_do_fail_over_mac
629 630 631
 *
 * Perform special MAC address swapping for fail_over_mac settings
 *
632
 * Called with RTNL
633 634 635 636 637 638 639 640 641 642 643
 */
static void bond_do_fail_over_mac(struct bonding *bond,
				  struct slave *new_active,
				  struct slave *old_active)
{
	u8 tmp_mac[ETH_ALEN];
	struct sockaddr saddr;
	int rv;

	switch (bond->params.fail_over_mac) {
	case BOND_FOM_ACTIVE:
644
		if (new_active)
645
			bond_set_dev_addr(bond->dev, new_active->dev);
646 647
		break;
	case BOND_FOM_FOLLOW:
648
		/* if new_active && old_active, swap them
649 650 651 652 653 654 655
		 * if just old_active, do nothing (going to no active slave)
		 * if just new_active, set new_active to bond's MAC
		 */
		if (!new_active)
			return;

		if (old_active) {
656
			ether_addr_copy(tmp_mac, new_active->dev->dev_addr);
657 658
			ether_addr_copy(saddr.sa_data,
					old_active->dev->dev_addr);
659 660
			saddr.sa_family = new_active->dev->type;
		} else {
661
			ether_addr_copy(saddr.sa_data, bond->dev->dev_addr);
662 663 664 665 666
			saddr.sa_family = bond->dev->type;
		}

		rv = dev_set_mac_address(new_active->dev, &saddr);
		if (rv) {
667 668
			netdev_err(bond->dev, "Error %d setting MAC of slave %s\n",
				   -rv, new_active->dev->name);
669 670 671 672 673 674
			goto out;
		}

		if (!old_active)
			goto out;

675
		ether_addr_copy(saddr.sa_data, tmp_mac);
676 677 678 679
		saddr.sa_family = old_active->dev->type;

		rv = dev_set_mac_address(old_active->dev, &saddr);
		if (rv)
680 681
			netdev_err(bond->dev, "Error %d setting MAC of slave %s\n",
				   -rv, new_active->dev->name);
682 683 684
out:
		break;
	default:
685 686
		netdev_err(bond->dev, "bond_do_fail_over_mac impossible: bad policy %d\n",
			   bond->params.fail_over_mac);
687 688 689 690 691
		break;
	}

}

692 693
static bool bond_should_change_active(struct bonding *bond)
{
694
	struct slave *prim = rtnl_dereference(bond->primary_slave);
695
	struct slave *curr = rtnl_dereference(bond->curr_active_slave);
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710

	if (!prim || !curr || curr->link != BOND_LINK_UP)
		return true;
	if (bond->force_primary) {
		bond->force_primary = false;
		return true;
	}
	if (bond->params.primary_reselect == BOND_PRI_RESELECT_BETTER &&
	    (prim->speed < curr->speed ||
	     (prim->speed == curr->speed && prim->duplex <= curr->duplex)))
		return false;
	if (bond->params.primary_reselect == BOND_PRI_RESELECT_FAILURE)
		return false;
	return true;
}
711

L
Linus Torvalds 已提交
712 713 714 715 716 717
/**
 * find_best_interface - select the best available slave to be the active one
 * @bond: our bonding struct
 */
static struct slave *bond_find_best_slave(struct bonding *bond)
{
718
	struct slave *slave, *bestslave = NULL, *primary;
719
	struct list_head *iter;
L
Linus Torvalds 已提交
720 721
	int mintime = bond->params.updelay;

722 723
	primary = rtnl_dereference(bond->primary_slave);
	if (primary && primary->link == BOND_LINK_UP &&
724
	    bond_should_change_active(bond))
725
		return primary;
L
Linus Torvalds 已提交
726

727 728 729
	bond_for_each_slave(bond, slave, iter) {
		if (slave->link == BOND_LINK_UP)
			return slave;
730
		if (slave->link == BOND_LINK_BACK && bond_slave_is_up(slave) &&
731 732 733
		    slave->delay < mintime) {
			mintime = slave->delay;
			bestslave = slave;
L
Linus Torvalds 已提交
734 735 736 737 738 739
		}
	}

	return bestslave;
}

740 741
static bool bond_should_notify_peers(struct bonding *bond)
{
742 743 744 745 746
	struct slave *slave;

	rcu_read_lock();
	slave = rcu_dereference(bond->curr_active_slave);
	rcu_read_unlock();
747

748 749
	netdev_dbg(bond->dev, "bond_should_notify_peers: slave %s\n",
		   slave ? slave->dev->name : "NULL");
750 751 752 753 754 755 756 757

	if (!slave || !bond->send_peer_notif ||
	    test_bit(__LINK_STATE_LINKWATCH_PENDING, &slave->dev->state))
		return false;

	return true;
}

L
Linus Torvalds 已提交
758 759 760 761 762 763 764 765 766 767 768 769 770
/**
 * change_active_interface - change the active slave into the specified one
 * @bond: our bonding struct
 * @new: the new slave to make the active one
 *
 * Set the new slave to the bond's settings and unset them on the old
 * curr_active_slave.
 * Setting include flags, mc-list, promiscuity, allmulti, etc.
 *
 * If @new's link state is %BOND_LINK_BACK we'll set it to %BOND_LINK_UP,
 * because it is apparently the best available slave we have, even though its
 * updelay hasn't timed out yet.
 *
771
 * Caller must hold RTNL.
L
Linus Torvalds 已提交
772
 */
773
void bond_change_active_slave(struct bonding *bond, struct slave *new_active)
L
Linus Torvalds 已提交
774
{
775 776
	struct slave *old_active;

777 778 779
	ASSERT_RTNL();

	old_active = rtnl_dereference(bond->curr_active_slave);
L
Linus Torvalds 已提交
780

S
Stephen Hemminger 已提交
781
	if (old_active == new_active)
L
Linus Torvalds 已提交
782 783 784
		return;

	if (new_active) {
785
		new_active->last_link_up = jiffies;
786

L
Linus Torvalds 已提交
787
		if (new_active->link == BOND_LINK_BACK) {
788
			if (bond_uses_primary(bond)) {
789 790 791
				netdev_info(bond->dev, "making interface %s the new active one %d ms earlier\n",
					    new_active->dev->name,
					    (bond->params.updelay - new_active->delay) * bond->params.miimon);
L
Linus Torvalds 已提交
792 793 794
			}

			new_active->delay = 0;
795
			bond_set_slave_link_state(new_active, BOND_LINK_UP);
L
Linus Torvalds 已提交
796

797
			if (BOND_MODE(bond) == BOND_MODE_8023AD)
L
Linus Torvalds 已提交
798 799
				bond_3ad_handle_link_change(new_active, BOND_LINK_UP);

800
			if (bond_is_lb(bond))
L
Linus Torvalds 已提交
801 802
				bond_alb_handle_link_change(bond, new_active, BOND_LINK_UP);
		} else {
803
			if (bond_uses_primary(bond)) {
804 805
				netdev_info(bond->dev, "making interface %s the new active one\n",
					    new_active->dev->name);
L
Linus Torvalds 已提交
806 807 808 809
			}
		}
	}

810
	if (bond_uses_primary(bond))
811
		bond_hw_addr_swap(bond, new_active, old_active);
L
Linus Torvalds 已提交
812

813
	if (bond_is_lb(bond)) {
L
Linus Torvalds 已提交
814
		bond_alb_handle_active_change(bond, new_active);
815
		if (old_active)
816 817
			bond_set_slave_inactive_flags(old_active,
						      BOND_SLAVE_NOTIFY_NOW);
818
		if (new_active)
819 820
			bond_set_slave_active_flags(new_active,
						    BOND_SLAVE_NOTIFY_NOW);
L
Linus Torvalds 已提交
821
	} else {
822
		rcu_assign_pointer(bond->curr_active_slave, new_active);
L
Linus Torvalds 已提交
823
	}
J
Jay Vosburgh 已提交
824

825
	if (BOND_MODE(bond) == BOND_MODE_ACTIVEBACKUP) {
S
Stephen Hemminger 已提交
826
		if (old_active)
827 828
			bond_set_slave_inactive_flags(old_active,
						      BOND_SLAVE_NOTIFY_NOW);
J
Jay Vosburgh 已提交
829 830

		if (new_active) {
831 832
			bool should_notify_peers = false;

833 834
			bond_set_slave_active_flags(new_active,
						    BOND_SLAVE_NOTIFY_NOW);
835

836 837 838
			if (bond->params.fail_over_mac)
				bond_do_fail_over_mac(bond, new_active,
						      old_active);
839

840 841 842 843 844 845 846
			if (netif_running(bond->dev)) {
				bond->send_peer_notif =
					bond->params.num_peer_notif;
				should_notify_peers =
					bond_should_notify_peers(bond);
			}

847
			call_netdevice_notifiers(NETDEV_BONDING_FAILOVER, bond->dev);
848
			if (should_notify_peers)
849 850
				call_netdevice_notifiers(NETDEV_NOTIFY_PEERS,
							 bond->dev);
851
		}
J
Jay Vosburgh 已提交
852
	}
853

854
	/* resend IGMP joins since active slave has changed or
855 856
	 * all were sent on curr_active_slave.
	 * resend only if bond is brought up with the affected
857 858
	 * bonding modes and the retransmission is enabled
	 */
859
	if (netif_running(bond->dev) && (bond->params.resend_igmp > 0) &&
860
	    ((bond_uses_primary(bond) && new_active) ||
861
	     BOND_MODE(bond) == BOND_MODE_ROUNDROBIN)) {
862
		bond->igmp_retrans = bond->params.resend_igmp;
863
		queue_delayed_work(bond->wq, &bond->mcast_work, 1);
864
	}
L
Linus Torvalds 已提交
865 866 867 868 869 870
}

/**
 * bond_select_active_slave - select a new active slave, if needed
 * @bond: our bonding struct
 *
S
Stephen Hemminger 已提交
871
 * This functions should be called when one of the following occurs:
L
Linus Torvalds 已提交
872 873 874 875
 * - The old curr_active_slave has been released or lost its link.
 * - The primary_slave has got its link back.
 * - A slave has got its link back and there's no old curr_active_slave.
 *
876
 * Caller must hold RTNL.
L
Linus Torvalds 已提交
877
 */
878
void bond_select_active_slave(struct bonding *bond)
L
Linus Torvalds 已提交
879 880
{
	struct slave *best_slave;
881
	int rv;
L
Linus Torvalds 已提交
882

883 884
	ASSERT_RTNL();

L
Linus Torvalds 已提交
885
	best_slave = bond_find_best_slave(bond);
886
	if (best_slave != rtnl_dereference(bond->curr_active_slave)) {
L
Linus Torvalds 已提交
887
		bond_change_active_slave(bond, best_slave);
888 889 890 891 892
		rv = bond_set_carrier(bond);
		if (!rv)
			return;

		if (netif_carrier_ok(bond->dev)) {
893
			netdev_info(bond->dev, "first active interface up!\n");
894
		} else {
895
			netdev_info(bond->dev, "now running without any active interface!\n");
896
		}
L
Linus Torvalds 已提交
897 898 899
	}
}

900
#ifdef CONFIG_NET_POLL_CONTROLLER
901
static inline int slave_enable_netpoll(struct slave *slave)
902
{
903 904
	struct netpoll *np;
	int err = 0;
905

906
	np = kzalloc(sizeof(*np), GFP_KERNEL);
907 908 909 910
	err = -ENOMEM;
	if (!np)
		goto out;

911
	err = __netpoll_setup(np, slave->dev);
912 913 914
	if (err) {
		kfree(np);
		goto out;
915
	}
916 917 918 919 920 921 922 923 924 925 926 927
	slave->np = np;
out:
	return err;
}
static inline void slave_disable_netpoll(struct slave *slave)
{
	struct netpoll *np = slave->np;

	if (!np)
		return;

	slave->np = NULL;
928
	__netpoll_free_async(np);
929
}
930 931 932

static void bond_poll_controller(struct net_device *bond_dev)
{
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
	struct bonding *bond = netdev_priv(bond_dev);
	struct slave *slave = NULL;
	struct list_head *iter;
	struct ad_info ad_info;
	struct netpoll_info *ni;
	const struct net_device_ops *ops;

	if (BOND_MODE(bond) == BOND_MODE_8023AD)
		if (bond_3ad_get_active_agg_info(bond, &ad_info))
			return;

	rcu_read_lock_bh();
	bond_for_each_slave_rcu(bond, slave, iter) {
		ops = slave->dev->netdev_ops;
		if (!bond_slave_is_up(slave) || !ops->ndo_poll_controller)
			continue;

		if (BOND_MODE(bond) == BOND_MODE_8023AD) {
			struct aggregator *agg =
			    SLAVE_AD_INFO(slave)->port.aggregator;

			if (agg &&
			    agg->aggregator_identifier != ad_info.aggregator_id)
				continue;
		}

		ni = rcu_dereference_bh(slave->dev->npinfo);
		if (down_trylock(&ni->dev_lock))
			continue;
		ops->ndo_poll_controller(slave->dev);
		up(&ni->dev_lock);
	}
	rcu_read_unlock_bh();
966 967
}

968
static void bond_netpoll_cleanup(struct net_device *bond_dev)
969
{
970
	struct bonding *bond = netdev_priv(bond_dev);
971
	struct list_head *iter;
972 973
	struct slave *slave;

974
	bond_for_each_slave(bond, slave, iter)
975
		if (bond_slave_is_up(slave))
976
			slave_disable_netpoll(slave);
977
}
978

979
static int bond_netpoll_setup(struct net_device *dev, struct netpoll_info *ni)
980 981
{
	struct bonding *bond = netdev_priv(dev);
982
	struct list_head *iter;
983
	struct slave *slave;
984
	int err = 0;
985

986
	bond_for_each_slave(bond, slave, iter) {
987 988
		err = slave_enable_netpoll(slave);
		if (err) {
989
			bond_netpoll_cleanup(dev);
990
			break;
991 992
		}
	}
993
	return err;
994
}
995 996 997 998 999 1000 1001 1002
#else
static inline int slave_enable_netpoll(struct slave *slave)
{
	return 0;
}
static inline void slave_disable_netpoll(struct slave *slave)
{
}
1003 1004 1005 1006 1007
static void bond_netpoll_cleanup(struct net_device *bond_dev)
{
}
#endif

L
Linus Torvalds 已提交
1008 1009
/*---------------------------------- IOCTL ----------------------------------*/

1010
static netdev_features_t bond_fix_features(struct net_device *dev,
1011
					   netdev_features_t features)
1012
{
1013
	struct bonding *bond = netdev_priv(dev);
1014
	struct list_head *iter;
1015
	netdev_features_t mask;
1016
	struct slave *slave;
1017

1018 1019 1020 1021 1022
	/* If any slave has the offload feature flag set,
	 * set the offload flag on the bond.
	 */
	mask = features | NETIF_F_HW_SWITCH_OFFLOAD;

1023
	features &= ~NETIF_F_ONE_FOR_ALL;
1024
	features |= NETIF_F_ALL_FOR_ALL;
1025

1026
	bond_for_each_slave(bond, slave, iter) {
1027 1028
		features = netdev_increment_features(features,
						     slave->dev->features,
1029 1030
						     mask);
	}
1031
	features = netdev_add_tso_features(features, mask);
1032 1033 1034 1035

	return features;
}

1036 1037 1038
#define BOND_VLAN_FEATURES	(NETIF_F_ALL_CSUM | NETIF_F_SG | \
				 NETIF_F_FRAGLIST | NETIF_F_ALL_TSO | \
				 NETIF_F_HIGHDMA | NETIF_F_LRO)
1039

1040
#define BOND_ENC_FEATURES	(NETIF_F_ALL_CSUM | NETIF_F_SG | NETIF_F_RXCSUM |\
E
Eric Dumazet 已提交
1041
				 NETIF_F_TSO)
1042

1043 1044
static void bond_compute_features(struct bonding *bond)
{
1045 1046
	unsigned int dst_release_flag = IFF_XMIT_DST_RELEASE |
					IFF_XMIT_DST_RELEASE_PERM;
1047
	netdev_features_t vlan_features = BOND_VLAN_FEATURES;
1048
	netdev_features_t enc_features  = BOND_ENC_FEATURES;
1049 1050 1051
	struct net_device *bond_dev = bond->dev;
	struct list_head *iter;
	struct slave *slave;
1052
	unsigned short max_hard_header_len = ETH_HLEN;
1053 1054
	unsigned int gso_max_size = GSO_MAX_SIZE;
	u16 gso_max_segs = GSO_MAX_SEGS;
1055

1056
	if (!bond_has_slaves(bond))
1057
		goto done;
1058
	vlan_features &= NETIF_F_ALL_FOR_ALL;
1059

1060
	bond_for_each_slave(bond, slave, iter) {
1061
		vlan_features = netdev_increment_features(vlan_features,
1062 1063
			slave->dev->vlan_features, BOND_VLAN_FEATURES);

1064 1065 1066
		enc_features = netdev_increment_features(enc_features,
							 slave->dev->hw_enc_features,
							 BOND_ENC_FEATURES);
1067
		dst_release_flag &= slave->dev->priv_flags;
1068 1069
		if (slave->dev->hard_header_len > max_hard_header_len)
			max_hard_header_len = slave->dev->hard_header_len;
1070 1071 1072

		gso_max_size = min(gso_max_size, slave->dev->gso_max_size);
		gso_max_segs = min(gso_max_segs, slave->dev->gso_max_segs);
1073
	}
1074

1075
done:
1076
	bond_dev->vlan_features = vlan_features;
E
Eric Dumazet 已提交
1077
	bond_dev->hw_enc_features = enc_features | NETIF_F_GSO_ENCAP_ALL;
1078
	bond_dev->hard_header_len = max_hard_header_len;
1079 1080
	bond_dev->gso_max_segs = gso_max_segs;
	netif_set_gso_max_size(bond_dev, gso_max_size);
1081

1082 1083 1084 1085
	bond_dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
	if ((bond_dev->priv_flags & IFF_XMIT_DST_RELEASE_PERM) &&
	    dst_release_flag == (IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM))
		bond_dev->priv_flags |= IFF_XMIT_DST_RELEASE;
1086

1087
	netdev_change_features(bond_dev);
1088 1089
}

1090 1091 1092
static void bond_setup_by_slave(struct net_device *bond_dev,
				struct net_device *slave_dev)
{
1093
	bond_dev->header_ops	    = slave_dev->header_ops;
1094 1095 1096 1097 1098 1099 1100 1101 1102

	bond_dev->type		    = slave_dev->type;
	bond_dev->hard_header_len   = slave_dev->hard_header_len;
	bond_dev->addr_len	    = slave_dev->addr_len;

	memcpy(bond_dev->broadcast, slave_dev->broadcast,
		slave_dev->addr_len);
}

1103
/* On bonding slaves other than the currently active slave, suppress
1104
 * duplicates except for alb non-mcast/bcast.
1105 1106
 */
static bool bond_should_deliver_exact_match(struct sk_buff *skb,
1107 1108
					    struct slave *slave,
					    struct bonding *bond)
1109
{
1110
	if (bond_is_slave_inactive(slave)) {
1111
		if (BOND_MODE(bond) == BOND_MODE_ALB &&
1112 1113 1114 1115 1116 1117 1118 1119
		    skb->pkt_type != PACKET_BROADCAST &&
		    skb->pkt_type != PACKET_MULTICAST)
			return false;
		return true;
	}
	return false;
}

1120
static rx_handler_result_t bond_handle_frame(struct sk_buff **pskb)
1121
{
1122
	struct sk_buff *skb = *pskb;
1123
	struct slave *slave;
1124
	struct bonding *bond;
1125 1126
	int (*recv_probe)(const struct sk_buff *, struct bonding *,
			  struct slave *);
1127
	int ret = RX_HANDLER_ANOTHER;
1128

1129 1130 1131 1132 1133
	skb = skb_share_check(skb, GFP_ATOMIC);
	if (unlikely(!skb))
		return RX_HANDLER_CONSUMED;

	*pskb = skb;
1134

J
Jiri Pirko 已提交
1135 1136
	slave = bond_slave_get_rcu(skb->dev);
	bond = slave->bond;
1137

1138 1139
	recv_probe = ACCESS_ONCE(bond->recv_probe);
	if (recv_probe) {
1140 1141 1142 1143
		ret = recv_probe(skb, bond, slave);
		if (ret == RX_HANDLER_CONSUMED) {
			consume_skb(skb);
			return ret;
1144 1145 1146
		}
	}

1147
	if (bond_should_deliver_exact_match(skb, slave, bond)) {
1148
		return RX_HANDLER_EXACT;
1149 1150
	}

J
Jiri Pirko 已提交
1151
	skb->dev = bond->dev;
1152

1153
	if (BOND_MODE(bond) == BOND_MODE_ALB &&
J
Jiri Pirko 已提交
1154
	    bond->dev->priv_flags & IFF_BRIDGE_PORT &&
1155 1156
	    skb->pkt_type == PACKET_HOST) {

1157 1158 1159
		if (unlikely(skb_cow_head(skb,
					  skb->data - skb_mac_header(skb)))) {
			kfree_skb(skb);
1160
			return RX_HANDLER_CONSUMED;
1161
		}
1162
		ether_addr_copy(eth_hdr(skb)->h_dest, bond->dev->dev_addr);
1163 1164
	}

1165
	return ret;
1166 1167
}

1168
static int bond_master_upper_dev_link(struct net_device *bond_dev,
1169 1170
				      struct net_device *slave_dev,
				      struct slave *slave)
1171 1172 1173
{
	int err;

1174
	err = netdev_master_upper_dev_link_private(slave_dev, bond_dev, slave);
1175 1176 1177
	if (err)
		return err;
	slave_dev->flags |= IFF_SLAVE;
1178
	rtmsg_ifinfo(RTM_NEWLINK, slave_dev, IFF_SLAVE, GFP_KERNEL);
1179 1180 1181 1182 1183 1184 1185 1186
	return 0;
}

static void bond_upper_dev_unlink(struct net_device *bond_dev,
				  struct net_device *slave_dev)
{
	netdev_upper_dev_unlink(slave_dev, bond_dev);
	slave_dev->flags &= ~IFF_SLAVE;
1187
	rtmsg_ifinfo(RTM_NEWLINK, slave_dev, IFF_SLAVE, GFP_KERNEL);
1188 1189
}

1190 1191 1192 1193 1194 1195 1196 1197
static struct slave *bond_alloc_slave(struct bonding *bond)
{
	struct slave *slave = NULL;

	slave = kzalloc(sizeof(struct slave), GFP_KERNEL);
	if (!slave)
		return NULL;

1198
	if (BOND_MODE(bond) == BOND_MODE_8023AD) {
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
		SLAVE_AD_INFO(slave) = kzalloc(sizeof(struct ad_slave_info),
					       GFP_KERNEL);
		if (!SLAVE_AD_INFO(slave)) {
			kfree(slave);
			return NULL;
		}
	}
	return slave;
}

static void bond_free_slave(struct slave *slave)
{
	struct bonding *bond = bond_get_bond_by_slave(slave);

1213
	if (BOND_MODE(bond) == BOND_MODE_8023AD)
1214 1215 1216 1217 1218
		kfree(SLAVE_AD_INFO(slave));

	kfree(slave);
}

1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
static void bond_fill_ifbond(struct bonding *bond, struct ifbond *info)
{
	info->bond_mode = BOND_MODE(bond);
	info->miimon = bond->params.miimon;
	info->num_slaves = bond->slave_cnt;
}

static void bond_fill_ifslave(struct slave *slave, struct ifslave *info)
{
	strcpy(info->slave_name, slave->dev->name);
	info->link = slave->link;
	info->state = bond_slave_state(slave);
	info->link_failure_count = slave->link_failure_count;
}

1234 1235
static void bond_netdev_notify(struct net_device *dev,
			       struct netdev_bonding_info *info)
1236 1237
{
	rtnl_lock();
1238
	netdev_bonding_info_change(dev, info);
1239 1240 1241 1242 1243 1244 1245 1246
	rtnl_unlock();
}

static void bond_netdev_notify_work(struct work_struct *_work)
{
	struct netdev_notify_work *w =
		container_of(_work, struct netdev_notify_work, work.work);

1247
	bond_netdev_notify(w->dev, &w->bonding_info);
1248
	dev_put(w->dev);
1249
	kfree(w);
1250 1251 1252 1253
}

void bond_queue_slave_event(struct slave *slave)
{
1254
	struct bonding *bond = slave->bond;
1255 1256 1257 1258 1259
	struct netdev_notify_work *nnw = kzalloc(sizeof(*nnw), GFP_ATOMIC);

	if (!nnw)
		return;

1260
	dev_hold(slave->dev);
1261
	nnw->dev = slave->dev;
1262 1263 1264
	bond_fill_ifslave(slave, &nnw->bonding_info.slave);
	bond_fill_ifbond(bond, &nnw->bonding_info.master);
	INIT_DELAYED_WORK(&nnw->work, bond_netdev_notify_work);
1265

1266
	queue_delayed_work(slave->bond->wq, &nnw->work, 0);
1267 1268
}

L
Linus Torvalds 已提交
1269
/* enslave device <slave> to bond device <master> */
1270
int bond_enslave(struct net_device *bond_dev, struct net_device *slave_dev)
L
Linus Torvalds 已提交
1271
{
1272
	struct bonding *bond = netdev_priv(bond_dev);
1273
	const struct net_device_ops *slave_ops = slave_dev->netdev_ops;
1274
	struct slave *new_slave = NULL, *prev_slave;
L
Linus Torvalds 已提交
1275 1276
	struct sockaddr addr;
	int link_reporting;
1277
	int res = 0, i;
L
Linus Torvalds 已提交
1278

1279 1280 1281
	if (!bond->params.use_carrier &&
	    slave_dev->ethtool_ops->get_link == NULL &&
	    slave_ops->ndo_do_ioctl == NULL) {
1282 1283
		netdev_warn(bond_dev, "no link monitoring support for %s\n",
			    slave_dev->name);
L
Linus Torvalds 已提交
1284 1285 1286 1287
	}

	/* already enslaved */
	if (slave_dev->flags & IFF_SLAVE) {
1288
		netdev_dbg(bond_dev, "Error: Device was already enslaved\n");
L
Linus Torvalds 已提交
1289 1290 1291
		return -EBUSY;
	}

1292
	if (bond_dev == slave_dev) {
1293
		netdev_err(bond_dev, "cannot enslave bond to itself.\n");
1294 1295 1296
		return -EPERM;
	}

L
Linus Torvalds 已提交
1297 1298 1299
	/* vlan challenged mutual exclusion */
	/* no need to lock since we're protected by rtnl_lock */
	if (slave_dev->features & NETIF_F_VLAN_CHALLENGED) {
1300 1301
		netdev_dbg(bond_dev, "%s is NETIF_F_VLAN_CHALLENGED\n",
			   slave_dev->name);
1302
		if (vlan_uses_dev(bond_dev)) {
1303 1304
			netdev_err(bond_dev, "Error: cannot enslave VLAN challenged slave %s on VLAN enabled bond %s\n",
				   slave_dev->name, bond_dev->name);
L
Linus Torvalds 已提交
1305 1306
			return -EPERM;
		} else {
1307 1308 1309
			netdev_warn(bond_dev, "enslaved VLAN challenged slave %s. Adding VLANs will be blocked as long as %s is part of bond %s\n",
				    slave_dev->name, slave_dev->name,
				    bond_dev->name);
L
Linus Torvalds 已提交
1310 1311
		}
	} else {
1312 1313
		netdev_dbg(bond_dev, "%s is !NETIF_F_VLAN_CHALLENGED\n",
			   slave_dev->name);
L
Linus Torvalds 已提交
1314 1315
	}

1316
	/* Old ifenslave binaries are no longer supported.  These can
S
Stephen Hemminger 已提交
1317
	 * be identified with moderate accuracy by the state of the slave:
1318 1319 1320 1321
	 * the current ifenslave will set the interface down prior to
	 * enslaving it; the old ifenslave will not.
	 */
	if ((slave_dev->flags & IFF_UP)) {
1322 1323
		netdev_err(bond_dev, "%s is up - this may be due to an out of date ifenslave\n",
			   slave_dev->name);
1324 1325 1326
		res = -EPERM;
		goto err_undo_flags;
	}
L
Linus Torvalds 已提交
1327

1328 1329 1330 1331 1332 1333 1334
	/* set bonding device ether type by slave - bonding netdevices are
	 * created with ether_setup, so when the slave type is not ARPHRD_ETHER
	 * there is a need to override some of the type dependent attribs/funcs.
	 *
	 * bond ether type mutual exclusion - don't allow slaves of dissimilar
	 * ether type (eg ARPHRD_ETHER and ARPHRD_INFINIBAND) share the same bond
	 */
1335
	if (!bond_has_slaves(bond)) {
1336
		if (bond_dev->type != slave_dev->type) {
1337 1338
			netdev_dbg(bond_dev, "change device type from %d to %d\n",
				   bond_dev->type, slave_dev->type);
1339

1340 1341
			res = call_netdevice_notifiers(NETDEV_PRE_TYPE_CHANGE,
						       bond_dev);
1342 1343
			res = notifier_to_errno(res);
			if (res) {
1344
				netdev_err(bond_dev, "refused to change device type\n");
1345 1346 1347
				res = -EBUSY;
				goto err_undo_flags;
			}
1348

1349
			/* Flush unicast and multicast addresses */
1350
			dev_uc_flush(bond_dev);
1351
			dev_mc_flush(bond_dev);
1352

1353 1354
			if (slave_dev->type != ARPHRD_ETHER)
				bond_setup_by_slave(bond_dev, slave_dev);
1355
			else {
1356
				ether_setup(bond_dev);
1357 1358
				bond_dev->priv_flags &= ~IFF_TX_SKB_SHARING;
			}
1359

1360 1361
			call_netdevice_notifiers(NETDEV_POST_TYPE_CHANGE,
						 bond_dev);
1362
		}
1363
	} else if (bond_dev->type != slave_dev->type) {
1364 1365
		netdev_err(bond_dev, "%s ether type (%d) is different from other slaves (%d), can not enslave it\n",
			   slave_dev->name, slave_dev->type, bond_dev->type);
J
Joe Perches 已提交
1366 1367
		res = -EINVAL;
		goto err_undo_flags;
1368 1369
	}

1370
	if (slave_ops->ndo_set_mac_address == NULL) {
1371
		netdev_warn(bond_dev, "The slave device specified does not support setting the MAC address\n");
1372 1373 1374
		if (BOND_MODE(bond) == BOND_MODE_ACTIVEBACKUP &&
		    bond->params.fail_over_mac != BOND_FOM_ACTIVE) {
			if (!bond_has_slaves(bond)) {
1375
				bond->params.fail_over_mac = BOND_FOM_ACTIVE;
1376
				netdev_warn(bond_dev, "Setting fail_over_mac to active for active-backup mode\n");
1377
			} else {
1378
				netdev_err(bond_dev, "The slave device specified does not support setting the MAC address, but fail_over_mac is not set to active\n");
1379 1380
				res = -EOPNOTSUPP;
				goto err_undo_flags;
1381
			}
1382
		}
L
Linus Torvalds 已提交
1383 1384
	}

1385 1386
	call_netdevice_notifiers(NETDEV_JOIN, slave_dev);

1387
	/* If this is the first slave, then we need to set the master's hardware
1388 1389
	 * address to be the same as the slave's.
	 */
1390
	if (!bond_has_slaves(bond) &&
1391
	    bond->dev->addr_assign_type == NET_ADDR_RANDOM)
1392
		bond_set_dev_addr(bond->dev, slave_dev);
1393

1394
	new_slave = bond_alloc_slave(bond);
L
Linus Torvalds 已提交
1395 1396 1397 1398
	if (!new_slave) {
		res = -ENOMEM;
		goto err_undo_flags;
	}
1399

1400 1401
	new_slave->bond = bond;
	new_slave->dev = slave_dev;
1402
	/* Set the new_slave's queue_id to be zero.  Queue ID mapping
1403 1404 1405 1406
	 * is set via sysfs or module option if desired.
	 */
	new_slave->queue_id = 0;

1407 1408 1409 1410
	/* Save slave's original mtu and then set it to match the bond */
	new_slave->original_mtu = slave_dev->mtu;
	res = dev_set_mtu(slave_dev, bond->dev->mtu);
	if (res) {
1411
		netdev_dbg(bond_dev, "Error %d calling dev_set_mtu\n", res);
1412 1413 1414
		goto err_free;
	}

1415
	/* Save slave's original ("permanent") mac address for modes
1416 1417 1418
	 * that need it, and for restoring it upon release, and then
	 * set it to the master's address
	 */
1419
	ether_addr_copy(new_slave->perm_hwaddr, slave_dev->dev_addr);
L
Linus Torvalds 已提交
1420

1421
	if (!bond->params.fail_over_mac ||
1422
	    BOND_MODE(bond) != BOND_MODE_ACTIVEBACKUP) {
1423
		/* Set slave to master's mac address.  The application already
1424 1425 1426 1427 1428 1429
		 * set the master's mac address to that of the first slave
		 */
		memcpy(addr.sa_data, bond_dev->dev_addr, bond_dev->addr_len);
		addr.sa_family = slave_dev->type;
		res = dev_set_mac_address(slave_dev, &addr);
		if (res) {
1430
			netdev_dbg(bond_dev, "Error %d calling set_mac_address\n", res);
1431
			goto err_restore_mtu;
1432
		}
1433
	}
L
Linus Torvalds 已提交
1434

1435 1436 1437
	/* open the slave since the application closed it */
	res = dev_open(slave_dev);
	if (res) {
1438
		netdev_dbg(bond_dev, "Opening slave %s failed\n", slave_dev->name);
1439
		goto err_restore_mac;
L
Linus Torvalds 已提交
1440 1441
	}

1442
	slave_dev->priv_flags |= IFF_BONDING;
1443 1444
	/* initialize slave stats */
	dev_get_stats(new_slave->dev, &new_slave->slave_stats);
L
Linus Torvalds 已提交
1445

1446
	if (bond_is_lb(bond)) {
L
Linus Torvalds 已提交
1447 1448 1449 1450
		/* bond_alb_init_slave() must be called before all other stages since
		 * it might fail and we do not want to have to undo everything
		 */
		res = bond_alb_init_slave(bond, new_slave);
S
Stephen Hemminger 已提交
1451
		if (res)
1452
			goto err_close;
L
Linus Torvalds 已提交
1453 1454
	}

1455
	/* If the mode uses primary, then the following is handled by
1456
	 * bond_change_active_slave().
L
Linus Torvalds 已提交
1457
	 */
1458
	if (!bond_uses_primary(bond)) {
L
Linus Torvalds 已提交
1459 1460
		/* set promiscuity level to new slave */
		if (bond_dev->flags & IFF_PROMISC) {
1461 1462 1463
			res = dev_set_promiscuity(slave_dev, 1);
			if (res)
				goto err_close;
L
Linus Torvalds 已提交
1464 1465 1466 1467
		}

		/* set allmulti level to new slave */
		if (bond_dev->flags & IFF_ALLMULTI) {
1468 1469 1470
			res = dev_set_allmulti(slave_dev, 1);
			if (res)
				goto err_close;
L
Linus Torvalds 已提交
1471 1472
		}

1473
		netif_addr_lock_bh(bond_dev);
1474 1475 1476 1477

		dev_mc_sync_multiple(slave_dev, bond_dev);
		dev_uc_sync_multiple(slave_dev, bond_dev);

1478
		netif_addr_unlock_bh(bond_dev);
L
Linus Torvalds 已提交
1479 1480
	}

1481
	if (BOND_MODE(bond) == BOND_MODE_8023AD) {
L
Linus Torvalds 已提交
1482 1483 1484
		/* add lacpdu mc addr to mc list */
		u8 lacpdu_multicast[ETH_ALEN] = MULTICAST_LACPDU_ADDR;

1485
		dev_mc_add(slave_dev, lacpdu_multicast);
L
Linus Torvalds 已提交
1486 1487
	}

1488 1489
	res = vlan_vids_add_by_dev(slave_dev, bond_dev);
	if (res) {
1490 1491
		netdev_err(bond_dev, "Couldn't add bond vlan ids to %s\n",
			   slave_dev->name);
1492 1493
		goto err_close;
	}
L
Linus Torvalds 已提交
1494

1495
	prev_slave = bond_last_slave(bond);
L
Linus Torvalds 已提交
1496 1497 1498 1499

	new_slave->delay = 0;
	new_slave->link_failure_count = 0;

1500 1501
	bond_update_speed_duplex(new_slave);

1502
	new_slave->last_rx = jiffies -
1503
		(msecs_to_jiffies(bond->params.arp_interval) + 1);
1504
	for (i = 0; i < BOND_MAX_ARP_TARGETS; i++)
1505
		new_slave->target_last_arp_rx[i] = new_slave->last_rx;
1506

L
Linus Torvalds 已提交
1507 1508 1509 1510
	if (bond->params.miimon && !bond->params.use_carrier) {
		link_reporting = bond_check_dev_link(bond, slave_dev, 1);

		if ((link_reporting == -1) && !bond->params.arp_interval) {
1511
			/* miimon is set but a bonded network driver
L
Linus Torvalds 已提交
1512 1513 1514 1515 1516 1517 1518
			 * does not support ETHTOOL/MII and
			 * arp_interval is not set.  Note: if
			 * use_carrier is enabled, we will never go
			 * here (because netif_carrier is always
			 * supported); thus, we don't need to change
			 * the messages for netif_carrier.
			 */
1519 1520
			netdev_warn(bond_dev, "MII and ETHTOOL support not available for interface %s, and arp_interval/arp_ip_target module parameters not specified, thus bonding will not detect link failures! see bonding.txt for details\n",
				    slave_dev->name);
L
Linus Torvalds 已提交
1521 1522
		} else if (link_reporting == -1) {
			/* unable get link status using mii/ethtool */
1523 1524
			netdev_warn(bond_dev, "can't get link status from interface %s; the network driver associated with this interface does not support MII or ETHTOOL link status reporting, thus miimon has no effect on this interface\n",
				    slave_dev->name);
L
Linus Torvalds 已提交
1525 1526 1527 1528
		}
	}

	/* check for initial state */
1529 1530 1531
	if (bond->params.miimon) {
		if (bond_check_dev_link(bond, slave_dev, 0) == BMSR_LSTATUS) {
			if (bond->params.updelay) {
1532 1533
				bond_set_slave_link_state(new_slave,
							  BOND_LINK_BACK);
1534 1535
				new_slave->delay = bond->params.updelay;
			} else {
1536 1537
				bond_set_slave_link_state(new_slave,
							  BOND_LINK_UP);
1538
			}
L
Linus Torvalds 已提交
1539
		} else {
1540
			bond_set_slave_link_state(new_slave, BOND_LINK_DOWN);
L
Linus Torvalds 已提交
1541
		}
1542
	} else if (bond->params.arp_interval) {
1543 1544 1545
		bond_set_slave_link_state(new_slave,
					  (netif_carrier_ok(slave_dev) ?
					  BOND_LINK_UP : BOND_LINK_DOWN));
L
Linus Torvalds 已提交
1546
	} else {
1547
		bond_set_slave_link_state(new_slave, BOND_LINK_UP);
L
Linus Torvalds 已提交
1548 1549
	}

1550
	if (new_slave->link != BOND_LINK_DOWN)
1551
		new_slave->last_link_up = jiffies;
1552 1553 1554
	netdev_dbg(bond_dev, "Initial state of slave_dev is BOND_LINK_%s\n",
		   new_slave->link == BOND_LINK_DOWN ? "DOWN" :
		   (new_slave->link == BOND_LINK_UP ? "UP" : "BACK"));
1555

1556
	if (bond_uses_primary(bond) && bond->params.primary[0]) {
L
Linus Torvalds 已提交
1557
		/* if there is a primary slave, remember it */
1558
		if (strcmp(bond->params.primary, new_slave->dev->name) == 0) {
1559
			rcu_assign_pointer(bond->primary_slave, new_slave);
1560 1561
			bond->force_primary = true;
		}
L
Linus Torvalds 已提交
1562 1563
	}

1564
	switch (BOND_MODE(bond)) {
L
Linus Torvalds 已提交
1565
	case BOND_MODE_ACTIVEBACKUP:
1566 1567
		bond_set_slave_inactive_flags(new_slave,
					      BOND_SLAVE_NOTIFY_NOW);
L
Linus Torvalds 已提交
1568 1569 1570 1571 1572 1573
		break;
	case BOND_MODE_8023AD:
		/* in 802.3ad mode, the internal mechanism
		 * will activate the slaves in the selected
		 * aggregator
		 */
1574
		bond_set_slave_inactive_flags(new_slave, BOND_SLAVE_NOTIFY_NOW);
L
Linus Torvalds 已提交
1575
		/* if this is the first slave */
1576
		if (!prev_slave) {
1577
			SLAVE_AD_INFO(new_slave)->id = 1;
L
Linus Torvalds 已提交
1578 1579 1580
			/* Initialize AD with the number of times that the AD timer is called in 1 second
			 * can be called only after the mac address of the bond is set
			 */
1581
			bond_3ad_initialize(bond, 1000/AD_TIMER_INTERVAL);
L
Linus Torvalds 已提交
1582
		} else {
1583 1584
			SLAVE_AD_INFO(new_slave)->id =
				SLAVE_AD_INFO(prev_slave)->id + 1;
L
Linus Torvalds 已提交
1585 1586 1587 1588 1589 1590
		}

		bond_3ad_bind_slave(new_slave);
		break;
	case BOND_MODE_TLB:
	case BOND_MODE_ALB:
J
Jiri Pirko 已提交
1591
		bond_set_active_slave(new_slave);
1592
		bond_set_slave_inactive_flags(new_slave, BOND_SLAVE_NOTIFY_NOW);
L
Linus Torvalds 已提交
1593 1594
		break;
	default:
1595
		netdev_dbg(bond_dev, "This slave is always active in trunk mode\n");
L
Linus Torvalds 已提交
1596 1597

		/* always active in trunk mode */
J
Jiri Pirko 已提交
1598
		bond_set_active_slave(new_slave);
L
Linus Torvalds 已提交
1599 1600 1601 1602 1603

		/* In trunking mode there is little meaning to curr_active_slave
		 * anyway (it holds no special properties of the bond device),
		 * so we can change it without calling change_active_interface()
		 */
1604 1605
		if (!rcu_access_pointer(bond->curr_active_slave) &&
		    new_slave->link == BOND_LINK_UP)
1606
			rcu_assign_pointer(bond->curr_active_slave, new_slave);
S
Stephen Hemminger 已提交
1607

L
Linus Torvalds 已提交
1608 1609 1610
		break;
	} /* switch(bond_mode) */

1611
#ifdef CONFIG_NET_POLL_CONTROLLER
S
stephen hemminger 已提交
1612
	slave_dev->npinfo = bond->dev->npinfo;
1613 1614
	if (slave_dev->npinfo) {
		if (slave_enable_netpoll(new_slave)) {
1615
			netdev_info(bond_dev, "master_dev is using netpoll, but new slave device does not support netpoll\n");
1616
			res = -EBUSY;
1617
			goto err_detach;
1618
		}
1619 1620
	}
#endif
1621

1622 1623 1624
	if (!(bond_dev->features & NETIF_F_LRO))
		dev_disable_lro(slave_dev);

J
Jiri Pirko 已提交
1625 1626 1627
	res = netdev_rx_handler_register(slave_dev, bond_handle_frame,
					 new_slave);
	if (res) {
1628
		netdev_dbg(bond_dev, "Error %d calling netdev_rx_handler_register\n", res);
1629
		goto err_detach;
J
Jiri Pirko 已提交
1630 1631
	}

1632 1633
	res = bond_master_upper_dev_link(bond_dev, slave_dev, new_slave);
	if (res) {
1634
		netdev_dbg(bond_dev, "Error %d calling bond_master_upper_dev_link\n", res);
1635 1636 1637
		goto err_unregister;
	}

1638 1639
	res = bond_sysfs_slave_add(new_slave);
	if (res) {
1640
		netdev_dbg(bond_dev, "Error %d calling bond_sysfs_slave_add\n", res);
1641 1642 1643
		goto err_upper_unlink;
	}

1644 1645 1646 1647
	bond->slave_cnt++;
	bond_compute_features(bond);
	bond_set_carrier(bond);

1648
	if (bond_uses_primary(bond)) {
1649
		block_netpoll_tx();
1650
		bond_select_active_slave(bond);
1651
		unblock_netpoll_tx();
1652
	}
1653

1654 1655 1656
	if (bond_mode_uses_xmit_hash(bond))
		bond_update_slave_arr(bond, NULL);

1657 1658 1659 1660
	netdev_info(bond_dev, "Enslaving %s as %s interface with %s link\n",
		    slave_dev->name,
		    bond_is_active_slave(new_slave) ? "an active" : "a backup",
		    new_slave->link != BOND_LINK_DOWN ? "an up" : "a down");
L
Linus Torvalds 已提交
1661 1662

	/* enslave is successful */
1663
	bond_queue_slave_event(new_slave);
L
Linus Torvalds 已提交
1664 1665 1666
	return 0;

/* Undo stages on error */
1667 1668 1669
err_upper_unlink:
	bond_upper_dev_unlink(bond_dev, slave_dev);

1670 1671 1672
err_unregister:
	netdev_rx_handler_unregister(slave_dev);

1673
err_detach:
1674
	if (!bond_uses_primary(bond))
1675 1676
		bond_hw_addr_flush(bond_dev, slave_dev);

1677
	vlan_vids_del_by_dev(slave_dev, bond_dev);
1678 1679
	if (rcu_access_pointer(bond->primary_slave) == new_slave)
		RCU_INIT_POINTER(bond->primary_slave, NULL);
1680
	if (rcu_access_pointer(bond->curr_active_slave) == new_slave) {
1681
		block_netpoll_tx();
1682
		bond_change_active_slave(bond, NULL);
1683
		bond_select_active_slave(bond);
1684
		unblock_netpoll_tx();
1685
	}
1686 1687
	/* either primary_slave or curr_active_slave might've changed */
	synchronize_rcu();
1688
	slave_disable_netpoll(new_slave);
1689

L
Linus Torvalds 已提交
1690
err_close:
1691
	slave_dev->priv_flags &= ~IFF_BONDING;
L
Linus Torvalds 已提交
1692 1693 1694
	dev_close(slave_dev);

err_restore_mac:
1695
	if (!bond->params.fail_over_mac ||
1696
	    BOND_MODE(bond) != BOND_MODE_ACTIVEBACKUP) {
1697 1698 1699 1700
		/* XXX TODO - fom follow mode needs to change master's
		 * MAC if this slave's MAC is in use by the bond, or at
		 * least print a warning.
		 */
1701
		ether_addr_copy(addr.sa_data, new_slave->perm_hwaddr);
1702 1703 1704
		addr.sa_family = slave_dev->type;
		dev_set_mac_address(slave_dev, &addr);
	}
L
Linus Torvalds 已提交
1705

1706 1707 1708
err_restore_mtu:
	dev_set_mtu(slave_dev, new_slave->original_mtu);

L
Linus Torvalds 已提交
1709
err_free:
1710
	bond_free_slave(new_slave);
L
Linus Torvalds 已提交
1711 1712

err_undo_flags:
1713
	/* Enslave of first slave has failed and we need to fix master's mac */
1714
	if (!bond_has_slaves(bond) &&
1715
	    ether_addr_equal_64bits(bond_dev->dev_addr, slave_dev->dev_addr))
1716
		eth_hw_addr_random(bond_dev);
S
Stephen Hemminger 已提交
1717

L
Linus Torvalds 已提交
1718 1719 1720
	return res;
}

1721
/* Try to release the slave device <slave> from the bond device <master>
L
Linus Torvalds 已提交
1722
 * It is legal to access curr_active_slave without a lock because all the function
1723
 * is RTNL-locked. If "all" is true it means that the function is being called
1724
 * while destroying a bond interface and all slaves are being released.
L
Linus Torvalds 已提交
1725 1726 1727 1728 1729 1730 1731
 *
 * The rules for slave state should be:
 *   for Active/Backup:
 *     Active stays on all backups go down
 *   for Bonded connections:
 *     The first up interface should be left on and all others downed.
 */
1732 1733 1734
static int __bond_release_one(struct net_device *bond_dev,
			      struct net_device *slave_dev,
			      bool all)
L
Linus Torvalds 已提交
1735
{
1736
	struct bonding *bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
1737 1738
	struct slave *slave, *oldcurrent;
	struct sockaddr addr;
1739
	int old_flags = bond_dev->flags;
1740
	netdev_features_t old_features = bond_dev->features;
L
Linus Torvalds 已提交
1741 1742 1743

	/* slave is not a slave or master is not master of this slave */
	if (!(slave_dev->flags & IFF_SLAVE) ||
1744
	    !netdev_has_upper_dev(slave_dev, bond_dev)) {
1745
		netdev_dbg(bond_dev, "cannot release %s\n",
1746
			   slave_dev->name);
L
Linus Torvalds 已提交
1747 1748 1749
		return -EINVAL;
	}

1750
	block_netpoll_tx();
L
Linus Torvalds 已提交
1751 1752 1753 1754

	slave = bond_get_slave_by_dev(bond, slave_dev);
	if (!slave) {
		/* not a slave of this bond */
1755 1756
		netdev_info(bond_dev, "%s not enslaved\n",
			    slave_dev->name);
1757
		unblock_netpoll_tx();
L
Linus Torvalds 已提交
1758 1759 1760
		return -EINVAL;
	}

1761 1762
	bond_sysfs_slave_del(slave);

1763 1764 1765
	/* recompute stats just before removing the slave */
	bond_get_stats(bond->dev, &bond->bond_stats);

1766
	bond_upper_dev_unlink(bond_dev, slave_dev);
J
Jiri Pirko 已提交
1767 1768 1769 1770 1771
	/* unregister rx_handler early so bond_handle_frame wouldn't be called
	 * for this slave anymore.
	 */
	netdev_rx_handler_unregister(slave_dev);

1772
	if (BOND_MODE(bond) == BOND_MODE_8023AD)
L
Linus Torvalds 已提交
1773 1774
		bond_3ad_unbind_slave(slave);

1775 1776 1777
	if (bond_mode_uses_xmit_hash(bond))
		bond_update_slave_arr(bond, slave);

1778 1779 1780
	netdev_info(bond_dev, "Releasing %s interface %s\n",
		    bond_is_active_slave(slave) ? "active" : "backup",
		    slave_dev->name);
L
Linus Torvalds 已提交
1781

1782
	oldcurrent = rcu_access_pointer(bond->curr_active_slave);
L
Linus Torvalds 已提交
1783

1784
	RCU_INIT_POINTER(bond->current_arp_slave, NULL);
L
Linus Torvalds 已提交
1785

1786
	if (!all && (!bond->params.fail_over_mac ||
1787
		     BOND_MODE(bond) != BOND_MODE_ACTIVEBACKUP)) {
1788
		if (ether_addr_equal_64bits(bond_dev->dev_addr, slave->perm_hwaddr) &&
1789
		    bond_has_slaves(bond))
1790 1791 1792
			netdev_warn(bond_dev, "the permanent HWaddr of %s - %pM - is still in use by %s - set the HWaddr of %s to a different address to avoid conflicts\n",
				    slave_dev->name, slave->perm_hwaddr,
				    bond_dev->name, slave_dev->name);
1793 1794
	}

1795 1796
	if (rtnl_dereference(bond->primary_slave) == slave)
		RCU_INIT_POINTER(bond->primary_slave, NULL);
L
Linus Torvalds 已提交
1797

1798
	if (oldcurrent == slave)
L
Linus Torvalds 已提交
1799 1800
		bond_change_active_slave(bond, NULL);

1801
	if (bond_is_lb(bond)) {
L
Linus Torvalds 已提交
1802 1803 1804 1805 1806 1807 1808 1809
		/* Must be called only after the slave has been
		 * detached from the list and the curr_active_slave
		 * has been cleared (if our_slave == old_current),
		 * but before a new active slave is selected.
		 */
		bond_alb_deinit_slave(bond, slave);
	}

1810
	if (all) {
1811
		RCU_INIT_POINTER(bond->curr_active_slave, NULL);
1812
	} else if (oldcurrent == slave) {
1813
		/* Note that we hold RTNL over this sequence, so there
1814 1815 1816
		 * is no concern that another slave add/remove event
		 * will interfere.
		 */
L
Linus Torvalds 已提交
1817
		bond_select_active_slave(bond);
1818 1819
	}

1820
	if (!bond_has_slaves(bond)) {
1821
		bond_set_carrier(bond);
1822
		eth_hw_addr_random(bond_dev);
L
Linus Torvalds 已提交
1823 1824
	}

1825
	unblock_netpoll_tx();
1826
	synchronize_rcu();
1827
	bond->slave_cnt--;
L
Linus Torvalds 已提交
1828

1829
	if (!bond_has_slaves(bond)) {
1830
		call_netdevice_notifiers(NETDEV_CHANGEADDR, bond->dev);
1831 1832
		call_netdevice_notifiers(NETDEV_RELEASE, bond->dev);
	}
1833

1834 1835 1836
	bond_compute_features(bond);
	if (!(bond_dev->features & NETIF_F_VLAN_CHALLENGED) &&
	    (old_features & NETIF_F_VLAN_CHALLENGED))
1837 1838
		netdev_info(bond_dev, "last VLAN challenged slave %s left bond %s - VLAN blocking is removed\n",
			    slave_dev->name, bond_dev->name);
1839

1840
	vlan_vids_del_by_dev(slave_dev, bond_dev);
L
Linus Torvalds 已提交
1841

1842
	/* If the mode uses primary, then this case was handled above by
1843
	 * bond_change_active_slave(..., NULL)
L
Linus Torvalds 已提交
1844
	 */
1845
	if (!bond_uses_primary(bond)) {
1846 1847 1848 1849 1850 1851 1852 1853
		/* unset promiscuity level from slave
		 * NOTE: The NETDEV_CHANGEADDR call above may change the value
		 * of the IFF_PROMISC flag in the bond_dev, but we need the
		 * value of that flag before that change, as that was the value
		 * when this slave was attached, so we cache at the start of the
		 * function and use it here. Same goes for ALLMULTI below
		 */
		if (old_flags & IFF_PROMISC)
L
Linus Torvalds 已提交
1854 1855 1856
			dev_set_promiscuity(slave_dev, -1);

		/* unset allmulti level from slave */
1857
		if (old_flags & IFF_ALLMULTI)
L
Linus Torvalds 已提交
1858 1859
			dev_set_allmulti(slave_dev, -1);

1860
		bond_hw_addr_flush(bond_dev, slave_dev);
L
Linus Torvalds 已提交
1861 1862
	}

1863
	slave_disable_netpoll(slave);
1864

L
Linus Torvalds 已提交
1865 1866 1867
	/* close slave before restoring its mac address */
	dev_close(slave_dev);

1868
	if (bond->params.fail_over_mac != BOND_FOM_ACTIVE ||
1869
	    BOND_MODE(bond) != BOND_MODE_ACTIVEBACKUP) {
1870
		/* restore original ("permanent") mac address */
1871
		ether_addr_copy(addr.sa_data, slave->perm_hwaddr);
1872 1873 1874
		addr.sa_family = slave_dev->type;
		dev_set_mac_address(slave_dev, &addr);
	}
L
Linus Torvalds 已提交
1875

1876 1877
	dev_set_mtu(slave_dev, slave->original_mtu);

1878
	slave_dev->priv_flags &= ~IFF_BONDING;
L
Linus Torvalds 已提交
1879

1880
	bond_free_slave(slave);
L
Linus Torvalds 已提交
1881

1882
	return 0;
L
Linus Torvalds 已提交
1883 1884
}

1885 1886 1887 1888 1889 1890
/* A wrapper used because of ndo_del_link */
int bond_release(struct net_device *bond_dev, struct net_device *slave_dev)
{
	return __bond_release_one(bond_dev, slave_dev, false);
}

1891 1892 1893
/* First release a slave and then destroy the bond if no more slaves are left.
 * Must be under rtnl_lock when this function is called.
 */
1894 1895
static int  bond_release_and_destroy(struct net_device *bond_dev,
				     struct net_device *slave_dev)
1896
{
1897
	struct bonding *bond = netdev_priv(bond_dev);
1898 1899 1900
	int ret;

	ret = bond_release(bond_dev, slave_dev);
1901
	if (ret == 0 && !bond_has_slaves(bond)) {
1902
		bond_dev->priv_flags |= IFF_DISABLE_NETPOLL;
1903 1904
		netdev_info(bond_dev, "Destroying bond %s\n",
			    bond_dev->name);
S
Stephen Hemminger 已提交
1905
		unregister_netdevice(bond_dev);
1906 1907 1908 1909
	}
	return ret;
}

L
Linus Torvalds 已提交
1910 1911
static int bond_info_query(struct net_device *bond_dev, struct ifbond *info)
{
1912
	struct bonding *bond = netdev_priv(bond_dev);
1913
	bond_fill_ifbond(bond, info);
L
Linus Torvalds 已提交
1914 1915 1916 1917 1918
	return 0;
}

static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *info)
{
1919
	struct bonding *bond = netdev_priv(bond_dev);
1920
	struct list_head *iter;
1921
	int i = 0, res = -ENODEV;
L
Linus Torvalds 已提交
1922 1923
	struct slave *slave;

1924
	bond_for_each_slave(bond, slave, iter) {
1925
		if (i++ == (int)info->slave_id) {
1926
			res = 0;
1927
			bond_fill_ifslave(slave, info);
L
Linus Torvalds 已提交
1928 1929 1930 1931
			break;
		}
	}

1932
	return res;
L
Linus Torvalds 已提交
1933 1934 1935 1936
}

/*-------------------------------- Monitoring -------------------------------*/

1937
/* called with rcu_read_lock() */
J
Jay Vosburgh 已提交
1938 1939
static int bond_miimon_inspect(struct bonding *bond)
{
1940
	int link_state, commit = 0;
1941
	struct list_head *iter;
J
Jay Vosburgh 已提交
1942
	struct slave *slave;
1943 1944
	bool ignore_updelay;

1945
	ignore_updelay = !rcu_dereference(bond->curr_active_slave);
L
Linus Torvalds 已提交
1946

1947
	bond_for_each_slave_rcu(bond, slave, iter) {
J
Jay Vosburgh 已提交
1948
		slave->new_link = BOND_LINK_NOCHANGE;
L
Linus Torvalds 已提交
1949

J
Jay Vosburgh 已提交
1950
		link_state = bond_check_dev_link(bond, slave->dev, 0);
L
Linus Torvalds 已提交
1951 1952

		switch (slave->link) {
J
Jay Vosburgh 已提交
1953 1954 1955
		case BOND_LINK_UP:
			if (link_state)
				continue;
L
Linus Torvalds 已提交
1956

1957
			bond_set_slave_link_state(slave, BOND_LINK_FAIL);
J
Jay Vosburgh 已提交
1958 1959
			slave->delay = bond->params.downdelay;
			if (slave->delay) {
1960 1961 1962 1963 1964 1965 1966
				netdev_info(bond->dev, "link status down for %sinterface %s, disabling it in %d ms\n",
					    (BOND_MODE(bond) ==
					     BOND_MODE_ACTIVEBACKUP) ?
					     (bond_is_active_slave(slave) ?
					      "active " : "backup ") : "",
					    slave->dev->name,
					    bond->params.downdelay * bond->params.miimon);
L
Linus Torvalds 已提交
1967
			}
J
Jay Vosburgh 已提交
1968 1969 1970
			/*FALLTHRU*/
		case BOND_LINK_FAIL:
			if (link_state) {
1971
				/* recovered before downdelay expired */
1972
				bond_set_slave_link_state(slave, BOND_LINK_UP);
1973
				slave->last_link_up = jiffies;
1974 1975 1976 1977
				netdev_info(bond->dev, "link status up again after %d ms for interface %s\n",
					    (bond->params.downdelay - slave->delay) *
					    bond->params.miimon,
					    slave->dev->name);
J
Jay Vosburgh 已提交
1978
				continue;
L
Linus Torvalds 已提交
1979
			}
J
Jay Vosburgh 已提交
1980 1981 1982 1983 1984

			if (slave->delay <= 0) {
				slave->new_link = BOND_LINK_DOWN;
				commit++;
				continue;
L
Linus Torvalds 已提交
1985 1986
			}

J
Jay Vosburgh 已提交
1987 1988 1989 1990 1991 1992 1993
			slave->delay--;
			break;

		case BOND_LINK_DOWN:
			if (!link_state)
				continue;

1994
			bond_set_slave_link_state(slave, BOND_LINK_BACK);
J
Jay Vosburgh 已提交
1995 1996 1997
			slave->delay = bond->params.updelay;

			if (slave->delay) {
1998 1999 2000 2001 2002
				netdev_info(bond->dev, "link status up for interface %s, enabling it in %d ms\n",
					    slave->dev->name,
					    ignore_updelay ? 0 :
					    bond->params.updelay *
					    bond->params.miimon);
J
Jay Vosburgh 已提交
2003 2004 2005 2006
			}
			/*FALLTHRU*/
		case BOND_LINK_BACK:
			if (!link_state) {
2007 2008
				bond_set_slave_link_state(slave,
							  BOND_LINK_DOWN);
2009 2010 2011 2012
				netdev_info(bond->dev, "link status down again after %d ms for interface %s\n",
					    (bond->params.updelay - slave->delay) *
					    bond->params.miimon,
					    slave->dev->name);
J
Jay Vosburgh 已提交
2013 2014 2015 2016

				continue;
			}

2017 2018 2019
			if (ignore_updelay)
				slave->delay = 0;

J
Jay Vosburgh 已提交
2020 2021 2022
			if (slave->delay <= 0) {
				slave->new_link = BOND_LINK_UP;
				commit++;
2023
				ignore_updelay = false;
J
Jay Vosburgh 已提交
2024
				continue;
L
Linus Torvalds 已提交
2025
			}
J
Jay Vosburgh 已提交
2026 2027

			slave->delay--;
L
Linus Torvalds 已提交
2028
			break;
J
Jay Vosburgh 已提交
2029 2030
		}
	}
L
Linus Torvalds 已提交
2031

J
Jay Vosburgh 已提交
2032 2033
	return commit;
}
L
Linus Torvalds 已提交
2034

J
Jay Vosburgh 已提交
2035 2036
static void bond_miimon_commit(struct bonding *bond)
{
2037
	struct list_head *iter;
2038
	struct slave *slave, *primary;
J
Jay Vosburgh 已提交
2039

2040
	bond_for_each_slave(bond, slave, iter) {
J
Jay Vosburgh 已提交
2041 2042 2043
		switch (slave->new_link) {
		case BOND_LINK_NOCHANGE:
			continue;
L
Linus Torvalds 已提交
2044

J
Jay Vosburgh 已提交
2045
		case BOND_LINK_UP:
2046
			bond_set_slave_link_state(slave, BOND_LINK_UP);
2047
			slave->last_link_up = jiffies;
J
Jay Vosburgh 已提交
2048

2049
			primary = rtnl_dereference(bond->primary_slave);
2050
			if (BOND_MODE(bond) == BOND_MODE_8023AD) {
J
Jay Vosburgh 已提交
2051
				/* prevent it from being the active one */
J
Jiri Pirko 已提交
2052
				bond_set_backup_slave(slave);
2053
			} else if (BOND_MODE(bond) != BOND_MODE_ACTIVEBACKUP) {
J
Jay Vosburgh 已提交
2054
				/* make it immediately active */
J
Jiri Pirko 已提交
2055
				bond_set_active_slave(slave);
2056
			} else if (slave != primary) {
J
Jay Vosburgh 已提交
2057
				/* prevent it from being the active one */
J
Jiri Pirko 已提交
2058
				bond_set_backup_slave(slave);
L
Linus Torvalds 已提交
2059 2060
			}

2061 2062 2063 2064
			netdev_info(bond->dev, "link status definitely up for interface %s, %u Mbps %s duplex\n",
				    slave->dev->name,
				    slave->speed == SPEED_UNKNOWN ? 0 : slave->speed,
				    slave->duplex ? "full" : "half");
L
Linus Torvalds 已提交
2065

J
Jay Vosburgh 已提交
2066
			/* notify ad that the link status has changed */
2067
			if (BOND_MODE(bond) == BOND_MODE_8023AD)
J
Jay Vosburgh 已提交
2068
				bond_3ad_handle_link_change(slave, BOND_LINK_UP);
2069

2070
			if (bond_is_lb(bond))
J
Jay Vosburgh 已提交
2071 2072
				bond_alb_handle_link_change(bond, slave,
							    BOND_LINK_UP);
L
Linus Torvalds 已提交
2073

2074 2075 2076
			if (BOND_MODE(bond) == BOND_MODE_XOR)
				bond_update_slave_arr(bond, NULL);

2077
			if (!bond->curr_active_slave || slave == primary)
J
Jay Vosburgh 已提交
2078
				goto do_failover;
L
Linus Torvalds 已提交
2079

J
Jay Vosburgh 已提交
2080
			continue;
2081

J
Jay Vosburgh 已提交
2082
		case BOND_LINK_DOWN:
J
Jay Vosburgh 已提交
2083 2084 2085
			if (slave->link_failure_count < UINT_MAX)
				slave->link_failure_count++;

2086
			bond_set_slave_link_state(slave, BOND_LINK_DOWN);
L
Linus Torvalds 已提交
2087

2088 2089
			if (BOND_MODE(bond) == BOND_MODE_ACTIVEBACKUP ||
			    BOND_MODE(bond) == BOND_MODE_8023AD)
2090 2091
				bond_set_slave_inactive_flags(slave,
							      BOND_SLAVE_NOTIFY_NOW);
J
Jay Vosburgh 已提交
2092

2093 2094
			netdev_info(bond->dev, "link status definitely down for interface %s, disabling it\n",
				    slave->dev->name);
J
Jay Vosburgh 已提交
2095

2096
			if (BOND_MODE(bond) == BOND_MODE_8023AD)
J
Jay Vosburgh 已提交
2097 2098 2099
				bond_3ad_handle_link_change(slave,
							    BOND_LINK_DOWN);

2100
			if (bond_is_lb(bond))
J
Jay Vosburgh 已提交
2101 2102 2103
				bond_alb_handle_link_change(bond, slave,
							    BOND_LINK_DOWN);

2104 2105 2106
			if (BOND_MODE(bond) == BOND_MODE_XOR)
				bond_update_slave_arr(bond, NULL);

2107
			if (slave == rcu_access_pointer(bond->curr_active_slave))
J
Jay Vosburgh 已提交
2108 2109 2110 2111 2112
				goto do_failover;

			continue;

		default:
2113 2114
			netdev_err(bond->dev, "invalid new link %d on slave %s\n",
				   slave->new_link, slave->dev->name);
J
Jay Vosburgh 已提交
2115 2116 2117 2118 2119 2120
			slave->new_link = BOND_LINK_NOCHANGE;

			continue;
		}

do_failover:
2121
		block_netpoll_tx();
J
Jay Vosburgh 已提交
2122
		bond_select_active_slave(bond);
2123
		unblock_netpoll_tx();
J
Jay Vosburgh 已提交
2124 2125 2126
	}

	bond_set_carrier(bond);
L
Linus Torvalds 已提交
2127 2128
}

2129
/* bond_mii_monitor
2130 2131
 *
 * Really a wrapper that splits the mii monitor into two phases: an
J
Jay Vosburgh 已提交
2132 2133 2134
 * inspection, then (if inspection indicates something needs to be done)
 * an acquisition of appropriate locks followed by a commit phase to
 * implement whatever link state changes are indicated.
2135
 */
2136
static void bond_mii_monitor(struct work_struct *work)
2137 2138 2139
{
	struct bonding *bond = container_of(work, struct bonding,
					    mii_work.work);
2140
	bool should_notify_peers = false;
2141
	unsigned long delay;
2142

2143 2144 2145
	delay = msecs_to_jiffies(bond->params.miimon);

	if (!bond_has_slaves(bond))
J
Jay Vosburgh 已提交
2146
		goto re_arm;
2147

2148 2149
	rcu_read_lock();

2150 2151
	should_notify_peers = bond_should_notify_peers(bond);

2152
	if (bond_miimon_inspect(bond)) {
2153
		rcu_read_unlock();
J
Jay Vosburgh 已提交
2154

2155 2156 2157 2158 2159 2160
		/* Race avoidance with bond_close cancel of workqueue */
		if (!rtnl_trylock()) {
			delay = 1;
			should_notify_peers = false;
			goto re_arm;
		}
2161

2162 2163 2164
		bond_miimon_commit(bond);

		rtnl_unlock();	/* might sleep, hold no other locks */
2165 2166
	} else
		rcu_read_unlock();
2167

J
Jay Vosburgh 已提交
2168
re_arm:
2169
	if (bond->params.miimon)
2170 2171 2172 2173 2174 2175 2176 2177
		queue_delayed_work(bond->wq, &bond->mii_work, delay);

	if (should_notify_peers) {
		if (!rtnl_trylock())
			return;
		call_netdevice_notifiers(NETDEV_NOTIFY_PEERS, bond->dev);
		rtnl_unlock();
	}
2178
}
J
Jay Vosburgh 已提交
2179

2180
static bool bond_has_this_ip(struct bonding *bond, __be32 ip)
2181
{
2182 2183 2184
	struct net_device *upper;
	struct list_head *iter;
	bool ret = false;
2185

2186
	if (ip == bond_confirm_addr(bond->dev, 0, ip))
2187
		return true;
2188

2189
	rcu_read_lock();
2190
	netdev_for_each_all_upper_dev_rcu(bond->dev, upper, iter) {
2191 2192 2193 2194
		if (ip == bond_confirm_addr(upper, 0, ip)) {
			ret = true;
			break;
		}
2195
	}
2196
	rcu_read_unlock();
2197

2198
	return ret;
2199 2200
}

2201
/* We go to the (large) trouble of VLAN tagging ARP frames because
J
Jay Vosburgh 已提交
2202 2203 2204
 * switches in VLAN mode (especially if ports are configured as
 * "native" to a VLAN) might not pass non-tagged frames.
 */
2205 2206
static void bond_arp_send(struct net_device *slave_dev, int arp_op,
			  __be32 dest_ip, __be32 src_ip,
2207
			  struct bond_vlan_tag *tags)
J
Jay Vosburgh 已提交
2208 2209
{
	struct sk_buff *skb;
2210
	struct bond_vlan_tag *outer_tag = tags;
J
Jay Vosburgh 已提交
2211

2212 2213
	netdev_dbg(slave_dev, "arp %d on slave %s: dst %pI4 src %pI4\n",
		   arp_op, slave_dev->name, &dest_ip, &src_ip);
S
Stephen Hemminger 已提交
2214

J
Jay Vosburgh 已提交
2215 2216 2217 2218
	skb = arp_create(arp_op, ETH_P_ARP, dest_ip, slave_dev, src_ip,
			 NULL, slave_dev->dev_addr, NULL);

	if (!skb) {
2219
		net_err_ratelimited("ARP packet allocation failed\n");
J
Jay Vosburgh 已提交
2220 2221
		return;
	}
2222

2223 2224 2225 2226 2227
	if (!tags || tags->vlan_proto == VLAN_N_VID)
		goto xmit;

	tags++;

2228
	/* Go through all the tags backwards and add them to the packet */
2229 2230 2231
	while (tags->vlan_proto != VLAN_N_VID) {
		if (!tags->vlan_id) {
			tags++;
2232
			continue;
2233
		}
2234

2235
		netdev_dbg(slave_dev, "inner tag: proto %X vid %X\n",
2236
			   ntohs(outer_tag->vlan_proto), tags->vlan_id);
2237 2238
		skb = vlan_insert_tag_set_proto(skb, tags->vlan_proto,
						tags->vlan_id);
2239 2240 2241 2242
		if (!skb) {
			net_err_ratelimited("failed to insert inner VLAN tag\n");
			return;
		}
2243 2244

		tags++;
2245 2246
	}
	/* Set the outer tag */
2247
	if (outer_tag->vlan_id) {
2248
		netdev_dbg(slave_dev, "outer tag: proto %X vid %X\n",
2249
			   ntohs(outer_tag->vlan_proto), outer_tag->vlan_id);
J
Jiri Pirko 已提交
2250 2251
		__vlan_hwaccel_put_tag(skb, outer_tag->vlan_proto,
				       outer_tag->vlan_id);
J
Jay Vosburgh 已提交
2252
	}
2253 2254

xmit:
J
Jay Vosburgh 已提交
2255 2256 2257
	arp_xmit(skb);
}

2258 2259 2260 2261 2262 2263
/* Validate the device path between the @start_dev and the @end_dev.
 * The path is valid if the @end_dev is reachable through device
 * stacking.
 * When the path is validated, collect any vlan information in the
 * path.
 */
2264 2265 2266
struct bond_vlan_tag *bond_verify_device_path(struct net_device *start_dev,
					      struct net_device *end_dev,
					      int level)
2267
{
2268
	struct bond_vlan_tag *tags;
2269 2270 2271
	struct net_device *upper;
	struct list_head  *iter;

2272 2273 2274 2275 2276 2277 2278
	if (start_dev == end_dev) {
		tags = kzalloc(sizeof(*tags) * (level + 1), GFP_ATOMIC);
		if (!tags)
			return ERR_PTR(-ENOMEM);
		tags[level].vlan_proto = VLAN_N_VID;
		return tags;
	}
2279 2280

	netdev_for_each_upper_dev_rcu(start_dev, upper, iter) {
2281 2282 2283 2284 2285
		tags = bond_verify_device_path(upper, end_dev, level + 1);
		if (IS_ERR_OR_NULL(tags)) {
			if (IS_ERR(tags))
				return tags;
			continue;
2286
		}
2287 2288 2289 2290 2291 2292
		if (is_vlan_dev(upper)) {
			tags[level].vlan_proto = vlan_dev_vlan_proto(upper);
			tags[level].vlan_id = vlan_dev_vlan_id(upper);
		}

		return tags;
2293 2294
	}

2295
	return NULL;
2296
}
J
Jay Vosburgh 已提交
2297

L
Linus Torvalds 已提交
2298 2299
static void bond_arp_send_all(struct bonding *bond, struct slave *slave)
{
J
Jay Vosburgh 已提交
2300
	struct rtable *rt;
2301
	struct bond_vlan_tag *tags;
2302
	__be32 *targets = bond->params.arp_targets, addr;
2303
	int i;
L
Linus Torvalds 已提交
2304

2305
	for (i = 0; i < BOND_MAX_ARP_TARGETS && targets[i]; i++) {
2306
		netdev_dbg(bond->dev, "basa: target %pI4\n", &targets[i]);
2307
		tags = NULL;
J
Jay Vosburgh 已提交
2308

2309
		/* Find out through which dev should the packet go */
2310 2311
		rt = ip_route_output(dev_net(bond->dev), targets[i], 0,
				     RTO_ONLINK, 0);
2312
		if (IS_ERR(rt)) {
2313 2314 2315
			/* there's no route to target - try to send arp
			 * probe to generate any traffic (arp_validate=0)
			 */
2316 2317 2318 2319
			if (bond->params.arp_validate)
				net_warn_ratelimited("%s: no route to arp_ip_target %pI4 and arp_validate is set\n",
						     bond->dev->name,
						     &targets[i]);
2320 2321
			bond_arp_send(slave->dev, ARPOP_REQUEST, targets[i],
				      0, tags);
J
Jay Vosburgh 已提交
2322 2323 2324
			continue;
		}

2325 2326 2327 2328 2329
		/* bond device itself */
		if (rt->dst.dev == bond->dev)
			goto found;

		rcu_read_lock();
2330
		tags = bond_verify_device_path(bond->dev, rt->dst.dev, 0);
2331
		rcu_read_unlock();
J
Jay Vosburgh 已提交
2332

2333
		if (!IS_ERR_OR_NULL(tags))
2334 2335
			goto found;

2336
		/* Not our device - skip */
2337 2338
		netdev_dbg(bond->dev, "no path to arp_ip_target %pI4 via rt.dev %s\n",
			   &targets[i], rt->dst.dev ? rt->dst.dev->name : "NULL");
2339

2340
		ip_rt_put(rt);
2341 2342 2343 2344 2345 2346
		continue;

found:
		addr = bond_confirm_addr(rt->dst.dev, targets[i], 0);
		ip_rt_put(rt);
		bond_arp_send(slave->dev, ARPOP_REQUEST, targets[i],
2347
			      addr, tags);
2348
		kfree(tags);
J
Jay Vosburgh 已提交
2349 2350 2351
	}
}

2352
static void bond_validate_arp(struct bonding *bond, struct slave *slave, __be32 sip, __be32 tip)
2353
{
2354 2355
	int i;

2356
	if (!sip || !bond_has_this_ip(bond, tip)) {
2357 2358
		netdev_dbg(bond->dev, "bva: sip %pI4 tip %pI4 not found\n",
			   &sip, &tip);
2359 2360
		return;
	}
2361

2362 2363
	i = bond_get_targets_ip(bond->params.arp_targets, sip);
	if (i == -1) {
2364 2365
		netdev_dbg(bond->dev, "bva: sip %pI4 not found in targets\n",
			   &sip);
2366
		return;
2367
	}
2368
	slave->last_rx = jiffies;
2369
	slave->target_last_arp_rx[i] = jiffies;
2370 2371
}

2372 2373
int bond_arp_rcv(const struct sk_buff *skb, struct bonding *bond,
		 struct slave *slave)
2374
{
2375
	struct arphdr *arp = (struct arphdr *)skb->data;
2376
	struct slave *curr_active_slave;
2377
	unsigned char *arp_ptr;
2378
	__be32 sip, tip;
2379
	int alen, is_arp = skb->protocol == __cpu_to_be16(ETH_P_ARP);
2380

2381
	if (!slave_do_arp_validate(bond, slave)) {
2382 2383
		if ((slave_do_arp_validate_only(bond) && is_arp) ||
		    !slave_do_arp_validate_only(bond))
2384
			slave->last_rx = jiffies;
2385
		return RX_HANDLER_ANOTHER;
2386 2387 2388
	} else if (!is_arp) {
		return RX_HANDLER_ANOTHER;
	}
2389

2390
	alen = arp_hdr_len(bond->dev);
2391

2392 2393
	netdev_dbg(bond->dev, "bond_arp_rcv: skb->dev %s\n",
		   skb->dev->name);
2394

2395 2396 2397 2398 2399 2400 2401
	if (alen > skb_headlen(skb)) {
		arp = kmalloc(alen, GFP_ATOMIC);
		if (!arp)
			goto out_unlock;
		if (skb_copy_bits(skb, 0, arp, alen) < 0)
			goto out_unlock;
	}
2402

2403
	if (arp->ar_hln != bond->dev->addr_len ||
2404 2405 2406 2407 2408 2409 2410 2411
	    skb->pkt_type == PACKET_OTHERHOST ||
	    skb->pkt_type == PACKET_LOOPBACK ||
	    arp->ar_hrd != htons(ARPHRD_ETHER) ||
	    arp->ar_pro != htons(ETH_P_IP) ||
	    arp->ar_pln != 4)
		goto out_unlock;

	arp_ptr = (unsigned char *)(arp + 1);
2412
	arp_ptr += bond->dev->addr_len;
2413
	memcpy(&sip, arp_ptr, 4);
2414
	arp_ptr += 4 + bond->dev->addr_len;
2415 2416
	memcpy(&tip, arp_ptr, 4);

2417 2418 2419 2420
	netdev_dbg(bond->dev, "bond_arp_rcv: %s/%d av %d sv %d sip %pI4 tip %pI4\n",
		   slave->dev->name, bond_slave_state(slave),
		     bond->params.arp_validate, slave_do_arp_validate(bond, slave),
		     &sip, &tip);
2421

2422 2423
	curr_active_slave = rcu_dereference(bond->curr_active_slave);

2424
	/* Backup slaves won't see the ARP reply, but do come through
2425 2426 2427 2428 2429
	 * here for each ARP probe (so we swap the sip/tip to validate
	 * the probe).  In a "redundant switch, common router" type of
	 * configuration, the ARP probe will (hopefully) travel from
	 * the active, through one switch, the router, then the other
	 * switch before reaching the backup.
2430 2431 2432 2433 2434
	 *
	 * We 'trust' the arp requests if there is an active slave and
	 * it received valid arp reply(s) after it became active. This
	 * is done to avoid endless looping when we can't reach the
	 * arp_ip_target and fool ourselves with our own arp requests.
2435
	 */
2436

J
Jiri Pirko 已提交
2437
	if (bond_is_active_slave(slave))
2438
		bond_validate_arp(bond, slave, sip, tip);
2439 2440 2441
	else if (curr_active_slave &&
		 time_after(slave_last_rx(bond, curr_active_slave),
			    curr_active_slave->last_link_up))
2442 2443 2444
		bond_validate_arp(bond, slave, tip, sip);

out_unlock:
2445 2446
	if (arp != (struct arphdr *)skb->data)
		kfree(arp);
2447
	return RX_HANDLER_ANOTHER;
2448 2449
}

2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463
/* function to verify if we're in the arp_interval timeslice, returns true if
 * (last_act - arp_interval) <= jiffies <= (last_act + mod * arp_interval +
 * arp_interval/2) . the arp_interval/2 is needed for really fast networks.
 */
static bool bond_time_in_interval(struct bonding *bond, unsigned long last_act,
				  int mod)
{
	int delta_in_ticks = msecs_to_jiffies(bond->params.arp_interval);

	return time_in_range(jiffies,
			     last_act - delta_in_ticks,
			     last_act + mod * delta_in_ticks + delta_in_ticks/2);
}

2464
/* This function is called regularly to monitor each slave's link
L
Linus Torvalds 已提交
2465 2466 2467 2468 2469
 * ensuring that traffic is being sent and received when arp monitoring
 * is used in load-balancing mode. if the adapter has been dormant, then an
 * arp is transmitted to generate traffic. see activebackup_arp_monitor for
 * arp monitoring in active backup mode.
 */
2470
static void bond_loadbalance_arp_mon(struct work_struct *work)
L
Linus Torvalds 已提交
2471
{
2472 2473
	struct bonding *bond = container_of(work, struct bonding,
					    arp_work.work);
L
Linus Torvalds 已提交
2474
	struct slave *slave, *oldcurrent;
2475
	struct list_head *iter;
2476
	int do_failover = 0, slave_state_changed = 0;
L
Linus Torvalds 已提交
2477

2478
	if (!bond_has_slaves(bond))
L
Linus Torvalds 已提交
2479 2480
		goto re_arm;

2481 2482
	rcu_read_lock();

2483
	oldcurrent = rcu_dereference(bond->curr_active_slave);
L
Linus Torvalds 已提交
2484 2485
	/* see if any of the previous devices are up now (i.e. they have
	 * xmt and rcv traffic). the curr_active_slave does not come into
2486 2487 2488
	 * the picture unless it is null. also, slave->last_link_up is not
	 * needed here because we send an arp on each slave and give a slave
	 * as long as it needs to get the tx/rx within the delta.
L
Linus Torvalds 已提交
2489 2490 2491
	 * TODO: what about up/down delay in arp mode? it wasn't here before
	 *       so it can wait
	 */
2492
	bond_for_each_slave_rcu(bond, slave, iter) {
2493 2494
		unsigned long trans_start = dev_trans_start(slave->dev);

L
Linus Torvalds 已提交
2495
		if (slave->link != BOND_LINK_UP) {
2496
			if (bond_time_in_interval(bond, trans_start, 1) &&
2497
			    bond_time_in_interval(bond, slave->last_rx, 1)) {
L
Linus Torvalds 已提交
2498 2499

				slave->link  = BOND_LINK_UP;
2500
				slave_state_changed = 1;
L
Linus Torvalds 已提交
2501 2502 2503 2504 2505 2506 2507

				/* primary_slave has no meaning in round-robin
				 * mode. the window of a slave being up and
				 * curr_active_slave being null after enslaving
				 * is closed.
				 */
				if (!oldcurrent) {
2508 2509
					netdev_info(bond->dev, "link status definitely up for interface %s\n",
						    slave->dev->name);
L
Linus Torvalds 已提交
2510 2511
					do_failover = 1;
				} else {
2512 2513
					netdev_info(bond->dev, "interface %s is now up\n",
						    slave->dev->name);
L
Linus Torvalds 已提交
2514 2515 2516 2517 2518 2519 2520 2521 2522
				}
			}
		} else {
			/* slave->link == BOND_LINK_UP */

			/* not all switches will respond to an arp request
			 * when the source ip is 0, so don't take the link down
			 * if we don't know our ip yet
			 */
2523
			if (!bond_time_in_interval(bond, trans_start, 2) ||
2524
			    !bond_time_in_interval(bond, slave->last_rx, 2)) {
L
Linus Torvalds 已提交
2525 2526

				slave->link  = BOND_LINK_DOWN;
2527
				slave_state_changed = 1;
L
Linus Torvalds 已提交
2528

S
Stephen Hemminger 已提交
2529
				if (slave->link_failure_count < UINT_MAX)
L
Linus Torvalds 已提交
2530 2531
					slave->link_failure_count++;

2532 2533
				netdev_info(bond->dev, "interface %s is now down\n",
					    slave->dev->name);
L
Linus Torvalds 已提交
2534

S
Stephen Hemminger 已提交
2535
				if (slave == oldcurrent)
L
Linus Torvalds 已提交
2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546
					do_failover = 1;
			}
		}

		/* note: if switch is in round-robin mode, all links
		 * must tx arp to ensure all links rx an arp - otherwise
		 * links may oscillate or not come up at all; if switch is
		 * in something like xor mode, there is nothing we can
		 * do - all replies will be rx'ed on same link causing slaves
		 * to be unstable during low/no traffic periods
		 */
2547
		if (bond_slave_is_up(slave))
L
Linus Torvalds 已提交
2548 2549 2550
			bond_arp_send_all(bond, slave);
	}

2551 2552
	rcu_read_unlock();

2553
	if (do_failover || slave_state_changed) {
2554 2555
		if (!rtnl_trylock())
			goto re_arm;
L
Linus Torvalds 已提交
2556

2557 2558
		if (slave_state_changed) {
			bond_slave_state_change(bond);
2559 2560
			if (BOND_MODE(bond) == BOND_MODE_XOR)
				bond_update_slave_arr(bond, NULL);
2561 2562
		}
		if (do_failover) {
2563 2564 2565 2566
			block_netpoll_tx();
			bond_select_active_slave(bond);
			unblock_netpoll_tx();
		}
2567
		rtnl_unlock();
L
Linus Torvalds 已提交
2568 2569 2570
	}

re_arm:
2571
	if (bond->params.arp_interval)
2572 2573
		queue_delayed_work(bond->wq, &bond->arp_work,
				   msecs_to_jiffies(bond->params.arp_interval));
L
Linus Torvalds 已提交
2574 2575
}

2576
/* Called to inspect slaves for active-backup mode ARP monitor link state
2577 2578 2579 2580
 * changes.  Sets new_link in slaves to specify what action should take
 * place for the slave.  Returns 0 if no changes are found, >0 if changes
 * to link states must be committed.
 *
2581
 * Called with rcu_read_lock held.
L
Linus Torvalds 已提交
2582
 */
2583
static int bond_ab_arp_inspect(struct bonding *bond)
L
Linus Torvalds 已提交
2584
{
2585
	unsigned long trans_start, last_rx;
2586
	struct list_head *iter;
2587 2588
	struct slave *slave;
	int commit = 0;
2589

2590
	bond_for_each_slave_rcu(bond, slave, iter) {
2591
		slave->new_link = BOND_LINK_NOCHANGE;
2592
		last_rx = slave_last_rx(bond, slave);
L
Linus Torvalds 已提交
2593

2594
		if (slave->link != BOND_LINK_UP) {
2595
			if (bond_time_in_interval(bond, last_rx, 1)) {
2596 2597 2598 2599 2600
				slave->new_link = BOND_LINK_UP;
				commit++;
			}
			continue;
		}
L
Linus Torvalds 已提交
2601

2602
		/* Give slaves 2*delta after being enslaved or made
2603 2604 2605
		 * active.  This avoids bouncing, as the last receive
		 * times need a full ARP monitor cycle to be updated.
		 */
2606
		if (bond_time_in_interval(bond, slave->last_link_up, 2))
2607 2608
			continue;

2609
		/* Backup slave is down if:
2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620
		 * - No current_arp_slave AND
		 * - more than 3*delta since last receive AND
		 * - the bond has an IP address
		 *
		 * Note: a non-null current_arp_slave indicates
		 * the curr_active_slave went down and we are
		 * searching for a new one; under this condition
		 * we only take the curr_active_slave down - this
		 * gives each slave a chance to tx/rx traffic
		 * before being taken out
		 */
J
Jiri Pirko 已提交
2621
		if (!bond_is_active_slave(slave) &&
2622
		    !rcu_access_pointer(bond->current_arp_slave) &&
2623
		    !bond_time_in_interval(bond, last_rx, 3)) {
2624 2625 2626 2627
			slave->new_link = BOND_LINK_DOWN;
			commit++;
		}

2628
		/* Active slave is down if:
2629 2630 2631 2632
		 * - more than 2*delta since transmitting OR
		 * - (more than 2*delta since receive AND
		 *    the bond has an IP address)
		 */
2633
		trans_start = dev_trans_start(slave->dev);
J
Jiri Pirko 已提交
2634
		if (bond_is_active_slave(slave) &&
2635 2636
		    (!bond_time_in_interval(bond, trans_start, 2) ||
		     !bond_time_in_interval(bond, last_rx, 2))) {
2637 2638 2639
			slave->new_link = BOND_LINK_DOWN;
			commit++;
		}
L
Linus Torvalds 已提交
2640 2641
	}

2642 2643
	return commit;
}
L
Linus Torvalds 已提交
2644

2645
/* Called to commit link state changes noted by inspection step of
2646 2647
 * active-backup mode ARP monitor.
 *
2648
 * Called with RTNL hold.
2649
 */
2650
static void bond_ab_arp_commit(struct bonding *bond)
2651
{
2652
	unsigned long trans_start;
2653
	struct list_head *iter;
2654
	struct slave *slave;
L
Linus Torvalds 已提交
2655

2656
	bond_for_each_slave(bond, slave, iter) {
2657 2658 2659
		switch (slave->new_link) {
		case BOND_LINK_NOCHANGE:
			continue;
2660

2661
		case BOND_LINK_UP:
2662
			trans_start = dev_trans_start(slave->dev);
2663 2664
			if (rtnl_dereference(bond->curr_active_slave) != slave ||
			    (!rtnl_dereference(bond->curr_active_slave) &&
2665
			     bond_time_in_interval(bond, trans_start, 1))) {
2666 2667 2668
				struct slave *current_arp_slave;

				current_arp_slave = rtnl_dereference(bond->current_arp_slave);
2669
				bond_set_slave_link_state(slave, BOND_LINK_UP);
2670
				if (current_arp_slave) {
2671
					bond_set_slave_inactive_flags(
2672
						current_arp_slave,
2673
						BOND_SLAVE_NOTIFY_NOW);
2674
					RCU_INIT_POINTER(bond->current_arp_slave, NULL);
2675
				}
2676

2677 2678
				netdev_info(bond->dev, "link status definitely up for interface %s\n",
					    slave->dev->name);
2679

2680
				if (!rtnl_dereference(bond->curr_active_slave) ||
2681
				    slave == rtnl_dereference(bond->primary_slave))
2682
					goto do_failover;
L
Linus Torvalds 已提交
2683

2684
			}
L
Linus Torvalds 已提交
2685

2686
			continue;
L
Linus Torvalds 已提交
2687

2688 2689 2690 2691
		case BOND_LINK_DOWN:
			if (slave->link_failure_count < UINT_MAX)
				slave->link_failure_count++;

2692
			bond_set_slave_link_state(slave, BOND_LINK_DOWN);
2693 2694
			bond_set_slave_inactive_flags(slave,
						      BOND_SLAVE_NOTIFY_NOW);
2695

2696 2697
			netdev_info(bond->dev, "link status definitely down for interface %s, disabling it\n",
				    slave->dev->name);
2698

2699
			if (slave == rtnl_dereference(bond->curr_active_slave)) {
2700
				RCU_INIT_POINTER(bond->current_arp_slave, NULL);
2701
				goto do_failover;
L
Linus Torvalds 已提交
2702
			}
2703 2704

			continue;
2705 2706

		default:
2707 2708
			netdev_err(bond->dev, "impossible: new_link %d on slave %s\n",
				   slave->new_link, slave->dev->name);
2709
			continue;
L
Linus Torvalds 已提交
2710 2711
		}

2712
do_failover:
2713
		block_netpoll_tx();
2714
		bond_select_active_slave(bond);
2715
		unblock_netpoll_tx();
2716
	}
L
Linus Torvalds 已提交
2717

2718 2719
	bond_set_carrier(bond);
}
L
Linus Torvalds 已提交
2720

2721
/* Send ARP probes for active-backup mode ARP monitor.
2722
 *
2723
 * Called with rcu_read_lock held.
2724
 */
2725
static bool bond_ab_arp_probe(struct bonding *bond)
2726
{
2727
	struct slave *slave, *before = NULL, *new_slave = NULL,
2728 2729
		     *curr_arp_slave = rcu_dereference(bond->current_arp_slave),
		     *curr_active_slave = rcu_dereference(bond->curr_active_slave);
2730 2731
	struct list_head *iter;
	bool found = false;
2732
	bool should_notify_rtnl = BOND_SLAVE_NOTIFY_LATER;
2733

2734
	if (curr_arp_slave && curr_active_slave)
2735 2736 2737
		netdev_info(bond->dev, "PROBE: c_arp %s && cas %s BAD\n",
			    curr_arp_slave->dev->name,
			    curr_active_slave->dev->name);
L
Linus Torvalds 已提交
2738

2739 2740
	if (curr_active_slave) {
		bond_arp_send_all(bond, curr_active_slave);
2741
		return should_notify_rtnl;
2742
	}
L
Linus Torvalds 已提交
2743

2744 2745 2746 2747
	/* if we don't have a curr_active_slave, search for the next available
	 * backup slave from the current_arp_slave and make it the candidate
	 * for becoming the curr_active_slave
	 */
L
Linus Torvalds 已提交
2748

2749
	if (!curr_arp_slave) {
2750 2751 2752
		curr_arp_slave = bond_first_slave_rcu(bond);
		if (!curr_arp_slave)
			return should_notify_rtnl;
2753
	}
L
Linus Torvalds 已提交
2754

2755
	bond_set_slave_inactive_flags(curr_arp_slave, BOND_SLAVE_NOTIFY_LATER);
2756

2757
	bond_for_each_slave_rcu(bond, slave, iter) {
2758
		if (!found && !before && bond_slave_is_up(slave))
2759
			before = slave;
L
Linus Torvalds 已提交
2760

2761
		if (found && !new_slave && bond_slave_is_up(slave))
2762
			new_slave = slave;
2763 2764 2765 2766 2767 2768
		/* if the link state is up at this point, we
		 * mark it down - this can happen if we have
		 * simultaneous link failures and
		 * reselect_active_interface doesn't make this
		 * one the current slave so it is still marked
		 * up when it is actually down
L
Linus Torvalds 已提交
2769
		 */
2770
		if (!bond_slave_is_up(slave) && slave->link == BOND_LINK_UP) {
2771
			bond_set_slave_link_state(slave, BOND_LINK_DOWN);
2772 2773
			if (slave->link_failure_count < UINT_MAX)
				slave->link_failure_count++;
L
Linus Torvalds 已提交
2774

2775
			bond_set_slave_inactive_flags(slave,
2776
						      BOND_SLAVE_NOTIFY_LATER);
2777

2778 2779
			netdev_info(bond->dev, "backup interface %s is now down\n",
				    slave->dev->name);
L
Linus Torvalds 已提交
2780
		}
2781
		if (slave == curr_arp_slave)
2782
			found = true;
2783
	}
2784 2785 2786 2787

	if (!new_slave && before)
		new_slave = before;

2788 2789
	if (!new_slave)
		goto check_state;
2790

2791
	bond_set_slave_link_state(new_slave, BOND_LINK_BACK);
2792
	bond_set_slave_active_flags(new_slave, BOND_SLAVE_NOTIFY_LATER);
2793
	bond_arp_send_all(bond, new_slave);
2794
	new_slave->last_link_up = jiffies;
2795
	rcu_assign_pointer(bond->current_arp_slave, new_slave);
2796

2797 2798 2799 2800 2801 2802 2803 2804
check_state:
	bond_for_each_slave_rcu(bond, slave, iter) {
		if (slave->should_notify) {
			should_notify_rtnl = BOND_SLAVE_NOTIFY_NOW;
			break;
		}
	}
	return should_notify_rtnl;
2805
}
L
Linus Torvalds 已提交
2806

2807
static void bond_activebackup_arp_mon(struct work_struct *work)
2808 2809 2810
{
	struct bonding *bond = container_of(work, struct bonding,
					    arp_work.work);
2811 2812
	bool should_notify_peers = false;
	bool should_notify_rtnl = false;
2813
	int delta_in_ticks;
L
Linus Torvalds 已提交
2814

2815 2816 2817
	delta_in_ticks = msecs_to_jiffies(bond->params.arp_interval);

	if (!bond_has_slaves(bond))
2818 2819
		goto re_arm;

2820
	rcu_read_lock();
2821

2822 2823
	should_notify_peers = bond_should_notify_peers(bond);

2824 2825 2826
	if (bond_ab_arp_inspect(bond)) {
		rcu_read_unlock();

2827 2828 2829 2830 2831 2832
		/* Race avoidance with bond_close flush of workqueue */
		if (!rtnl_trylock()) {
			delta_in_ticks = 1;
			should_notify_peers = false;
			goto re_arm;
		}
2833

2834
		bond_ab_arp_commit(bond);
2835

2836
		rtnl_unlock();
2837
		rcu_read_lock();
2838 2839
	}

2840 2841
	should_notify_rtnl = bond_ab_arp_probe(bond);
	rcu_read_unlock();
2842

2843 2844
re_arm:
	if (bond->params.arp_interval)
2845 2846
		queue_delayed_work(bond->wq, &bond->arp_work, delta_in_ticks);

2847
	if (should_notify_peers || should_notify_rtnl) {
2848 2849
		if (!rtnl_trylock())
			return;
2850 2851 2852 2853 2854 2855 2856

		if (should_notify_peers)
			call_netdevice_notifiers(NETDEV_NOTIFY_PEERS,
						 bond->dev);
		if (should_notify_rtnl)
			bond_slave_state_notify(bond);

2857 2858
		rtnl_unlock();
	}
L
Linus Torvalds 已提交
2859 2860 2861 2862
}

/*-------------------------- netdev event handling --------------------------*/

2863
/* Change device name */
L
Linus Torvalds 已提交
2864 2865 2866 2867
static int bond_event_changename(struct bonding *bond)
{
	bond_remove_proc_entry(bond);
	bond_create_proc_entry(bond);
2868

2869 2870
	bond_debug_reregister(bond);

L
Linus Torvalds 已提交
2871 2872 2873
	return NOTIFY_DONE;
}

S
Stephen Hemminger 已提交
2874 2875
static int bond_master_netdev_event(unsigned long event,
				    struct net_device *bond_dev)
L
Linus Torvalds 已提交
2876
{
2877
	struct bonding *event_bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
2878 2879 2880 2881

	switch (event) {
	case NETDEV_CHANGENAME:
		return bond_event_changename(event_bond);
2882 2883 2884 2885 2886 2887
	case NETDEV_UNREGISTER:
		bond_remove_proc_entry(event_bond);
		break;
	case NETDEV_REGISTER:
		bond_create_proc_entry(event_bond);
		break;
2888 2889 2890 2891
	case NETDEV_NOTIFY_PEERS:
		if (event_bond->send_peer_notif)
			event_bond->send_peer_notif--;
		break;
L
Linus Torvalds 已提交
2892 2893 2894 2895 2896 2897 2898
	default:
		break;
	}

	return NOTIFY_DONE;
}

S
Stephen Hemminger 已提交
2899 2900
static int bond_slave_netdev_event(unsigned long event,
				   struct net_device *slave_dev)
L
Linus Torvalds 已提交
2901
{
2902
	struct slave *slave = bond_slave_get_rtnl(slave_dev), *primary;
2903 2904
	struct bonding *bond;
	struct net_device *bond_dev;
2905 2906
	u32 old_speed;
	u8 old_duplex;
L
Linus Torvalds 已提交
2907

2908 2909 2910 2911 2912 2913 2914 2915
	/* A netdev event can be generated while enslaving a device
	 * before netdev_rx_handler_register is called in which case
	 * slave will be NULL
	 */
	if (!slave)
		return NOTIFY_DONE;
	bond_dev = slave->bond->dev;
	bond = slave->bond;
2916
	primary = rtnl_dereference(bond->primary_slave);
2917

L
Linus Torvalds 已提交
2918 2919
	switch (event) {
	case NETDEV_UNREGISTER:
2920
		if (bond_dev->type != ARPHRD_ETHER)
2921 2922 2923
			bond_release_and_destroy(bond_dev, slave_dev);
		else
			bond_release(bond_dev, slave_dev);
L
Linus Torvalds 已提交
2924
		break;
2925
	case NETDEV_UP:
L
Linus Torvalds 已提交
2926
	case NETDEV_CHANGE:
2927 2928
		old_speed = slave->speed;
		old_duplex = slave->duplex;
2929

2930
		bond_update_speed_duplex(slave);
2931

2932
		if (BOND_MODE(bond) == BOND_MODE_8023AD) {
2933 2934 2935 2936
			if (old_speed != slave->speed)
				bond_3ad_adapter_speed_changed(slave);
			if (old_duplex != slave->duplex)
				bond_3ad_adapter_duplex_changed(slave);
2937
		}
M
Mahesh Bandewar 已提交
2938 2939
		/* Fallthrough */
	case NETDEV_DOWN:
2940 2941 2942 2943 2944 2945 2946 2947 2948 2949
		/* Refresh slave-array if applicable!
		 * If the setup does not use miimon or arpmon (mode-specific!),
		 * then these events will not cause the slave-array to be
		 * refreshed. This will cause xmit to use a slave that is not
		 * usable. Avoid such situation by refeshing the array at these
		 * events. If these (miimon/arpmon) parameters are configured
		 * then array gets refreshed twice and that should be fine!
		 */
		if (bond_mode_uses_xmit_hash(bond))
			bond_update_slave_arr(bond, NULL);
L
Linus Torvalds 已提交
2950 2951
		break;
	case NETDEV_CHANGEMTU:
2952
		/* TODO: Should slaves be allowed to
L
Linus Torvalds 已提交
2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964
		 * independently alter their MTU?  For
		 * an active-backup bond, slaves need
		 * not be the same type of device, so
		 * MTUs may vary.  For other modes,
		 * slaves arguably should have the
		 * same MTUs. To do this, we'd need to
		 * take over the slave's change_mtu
		 * function for the duration of their
		 * servitude.
		 */
		break;
	case NETDEV_CHANGENAME:
2965
		/* we don't care if we don't have primary set */
2966
		if (!bond_uses_primary(bond) ||
2967 2968 2969
		    !bond->params.primary[0])
			break;

2970
		if (slave == primary) {
2971
			/* slave's name changed - he's no longer primary */
2972
			RCU_INIT_POINTER(bond->primary_slave, NULL);
2973 2974
		} else if (!strcmp(slave_dev->name, bond->params.primary)) {
			/* we have a new primary slave */
2975
			rcu_assign_pointer(bond->primary_slave, slave);
2976 2977 2978 2979
		} else { /* we didn't change primary - exit */
			break;
		}

2980
		netdev_info(bond->dev, "Primary slave changed to %s, reselecting active slave\n",
2981
			    primary ? slave_dev->name : "none");
2982 2983

		block_netpoll_tx();
2984
		bond_select_active_slave(bond);
2985
		unblock_netpoll_tx();
L
Linus Torvalds 已提交
2986
		break;
2987 2988 2989
	case NETDEV_FEAT_CHANGE:
		bond_compute_features(bond);
		break;
2990 2991 2992 2993
	case NETDEV_RESEND_IGMP:
		/* Propagate to master device */
		call_netdevice_notifiers(event, slave->bond->dev);
		break;
L
Linus Torvalds 已提交
2994 2995 2996 2997 2998 2999 3000
	default:
		break;
	}

	return NOTIFY_DONE;
}

3001
/* bond_netdev_event: handle netdev notifier chain events.
L
Linus Torvalds 已提交
3002 3003
 *
 * This function receives events for the netdev chain.  The caller (an
3004
 * ioctl handler calling blocking_notifier_call_chain) holds the necessary
L
Linus Torvalds 已提交
3005 3006 3007
 * locks for us to safely manipulate the slave devices (RTNL lock,
 * dev_probe_lock).
 */
S
Stephen Hemminger 已提交
3008 3009
static int bond_netdev_event(struct notifier_block *this,
			     unsigned long event, void *ptr)
L
Linus Torvalds 已提交
3010
{
3011
	struct net_device *event_dev = netdev_notifier_info_to_dev(ptr);
L
Linus Torvalds 已提交
3012

3013
	netdev_dbg(event_dev, "event: %lx\n", event);
L
Linus Torvalds 已提交
3014

3015 3016 3017
	if (!(event_dev->priv_flags & IFF_BONDING))
		return NOTIFY_DONE;

L
Linus Torvalds 已提交
3018
	if (event_dev->flags & IFF_MASTER) {
3019
		netdev_dbg(event_dev, "IFF_MASTER\n");
L
Linus Torvalds 已提交
3020 3021 3022 3023
		return bond_master_netdev_event(event, event_dev);
	}

	if (event_dev->flags & IFF_SLAVE) {
3024
		netdev_dbg(event_dev, "IFF_SLAVE\n");
L
Linus Torvalds 已提交
3025 3026 3027 3028 3029 3030 3031 3032 3033 3034
		return bond_slave_netdev_event(event, event_dev);
	}

	return NOTIFY_DONE;
}

static struct notifier_block bond_netdev_notifier = {
	.notifier_call = bond_netdev_event,
};

3035 3036
/*---------------------------- Hashing Policies -----------------------------*/

3037 3038
/* L2 hash helper */
static inline u32 bond_eth_hash(struct sk_buff *skb)
3039
{
3040
	struct ethhdr *ep, hdr_tmp;
3041

3042 3043 3044
	ep = skb_header_pointer(skb, 0, sizeof(hdr_tmp), &hdr_tmp);
	if (ep)
		return ep->h_dest[5] ^ ep->h_source[5] ^ ep->h_proto;
3045 3046 3047
	return 0;
}

3048 3049 3050
/* Extract the appropriate headers based on bond's xmit policy */
static bool bond_flow_dissect(struct bonding *bond, struct sk_buff *skb,
			      struct flow_keys *fk)
3051
{
3052
	const struct ipv6hdr *iph6;
3053
	const struct iphdr *iph;
3054
	int noff, proto = -1;
3055

3056 3057 3058 3059 3060 3061
	if (bond->params.xmit_policy > BOND_XMIT_POLICY_LAYER23)
		return skb_flow_dissect(skb, fk);

	fk->ports = 0;
	noff = skb_network_offset(skb);
	if (skb->protocol == htons(ETH_P_IP)) {
3062
		if (unlikely(!pskb_may_pull(skb, noff + sizeof(*iph))))
3063
			return false;
3064
		iph = ip_hdr(skb);
3065 3066 3067 3068 3069 3070
		fk->src = iph->saddr;
		fk->dst = iph->daddr;
		noff += iph->ihl << 2;
		if (!ip_is_fragment(iph))
			proto = iph->protocol;
	} else if (skb->protocol == htons(ETH_P_IPV6)) {
3071
		if (unlikely(!pskb_may_pull(skb, noff + sizeof(*iph6))))
3072 3073 3074 3075 3076 3077 3078 3079
			return false;
		iph6 = ipv6_hdr(skb);
		fk->src = (__force __be32)ipv6_addr_hash(&iph6->saddr);
		fk->dst = (__force __be32)ipv6_addr_hash(&iph6->daddr);
		noff += sizeof(*iph6);
		proto = iph6->nexthdr;
	} else {
		return false;
3080
	}
3081 3082
	if (bond->params.xmit_policy == BOND_XMIT_POLICY_LAYER34 && proto >= 0)
		fk->ports = skb_flow_get_ports(skb, noff, proto);
3083

3084
	return true;
3085 3086
}

3087 3088 3089 3090 3091 3092 3093
/**
 * bond_xmit_hash - generate a hash value based on the xmit policy
 * @bond: bonding device
 * @skb: buffer to use for headers
 *
 * This function will extract the necessary headers from the skb buffer and use
 * them to generate a hash based on the xmit_policy set in the bonding device
3094
 */
3095
u32 bond_xmit_hash(struct bonding *bond, struct sk_buff *skb)
3096
{
3097 3098
	struct flow_keys flow;
	u32 hash;
3099

3100 3101
	if (bond->params.xmit_policy == BOND_XMIT_POLICY_LAYER2 ||
	    !bond_flow_dissect(bond, skb, &flow))
3102
		return bond_eth_hash(skb);
3103

3104 3105 3106 3107 3108 3109 3110 3111 3112
	if (bond->params.xmit_policy == BOND_XMIT_POLICY_LAYER23 ||
	    bond->params.xmit_policy == BOND_XMIT_POLICY_ENCAP23)
		hash = bond_eth_hash(skb);
	else
		hash = (__force u32)flow.ports;
	hash ^= (__force u32)flow.dst ^ (__force u32)flow.src;
	hash ^= (hash >> 16);
	hash ^= (hash >> 8);

3113
	return hash;
3114 3115
}

L
Linus Torvalds 已提交
3116 3117
/*-------------------------- Device entry points ----------------------------*/

3118 3119 3120 3121 3122 3123
static void bond_work_init_all(struct bonding *bond)
{
	INIT_DELAYED_WORK(&bond->mcast_work,
			  bond_resend_igmp_join_requests_delayed);
	INIT_DELAYED_WORK(&bond->alb_work, bond_alb_monitor);
	INIT_DELAYED_WORK(&bond->mii_work, bond_mii_monitor);
3124
	if (BOND_MODE(bond) == BOND_MODE_ACTIVEBACKUP)
3125 3126 3127 3128
		INIT_DELAYED_WORK(&bond->arp_work, bond_activebackup_arp_mon);
	else
		INIT_DELAYED_WORK(&bond->arp_work, bond_loadbalance_arp_mon);
	INIT_DELAYED_WORK(&bond->ad_work, bond_3ad_state_machine_handler);
3129
	INIT_DELAYED_WORK(&bond->slave_arr_work, bond_slave_arr_handler);
3130 3131 3132 3133 3134 3135 3136 3137 3138
}

static void bond_work_cancel_all(struct bonding *bond)
{
	cancel_delayed_work_sync(&bond->mii_work);
	cancel_delayed_work_sync(&bond->arp_work);
	cancel_delayed_work_sync(&bond->alb_work);
	cancel_delayed_work_sync(&bond->ad_work);
	cancel_delayed_work_sync(&bond->mcast_work);
3139
	cancel_delayed_work_sync(&bond->slave_arr_work);
3140 3141
}

L
Linus Torvalds 已提交
3142 3143
static int bond_open(struct net_device *bond_dev)
{
3144
	struct bonding *bond = netdev_priv(bond_dev);
3145
	struct list_head *iter;
3146
	struct slave *slave;
L
Linus Torvalds 已提交
3147

3148
	/* reset slave->backup and slave->inactive */
3149
	if (bond_has_slaves(bond)) {
3150
		bond_for_each_slave(bond, slave, iter) {
3151 3152
			if (bond_uses_primary(bond) &&
			    slave != rcu_access_pointer(bond->curr_active_slave)) {
3153 3154
				bond_set_slave_inactive_flags(slave,
							      BOND_SLAVE_NOTIFY_NOW);
3155
			} else if (BOND_MODE(bond) != BOND_MODE_8023AD) {
3156 3157
				bond_set_slave_active_flags(slave,
							    BOND_SLAVE_NOTIFY_NOW);
3158 3159 3160 3161
			}
		}
	}

3162
	bond_work_init_all(bond);
3163

3164
	if (bond_is_lb(bond)) {
L
Linus Torvalds 已提交
3165 3166 3167
		/* bond_alb_initialize must be called before the timer
		 * is started.
		 */
3168
		if (bond_alb_initialize(bond, (BOND_MODE(bond) == BOND_MODE_ALB)))
3169
			return -ENOMEM;
3170 3171
		if (bond->params.tlb_dynamic_lb)
			queue_delayed_work(bond->wq, &bond->alb_work, 0);
L
Linus Torvalds 已提交
3172 3173
	}

3174
	if (bond->params.miimon)  /* link check interval, in milliseconds. */
3175
		queue_delayed_work(bond->wq, &bond->mii_work, 0);
L
Linus Torvalds 已提交
3176 3177

	if (bond->params.arp_interval) {  /* arp interval, in milliseconds. */
3178
		queue_delayed_work(bond->wq, &bond->arp_work, 0);
3179
		bond->recv_probe = bond_arp_rcv;
L
Linus Torvalds 已提交
3180 3181
	}

3182
	if (BOND_MODE(bond) == BOND_MODE_8023AD) {
3183
		queue_delayed_work(bond->wq, &bond->ad_work, 0);
L
Linus Torvalds 已提交
3184
		/* register to receive LACPDUs */
3185
		bond->recv_probe = bond_3ad_lacpdu_recv;
3186
		bond_3ad_initiate_agg_selection(bond, 1);
L
Linus Torvalds 已提交
3187 3188
	}

3189 3190 3191
	if (bond_mode_uses_xmit_hash(bond))
		bond_update_slave_arr(bond, NULL);

L
Linus Torvalds 已提交
3192 3193 3194 3195 3196
	return 0;
}

static int bond_close(struct net_device *bond_dev)
{
3197
	struct bonding *bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
3198

3199
	bond_work_cancel_all(bond);
3200
	bond->send_peer_notif = 0;
3201
	if (bond_is_lb(bond))
L
Linus Torvalds 已提交
3202
		bond_alb_deinitialize(bond);
3203
	bond->recv_probe = NULL;
L
Linus Torvalds 已提交
3204 3205 3206 3207

	return 0;
}

3208 3209
static struct rtnl_link_stats64 *bond_get_stats(struct net_device *bond_dev,
						struct rtnl_link_stats64 *stats)
L
Linus Torvalds 已提交
3210
{
3211
	struct bonding *bond = netdev_priv(bond_dev);
3212
	struct rtnl_link_stats64 temp;
3213
	struct list_head *iter;
L
Linus Torvalds 已提交
3214 3215
	struct slave *slave;

3216
	memcpy(stats, &bond->bond_stats, sizeof(*stats));
L
Linus Torvalds 已提交
3217

3218
	bond_for_each_slave(bond, slave, iter) {
3219
		const struct rtnl_link_stats64 *sstats =
3220
			dev_get_stats(slave->dev, &temp);
3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252
		struct rtnl_link_stats64 *pstats = &slave->slave_stats;

		stats->rx_packets +=  sstats->rx_packets - pstats->rx_packets;
		stats->rx_bytes += sstats->rx_bytes - pstats->rx_bytes;
		stats->rx_errors += sstats->rx_errors - pstats->rx_errors;
		stats->rx_dropped += sstats->rx_dropped - pstats->rx_dropped;

		stats->tx_packets += sstats->tx_packets - pstats->tx_packets;;
		stats->tx_bytes += sstats->tx_bytes - pstats->tx_bytes;
		stats->tx_errors += sstats->tx_errors - pstats->tx_errors;
		stats->tx_dropped += sstats->tx_dropped - pstats->tx_dropped;

		stats->multicast += sstats->multicast - pstats->multicast;
		stats->collisions += sstats->collisions - pstats->collisions;

		stats->rx_length_errors += sstats->rx_length_errors - pstats->rx_length_errors;
		stats->rx_over_errors += sstats->rx_over_errors - pstats->rx_over_errors;
		stats->rx_crc_errors += sstats->rx_crc_errors - pstats->rx_crc_errors;
		stats->rx_frame_errors += sstats->rx_frame_errors - pstats->rx_frame_errors;
		stats->rx_fifo_errors += sstats->rx_fifo_errors - pstats->rx_fifo_errors;
		stats->rx_missed_errors += sstats->rx_missed_errors - pstats->rx_missed_errors;

		stats->tx_aborted_errors += sstats->tx_aborted_errors - pstats->tx_aborted_errors;
		stats->tx_carrier_errors += sstats->tx_carrier_errors - pstats->tx_carrier_errors;
		stats->tx_fifo_errors += sstats->tx_fifo_errors - pstats->tx_fifo_errors;
		stats->tx_heartbeat_errors += sstats->tx_heartbeat_errors - pstats->tx_heartbeat_errors;
		stats->tx_window_errors += sstats->tx_window_errors - pstats->tx_window_errors;

		/* save off the slave stats for the next run */
		memcpy(pstats, sstats, sizeof(*sstats));
	}
	memcpy(&bond->bond_stats, stats, sizeof(*stats));
L
Linus Torvalds 已提交
3253 3254 3255 3256 3257 3258

	return stats;
}

static int bond_do_ioctl(struct net_device *bond_dev, struct ifreq *ifr, int cmd)
{
3259
	struct bonding *bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
3260 3261 3262 3263 3264 3265
	struct net_device *slave_dev = NULL;
	struct ifbond k_binfo;
	struct ifbond __user *u_binfo = NULL;
	struct ifslave k_sinfo;
	struct ifslave __user *u_sinfo = NULL;
	struct mii_ioctl_data *mii = NULL;
3266
	struct bond_opt_value newval;
3267
	struct net *net;
L
Linus Torvalds 已提交
3268 3269
	int res = 0;

3270
	netdev_dbg(bond_dev, "bond_ioctl: cmd=%d\n", cmd);
L
Linus Torvalds 已提交
3271 3272 3273 3274

	switch (cmd) {
	case SIOCGMIIPHY:
		mii = if_mii(ifr);
S
Stephen Hemminger 已提交
3275
		if (!mii)
L
Linus Torvalds 已提交
3276
			return -EINVAL;
S
Stephen Hemminger 已提交
3277

L
Linus Torvalds 已提交
3278 3279 3280
		mii->phy_id = 0;
		/* Fall Through */
	case SIOCGMIIREG:
3281
		/* We do this again just in case we were called by SIOCGMIIREG
L
Linus Torvalds 已提交
3282 3283 3284
		 * instead of SIOCGMIIPHY.
		 */
		mii = if_mii(ifr);
S
Stephen Hemminger 已提交
3285
		if (!mii)
L
Linus Torvalds 已提交
3286
			return -EINVAL;
S
Stephen Hemminger 已提交
3287

L
Linus Torvalds 已提交
3288 3289
		if (mii->reg_num == 1) {
			mii->val_out = 0;
S
Stephen Hemminger 已提交
3290
			if (netif_carrier_ok(bond->dev))
L
Linus Torvalds 已提交
3291 3292 3293 3294 3295 3296 3297 3298
				mii->val_out = BMSR_LSTATUS;
		}

		return 0;
	case BOND_INFO_QUERY_OLD:
	case SIOCBONDINFOQUERY:
		u_binfo = (struct ifbond __user *)ifr->ifr_data;

S
Stephen Hemminger 已提交
3299
		if (copy_from_user(&k_binfo, u_binfo, sizeof(ifbond)))
L
Linus Torvalds 已提交
3300 3301 3302
			return -EFAULT;

		res = bond_info_query(bond_dev, &k_binfo);
S
Stephen Hemminger 已提交
3303 3304 3305
		if (res == 0 &&
		    copy_to_user(u_binfo, &k_binfo, sizeof(ifbond)))
			return -EFAULT;
L
Linus Torvalds 已提交
3306 3307 3308 3309 3310 3311

		return res;
	case BOND_SLAVE_INFO_QUERY_OLD:
	case SIOCBONDSLAVEINFOQUERY:
		u_sinfo = (struct ifslave __user *)ifr->ifr_data;

S
Stephen Hemminger 已提交
3312
		if (copy_from_user(&k_sinfo, u_sinfo, sizeof(ifslave)))
L
Linus Torvalds 已提交
3313 3314 3315
			return -EFAULT;

		res = bond_slave_info_query(bond_dev, &k_sinfo);
S
Stephen Hemminger 已提交
3316 3317 3318
		if (res == 0 &&
		    copy_to_user(u_sinfo, &k_sinfo, sizeof(ifslave)))
			return -EFAULT;
L
Linus Torvalds 已提交
3319 3320 3321 3322 3323 3324

		return res;
	default:
		break;
	}

3325 3326 3327
	net = dev_net(bond_dev);

	if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
L
Linus Torvalds 已提交
3328 3329
		return -EPERM;

3330
	slave_dev = __dev_get_by_name(net, ifr->ifr_slave);
L
Linus Torvalds 已提交
3331

3332
	netdev_dbg(bond_dev, "slave_dev=%p:\n", slave_dev);
L
Linus Torvalds 已提交
3333

S
Stephen Hemminger 已提交
3334
	if (!slave_dev)
3335
		return -ENODEV;
L
Linus Torvalds 已提交
3336

3337
	netdev_dbg(bond_dev, "slave_dev->name=%s:\n", slave_dev->name);
3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353
	switch (cmd) {
	case BOND_ENSLAVE_OLD:
	case SIOCBONDENSLAVE:
		res = bond_enslave(bond_dev, slave_dev);
		break;
	case BOND_RELEASE_OLD:
	case SIOCBONDRELEASE:
		res = bond_release(bond_dev, slave_dev);
		break;
	case BOND_SETHWADDR_OLD:
	case SIOCBONDSETHWADDR:
		bond_set_dev_addr(bond_dev, slave_dev);
		res = 0;
		break;
	case BOND_CHANGE_ACTIVE_OLD:
	case SIOCBONDCHANGEACTIVE:
3354 3355
		bond_opt_initstr(&newval, slave_dev->name);
		res = __bond_opt_set(bond, BOND_OPT_ACTIVE_SLAVE, &newval);
3356 3357 3358
		break;
	default:
		res = -EOPNOTSUPP;
L
Linus Torvalds 已提交
3359 3360 3361 3362 3363
	}

	return res;
}

3364
static void bond_change_rx_flags(struct net_device *bond_dev, int change)
L
Linus Torvalds 已提交
3365
{
3366
	struct bonding *bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
3367

3368 3369 3370
	if (change & IFF_PROMISC)
		bond_set_promiscuity(bond,
				     bond_dev->flags & IFF_PROMISC ? 1 : -1);
S
Stephen Hemminger 已提交
3371

3372 3373 3374 3375
	if (change & IFF_ALLMULTI)
		bond_set_allmulti(bond,
				  bond_dev->flags & IFF_ALLMULTI ? 1 : -1);
}
L
Linus Torvalds 已提交
3376

3377
static void bond_set_rx_mode(struct net_device *bond_dev)
3378 3379
{
	struct bonding *bond = netdev_priv(bond_dev);
3380
	struct list_head *iter;
3381
	struct slave *slave;
L
Linus Torvalds 已提交
3382

3383
	rcu_read_lock();
3384
	if (bond_uses_primary(bond)) {
3385
		slave = rcu_dereference(bond->curr_active_slave);
3386 3387 3388 3389 3390
		if (slave) {
			dev_uc_sync(slave->dev, bond_dev);
			dev_mc_sync(slave->dev, bond_dev);
		}
	} else {
3391
		bond_for_each_slave_rcu(bond, slave, iter) {
3392 3393 3394
			dev_uc_sync_multiple(slave->dev, bond_dev);
			dev_mc_sync_multiple(slave->dev, bond_dev);
		}
L
Linus Torvalds 已提交
3395
	}
3396
	rcu_read_unlock();
L
Linus Torvalds 已提交
3397 3398
}

3399
static int bond_neigh_init(struct neighbour *n)
3400
{
3401 3402 3403
	struct bonding *bond = netdev_priv(n->dev);
	const struct net_device_ops *slave_ops;
	struct neigh_parms parms;
3404
	struct slave *slave;
3405 3406
	int ret;

3407
	slave = bond_first_slave(bond);
3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419
	if (!slave)
		return 0;
	slave_ops = slave->dev->netdev_ops;
	if (!slave_ops->ndo_neigh_setup)
		return 0;

	parms.neigh_setup = NULL;
	parms.neigh_cleanup = NULL;
	ret = slave_ops->ndo_neigh_setup(slave->dev, &parms);
	if (ret)
		return ret;

3420
	/* Assign slave's neigh_cleanup to neighbour in case cleanup is called
3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432
	 * after the last slave has been detached.  Assumes that all slaves
	 * utilize the same neigh_cleanup (true at this writing as only user
	 * is ipoib).
	 */
	n->parms->neigh_cleanup = parms.neigh_cleanup;

	if (!parms.neigh_setup)
		return 0;

	return parms.neigh_setup(n);
}

3433
/* The bonding ndo_neigh_setup is called at init time beofre any
3434 3435
 * slave exists. So we must declare proxy setup function which will
 * be used at run time to resolve the actual slave neigh param setup.
3436 3437 3438 3439
 *
 * It's also called by master devices (such as vlans) to setup their
 * underlying devices. In that case - do nothing, we're already set up from
 * our init.
3440 3441 3442 3443
 */
static int bond_neigh_setup(struct net_device *dev,
			    struct neigh_parms *parms)
{
3444 3445 3446
	/* modify only our neigh_parms */
	if (parms->dev == dev)
		parms->neigh_setup = bond_neigh_init;
3447 3448 3449 3450

	return 0;
}

3451
/* Change the MTU of all of a master's slaves to match the master */
L
Linus Torvalds 已提交
3452 3453
static int bond_change_mtu(struct net_device *bond_dev, int new_mtu)
{
3454
	struct bonding *bond = netdev_priv(bond_dev);
3455
	struct slave *slave, *rollback_slave;
3456
	struct list_head *iter;
L
Linus Torvalds 已提交
3457 3458
	int res = 0;

3459
	netdev_dbg(bond_dev, "bond=%p, new_mtu=%d\n", bond, new_mtu);
L
Linus Torvalds 已提交
3460

3461
	bond_for_each_slave(bond, slave, iter) {
3462 3463
		netdev_dbg(bond_dev, "s %p c_m %p\n",
			   slave, slave->dev->netdev_ops->ndo_change_mtu);
3464

L
Linus Torvalds 已提交
3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475
		res = dev_set_mtu(slave->dev, new_mtu);

		if (res) {
			/* If we failed to set the slave's mtu to the new value
			 * we must abort the operation even in ACTIVE_BACKUP
			 * mode, because if we allow the backup slaves to have
			 * different mtu values than the active slave we'll
			 * need to change their mtu when doing a failover. That
			 * means changing their mtu from timer context, which
			 * is probably not a good idea.
			 */
3476 3477
			netdev_dbg(bond_dev, "err %d %s\n", res,
				   slave->dev->name);
L
Linus Torvalds 已提交
3478 3479 3480 3481 3482 3483 3484 3485 3486 3487
			goto unwind;
		}
	}

	bond_dev->mtu = new_mtu;

	return 0;

unwind:
	/* unwind from head to the slave that failed */
3488
	bond_for_each_slave(bond, rollback_slave, iter) {
L
Linus Torvalds 已提交
3489 3490
		int tmp_res;

3491 3492 3493 3494
		if (rollback_slave == slave)
			break;

		tmp_res = dev_set_mtu(rollback_slave->dev, bond_dev->mtu);
L
Linus Torvalds 已提交
3495
		if (tmp_res) {
3496 3497
			netdev_dbg(bond_dev, "unwind err %d dev %s\n",
				   tmp_res, rollback_slave->dev->name);
L
Linus Torvalds 已提交
3498 3499 3500 3501 3502 3503
		}
	}

	return res;
}

3504
/* Change HW address
L
Linus Torvalds 已提交
3505 3506 3507 3508 3509 3510 3511
 *
 * Note that many devices must be down to change the HW address, and
 * downing the master releases all slaves.  We can make bonds full of
 * bonding devices to test this, however.
 */
static int bond_set_mac_address(struct net_device *bond_dev, void *addr)
{
3512
	struct bonding *bond = netdev_priv(bond_dev);
3513
	struct slave *slave, *rollback_slave;
L
Linus Torvalds 已提交
3514
	struct sockaddr *sa = addr, tmp_sa;
3515
	struct list_head *iter;
L
Linus Torvalds 已提交
3516 3517
	int res = 0;

3518
	if (BOND_MODE(bond) == BOND_MODE_ALB)
3519 3520 3521
		return bond_alb_set_mac_address(bond_dev, addr);


3522
	netdev_dbg(bond_dev, "bond=%p\n", bond);
L
Linus Torvalds 已提交
3523

3524 3525
	/* If fail_over_mac is enabled, do nothing and return success.
	 * Returning an error causes ifenslave to fail.
3526
	 */
3527
	if (bond->params.fail_over_mac &&
3528
	    BOND_MODE(bond) == BOND_MODE_ACTIVEBACKUP)
3529
		return 0;
3530

S
Stephen Hemminger 已提交
3531
	if (!is_valid_ether_addr(sa->sa_data))
L
Linus Torvalds 已提交
3532 3533
		return -EADDRNOTAVAIL;

3534
	bond_for_each_slave(bond, slave, iter) {
3535
		netdev_dbg(bond_dev, "slave %p %s\n", slave, slave->dev->name);
L
Linus Torvalds 已提交
3536 3537 3538 3539 3540 3541 3542 3543
		res = dev_set_mac_address(slave->dev, addr);
		if (res) {
			/* TODO: consider downing the slave
			 * and retry ?
			 * User should expect communications
			 * breakage anyway until ARP finish
			 * updating, so...
			 */
3544
			netdev_dbg(bond_dev, "err %d %s\n", res, slave->dev->name);
L
Linus Torvalds 已提交
3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557
			goto unwind;
		}
	}

	/* success */
	memcpy(bond_dev->dev_addr, sa->sa_data, bond_dev->addr_len);
	return 0;

unwind:
	memcpy(tmp_sa.sa_data, bond_dev->dev_addr, bond_dev->addr_len);
	tmp_sa.sa_family = bond_dev->type;

	/* unwind from head to the slave that failed */
3558
	bond_for_each_slave(bond, rollback_slave, iter) {
L
Linus Torvalds 已提交
3559 3560
		int tmp_res;

3561 3562 3563 3564
		if (rollback_slave == slave)
			break;

		tmp_res = dev_set_mac_address(rollback_slave->dev, &tmp_sa);
L
Linus Torvalds 已提交
3565
		if (tmp_res) {
3566 3567
			netdev_dbg(bond_dev, "unwind err %d dev %s\n",
				   tmp_res, rollback_slave->dev->name);
L
Linus Torvalds 已提交
3568 3569 3570 3571 3572 3573
		}
	}

	return res;
}

3574 3575 3576 3577 3578 3579 3580 3581 3582 3583
/**
 * bond_xmit_slave_id - transmit skb through slave with slave_id
 * @bond: bonding device that is transmitting
 * @skb: buffer to transmit
 * @slave_id: slave id up to slave_cnt-1 through which to transmit
 *
 * This function tries to transmit through slave with slave_id but in case
 * it fails, it tries to find the first available slave for transmission.
 * The skb is consumed in all cases, thus the function is void.
 */
3584
static void bond_xmit_slave_id(struct bonding *bond, struct sk_buff *skb, int slave_id)
3585
{
3586
	struct list_head *iter;
3587 3588 3589 3590
	struct slave *slave;
	int i = slave_id;

	/* Here we start from the slave with slave_id */
3591
	bond_for_each_slave_rcu(bond, slave, iter) {
3592
		if (--i < 0) {
3593
			if (bond_slave_can_tx(slave)) {
3594 3595 3596 3597 3598 3599 3600 3601
				bond_dev_queue_xmit(bond, skb, slave->dev);
				return;
			}
		}
	}

	/* Here we start from the first slave up to slave_id */
	i = slave_id;
3602
	bond_for_each_slave_rcu(bond, slave, iter) {
3603 3604
		if (--i < 0)
			break;
3605
		if (bond_slave_can_tx(slave)) {
3606 3607 3608 3609 3610
			bond_dev_queue_xmit(bond, skb, slave->dev);
			return;
		}
	}
	/* no slave that can tx has been found */
E
Eric Dumazet 已提交
3611
	bond_tx_drop(bond->dev, skb);
3612 3613
}

3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624
/**
 * bond_rr_gen_slave_id - generate slave id based on packets_per_slave
 * @bond: bonding device to use
 *
 * Based on the value of the bonding device's packets_per_slave parameter
 * this function generates a slave id, which is usually used as the next
 * slave to transmit through.
 */
static u32 bond_rr_gen_slave_id(struct bonding *bond)
{
	u32 slave_id;
3625 3626
	struct reciprocal_value reciprocal_packets_per_slave;
	int packets_per_slave = bond->params.packets_per_slave;
3627 3628 3629 3630 3631 3632 3633 3634 3635

	switch (packets_per_slave) {
	case 0:
		slave_id = prandom_u32();
		break;
	case 1:
		slave_id = bond->rr_tx_counter;
		break;
	default:
3636 3637
		reciprocal_packets_per_slave =
			bond->params.reciprocal_packets_per_slave;
3638
		slave_id = reciprocal_divide(bond->rr_tx_counter,
3639
					     reciprocal_packets_per_slave);
3640 3641 3642 3643 3644 3645 3646
		break;
	}
	bond->rr_tx_counter++;

	return slave_id;
}

L
Linus Torvalds 已提交
3647 3648
static int bond_xmit_roundrobin(struct sk_buff *skb, struct net_device *bond_dev)
{
3649
	struct bonding *bond = netdev_priv(bond_dev);
3650
	struct iphdr *iph = ip_hdr(skb);
3651
	struct slave *slave;
3652
	u32 slave_id;
L
Linus Torvalds 已提交
3653

3654
	/* Start with the curr_active_slave that joined the bond as the
3655 3656 3657 3658
	 * default for sending IGMP traffic.  For failover purposes one
	 * needs to maintain some consistency for the interface that will
	 * send the join/membership reports.  The curr_active_slave found
	 * will send all of this type of traffic.
3659
	 */
3660
	if (iph->protocol == IPPROTO_IGMP && skb->protocol == htons(ETH_P_IP)) {
3661
		slave = rcu_dereference(bond->curr_active_slave);
3662
		if (slave)
3663 3664 3665
			bond_dev_queue_xmit(bond, skb, slave->dev);
		else
			bond_xmit_slave_id(bond, skb, 0);
3666
	} else {
3667 3668 3669 3670 3671 3672
		int slave_cnt = ACCESS_ONCE(bond->slave_cnt);

		if (likely(slave_cnt)) {
			slave_id = bond_rr_gen_slave_id(bond);
			bond_xmit_slave_id(bond, skb, slave_id % slave_cnt);
		} else {
E
Eric Dumazet 已提交
3673
			bond_tx_drop(bond_dev, skb);
3674
		}
L
Linus Torvalds 已提交
3675
	}
3676

3677
	return NETDEV_TX_OK;
L
Linus Torvalds 已提交
3678 3679
}

3680
/* In active-backup mode, we know that bond->curr_active_slave is always valid if
L
Linus Torvalds 已提交
3681 3682 3683 3684
 * the bond has a usable interface.
 */
static int bond_xmit_activebackup(struct sk_buff *skb, struct net_device *bond_dev)
{
3685
	struct bonding *bond = netdev_priv(bond_dev);
3686
	struct slave *slave;
L
Linus Torvalds 已提交
3687

3688
	slave = rcu_dereference(bond->curr_active_slave);
3689
	if (slave)
3690 3691
		bond_dev_queue_xmit(bond, skb, slave->dev);
	else
E
Eric Dumazet 已提交
3692
		bond_tx_drop(bond_dev, skb);
3693

3694
	return NETDEV_TX_OK;
L
Linus Torvalds 已提交
3695 3696
}

3697 3698 3699
/* Use this to update slave_array when (a) it's not appropriate to update
 * slave_array right away (note that update_slave_array() may sleep)
 * and / or (b) RTNL is not held.
L
Linus Torvalds 已提交
3700
 */
3701
void bond_slave_arr_work_rearm(struct bonding *bond, unsigned long delay)
L
Linus Torvalds 已提交
3702
{
3703 3704
	queue_delayed_work(bond->wq, &bond->slave_arr_work, delay);
}
L
Linus Torvalds 已提交
3705

3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822
/* Slave array work handler. Holds only RTNL */
static void bond_slave_arr_handler(struct work_struct *work)
{
	struct bonding *bond = container_of(work, struct bonding,
					    slave_arr_work.work);
	int ret;

	if (!rtnl_trylock())
		goto err;

	ret = bond_update_slave_arr(bond, NULL);
	rtnl_unlock();
	if (ret) {
		pr_warn_ratelimited("Failed to update slave array from WT\n");
		goto err;
	}
	return;

err:
	bond_slave_arr_work_rearm(bond, 1);
}

/* Build the usable slaves array in control path for modes that use xmit-hash
 * to determine the slave interface -
 * (a) BOND_MODE_8023AD
 * (b) BOND_MODE_XOR
 * (c) BOND_MODE_TLB && tlb_dynamic_lb == 0
 *
 * The caller is expected to hold RTNL only and NO other lock!
 */
int bond_update_slave_arr(struct bonding *bond, struct slave *skipslave)
{
	struct slave *slave;
	struct list_head *iter;
	struct bond_up_slave *new_arr, *old_arr;
	int slaves_in_agg;
	int agg_id = 0;
	int ret = 0;

#ifdef CONFIG_LOCKDEP
	WARN_ON(lockdep_is_held(&bond->mode_lock));
#endif

	new_arr = kzalloc(offsetof(struct bond_up_slave, arr[bond->slave_cnt]),
			  GFP_KERNEL);
	if (!new_arr) {
		ret = -ENOMEM;
		pr_err("Failed to build slave-array.\n");
		goto out;
	}
	if (BOND_MODE(bond) == BOND_MODE_8023AD) {
		struct ad_info ad_info;

		if (bond_3ad_get_active_agg_info(bond, &ad_info)) {
			pr_debug("bond_3ad_get_active_agg_info failed\n");
			kfree_rcu(new_arr, rcu);
			/* No active aggragator means it's not safe to use
			 * the previous array.
			 */
			old_arr = rtnl_dereference(bond->slave_arr);
			if (old_arr) {
				RCU_INIT_POINTER(bond->slave_arr, NULL);
				kfree_rcu(old_arr, rcu);
			}
			goto out;
		}
		slaves_in_agg = ad_info.ports;
		agg_id = ad_info.aggregator_id;
	}
	bond_for_each_slave(bond, slave, iter) {
		if (BOND_MODE(bond) == BOND_MODE_8023AD) {
			struct aggregator *agg;

			agg = SLAVE_AD_INFO(slave)->port.aggregator;
			if (!agg || agg->aggregator_identifier != agg_id)
				continue;
		}
		if (!bond_slave_can_tx(slave))
			continue;
		if (skipslave == slave)
			continue;
		new_arr->arr[new_arr->count++] = slave;
	}

	old_arr = rtnl_dereference(bond->slave_arr);
	rcu_assign_pointer(bond->slave_arr, new_arr);
	if (old_arr)
		kfree_rcu(old_arr, rcu);
out:
	if (ret != 0 && skipslave) {
		int idx;

		/* Rare situation where caller has asked to skip a specific
		 * slave but allocation failed (most likely!). BTW this is
		 * only possible when the call is initiated from
		 * __bond_release_one(). In this situation; overwrite the
		 * skipslave entry in the array with the last entry from the
		 * array to avoid a situation where the xmit path may choose
		 * this to-be-skipped slave to send a packet out.
		 */
		old_arr = rtnl_dereference(bond->slave_arr);
		for (idx = 0; idx < old_arr->count; idx++) {
			if (skipslave == old_arr->arr[idx]) {
				old_arr->arr[idx] =
				    old_arr->arr[old_arr->count-1];
				old_arr->count--;
				break;
			}
		}
	}
	return ret;
}

/* Use this Xmit function for 3AD as well as XOR modes. The current
 * usable slave array is formed in the control path. The xmit function
 * just calculates hash and sends the packet out.
 */
3823
static int bond_3ad_xor_xmit(struct sk_buff *skb, struct net_device *dev)
3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835
{
	struct bonding *bond = netdev_priv(dev);
	struct slave *slave;
	struct bond_up_slave *slaves;
	unsigned int count;

	slaves = rcu_dereference(bond->slave_arr);
	count = slaves ? ACCESS_ONCE(slaves->count) : 0;
	if (likely(count)) {
		slave = slaves->arr[bond_xmit_hash(bond, skb) % count];
		bond_dev_queue_xmit(bond, skb, slave->dev);
	} else {
E
Eric Dumazet 已提交
3836
		bond_tx_drop(dev, skb);
3837
	}
3838

3839
	return NETDEV_TX_OK;
L
Linus Torvalds 已提交
3840 3841
}

3842
/* in broadcast mode, we send everything to all usable interfaces. */
L
Linus Torvalds 已提交
3843 3844
static int bond_xmit_broadcast(struct sk_buff *skb, struct net_device *bond_dev)
{
3845
	struct bonding *bond = netdev_priv(bond_dev);
3846
	struct slave *slave = NULL;
3847
	struct list_head *iter;
L
Linus Torvalds 已提交
3848

3849
	bond_for_each_slave_rcu(bond, slave, iter) {
3850 3851
		if (bond_is_last_slave(bond, slave))
			break;
3852
		if (bond_slave_is_up(slave) && slave->link == BOND_LINK_UP) {
3853
			struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
L
Linus Torvalds 已提交
3854

3855
			if (!skb2) {
3856 3857
				net_err_ratelimited("%s: Error: %s: skb_clone() failed\n",
						    bond_dev->name, __func__);
3858
				continue;
L
Linus Torvalds 已提交
3859
			}
3860
			bond_dev_queue_xmit(bond, skb2, slave->dev);
L
Linus Torvalds 已提交
3861 3862
		}
	}
3863
	if (slave && bond_slave_is_up(slave) && slave->link == BOND_LINK_UP)
3864 3865
		bond_dev_queue_xmit(bond, skb, slave->dev);
	else
E
Eric Dumazet 已提交
3866
		bond_tx_drop(bond_dev, skb);
S
Stephen Hemminger 已提交
3867

3868
	return NETDEV_TX_OK;
L
Linus Torvalds 已提交
3869 3870 3871 3872
}

/*------------------------- Device initialization ---------------------------*/

3873
/* Lookup the slave that corresponds to a qid */
3874 3875 3876 3877
static inline int bond_slave_override(struct bonding *bond,
				      struct sk_buff *skb)
{
	struct slave *slave = NULL;
3878
	struct list_head *iter;
3879

3880 3881
	if (!skb->queue_mapping)
		return 1;
3882 3883

	/* Find out if any slaves have the same mapping as this skb. */
3884 3885
	bond_for_each_slave_rcu(bond, slave, iter) {
		if (slave->queue_id == skb->queue_mapping) {
3886 3887
			if (bond_slave_is_up(slave) &&
			    slave->link == BOND_LINK_UP) {
3888 3889 3890 3891
				bond_dev_queue_xmit(bond, skb, slave->dev);
				return 0;
			}
			/* If the slave isn't UP, use default transmit policy. */
3892 3893 3894 3895
			break;
		}
	}

3896
	return 1;
3897 3898
}

3899

3900
static u16 bond_select_queue(struct net_device *dev, struct sk_buff *skb,
3901
			     void *accel_priv, select_queue_fallback_t fallback)
3902
{
3903
	/* This helper function exists to help dev_pick_tx get the correct
P
Phil Oester 已提交
3904
	 * destination queue.  Using a helper function skips a call to
3905 3906 3907
	 * skb_tx_hash and will put the skbs in the queue we expect on their
	 * way down to the bonding driver.
	 */
P
Phil Oester 已提交
3908 3909
	u16 txq = skb_rx_queue_recorded(skb) ? skb_get_rx_queue(skb) : 0;

3910
	/* Save the original txq to restore before passing to the driver */
3911
	qdisc_skb_cb(skb)->slave_dev_queue_mapping = skb->queue_mapping;
3912

P
Phil Oester 已提交
3913
	if (unlikely(txq >= dev->real_num_tx_queues)) {
3914
		do {
P
Phil Oester 已提交
3915
			txq -= dev->real_num_tx_queues;
3916
		} while (txq >= dev->real_num_tx_queues);
P
Phil Oester 已提交
3917 3918
	}
	return txq;
3919 3920
}

3921
static netdev_tx_t __bond_start_xmit(struct sk_buff *skb, struct net_device *dev)
3922
{
3923 3924
	struct bonding *bond = netdev_priv(dev);

3925 3926 3927
	if (bond_should_override_tx_queue(bond) &&
	    !bond_slave_override(bond, skb))
		return NETDEV_TX_OK;
3928

3929
	switch (BOND_MODE(bond)) {
3930 3931 3932 3933
	case BOND_MODE_ROUNDROBIN:
		return bond_xmit_roundrobin(skb, dev);
	case BOND_MODE_ACTIVEBACKUP:
		return bond_xmit_activebackup(skb, dev);
3934
	case BOND_MODE_8023AD:
3935
	case BOND_MODE_XOR:
3936
		return bond_3ad_xor_xmit(skb, dev);
3937 3938 3939 3940
	case BOND_MODE_BROADCAST:
		return bond_xmit_broadcast(skb, dev);
	case BOND_MODE_ALB:
		return bond_alb_xmit(skb, dev);
3941 3942
	case BOND_MODE_TLB:
		return bond_tlb_xmit(skb, dev);
3943 3944
	default:
		/* Should never happen, mode already checked */
3945
		netdev_err(dev, "Unknown bonding mode %d\n", BOND_MODE(bond));
3946
		WARN_ON_ONCE(1);
E
Eric Dumazet 已提交
3947
		bond_tx_drop(dev, skb);
3948 3949 3950 3951
		return NETDEV_TX_OK;
	}
}

3952 3953 3954 3955 3956
static netdev_tx_t bond_start_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct bonding *bond = netdev_priv(dev);
	netdev_tx_t ret = NETDEV_TX_OK;

3957
	/* If we risk deadlock from transmitting this in the
3958 3959
	 * netpoll path, tell netpoll to queue the frame for later tx
	 */
3960
	if (unlikely(is_netpoll_tx_blocked(dev)))
3961 3962
		return NETDEV_TX_BUSY;

3963
	rcu_read_lock();
3964
	if (bond_has_slaves(bond))
3965 3966
		ret = __bond_start_xmit(skb, dev);
	else
E
Eric Dumazet 已提交
3967
		bond_tx_drop(dev, skb);
3968
	rcu_read_unlock();
3969 3970 3971

	return ret;
}
3972

3973 3974 3975 3976 3977
static int bond_ethtool_get_settings(struct net_device *bond_dev,
				     struct ethtool_cmd *ecmd)
{
	struct bonding *bond = netdev_priv(bond_dev);
	unsigned long speed = 0;
3978
	struct list_head *iter;
3979
	struct slave *slave;
3980 3981 3982 3983

	ecmd->duplex = DUPLEX_UNKNOWN;
	ecmd->port = PORT_OTHER;

3984
	/* Since bond_slave_can_tx returns false for all inactive or down slaves, we
3985 3986 3987 3988
	 * do not need to check mode.  Though link speed might not represent
	 * the true receive or transmit bandwidth (not all modes are symmetric)
	 * this is an accurate maximum.
	 */
3989
	bond_for_each_slave(bond, slave, iter) {
3990
		if (bond_slave_can_tx(slave)) {
3991 3992 3993 3994 3995 3996 3997 3998
			if (slave->speed != SPEED_UNKNOWN)
				speed += slave->speed;
			if (ecmd->duplex == DUPLEX_UNKNOWN &&
			    slave->duplex != DUPLEX_UNKNOWN)
				ecmd->duplex = slave->duplex;
		}
	}
	ethtool_cmd_speed_set(ecmd, speed ? : SPEED_UNKNOWN);
3999

4000 4001 4002
	return 0;
}

4003
static void bond_ethtool_get_drvinfo(struct net_device *bond_dev,
4004
				     struct ethtool_drvinfo *drvinfo)
4005
{
4006 4007 4008 4009
	strlcpy(drvinfo->driver, DRV_NAME, sizeof(drvinfo->driver));
	strlcpy(drvinfo->version, DRV_VERSION, sizeof(drvinfo->version));
	snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version), "%d",
		 BOND_ABI_VERSION);
4010 4011
}

4012
static const struct ethtool_ops bond_ethtool_ops = {
4013
	.get_drvinfo		= bond_ethtool_get_drvinfo,
4014
	.get_settings		= bond_ethtool_get_settings,
4015
	.get_link		= ethtool_op_get_link,
4016 4017
};

4018
static const struct net_device_ops bond_netdev_ops = {
4019
	.ndo_init		= bond_init,
S
Stephen Hemminger 已提交
4020
	.ndo_uninit		= bond_uninit,
4021 4022
	.ndo_open		= bond_open,
	.ndo_stop		= bond_close,
4023
	.ndo_start_xmit		= bond_start_xmit,
4024
	.ndo_select_queue	= bond_select_queue,
4025
	.ndo_get_stats64	= bond_get_stats,
4026
	.ndo_do_ioctl		= bond_do_ioctl,
4027
	.ndo_change_rx_flags	= bond_change_rx_flags,
4028
	.ndo_set_rx_mode	= bond_set_rx_mode,
4029
	.ndo_change_mtu		= bond_change_mtu,
J
Jiri Pirko 已提交
4030
	.ndo_set_mac_address	= bond_set_mac_address,
4031
	.ndo_neigh_setup	= bond_neigh_setup,
J
Jiri Pirko 已提交
4032
	.ndo_vlan_rx_add_vid	= bond_vlan_rx_add_vid,
4033
	.ndo_vlan_rx_kill_vid	= bond_vlan_rx_kill_vid,
4034
#ifdef CONFIG_NET_POLL_CONTROLLER
4035
	.ndo_netpoll_setup	= bond_netpoll_setup,
4036 4037 4038
	.ndo_netpoll_cleanup	= bond_netpoll_cleanup,
	.ndo_poll_controller	= bond_poll_controller,
#endif
J
Jiri Pirko 已提交
4039 4040
	.ndo_add_slave		= bond_enslave,
	.ndo_del_slave		= bond_release,
4041
	.ndo_fix_features	= bond_fix_features,
4042
	.ndo_bridge_setlink	= switchdev_port_bridge_setlink,
4043
	.ndo_bridge_getlink	= switchdev_port_bridge_getlink,
4044
	.ndo_bridge_dellink	= switchdev_port_bridge_dellink,
4045
	.ndo_features_check	= passthru_features_check,
4046 4047
};

4048 4049 4050 4051
static const struct device_type bond_type = {
	.name = "bond",
};

4052 4053 4054 4055 4056 4057 4058 4059
static void bond_destructor(struct net_device *bond_dev)
{
	struct bonding *bond = netdev_priv(bond_dev);
	if (bond->wq)
		destroy_workqueue(bond->wq);
	free_netdev(bond_dev);
}

4060
void bond_setup(struct net_device *bond_dev)
L
Linus Torvalds 已提交
4061
{
4062
	struct bonding *bond = netdev_priv(bond_dev);
L
Linus Torvalds 已提交
4063

4064
	spin_lock_init(&bond->mode_lock);
4065
	bond->params = bonding_defaults;
L
Linus Torvalds 已提交
4066 4067 4068 4069 4070

	/* Initialize pointers */
	bond->dev = bond_dev;

	/* Initialize the device entry points */
4071
	ether_setup(bond_dev);
4072
	bond_dev->netdev_ops = &bond_netdev_ops;
4073
	bond_dev->ethtool_ops = &bond_ethtool_ops;
L
Linus Torvalds 已提交
4074

4075
	bond_dev->destructor = bond_destructor;
L
Linus Torvalds 已提交
4076

4077 4078
	SET_NETDEV_DEVTYPE(bond_dev, &bond_type);

L
Linus Torvalds 已提交
4079 4080 4081
	/* Initialize the device options */
	bond_dev->tx_queue_len = 0;
	bond_dev->flags |= IFF_MASTER|IFF_MULTICAST;
4082
	bond_dev->priv_flags |= IFF_BONDING | IFF_UNICAST_FLT;
4083
	bond_dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_TX_SKB_SHARING);
4084

4085
	/* don't acquire bond device's netif_tx_lock when transmitting */
L
Linus Torvalds 已提交
4086 4087 4088 4089 4090 4091 4092 4093 4094
	bond_dev->features |= NETIF_F_LLTX;

	/* By default, we declare the bond to be fully
	 * VLAN hardware accelerated capable. Special
	 * care is taken in the various xmit functions
	 * when there are slaves that are not hw accel
	 * capable
	 */

4095 4096 4097
	/* Don't allow bond devices to change network namespaces. */
	bond_dev->features |= NETIF_F_NETNS_LOCAL;

4098
	bond_dev->hw_features = BOND_VLAN_FEATURES |
4099 4100 4101
				NETIF_F_HW_VLAN_CTAG_TX |
				NETIF_F_HW_VLAN_CTAG_RX |
				NETIF_F_HW_VLAN_CTAG_FILTER;
4102

4103
	bond_dev->hw_features &= ~(NETIF_F_ALL_CSUM & ~NETIF_F_HW_CSUM);
E
Eric Dumazet 已提交
4104
	bond_dev->hw_features |= NETIF_F_GSO_ENCAP_ALL;
4105
	bond_dev->features |= bond_dev->hw_features;
L
Linus Torvalds 已提交
4106 4107
}

4108 4109 4110
/* Destroy a bonding device.
 * Must be under rtnl_lock when this function is called.
 */
4111
static void bond_uninit(struct net_device *bond_dev)
J
Jay Vosburgh 已提交
4112
{
4113
	struct bonding *bond = netdev_priv(bond_dev);
4114 4115
	struct list_head *iter;
	struct slave *slave;
4116
	struct bond_up_slave *arr;
J
Jay Vosburgh 已提交
4117

4118 4119
	bond_netpoll_cleanup(bond_dev);

4120
	/* Release the bonded slaves */
4121
	bond_for_each_slave(bond, slave, iter)
4122
		__bond_release_one(bond_dev, slave->dev, true);
4123
	netdev_info(bond_dev, "Released all slaves\n");
4124

4125 4126 4127 4128 4129 4130
	arr = rtnl_dereference(bond->slave_arr);
	if (arr) {
		RCU_INIT_POINTER(bond->slave_arr, NULL);
		kfree_rcu(arr, rcu);
	}

J
Jay Vosburgh 已提交
4131 4132
	list_del(&bond->bond_list);

4133
	bond_debug_unregister(bond);
J
Jay Vosburgh 已提交
4134 4135
}

L
Linus Torvalds 已提交
4136 4137 4138 4139
/*------------------------- Module initialization ---------------------------*/

static int bond_check_params(struct bond_params *params)
{
4140
	int arp_validate_value, fail_over_mac_value, primary_reselect_value, i;
4141 4142
	struct bond_opt_value newval;
	const struct bond_opt_value *valptr;
4143
	int arp_all_targets_value;
4144
	u16 ad_actor_sys_prio = 0;
4145
	u16 ad_user_port_key = 0;
4146

4147
	/* Convert string parameters. */
L
Linus Torvalds 已提交
4148
	if (mode) {
4149 4150 4151 4152
		bond_opt_initstr(&newval, mode);
		valptr = bond_opt_parse(bond_opt_get(BOND_OPT_MODE), &newval);
		if (!valptr) {
			pr_err("Error: Invalid bonding mode \"%s\"\n", mode);
L
Linus Torvalds 已提交
4153 4154
			return -EINVAL;
		}
4155
		bond_mode = valptr->value;
L
Linus Torvalds 已提交
4156 4157
	}

4158 4159
	if (xmit_hash_policy) {
		if ((bond_mode != BOND_MODE_XOR) &&
4160 4161
		    (bond_mode != BOND_MODE_8023AD) &&
		    (bond_mode != BOND_MODE_TLB)) {
J
Joe Perches 已提交
4162
			pr_info("xmit_hash_policy param is irrelevant in mode %s\n",
J
Joe Perches 已提交
4163
				bond_mode_name(bond_mode));
4164
		} else {
4165 4166 4167 4168
			bond_opt_initstr(&newval, xmit_hash_policy);
			valptr = bond_opt_parse(bond_opt_get(BOND_OPT_XMIT_HASH),
						&newval);
			if (!valptr) {
J
Joe Perches 已提交
4169
				pr_err("Error: Invalid xmit_hash_policy \"%s\"\n",
4170 4171 4172
				       xmit_hash_policy);
				return -EINVAL;
			}
4173
			xmit_hashtype = valptr->value;
4174 4175 4176
		}
	}

L
Linus Torvalds 已提交
4177 4178
	if (lacp_rate) {
		if (bond_mode != BOND_MODE_8023AD) {
J
Joe Perches 已提交
4179 4180
			pr_info("lacp_rate param is irrelevant in mode %s\n",
				bond_mode_name(bond_mode));
L
Linus Torvalds 已提交
4181
		} else {
4182 4183 4184 4185
			bond_opt_initstr(&newval, lacp_rate);
			valptr = bond_opt_parse(bond_opt_get(BOND_OPT_LACP_RATE),
						&newval);
			if (!valptr) {
J
Joe Perches 已提交
4186
				pr_err("Error: Invalid lacp rate \"%s\"\n",
4187
				       lacp_rate);
L
Linus Torvalds 已提交
4188 4189
				return -EINVAL;
			}
4190
			lacp_fast = valptr->value;
L
Linus Torvalds 已提交
4191 4192 4193
		}
	}

4194
	if (ad_select) {
4195
		bond_opt_initstr(&newval, ad_select);
4196 4197 4198 4199
		valptr = bond_opt_parse(bond_opt_get(BOND_OPT_AD_SELECT),
					&newval);
		if (!valptr) {
			pr_err("Error: Invalid ad_select \"%s\"\n", ad_select);
4200 4201
			return -EINVAL;
		}
4202 4203
		params->ad_select = valptr->value;
		if (bond_mode != BOND_MODE_8023AD)
4204
			pr_warn("ad_select param only affects 802.3ad mode\n");
4205 4206 4207 4208
	} else {
		params->ad_select = BOND_AD_STABLE;
	}

4209
	if (max_bonds < 0) {
4210 4211
		pr_warn("Warning: max_bonds (%d) not in range %d-%d, so it was reset to BOND_DEFAULT_MAX_BONDS (%d)\n",
			max_bonds, 0, INT_MAX, BOND_DEFAULT_MAX_BONDS);
L
Linus Torvalds 已提交
4212 4213 4214 4215
		max_bonds = BOND_DEFAULT_MAX_BONDS;
	}

	if (miimon < 0) {
4216 4217
		pr_warn("Warning: miimon module parameter (%d), not in range 0-%d, so it was reset to 0\n",
			miimon, INT_MAX);
4218
		miimon = 0;
L
Linus Torvalds 已提交
4219 4220 4221
	}

	if (updelay < 0) {
4222 4223
		pr_warn("Warning: updelay module parameter (%d), not in range 0-%d, so it was reset to 0\n",
			updelay, INT_MAX);
L
Linus Torvalds 已提交
4224 4225 4226 4227
		updelay = 0;
	}

	if (downdelay < 0) {
4228 4229
		pr_warn("Warning: downdelay module parameter (%d), not in range 0-%d, so it was reset to 0\n",
			downdelay, INT_MAX);
L
Linus Torvalds 已提交
4230 4231 4232 4233
		downdelay = 0;
	}

	if ((use_carrier != 0) && (use_carrier != 1)) {
4234 4235
		pr_warn("Warning: use_carrier module parameter (%d), not of valid value (0/1), so it was set to 1\n",
			use_carrier);
L
Linus Torvalds 已提交
4236 4237 4238
		use_carrier = 1;
	}

4239
	if (num_peer_notif < 0 || num_peer_notif > 255) {
4240 4241
		pr_warn("Warning: num_grat_arp/num_unsol_na (%d) not in range 0-255 so it was reset to 1\n",
			num_peer_notif);
4242 4243 4244
		num_peer_notif = 1;
	}

4245
	/* reset values for 802.3ad/TLB/ALB */
4246
	if (!bond_mode_uses_arp(bond_mode)) {
L
Linus Torvalds 已提交
4247
		if (!miimon) {
4248 4249
			pr_warn("Warning: miimon must be specified, otherwise bonding will not detect link failure, speed and duplex which are essential for 802.3ad operation\n");
			pr_warn("Forcing miimon to 100msec\n");
4250
			miimon = BOND_DEFAULT_MIIMON;
L
Linus Torvalds 已提交
4251 4252 4253
		}
	}

4254
	if (tx_queues < 1 || tx_queues > 255) {
4255 4256
		pr_warn("Warning: tx_queues (%d) should be between 1 and 255, resetting to %d\n",
			tx_queues, BOND_DEFAULT_TX_QUEUES);
4257 4258 4259
		tx_queues = BOND_DEFAULT_TX_QUEUES;
	}

4260
	if ((all_slaves_active != 0) && (all_slaves_active != 1)) {
4261 4262
		pr_warn("Warning: all_slaves_active module parameter (%d), not of valid value (0/1), so it was set to 0\n",
			all_slaves_active);
4263 4264 4265
		all_slaves_active = 0;
	}

4266
	if (resend_igmp < 0 || resend_igmp > 255) {
4267 4268
		pr_warn("Warning: resend_igmp (%d) should be between 0 and 255, resetting to %d\n",
			resend_igmp, BOND_DEFAULT_RESEND_IGMP);
4269 4270 4271
		resend_igmp = BOND_DEFAULT_RESEND_IGMP;
	}

4272 4273
	bond_opt_initval(&newval, packets_per_slave);
	if (!bond_opt_parse(bond_opt_get(BOND_OPT_PACKETS_PER_SLAVE), &newval)) {
4274 4275 4276 4277 4278
		pr_warn("Warning: packets_per_slave (%d) should be between 0 and %u resetting to 1\n",
			packets_per_slave, USHRT_MAX);
		packets_per_slave = 1;
	}

L
Linus Torvalds 已提交
4279
	if (bond_mode == BOND_MODE_ALB) {
J
Joe Perches 已提交
4280 4281
		pr_notice("In ALB mode you might experience client disconnections upon reconnection of a link if the bonding module updelay parameter (%d msec) is incompatible with the forwarding delay time of the switch\n",
			  updelay);
L
Linus Torvalds 已提交
4282 4283 4284 4285 4286 4287 4288
	}

	if (!miimon) {
		if (updelay || downdelay) {
			/* just warn the user the up/down delay will have
			 * no effect since miimon is zero...
			 */
4289 4290
			pr_warn("Warning: miimon module parameter not set and updelay (%d) or downdelay (%d) module parameter is set; updelay and downdelay have no effect unless miimon is set\n",
				updelay, downdelay);
L
Linus Torvalds 已提交
4291 4292 4293 4294
		}
	} else {
		/* don't allow arp monitoring */
		if (arp_interval) {
4295 4296
			pr_warn("Warning: miimon (%d) and arp_interval (%d) can't be used simultaneously, disabling ARP monitoring\n",
				miimon, arp_interval);
L
Linus Torvalds 已提交
4297 4298 4299 4300
			arp_interval = 0;
		}

		if ((updelay % miimon) != 0) {
4301 4302
			pr_warn("Warning: updelay (%d) is not a multiple of miimon (%d), updelay rounded to %d ms\n",
				updelay, miimon, (updelay / miimon) * miimon);
L
Linus Torvalds 已提交
4303 4304 4305 4306 4307
		}

		updelay /= miimon;

		if ((downdelay % miimon) != 0) {
4308 4309 4310
			pr_warn("Warning: downdelay (%d) is not a multiple of miimon (%d), downdelay rounded to %d ms\n",
				downdelay, miimon,
				(downdelay / miimon) * miimon);
L
Linus Torvalds 已提交
4311 4312 4313 4314 4315 4316
		}

		downdelay /= miimon;
	}

	if (arp_interval < 0) {
4317 4318
		pr_warn("Warning: arp_interval module parameter (%d), not in range 0-%d, so it was reset to 0\n",
			arp_interval, INT_MAX);
4319
		arp_interval = 0;
L
Linus Torvalds 已提交
4320 4321
	}

4322 4323
	for (arp_ip_count = 0, i = 0;
	     (arp_ip_count < BOND_MAX_ARP_TARGETS) && arp_ip_target[i]; i++) {
4324
		__be32 ip;
4325 4326

		/* not a complete check, but good enough to catch mistakes */
4327
		if (!in4_pton(arp_ip_target[i], -1, (u8 *)&ip, -1, NULL) ||
4328
		    !bond_is_ip_target_ok(ip)) {
4329 4330
			pr_warn("Warning: bad arp_ip_target module parameter (%s), ARP monitoring will not be performed\n",
				arp_ip_target[i]);
L
Linus Torvalds 已提交
4331 4332
			arp_interval = 0;
		} else {
4333 4334 4335
			if (bond_get_targets_ip(arp_target, ip) == -1)
				arp_target[arp_ip_count++] = ip;
			else
4336 4337
				pr_warn("Warning: duplicate address %pI4 in arp_ip_target, skipping\n",
					&ip);
L
Linus Torvalds 已提交
4338 4339 4340 4341 4342
		}
	}

	if (arp_interval && !arp_ip_count) {
		/* don't allow arping if no arp_ip_target given... */
4343 4344
		pr_warn("Warning: arp_interval module parameter (%d) specified without providing an arp_ip_target parameter, arp_interval was reset to 0\n",
			arp_interval);
L
Linus Torvalds 已提交
4345 4346 4347
		arp_interval = 0;
	}

4348 4349
	if (arp_validate) {
		if (!arp_interval) {
J
Joe Perches 已提交
4350
			pr_err("arp_validate requires arp_interval\n");
4351 4352 4353
			return -EINVAL;
		}

4354 4355 4356 4357
		bond_opt_initstr(&newval, arp_validate);
		valptr = bond_opt_parse(bond_opt_get(BOND_OPT_ARP_VALIDATE),
					&newval);
		if (!valptr) {
J
Joe Perches 已提交
4358
			pr_err("Error: invalid arp_validate \"%s\"\n",
4359
			       arp_validate);
4360 4361
			return -EINVAL;
		}
4362 4363
		arp_validate_value = valptr->value;
	} else {
4364
		arp_validate_value = 0;
4365
	}
4366

4367 4368
	arp_all_targets_value = 0;
	if (arp_all_targets) {
4369 4370 4371 4372
		bond_opt_initstr(&newval, arp_all_targets);
		valptr = bond_opt_parse(bond_opt_get(BOND_OPT_ARP_ALL_TARGETS),
					&newval);
		if (!valptr) {
4373 4374 4375
			pr_err("Error: invalid arp_all_targets_value \"%s\"\n",
			       arp_all_targets);
			arp_all_targets_value = 0;
4376 4377
		} else {
			arp_all_targets_value = valptr->value;
4378 4379 4380
		}
	}

L
Linus Torvalds 已提交
4381
	if (miimon) {
J
Joe Perches 已提交
4382
		pr_info("MII link monitoring set to %d ms\n", miimon);
L
Linus Torvalds 已提交
4383
	} else if (arp_interval) {
4384 4385
		valptr = bond_opt_get_val(BOND_OPT_ARP_VALIDATE,
					  arp_validate_value);
J
Joe Perches 已提交
4386
		pr_info("ARP monitoring set to %d ms, validate %s, with %d target(s):",
4387
			arp_interval, valptr->string, arp_ip_count);
L
Linus Torvalds 已提交
4388 4389

		for (i = 0; i < arp_ip_count; i++)
J
Joe Perches 已提交
4390
			pr_cont(" %s", arp_ip_target[i]);
L
Linus Torvalds 已提交
4391

J
Joe Perches 已提交
4392
		pr_cont("\n");
L
Linus Torvalds 已提交
4393

4394
	} else if (max_bonds) {
L
Linus Torvalds 已提交
4395 4396 4397
		/* miimon and arp_interval not set, we need one so things
		 * work as expected, see bonding.txt for details
		 */
J
Joe Perches 已提交
4398
		pr_debug("Warning: either miimon or arp_interval and arp_ip_target module parameters must be specified, otherwise bonding will not detect link failures! see bonding.txt for details\n");
L
Linus Torvalds 已提交
4399 4400
	}

4401
	if (primary && !bond_mode_uses_primary(bond_mode)) {
L
Linus Torvalds 已提交
4402 4403 4404
		/* currently, using a primary only makes sense
		 * in active backup, TLB or ALB modes
		 */
4405 4406
		pr_warn("Warning: %s primary device specified but has no effect in %s mode\n",
			primary, bond_mode_name(bond_mode));
L
Linus Torvalds 已提交
4407 4408 4409
		primary = NULL;
	}

4410
	if (primary && primary_reselect) {
4411 4412 4413 4414
		bond_opt_initstr(&newval, primary_reselect);
		valptr = bond_opt_parse(bond_opt_get(BOND_OPT_PRIMARY_RESELECT),
					&newval);
		if (!valptr) {
J
Joe Perches 已提交
4415
			pr_err("Error: Invalid primary_reselect \"%s\"\n",
4416
			       primary_reselect);
4417 4418
			return -EINVAL;
		}
4419
		primary_reselect_value = valptr->value;
4420 4421 4422 4423
	} else {
		primary_reselect_value = BOND_PRI_RESELECT_ALWAYS;
	}

4424
	if (fail_over_mac) {
4425 4426 4427 4428
		bond_opt_initstr(&newval, fail_over_mac);
		valptr = bond_opt_parse(bond_opt_get(BOND_OPT_FAIL_OVER_MAC),
					&newval);
		if (!valptr) {
J
Joe Perches 已提交
4429
			pr_err("Error: invalid fail_over_mac \"%s\"\n",
4430
			       fail_over_mac);
4431 4432
			return -EINVAL;
		}
4433
		fail_over_mac_value = valptr->value;
4434
		if (bond_mode != BOND_MODE_ACTIVEBACKUP)
4435
			pr_warn("Warning: fail_over_mac only affects active-backup mode\n");
4436 4437 4438
	} else {
		fail_over_mac_value = BOND_FOM_NONE;
	}
4439

4440 4441 4442 4443 4444 4445 4446 4447 4448 4449
	bond_opt_initstr(&newval, "default");
	valptr = bond_opt_parse(
			bond_opt_get(BOND_OPT_AD_ACTOR_SYS_PRIO),
				     &newval);
	if (!valptr) {
		pr_err("Error: No ad_actor_sys_prio default value");
		return -EINVAL;
	}
	ad_actor_sys_prio = valptr->value;

4450 4451 4452 4453 4454 4455 4456 4457
	valptr = bond_opt_parse(bond_opt_get(BOND_OPT_AD_USER_PORT_KEY),
				&newval);
	if (!valptr) {
		pr_err("Error: No ad_user_port_key default value");
		return -EINVAL;
	}
	ad_user_port_key = valptr->value;

4458
	if (lp_interval == 0) {
4459 4460
		pr_warn("Warning: ip_interval must be between 1 and %d, so it was reset to %d\n",
			INT_MAX, BOND_ALB_DEFAULT_LP_INTERVAL);
4461 4462 4463
		lp_interval = BOND_ALB_DEFAULT_LP_INTERVAL;
	}

L
Linus Torvalds 已提交
4464 4465
	/* fill params struct with the proper values */
	params->mode = bond_mode;
4466
	params->xmit_policy = xmit_hashtype;
L
Linus Torvalds 已提交
4467
	params->miimon = miimon;
4468
	params->num_peer_notif = num_peer_notif;
L
Linus Torvalds 已提交
4469
	params->arp_interval = arp_interval;
4470
	params->arp_validate = arp_validate_value;
4471
	params->arp_all_targets = arp_all_targets_value;
L
Linus Torvalds 已提交
4472 4473 4474 4475 4476
	params->updelay = updelay;
	params->downdelay = downdelay;
	params->use_carrier = use_carrier;
	params->lacp_fast = lacp_fast;
	params->primary[0] = 0;
4477
	params->primary_reselect = primary_reselect_value;
4478
	params->fail_over_mac = fail_over_mac_value;
4479
	params->tx_queues = tx_queues;
4480
	params->all_slaves_active = all_slaves_active;
4481
	params->resend_igmp = resend_igmp;
4482
	params->min_links = min_links;
4483
	params->lp_interval = lp_interval;
4484
	params->packets_per_slave = packets_per_slave;
4485
	params->tlb_dynamic_lb = 1; /* Default value */
4486
	params->ad_actor_sys_prio = ad_actor_sys_prio;
4487
	eth_zero_addr(params->ad_actor_system);
4488
	params->ad_user_port_key = ad_user_port_key;
4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499
	if (packets_per_slave > 0) {
		params->reciprocal_packets_per_slave =
			reciprocal_value(packets_per_slave);
	} else {
		/* reciprocal_packets_per_slave is unused if
		 * packets_per_slave is 0 or 1, just initialize it
		 */
		params->reciprocal_packets_per_slave =
			(struct reciprocal_value) { 0 };
	}

L
Linus Torvalds 已提交
4500 4501 4502 4503 4504 4505 4506 4507 4508 4509
	if (primary) {
		strncpy(params->primary, primary, IFNAMSIZ);
		params->primary[IFNAMSIZ - 1] = 0;
	}

	memcpy(params->arp_targets, arp_target, sizeof(arp_target));

	return 0;
}

4510
static struct lock_class_key bonding_netdev_xmit_lock_key;
4511
static struct lock_class_key bonding_netdev_addr_lock_key;
4512
static struct lock_class_key bonding_tx_busylock_key;
4513

4514 4515 4516
static void bond_set_lockdep_class_one(struct net_device *dev,
				       struct netdev_queue *txq,
				       void *_unused)
4517 4518 4519 4520 4521 4522 4523
{
	lockdep_set_class(&txq->_xmit_lock,
			  &bonding_netdev_xmit_lock_key);
}

static void bond_set_lockdep_class(struct net_device *dev)
{
4524 4525
	lockdep_set_class(&dev->addr_list_lock,
			  &bonding_netdev_addr_lock_key);
4526
	netdev_for_each_tx_queue(dev, bond_set_lockdep_class_one, NULL);
4527
	dev->qdisc_tx_busylock = &bonding_tx_busylock_key;
4528 4529
}

4530
/* Called from registration process */
4531 4532 4533
static int bond_init(struct net_device *bond_dev)
{
	struct bonding *bond = netdev_priv(bond_dev);
4534
	struct bond_net *bn = net_generic(dev_net(bond_dev), bond_net_id);
4535

4536
	netdev_dbg(bond_dev, "Begin bond_init\n");
4537 4538 4539 4540 4541 4542 4543

	bond->wq = create_singlethread_workqueue(bond_dev->name);
	if (!bond->wq)
		return -ENOMEM;

	bond_set_lockdep_class(bond_dev);

4544
	list_add_tail(&bond->bond_list, &bn->dev_list);
4545

4546
	bond_prepare_sysfs_group(bond);
4547

4548 4549
	bond_debug_register(bond);

4550 4551
	/* Ensure valid dev_addr */
	if (is_zero_ether_addr(bond_dev->dev_addr) &&
4552
	    bond_dev->addr_assign_type == NET_ADDR_PERM)
4553 4554
		eth_hw_addr_random(bond_dev);

4555 4556 4557
	return 0;
}

4558
unsigned int bond_get_num_tx_queues(void)
4559
{
4560
	return tx_queues;
4561 4562
}

4563
/* Create a new bond based on the specified name and bonding parameters.
4564
 * If name is NULL, obtain a suitable "bond%d" name for us.
4565 4566 4567
 * Caller must NOT hold rtnl_lock; we need to release it here before we
 * set up our sysfs entries.
 */
4568
int bond_create(struct net *net, const char *name)
4569 4570
{
	struct net_device *bond_dev;
4571 4572
	struct bonding *bond;
	struct alb_bond_info *bond_info;
4573 4574 4575
	int res;

	rtnl_lock();
4576

4577
	bond_dev = alloc_netdev_mq(sizeof(struct bonding),
4578
				   name ? name : "bond%d", NET_NAME_UNKNOWN,
4579
				   bond_setup, tx_queues);
4580
	if (!bond_dev) {
J
Joe Perches 已提交
4581
		pr_err("%s: eek! can't alloc netdev!\n", name);
4582 4583
		rtnl_unlock();
		return -ENOMEM;
4584 4585
	}

4586 4587 4588 4589 4590 4591 4592 4593
	/*
	 * Initialize rx_hashtbl_used_head to RLB_NULL_INDEX.
	 * It is set to 0 by default which is wrong.
	 */
	bond = netdev_priv(bond_dev);
	bond_info = &(BOND_ALB_INFO(bond));
	bond_info->rx_hashtbl_used_head = RLB_NULL_INDEX;

4594
	dev_net_set(bond_dev, net);
4595 4596
	bond_dev->rtnl_link_ops = &bond_link_ops;

4597
	res = register_netdevice(bond_dev);
4598

4599 4600
	netif_carrier_off(bond_dev);

4601
	rtnl_unlock();
4602 4603
	if (res < 0)
		bond_destructor(bond_dev);
E
Eric W. Biederman 已提交
4604
	return res;
4605 4606
}

4607
static int __net_init bond_net_init(struct net *net)
4608
{
4609
	struct bond_net *bn = net_generic(net, bond_net_id);
4610 4611 4612 4613 4614

	bn->net = net;
	INIT_LIST_HEAD(&bn->dev_list);

	bond_create_proc_dir(bn);
4615
	bond_create_sysfs(bn);
4616

4617
	return 0;
4618 4619
}

4620
static void __net_exit bond_net_exit(struct net *net)
4621
{
4622
	struct bond_net *bn = net_generic(net, bond_net_id);
4623 4624
	struct bonding *bond, *tmp_bond;
	LIST_HEAD(list);
4625

4626
	bond_destroy_sysfs(bn);
4627 4628 4629 4630 4631 4632 4633

	/* Kill off any bonds created after unregistering bond rtnl ops */
	rtnl_lock();
	list_for_each_entry_safe(bond, tmp_bond, &bn->dev_list, bond_list)
		unregister_netdevice_queue(bond->dev, &list);
	unregister_netdevice_many(&list);
	rtnl_unlock();
4634 4635

	bond_destroy_proc_dir(bn);
4636 4637 4638 4639 4640
}

static struct pernet_operations bond_net_ops = {
	.init = bond_net_init,
	.exit = bond_net_exit,
4641 4642
	.id   = &bond_net_id,
	.size = sizeof(struct bond_net),
4643 4644
};

L
Linus Torvalds 已提交
4645 4646 4647 4648 4649
static int __init bonding_init(void)
{
	int i;
	int res;

4650
	pr_info("%s", bond_version);
L
Linus Torvalds 已提交
4651

4652
	res = bond_check_params(&bonding_defaults);
S
Stephen Hemminger 已提交
4653
	if (res)
4654
		goto out;
L
Linus Torvalds 已提交
4655

4656
	res = register_pernet_subsys(&bond_net_ops);
4657 4658
	if (res)
		goto out;
4659

4660
	res = bond_netlink_init();
4661
	if (res)
4662
		goto err_link;
4663

4664 4665
	bond_create_debugfs();

L
Linus Torvalds 已提交
4666
	for (i = 0; i < max_bonds; i++) {
4667
		res = bond_create(&init_net, NULL);
4668 4669
		if (res)
			goto err;
L
Linus Torvalds 已提交
4670 4671 4672
	}

	register_netdevice_notifier(&bond_netdev_notifier);
4673
out:
L
Linus Torvalds 已提交
4674
	return res;
4675
err:
4676
	bond_destroy_debugfs();
4677
	bond_netlink_fini();
4678
err_link:
4679
	unregister_pernet_subsys(&bond_net_ops);
4680
	goto out;
4681

L
Linus Torvalds 已提交
4682 4683 4684 4685 4686 4687
}

static void __exit bonding_exit(void)
{
	unregister_netdevice_notifier(&bond_netdev_notifier);

4688
	bond_destroy_debugfs();
4689

4690
	bond_netlink_fini();
4691
	unregister_pernet_subsys(&bond_net_ops);
4692 4693

#ifdef CONFIG_NET_POLL_CONTROLLER
4694
	/* Make sure we don't have an imbalance on our netpoll blocking */
4695
	WARN_ON(atomic_read(&netpoll_block_tx));
4696
#endif
L
Linus Torvalds 已提交
4697 4698 4699 4700 4701 4702 4703 4704
}

module_init(bonding_init);
module_exit(bonding_exit);
MODULE_LICENSE("GPL");
MODULE_VERSION(DRV_VERSION);
MODULE_DESCRIPTION(DRV_DESCRIPTION ", v" DRV_VERSION);
MODULE_AUTHOR("Thomas Davis, tadavis@lbl.gov and many others");