dev.c 188.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 * 	NET3	Protocol independent device support routines.
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 *	Derived from the non IP parts of dev.c 1.0.19
10
 * 		Authors:	Ross Biro
L
Linus Torvalds 已提交
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 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
 *				Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *				Mark Evans, <evansmp@uhura.aston.ac.uk>
 *
 *	Additional Authors:
 *		Florian la Roche <rzsfl@rz.uni-sb.de>
 *		Alan Cox <gw4pts@gw4pts.ampr.org>
 *		David Hinds <dahinds@users.sourceforge.net>
 *		Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
 *		Adam Sulmicki <adam@cfar.umd.edu>
 *              Pekka Riikonen <priikone@poesidon.pspt.fi>
 *
 *	Changes:
 *              D.J. Barrow     :       Fixed bug where dev->refcnt gets set
 *              			to 2 if register_netdev gets called
 *              			before net_dev_init & also removed a
 *              			few lines of code in the process.
 *		Alan Cox	:	device private ioctl copies fields back.
 *		Alan Cox	:	Transmit queue code does relevant
 *					stunts to keep the queue safe.
 *		Alan Cox	:	Fixed double lock.
 *		Alan Cox	:	Fixed promisc NULL pointer trap
 *		????????	:	Support the full private ioctl range
 *		Alan Cox	:	Moved ioctl permission check into
 *					drivers
 *		Tim Kordas	:	SIOCADDMULTI/SIOCDELMULTI
 *		Alan Cox	:	100 backlog just doesn't cut it when
 *					you start doing multicast video 8)
 *		Alan Cox	:	Rewrote net_bh and list manager.
 *		Alan Cox	: 	Fix ETH_P_ALL echoback lengths.
 *		Alan Cox	:	Took out transmit every packet pass
 *					Saved a few bytes in the ioctl handler
 *		Alan Cox	:	Network driver sets packet type before
 *					calling netif_rx. Saves a function
 *					call a packet.
 *		Alan Cox	:	Hashed net_bh()
 *		Richard Kooijman:	Timestamp fixes.
 *		Alan Cox	:	Wrong field in SIOCGIFDSTADDR
 *		Alan Cox	:	Device lock protection.
 *		Alan Cox	: 	Fixed nasty side effect of device close
 *					changes.
 *		Rudi Cilibrasi	:	Pass the right thing to
 *					set_mac_address()
 *		Dave Miller	:	32bit quantity for the device lock to
 *					make it work out on a Sparc.
 *		Bjorn Ekwall	:	Added KERNELD hack.
 *		Alan Cox	:	Cleaned up the backlog initialise.
 *		Craig Metz	:	SIOCGIFCONF fix if space for under
 *					1 device.
 *	    Thomas Bogendoerfer :	Return ENODEV for dev_open, if there
 *					is no device open function.
 *		Andi Kleen	:	Fix error reporting for SIOCGIFCONF
 *	    Michael Chastain	:	Fix signed/unsigned for SIOCGIFCONF
 *		Cyrus Durgin	:	Cleaned for KMOD
 *		Adam Sulmicki   :	Bug Fix : Network Device Unload
 *					A network device unload needs to purge
 *					the backlog queue.
 *	Paul Rusty Russell	:	SIOCSIFNAME
 *              Pekka Riikonen  :	Netdev boot-time settings code
 *              Andrew Morton   :       Make unregister_netdevice wait
 *              			indefinitely on dev->refcnt
 * 		J Hadi Salim	:	- Backlog queue sampling
 *				        - netif_rx() feedback
 */

#include <asm/uaccess.h>
#include <linux/bitops.h>
77
#include <linux/capability.h>
L
Linus Torvalds 已提交
78 79 80
#include <linux/cpu.h>
#include <linux/types.h>
#include <linux/kernel.h>
81
#include <linux/hash.h>
82
#include <linux/slab.h>
L
Linus Torvalds 已提交
83
#include <linux/sched.h>
A
Arjan van de Ven 已提交
84
#include <linux/mutex.h>
L
Linus Torvalds 已提交
85 86 87 88 89 90 91 92 93
#include <linux/string.h>
#include <linux/mm.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/errno.h>
#include <linux/interrupt.h>
#include <linux/if_ether.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
94
#include <linux/ethtool.h>
L
Linus Torvalds 已提交
95 96
#include <linux/notifier.h>
#include <linux/skbuff.h>
97
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
98 99 100 101 102 103
#include <net/sock.h>
#include <linux/rtnetlink.h>
#include <linux/stat.h>
#include <net/dst.h>
#include <net/pkt_sched.h>
#include <net/checksum.h>
104
#include <net/xfrm.h>
L
Linus Torvalds 已提交
105 106 107 108 109 110 111 112
#include <linux/highmem.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/netpoll.h>
#include <linux/rcupdate.h>
#include <linux/delay.h>
#include <net/iw_handler.h>
#include <asm/current.h>
S
Steve Grubb 已提交
113
#include <linux/audit.h>
114
#include <linux/dmaengine.h>
115
#include <linux/err.h>
116
#include <linux/ctype.h>
117
#include <linux/if_arp.h>
118
#include <linux/if_vlan.h>
119
#include <linux/ip.h>
120
#include <net/ip.h>
121
#include <net/mpls.h>
122 123
#include <linux/ipv6.h>
#include <linux/in.h>
D
David S. Miller 已提交
124 125
#include <linux/jhash.h>
#include <linux/random.h>
126
#include <trace/events/napi.h>
127
#include <trace/events/net.h>
128
#include <trace/events/skb.h>
129
#include <linux/pci.h>
130
#include <linux/inetdevice.h>
131
#include <linux/cpu_rmap.h>
132
#include <linux/static_key.h>
E
Eliezer Tamir 已提交
133
#include <linux/hashtable.h>
134
#include <linux/vmalloc.h>
135
#include <linux/if_macvlan.h>
136
#include <linux/errqueue.h>
137
#include <linux/hrtimer.h>
L
Linus Torvalds 已提交
138

139 140
#include "net-sysfs.h"

141 142 143
/* Instead of increasing this, you should create a hash table. */
#define MAX_GRO_SKBS 8

H
Herbert Xu 已提交
144 145 146
/* This should be increased if a protocol with a bigger head is added. */
#define GRO_MAX_HEAD (MAX_HEADER + 128)

L
Linus Torvalds 已提交
147
static DEFINE_SPINLOCK(ptype_lock);
148
static DEFINE_SPINLOCK(offload_lock);
149 150
struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
struct list_head ptype_all __read_mostly;	/* Taps */
151
static struct list_head offload_base __read_mostly;
L
Linus Torvalds 已提交
152

153
static int netif_rx_internal(struct sk_buff *skb);
154 155 156
static int call_netdevice_notifiers_info(unsigned long val,
					 struct net_device *dev,
					 struct netdev_notifier_info *info);
157

L
Linus Torvalds 已提交
158
/*
159
 * The @dev_base_head list is protected by @dev_base_lock and the rtnl
L
Linus Torvalds 已提交
160 161
 * semaphore.
 *
162
 * Pure readers hold dev_base_lock for reading, or rcu_read_lock()
L
Linus Torvalds 已提交
163 164
 *
 * Writers must hold the rtnl semaphore while they loop through the
165
 * dev_base_head list, and hold dev_base_lock for writing when they do the
L
Linus Torvalds 已提交
166 167 168 169 170 171 172 173 174 175 176 177 178 179
 * actual updates.  This allows pure readers to access the list even
 * while a writer is preparing to update it.
 *
 * To put it another way, dev_base_lock is held for writing only to
 * protect against pure readers; the rtnl semaphore provides the
 * protection against other writers.
 *
 * See, for example usages, register_netdevice() and
 * unregister_netdevice(), which must be called with the rtnl
 * semaphore held.
 */
DEFINE_RWLOCK(dev_base_lock);
EXPORT_SYMBOL(dev_base_lock);

E
Eliezer Tamir 已提交
180 181 182 183 184 185
/* protects napi_hash addition/deletion and napi_gen_id */
static DEFINE_SPINLOCK(napi_hash_lock);

static unsigned int napi_gen_id;
static DEFINE_HASHTABLE(napi_hash, 8);

186
static seqcount_t devnet_rename_seq;
187

188 189 190 191 192
static inline void dev_base_seq_inc(struct net *net)
{
	while (++net->dev_base_seq == 0);
}

193
static inline struct hlist_head *dev_name_hash(struct net *net, const char *name)
L
Linus Torvalds 已提交
194
{
195 196
	unsigned int hash = full_name_hash(name, strnlen(name, IFNAMSIZ));

197
	return &net->dev_name_head[hash_32(hash, NETDEV_HASHBITS)];
L
Linus Torvalds 已提交
198 199
}

200
static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex)
L
Linus Torvalds 已提交
201
{
202
	return &net->dev_index_head[ifindex & (NETDEV_HASHENTRIES - 1)];
L
Linus Torvalds 已提交
203 204
}

E
Eric Dumazet 已提交
205
static inline void rps_lock(struct softnet_data *sd)
206 207
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
208
	spin_lock(&sd->input_pkt_queue.lock);
209 210 211
#endif
}

E
Eric Dumazet 已提交
212
static inline void rps_unlock(struct softnet_data *sd)
213 214
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
215
	spin_unlock(&sd->input_pkt_queue.lock);
216 217 218
#endif
}

219
/* Device list insertion */
220
static void list_netdevice(struct net_device *dev)
221
{
222
	struct net *net = dev_net(dev);
223 224 225 226

	ASSERT_RTNL();

	write_lock_bh(&dev_base_lock);
227
	list_add_tail_rcu(&dev->dev_list, &net->dev_base_head);
228
	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net, dev->name));
229 230
	hlist_add_head_rcu(&dev->index_hlist,
			   dev_index_hash(net, dev->ifindex));
231
	write_unlock_bh(&dev_base_lock);
232 233

	dev_base_seq_inc(net);
234 235
}

236 237 238
/* Device list removal
 * caller must respect a RCU grace period before freeing/reusing dev
 */
239 240 241 242 243 244
static void unlist_netdevice(struct net_device *dev)
{
	ASSERT_RTNL();

	/* Unlink dev from the device chain */
	write_lock_bh(&dev_base_lock);
245
	list_del_rcu(&dev->dev_list);
246
	hlist_del_rcu(&dev->name_hlist);
247
	hlist_del_rcu(&dev->index_hlist);
248
	write_unlock_bh(&dev_base_lock);
249 250

	dev_base_seq_inc(dev_net(dev));
251 252
}

L
Linus Torvalds 已提交
253 254 255 256
/*
 *	Our notifier list
 */

257
static RAW_NOTIFIER_HEAD(netdev_chain);
L
Linus Torvalds 已提交
258 259 260 261 262

/*
 *	Device drivers call our routines to queue packets here. We empty the
 *	queue in the local softnet handler.
 */
263

264
DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
E
Eric Dumazet 已提交
265
EXPORT_PER_CPU_SYMBOL(softnet_data);
L
Linus Torvalds 已提交
266

267
#ifdef CONFIG_LOCKDEP
268
/*
269
 * register_netdevice() inits txq->_xmit_lock and sets lockdep class
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
 * according to dev->type
 */
static const unsigned short netdev_lock_type[] =
	{ARPHRD_NETROM, ARPHRD_ETHER, ARPHRD_EETHER, ARPHRD_AX25,
	 ARPHRD_PRONET, ARPHRD_CHAOS, ARPHRD_IEEE802, ARPHRD_ARCNET,
	 ARPHRD_APPLETLK, ARPHRD_DLCI, ARPHRD_ATM, ARPHRD_METRICOM,
	 ARPHRD_IEEE1394, ARPHRD_EUI64, ARPHRD_INFINIBAND, ARPHRD_SLIP,
	 ARPHRD_CSLIP, ARPHRD_SLIP6, ARPHRD_CSLIP6, ARPHRD_RSRVD,
	 ARPHRD_ADAPT, ARPHRD_ROSE, ARPHRD_X25, ARPHRD_HWX25,
	 ARPHRD_PPP, ARPHRD_CISCO, ARPHRD_LAPB, ARPHRD_DDCMP,
	 ARPHRD_RAWHDLC, ARPHRD_TUNNEL, ARPHRD_TUNNEL6, ARPHRD_FRAD,
	 ARPHRD_SKIP, ARPHRD_LOOPBACK, ARPHRD_LOCALTLK, ARPHRD_FDDI,
	 ARPHRD_BIF, ARPHRD_SIT, ARPHRD_IPDDP, ARPHRD_IPGRE,
	 ARPHRD_PIMREG, ARPHRD_HIPPI, ARPHRD_ASH, ARPHRD_ECONET,
	 ARPHRD_IRDA, ARPHRD_FCPP, ARPHRD_FCAL, ARPHRD_FCPL,
285 286 287
	 ARPHRD_FCFABRIC, ARPHRD_IEEE80211, ARPHRD_IEEE80211_PRISM,
	 ARPHRD_IEEE80211_RADIOTAP, ARPHRD_PHONET, ARPHRD_PHONET_PIPE,
	 ARPHRD_IEEE802154, ARPHRD_VOID, ARPHRD_NONE};
288

289
static const char *const netdev_lock_name[] =
290 291 292 293 294 295 296 297 298 299 300 301
	{"_xmit_NETROM", "_xmit_ETHER", "_xmit_EETHER", "_xmit_AX25",
	 "_xmit_PRONET", "_xmit_CHAOS", "_xmit_IEEE802", "_xmit_ARCNET",
	 "_xmit_APPLETLK", "_xmit_DLCI", "_xmit_ATM", "_xmit_METRICOM",
	 "_xmit_IEEE1394", "_xmit_EUI64", "_xmit_INFINIBAND", "_xmit_SLIP",
	 "_xmit_CSLIP", "_xmit_SLIP6", "_xmit_CSLIP6", "_xmit_RSRVD",
	 "_xmit_ADAPT", "_xmit_ROSE", "_xmit_X25", "_xmit_HWX25",
	 "_xmit_PPP", "_xmit_CISCO", "_xmit_LAPB", "_xmit_DDCMP",
	 "_xmit_RAWHDLC", "_xmit_TUNNEL", "_xmit_TUNNEL6", "_xmit_FRAD",
	 "_xmit_SKIP", "_xmit_LOOPBACK", "_xmit_LOCALTLK", "_xmit_FDDI",
	 "_xmit_BIF", "_xmit_SIT", "_xmit_IPDDP", "_xmit_IPGRE",
	 "_xmit_PIMREG", "_xmit_HIPPI", "_xmit_ASH", "_xmit_ECONET",
	 "_xmit_IRDA", "_xmit_FCPP", "_xmit_FCAL", "_xmit_FCPL",
302 303 304
	 "_xmit_FCFABRIC", "_xmit_IEEE80211", "_xmit_IEEE80211_PRISM",
	 "_xmit_IEEE80211_RADIOTAP", "_xmit_PHONET", "_xmit_PHONET_PIPE",
	 "_xmit_IEEE802154", "_xmit_VOID", "_xmit_NONE"};
305 306

static struct lock_class_key netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)];
307
static struct lock_class_key netdev_addr_lock_key[ARRAY_SIZE(netdev_lock_type)];
308 309 310 311 312 313 314 315 316 317 318 319

static inline unsigned short netdev_lock_pos(unsigned short dev_type)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(netdev_lock_type); i++)
		if (netdev_lock_type[i] == dev_type)
			return i;
	/* the last key is used by default */
	return ARRAY_SIZE(netdev_lock_type) - 1;
}

320 321
static inline void netdev_set_xmit_lockdep_class(spinlock_t *lock,
						 unsigned short dev_type)
322 323 324 325 326 327 328
{
	int i;

	i = netdev_lock_pos(dev_type);
	lockdep_set_class_and_name(lock, &netdev_xmit_lock_key[i],
				   netdev_lock_name[i]);
}
329 330 331 332 333 334 335 336 337 338

static inline void netdev_set_addr_lockdep_class(struct net_device *dev)
{
	int i;

	i = netdev_lock_pos(dev->type);
	lockdep_set_class_and_name(&dev->addr_list_lock,
				   &netdev_addr_lock_key[i],
				   netdev_lock_name[i]);
}
339
#else
340 341 342 343 344
static inline void netdev_set_xmit_lockdep_class(spinlock_t *lock,
						 unsigned short dev_type)
{
}
static inline void netdev_set_addr_lockdep_class(struct net_device *dev)
345 346 347
{
}
#endif
L
Linus Torvalds 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370

/*******************************************************************************

		Protocol management and registration routines

*******************************************************************************/

/*
 *	Add a protocol ID to the list. Now that the input handler is
 *	smarter we can dispense with all the messy stuff that used to be
 *	here.
 *
 *	BEWARE!!! Protocol handlers, mangling input packets,
 *	MUST BE last in hash buckets and checking protocol handlers
 *	MUST start from promiscuous ptype_all chain in net_bh.
 *	It is true now, do not change it.
 *	Explanation follows: if protocol handler, mangling packet, will
 *	be the first on list, it is not able to sense, that packet
 *	is cloned and should be copied-on-write, so that it will
 *	change it and subsequent readers will get broken packet.
 *							--ANK (980803)
 */

371 372 373
static inline struct list_head *ptype_head(const struct packet_type *pt)
{
	if (pt->type == htons(ETH_P_ALL))
374
		return pt->dev ? &pt->dev->ptype_all : &ptype_all;
375
	else
376 377
		return pt->dev ? &pt->dev->ptype_specific :
				 &ptype_base[ntohs(pt->type) & PTYPE_HASH_MASK];
378 379
}

L
Linus Torvalds 已提交
380 381 382 383 384 385 386 387
/**
 *	dev_add_pack - add packet handler
 *	@pt: packet type declaration
 *
 *	Add a protocol handler to the networking stack. The passed &packet_type
 *	is linked into kernel lists and may not be freed until it has been
 *	removed from the kernel lists.
 *
388
 *	This call does not sleep therefore it can not
L
Linus Torvalds 已提交
389 390 391 392 393 394
 *	guarantee all CPU's that are in middle of receiving packets
 *	will see the new packet type (until the next received packet).
 */

void dev_add_pack(struct packet_type *pt)
{
395
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
396

397 398 399
	spin_lock(&ptype_lock);
	list_add_rcu(&pt->list, head);
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
400
}
E
Eric Dumazet 已提交
401
EXPORT_SYMBOL(dev_add_pack);
L
Linus Torvalds 已提交
402 403 404 405 406 407 408 409

/**
 *	__dev_remove_pack	 - remove packet handler
 *	@pt: packet type declaration
 *
 *	Remove a protocol handler that was previously added to the kernel
 *	protocol handlers by dev_add_pack(). The passed &packet_type is removed
 *	from the kernel lists and can be freed or reused once this function
410
 *	returns.
L
Linus Torvalds 已提交
411 412 413 414 415 416 417
 *
 *      The packet type might still be in use by receivers
 *	and must not be freed until after all the CPU's have gone
 *	through a quiescent state.
 */
void __dev_remove_pack(struct packet_type *pt)
{
418
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
419 420
	struct packet_type *pt1;

421
	spin_lock(&ptype_lock);
L
Linus Torvalds 已提交
422 423 424 425 426 427 428 429

	list_for_each_entry(pt1, head, list) {
		if (pt == pt1) {
			list_del_rcu(&pt->list);
			goto out;
		}
	}

430
	pr_warn("dev_remove_pack: %p not found\n", pt);
L
Linus Torvalds 已提交
431
out:
432
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
433
}
E
Eric Dumazet 已提交
434 435
EXPORT_SYMBOL(__dev_remove_pack);

L
Linus Torvalds 已提交
436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
/**
 *	dev_remove_pack	 - remove packet handler
 *	@pt: packet type declaration
 *
 *	Remove a protocol handler that was previously added to the kernel
 *	protocol handlers by dev_add_pack(). The passed &packet_type is removed
 *	from the kernel lists and can be freed or reused once this function
 *	returns.
 *
 *	This call sleeps to guarantee that no CPU is looking at the packet
 *	type after return.
 */
void dev_remove_pack(struct packet_type *pt)
{
	__dev_remove_pack(pt);
451

L
Linus Torvalds 已提交
452 453
	synchronize_net();
}
E
Eric Dumazet 已提交
454
EXPORT_SYMBOL(dev_remove_pack);
L
Linus Torvalds 已提交
455

456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491

/**
 *	dev_add_offload - register offload handlers
 *	@po: protocol offload declaration
 *
 *	Add protocol offload handlers to the networking stack. The passed
 *	&proto_offload is linked into kernel lists and may not be freed until
 *	it has been removed from the kernel lists.
 *
 *	This call does not sleep therefore it can not
 *	guarantee all CPU's that are in middle of receiving packets
 *	will see the new offload handlers (until the next received packet).
 */
void dev_add_offload(struct packet_offload *po)
{
	struct list_head *head = &offload_base;

	spin_lock(&offload_lock);
	list_add_rcu(&po->list, head);
	spin_unlock(&offload_lock);
}
EXPORT_SYMBOL(dev_add_offload);

/**
 *	__dev_remove_offload	 - remove offload handler
 *	@po: packet offload declaration
 *
 *	Remove a protocol offload handler that was previously added to the
 *	kernel offload handlers by dev_add_offload(). The passed &offload_type
 *	is removed from the kernel lists and can be freed or reused once this
 *	function returns.
 *
 *      The packet type might still be in use by receivers
 *	and must not be freed until after all the CPU's have gone
 *	through a quiescent state.
 */
S
stephen hemminger 已提交
492
static void __dev_remove_offload(struct packet_offload *po)
493 494 495 496
{
	struct list_head *head = &offload_base;
	struct packet_offload *po1;

497
	spin_lock(&offload_lock);
498 499 500 501 502 503 504 505 506 507

	list_for_each_entry(po1, head, list) {
		if (po == po1) {
			list_del_rcu(&po->list);
			goto out;
		}
	}

	pr_warn("dev_remove_offload: %p not found\n", po);
out:
508
	spin_unlock(&offload_lock);
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530
}

/**
 *	dev_remove_offload	 - remove packet offload handler
 *	@po: packet offload declaration
 *
 *	Remove a packet offload handler that was previously added to the kernel
 *	offload handlers by dev_add_offload(). The passed &offload_type is
 *	removed from the kernel lists and can be freed or reused once this
 *	function returns.
 *
 *	This call sleeps to guarantee that no CPU is looking at the packet
 *	type after return.
 */
void dev_remove_offload(struct packet_offload *po)
{
	__dev_remove_offload(po);

	synchronize_net();
}
EXPORT_SYMBOL(dev_remove_offload);

L
Linus Torvalds 已提交
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
/******************************************************************************

		      Device Boot-time Settings Routines

*******************************************************************************/

/* Boot time configuration table */
static struct netdev_boot_setup dev_boot_setup[NETDEV_BOOT_SETUP_MAX];

/**
 *	netdev_boot_setup_add	- add new setup entry
 *	@name: name of the device
 *	@map: configured settings for the device
 *
 *	Adds new setup entry to the dev_boot_setup list.  The function
 *	returns 0 on error and 1 on success.  This is a generic routine to
 *	all netdevices.
 */
static int netdev_boot_setup_add(char *name, struct ifmap *map)
{
	struct netdev_boot_setup *s;
	int i;

	s = dev_boot_setup;
	for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++) {
		if (s[i].name[0] == '\0' || s[i].name[0] == ' ') {
			memset(s[i].name, 0, sizeof(s[i].name));
558
			strlcpy(s[i].name, name, IFNAMSIZ);
L
Linus Torvalds 已提交
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
			memcpy(&s[i].map, map, sizeof(s[i].map));
			break;
		}
	}

	return i >= NETDEV_BOOT_SETUP_MAX ? 0 : 1;
}

/**
 *	netdev_boot_setup_check	- check boot time settings
 *	@dev: the netdevice
 *
 * 	Check boot time settings for the device.
 *	The found settings are set for the device to be used
 *	later in the device probing.
 *	Returns 0 if no settings found, 1 if they are.
 */
int netdev_boot_setup_check(struct net_device *dev)
{
	struct netdev_boot_setup *s = dev_boot_setup;
	int i;

	for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++) {
		if (s[i].name[0] != '\0' && s[i].name[0] != ' ' &&
583
		    !strcmp(dev->name, s[i].name)) {
L
Linus Torvalds 已提交
584 585 586 587 588 589 590 591 592
			dev->irq 	= s[i].map.irq;
			dev->base_addr 	= s[i].map.base_addr;
			dev->mem_start 	= s[i].map.mem_start;
			dev->mem_end 	= s[i].map.mem_end;
			return 1;
		}
	}
	return 0;
}
E
Eric Dumazet 已提交
593
EXPORT_SYMBOL(netdev_boot_setup_check);
L
Linus Torvalds 已提交
594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617


/**
 *	netdev_boot_base	- get address from boot time settings
 *	@prefix: prefix for network device
 *	@unit: id for network device
 *
 * 	Check boot time settings for the base address of device.
 *	The found settings are set for the device to be used
 *	later in the device probing.
 *	Returns 0 if no settings found.
 */
unsigned long netdev_boot_base(const char *prefix, int unit)
{
	const struct netdev_boot_setup *s = dev_boot_setup;
	char name[IFNAMSIZ];
	int i;

	sprintf(name, "%s%d", prefix, unit);

	/*
	 * If device already registered then return base of 1
	 * to indicate not to probe for this interface
	 */
618
	if (__dev_get_by_name(&init_net, name))
L
Linus Torvalds 已提交
619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
		return 1;

	for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++)
		if (!strcmp(name, s[i].name))
			return s[i].map.base_addr;
	return 0;
}

/*
 * Saves at boot time configured settings for any netdevice.
 */
int __init netdev_boot_setup(char *str)
{
	int ints[5];
	struct ifmap map;

	str = get_options(str, ARRAY_SIZE(ints), ints);
	if (!str || !*str)
		return 0;

	/* Save settings */
	memset(&map, 0, sizeof(map));
	if (ints[0] > 0)
		map.irq = ints[1];
	if (ints[0] > 1)
		map.base_addr = ints[2];
	if (ints[0] > 2)
		map.mem_start = ints[3];
	if (ints[0] > 3)
		map.mem_end = ints[4];

	/* Add new entry to the list */
	return netdev_boot_setup_add(str, &map);
}

__setup("netdev=", netdev_boot_setup);

/*******************************************************************************

			    Device Interface Subroutines

*******************************************************************************/

662 663 664 665 666 667 668 669 670 671 672 673 674
/**
 *	dev_get_iflink	- get 'iflink' value of a interface
 *	@dev: targeted interface
 *
 *	Indicates the ifindex the interface is linked to.
 *	Physical interfaces have the same 'ifindex' and 'iflink' values.
 */

int dev_get_iflink(const struct net_device *dev)
{
	if (dev->netdev_ops && dev->netdev_ops->ndo_get_iflink)
		return dev->netdev_ops->ndo_get_iflink(dev);

675 676 677 678
	/* If dev->rtnl_link_ops is set, it's a virtual interface. */
	if (dev->rtnl_link_ops)
		return 0;

679
	return dev->ifindex;
680 681 682
}
EXPORT_SYMBOL(dev_get_iflink);

L
Linus Torvalds 已提交
683 684
/**
 *	__dev_get_by_name	- find a device by its name
685
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
686 687 688 689 690 691 692 693 694
 *	@name: name to find
 *
 *	Find an interface by name. Must be called under RTNL semaphore
 *	or @dev_base_lock. If the name is found a pointer to the device
 *	is returned. If the name is not found then %NULL is returned. The
 *	reference counters are not incremented so the caller must be
 *	careful with locks.
 */

695
struct net_device *__dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
696
{
E
Eric Dumazet 已提交
697 698
	struct net_device *dev;
	struct hlist_head *head = dev_name_hash(net, name);
L
Linus Torvalds 已提交
699

700
	hlist_for_each_entry(dev, head, name_hlist)
L
Linus Torvalds 已提交
701 702
		if (!strncmp(dev->name, name, IFNAMSIZ))
			return dev;
E
Eric Dumazet 已提交
703

L
Linus Torvalds 已提交
704 705
	return NULL;
}
E
Eric Dumazet 已提交
706
EXPORT_SYMBOL(__dev_get_by_name);
L
Linus Torvalds 已提交
707

708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724
/**
 *	dev_get_by_name_rcu	- find a device by its name
 *	@net: the applicable net namespace
 *	@name: name to find
 *
 *	Find an interface by name.
 *	If the name is found a pointer to the device is returned.
 * 	If the name is not found then %NULL is returned.
 *	The reference counters are not incremented so the caller must be
 *	careful with locks. The caller must hold RCU lock.
 */

struct net_device *dev_get_by_name_rcu(struct net *net, const char *name)
{
	struct net_device *dev;
	struct hlist_head *head = dev_name_hash(net, name);

725
	hlist_for_each_entry_rcu(dev, head, name_hlist)
726 727 728 729 730 731 732
		if (!strncmp(dev->name, name, IFNAMSIZ))
			return dev;

	return NULL;
}
EXPORT_SYMBOL(dev_get_by_name_rcu);

L
Linus Torvalds 已提交
733 734
/**
 *	dev_get_by_name		- find a device by its name
735
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
736 737 738 739 740 741 742 743 744
 *	@name: name to find
 *
 *	Find an interface by name. This can be called from any
 *	context and does its own locking. The returned handle has
 *	the usage count incremented and the caller must use dev_put() to
 *	release it when it is no longer needed. %NULL is returned if no
 *	matching device is found.
 */

745
struct net_device *dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
746 747 748
{
	struct net_device *dev;

749 750
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
L
Linus Torvalds 已提交
751 752
	if (dev)
		dev_hold(dev);
753
	rcu_read_unlock();
L
Linus Torvalds 已提交
754 755
	return dev;
}
E
Eric Dumazet 已提交
756
EXPORT_SYMBOL(dev_get_by_name);
L
Linus Torvalds 已提交
757 758 759

/**
 *	__dev_get_by_index - find a device by its ifindex
760
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
761 762 763 764 765 766 767 768 769
 *	@ifindex: index of device
 *
 *	Search for an interface by index. Returns %NULL if the device
 *	is not found or a pointer to the device. The device has not
 *	had its reference counter increased so the caller must be careful
 *	about locking. The caller must hold either the RTNL semaphore
 *	or @dev_base_lock.
 */

770
struct net_device *__dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
771
{
E
Eric Dumazet 已提交
772 773
	struct net_device *dev;
	struct hlist_head *head = dev_index_hash(net, ifindex);
L
Linus Torvalds 已提交
774

775
	hlist_for_each_entry(dev, head, index_hlist)
L
Linus Torvalds 已提交
776 777
		if (dev->ifindex == ifindex)
			return dev;
E
Eric Dumazet 已提交
778

L
Linus Torvalds 已提交
779 780
	return NULL;
}
E
Eric Dumazet 已提交
781
EXPORT_SYMBOL(__dev_get_by_index);
L
Linus Torvalds 已提交
782

783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
/**
 *	dev_get_by_index_rcu - find a device by its ifindex
 *	@net: the applicable net namespace
 *	@ifindex: index of device
 *
 *	Search for an interface by index. Returns %NULL if the device
 *	is not found or a pointer to the device. The device has not
 *	had its reference counter increased so the caller must be careful
 *	about locking. The caller must hold RCU lock.
 */

struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex)
{
	struct net_device *dev;
	struct hlist_head *head = dev_index_hash(net, ifindex);

799
	hlist_for_each_entry_rcu(dev, head, index_hlist)
800 801 802 803 804 805 806
		if (dev->ifindex == ifindex)
			return dev;

	return NULL;
}
EXPORT_SYMBOL(dev_get_by_index_rcu);

L
Linus Torvalds 已提交
807 808 809

/**
 *	dev_get_by_index - find a device by its ifindex
810
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
811 812 813 814 815 816 817 818
 *	@ifindex: index of device
 *
 *	Search for an interface by index. Returns NULL if the device
 *	is not found or a pointer to the device. The device returned has
 *	had a reference added and the pointer is safe until the user calls
 *	dev_put to indicate they have finished with it.
 */

819
struct net_device *dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
820 821 822
{
	struct net_device *dev;

823 824
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifindex);
L
Linus Torvalds 已提交
825 826
	if (dev)
		dev_hold(dev);
827
	rcu_read_unlock();
L
Linus Torvalds 已提交
828 829
	return dev;
}
E
Eric Dumazet 已提交
830
EXPORT_SYMBOL(dev_get_by_index);
L
Linus Torvalds 已提交
831

832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
/**
 *	netdev_get_name - get a netdevice name, knowing its ifindex.
 *	@net: network namespace
 *	@name: a pointer to the buffer where the name will be stored.
 *	@ifindex: the ifindex of the interface to get the name from.
 *
 *	The use of raw_seqcount_begin() and cond_resched() before
 *	retrying is required as we want to give the writers a chance
 *	to complete when CONFIG_PREEMPT is not set.
 */
int netdev_get_name(struct net *net, char *name, int ifindex)
{
	struct net_device *dev;
	unsigned int seq;

retry:
	seq = raw_seqcount_begin(&devnet_rename_seq);
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifindex);
	if (!dev) {
		rcu_read_unlock();
		return -ENODEV;
	}

	strcpy(name, dev->name);
	rcu_read_unlock();
	if (read_seqcount_retry(&devnet_rename_seq, seq)) {
		cond_resched();
		goto retry;
	}

	return 0;
}

L
Linus Torvalds 已提交
866
/**
867
 *	dev_getbyhwaddr_rcu - find a device by its hardware address
868
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
869 870 871 872
 *	@type: media type of device
 *	@ha: hardware address
 *
 *	Search for an interface by MAC address. Returns NULL if the device
E
Eric Dumazet 已提交
873 874
 *	is not found or a pointer to the device.
 *	The caller must hold RCU or RTNL.
875
 *	The returned device has not had its ref count increased
L
Linus Torvalds 已提交
876 877 878 879
 *	and the caller must therefore be careful about locking
 *
 */

880 881
struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
				       const char *ha)
L
Linus Torvalds 已提交
882 883 884
{
	struct net_device *dev;

885
	for_each_netdev_rcu(net, dev)
L
Linus Torvalds 已提交
886 887
		if (dev->type == type &&
		    !memcmp(dev->dev_addr, ha, dev->addr_len))
888 889 890
			return dev;

	return NULL;
L
Linus Torvalds 已提交
891
}
892
EXPORT_SYMBOL(dev_getbyhwaddr_rcu);
893

894
struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type)
L
Linus Torvalds 已提交
895 896 897
{
	struct net_device *dev;

898
	ASSERT_RTNL();
899
	for_each_netdev(net, dev)
900
		if (dev->type == type)
901 902 903
			return dev;

	return NULL;
904 905 906
}
EXPORT_SYMBOL(__dev_getfirstbyhwtype);

907
struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
908
{
909
	struct net_device *dev, *ret = NULL;
910

911 912 913 914 915 916 917 918 919
	rcu_read_lock();
	for_each_netdev_rcu(net, dev)
		if (dev->type == type) {
			dev_hold(dev);
			ret = dev;
			break;
		}
	rcu_read_unlock();
	return ret;
L
Linus Torvalds 已提交
920 921 922 923
}
EXPORT_SYMBOL(dev_getfirstbyhwtype);

/**
924
 *	__dev_get_by_flags - find any device with given flags
925
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
926 927 928 929
 *	@if_flags: IFF_* values
 *	@mask: bitmask of bits in if_flags to check
 *
 *	Search for any interface with the given flags. Returns NULL if a device
E
Eric Dumazet 已提交
930
 *	is not found or a pointer to the device. Must be called inside
931
 *	rtnl_lock(), and result refcount is unchanged.
L
Linus Torvalds 已提交
932 933
 */

934 935
struct net_device *__dev_get_by_flags(struct net *net, unsigned short if_flags,
				      unsigned short mask)
L
Linus Torvalds 已提交
936
{
937
	struct net_device *dev, *ret;
L
Linus Torvalds 已提交
938

939 940
	ASSERT_RTNL();

941
	ret = NULL;
942
	for_each_netdev(net, dev) {
L
Linus Torvalds 已提交
943
		if (((dev->flags ^ if_flags) & mask) == 0) {
944
			ret = dev;
L
Linus Torvalds 已提交
945 946 947
			break;
		}
	}
948
	return ret;
L
Linus Torvalds 已提交
949
}
950
EXPORT_SYMBOL(__dev_get_by_flags);
L
Linus Torvalds 已提交
951 952 953 954 955 956

/**
 *	dev_valid_name - check if name is okay for network device
 *	@name: name string
 *
 *	Network device names need to be valid file names to
957 958
 *	to allow sysfs to work.  We also disallow any kind of
 *	whitespace.
L
Linus Torvalds 已提交
959
 */
960
bool dev_valid_name(const char *name)
L
Linus Torvalds 已提交
961
{
962
	if (*name == '\0')
963
		return false;
964
	if (strlen(name) >= IFNAMSIZ)
965
		return false;
966
	if (!strcmp(name, ".") || !strcmp(name, ".."))
967
		return false;
968 969

	while (*name) {
970
		if (*name == '/' || *name == ':' || isspace(*name))
971
			return false;
972 973
		name++;
	}
974
	return true;
L
Linus Torvalds 已提交
975
}
E
Eric Dumazet 已提交
976
EXPORT_SYMBOL(dev_valid_name);
L
Linus Torvalds 已提交
977 978

/**
979 980
 *	__dev_alloc_name - allocate a name for a device
 *	@net: network namespace to allocate the device name in
L
Linus Torvalds 已提交
981
 *	@name: name format string
982
 *	@buf:  scratch buffer and result name string
L
Linus Torvalds 已提交
983 984
 *
 *	Passed a format string - eg "lt%d" it will try and find a suitable
S
Stephen Hemminger 已提交
985 986 987 988 989 990
 *	id. It scans list of devices to build up a free map, then chooses
 *	the first empty slot. The caller must hold the dev_base or rtnl lock
 *	while allocating the name and adding the device in order to avoid
 *	duplicates.
 *	Limited to bits_per_byte * page size devices (ie 32K on most platforms).
 *	Returns the number of the unit assigned or a negative errno code.
L
Linus Torvalds 已提交
991 992
 */

993
static int __dev_alloc_name(struct net *net, const char *name, char *buf)
L
Linus Torvalds 已提交
994 995 996 997
{
	int i = 0;
	const char *p;
	const int max_netdevices = 8*PAGE_SIZE;
S
Stephen Hemminger 已提交
998
	unsigned long *inuse;
L
Linus Torvalds 已提交
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
	struct net_device *d;

	p = strnchr(name, IFNAMSIZ-1, '%');
	if (p) {
		/*
		 * Verify the string as this thing may have come from
		 * the user.  There must be either one "%d" and no other "%"
		 * characters.
		 */
		if (p[1] != 'd' || strchr(p + 2, '%'))
			return -EINVAL;

		/* Use one page as a bit array of possible slots */
S
Stephen Hemminger 已提交
1012
		inuse = (unsigned long *) get_zeroed_page(GFP_ATOMIC);
L
Linus Torvalds 已提交
1013 1014 1015
		if (!inuse)
			return -ENOMEM;

1016
		for_each_netdev(net, d) {
L
Linus Torvalds 已提交
1017 1018 1019 1020 1021 1022
			if (!sscanf(d->name, name, &i))
				continue;
			if (i < 0 || i >= max_netdevices)
				continue;

			/*  avoid cases where sscanf is not exact inverse of printf */
1023
			snprintf(buf, IFNAMSIZ, name, i);
L
Linus Torvalds 已提交
1024 1025 1026 1027 1028 1029 1030 1031
			if (!strncmp(buf, d->name, IFNAMSIZ))
				set_bit(i, inuse);
		}

		i = find_first_zero_bit(inuse, max_netdevices);
		free_page((unsigned long) inuse);
	}

1032 1033
	if (buf != name)
		snprintf(buf, IFNAMSIZ, name, i);
1034
	if (!__dev_get_by_name(net, buf))
L
Linus Torvalds 已提交
1035 1036 1037 1038 1039 1040 1041 1042 1043
		return i;

	/* It is possible to run out of possible slots
	 * when the name is long and there isn't enough space left
	 * for the digits, or if all bits are used.
	 */
	return -ENFILE;
}

1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
/**
 *	dev_alloc_name - allocate a name for a device
 *	@dev: device
 *	@name: name format string
 *
 *	Passed a format string - eg "lt%d" it will try and find a suitable
 *	id. It scans list of devices to build up a free map, then chooses
 *	the first empty slot. The caller must hold the dev_base or rtnl lock
 *	while allocating the name and adding the device in order to avoid
 *	duplicates.
 *	Limited to bits_per_byte * page size devices (ie 32K on most platforms).
 *	Returns the number of the unit assigned or a negative errno code.
 */

int dev_alloc_name(struct net_device *dev, const char *name)
{
	char buf[IFNAMSIZ];
	struct net *net;
	int ret;

1064 1065
	BUG_ON(!dev_net(dev));
	net = dev_net(dev);
1066 1067 1068 1069 1070
	ret = __dev_alloc_name(net, name, buf);
	if (ret >= 0)
		strlcpy(dev->name, buf, IFNAMSIZ);
	return ret;
}
E
Eric Dumazet 已提交
1071
EXPORT_SYMBOL(dev_alloc_name);
1072

1073 1074 1075
static int dev_alloc_name_ns(struct net *net,
			     struct net_device *dev,
			     const char *name)
1076
{
1077 1078
	char buf[IFNAMSIZ];
	int ret;
1079

1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
	ret = __dev_alloc_name(net, name, buf);
	if (ret >= 0)
		strlcpy(dev->name, buf, IFNAMSIZ);
	return ret;
}

static int dev_get_valid_name(struct net *net,
			      struct net_device *dev,
			      const char *name)
{
	BUG_ON(!net);
1091

1092 1093 1094
	if (!dev_valid_name(name))
		return -EINVAL;

1095
	if (strchr(name, '%'))
1096
		return dev_alloc_name_ns(net, dev, name);
1097 1098
	else if (__dev_get_by_name(net, name))
		return -EEXIST;
1099 1100
	else if (dev->name != name)
		strlcpy(dev->name, name, IFNAMSIZ);
1101 1102 1103

	return 0;
}
L
Linus Torvalds 已提交
1104 1105 1106 1107 1108 1109 1110 1111 1112

/**
 *	dev_change_name - change name of a device
 *	@dev: device
 *	@newname: name (or format string) must be at least IFNAMSIZ
 *
 *	Change name of a device, can pass format strings "eth%d".
 *	for wildcarding.
 */
1113
int dev_change_name(struct net_device *dev, const char *newname)
L
Linus Torvalds 已提交
1114
{
1115
	unsigned char old_assign_type;
1116
	char oldname[IFNAMSIZ];
L
Linus Torvalds 已提交
1117
	int err = 0;
1118
	int ret;
1119
	struct net *net;
L
Linus Torvalds 已提交
1120 1121

	ASSERT_RTNL();
1122
	BUG_ON(!dev_net(dev));
L
Linus Torvalds 已提交
1123

1124
	net = dev_net(dev);
L
Linus Torvalds 已提交
1125 1126 1127
	if (dev->flags & IFF_UP)
		return -EBUSY;

1128
	write_seqcount_begin(&devnet_rename_seq);
1129 1130

	if (strncmp(newname, dev->name, IFNAMSIZ) == 0) {
1131
		write_seqcount_end(&devnet_rename_seq);
1132
		return 0;
1133
	}
1134

1135 1136
	memcpy(oldname, dev->name, IFNAMSIZ);

1137
	err = dev_get_valid_name(net, dev, newname);
1138
	if (err < 0) {
1139
		write_seqcount_end(&devnet_rename_seq);
1140
		return err;
1141
	}
L
Linus Torvalds 已提交
1142

1143 1144 1145
	if (oldname[0] && !strchr(oldname, '%'))
		netdev_info(dev, "renamed from %s\n", oldname);

1146 1147 1148
	old_assign_type = dev->name_assign_type;
	dev->name_assign_type = NET_NAME_RENAMED;

1149
rollback:
1150 1151 1152
	ret = device_rename(&dev->dev, dev->name);
	if (ret) {
		memcpy(dev->name, oldname, IFNAMSIZ);
1153
		dev->name_assign_type = old_assign_type;
1154
		write_seqcount_end(&devnet_rename_seq);
1155
		return ret;
1156
	}
1157

1158
	write_seqcount_end(&devnet_rename_seq);
1159

1160 1161
	netdev_adjacent_rename_links(dev, oldname);

1162
	write_lock_bh(&dev_base_lock);
1163
	hlist_del_rcu(&dev->name_hlist);
1164 1165 1166 1167 1168 1169
	write_unlock_bh(&dev_base_lock);

	synchronize_rcu();

	write_lock_bh(&dev_base_lock);
	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net, dev->name));
1170 1171
	write_unlock_bh(&dev_base_lock);

1172
	ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev);
1173 1174 1175
	ret = notifier_to_errno(ret);

	if (ret) {
1176 1177
		/* err >= 0 after dev_alloc_name() or stores the first errno */
		if (err >= 0) {
1178
			err = ret;
1179
			write_seqcount_begin(&devnet_rename_seq);
1180
			memcpy(dev->name, oldname, IFNAMSIZ);
1181
			memcpy(oldname, newname, IFNAMSIZ);
1182 1183
			dev->name_assign_type = old_assign_type;
			old_assign_type = NET_NAME_RENAMED;
1184
			goto rollback;
1185
		} else {
1186
			pr_err("%s: name change rollback failed: %d\n",
1187
			       dev->name, ret);
1188 1189
		}
	}
L
Linus Torvalds 已提交
1190 1191 1192 1193

	return err;
}

1194 1195 1196 1197
/**
 *	dev_set_alias - change ifalias of a device
 *	@dev: device
 *	@alias: name up to IFALIASZ
1198
 *	@len: limit of bytes to copy from info
1199 1200 1201 1202 1203
 *
 *	Set ifalias for a device,
 */
int dev_set_alias(struct net_device *dev, const char *alias, size_t len)
{
1204 1205
	char *new_ifalias;

1206 1207 1208 1209 1210
	ASSERT_RTNL();

	if (len >= IFALIASZ)
		return -EINVAL;

1211
	if (!len) {
1212 1213
		kfree(dev->ifalias);
		dev->ifalias = NULL;
1214 1215 1216
		return 0;
	}

1217 1218
	new_ifalias = krealloc(dev->ifalias, len + 1, GFP_KERNEL);
	if (!new_ifalias)
1219
		return -ENOMEM;
1220
	dev->ifalias = new_ifalias;
1221 1222 1223 1224 1225 1226

	strlcpy(dev->ifalias, alias, len+1);
	return len;
}


1227
/**
S
Stephen Hemminger 已提交
1228
 *	netdev_features_change - device changes features
1229 1230 1231 1232 1233 1234
 *	@dev: device to cause notification
 *
 *	Called to indicate a device has changed features.
 */
void netdev_features_change(struct net_device *dev)
{
1235
	call_netdevice_notifiers(NETDEV_FEAT_CHANGE, dev);
1236 1237 1238
}
EXPORT_SYMBOL(netdev_features_change);

L
Linus Torvalds 已提交
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
/**
 *	netdev_state_change - device changes state
 *	@dev: device to cause notification
 *
 *	Called to indicate a device has changed state. This function calls
 *	the notifier chains for netdev_chain and sends a NEWLINK message
 *	to the routing socket.
 */
void netdev_state_change(struct net_device *dev)
{
	if (dev->flags & IFF_UP) {
1250 1251 1252 1253 1254
		struct netdev_notifier_change_info change_info;

		change_info.flags_changed = 0;
		call_netdevice_notifiers_info(NETDEV_CHANGE, dev,
					      &change_info.info);
1255
		rtmsg_ifinfo(RTM_NEWLINK, dev, 0, GFP_KERNEL);
L
Linus Torvalds 已提交
1256 1257
	}
}
E
Eric Dumazet 已提交
1258
EXPORT_SYMBOL(netdev_state_change);
L
Linus Torvalds 已提交
1259

1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
/**
 * 	netdev_notify_peers - notify network peers about existence of @dev
 * 	@dev: network device
 *
 * Generate traffic such that interested network peers are aware of
 * @dev, such as by generating a gratuitous ARP. This may be used when
 * a device wants to inform the rest of the network about some sort of
 * reconfiguration such as a failover event or virtual machine
 * migration.
 */
void netdev_notify_peers(struct net_device *dev)
1271
{
1272 1273 1274
	rtnl_lock();
	call_netdevice_notifiers(NETDEV_NOTIFY_PEERS, dev);
	rtnl_unlock();
1275
}
1276
EXPORT_SYMBOL(netdev_notify_peers);
1277

1278
static int __dev_open(struct net_device *dev)
L
Linus Torvalds 已提交
1279
{
1280
	const struct net_device_ops *ops = dev->netdev_ops;
1281
	int ret;
L
Linus Torvalds 已提交
1282

1283 1284
	ASSERT_RTNL();

L
Linus Torvalds 已提交
1285 1286 1287
	if (!netif_device_present(dev))
		return -ENODEV;

1288 1289 1290 1291
	/* Block netpoll from trying to do any rx path servicing.
	 * If we don't do this there is a chance ndo_poll_controller
	 * or ndo_poll may be running while we open the device
	 */
1292
	netpoll_poll_disable(dev);
1293

1294 1295 1296 1297 1298
	ret = call_netdevice_notifiers(NETDEV_PRE_UP, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		return ret;

L
Linus Torvalds 已提交
1299
	set_bit(__LINK_STATE_START, &dev->state);
1300

1301 1302
	if (ops->ndo_validate_addr)
		ret = ops->ndo_validate_addr(dev);
1303

1304 1305
	if (!ret && ops->ndo_open)
		ret = ops->ndo_open(dev);
L
Linus Torvalds 已提交
1306

1307
	netpoll_poll_enable(dev);
1308

1309 1310 1311
	if (ret)
		clear_bit(__LINK_STATE_START, &dev->state);
	else {
L
Linus Torvalds 已提交
1312
		dev->flags |= IFF_UP;
1313
		dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
1314
		dev_activate(dev);
1315
		add_device_randomness(dev->dev_addr, dev->addr_len);
L
Linus Torvalds 已提交
1316
	}
1317

L
Linus Torvalds 已提交
1318 1319 1320 1321
	return ret;
}

/**
1322 1323
 *	dev_open	- prepare an interface for use.
 *	@dev:	device to open
L
Linus Torvalds 已提交
1324
 *
1325 1326 1327 1328 1329 1330 1331
 *	Takes a device from down to up state. The device's private open
 *	function is invoked and then the multicast lists are loaded. Finally
 *	the device is moved into the up state and a %NETDEV_UP message is
 *	sent to the netdev notifier chain.
 *
 *	Calling this function on an active interface is a nop. On a failure
 *	a negative errno code is returned.
L
Linus Torvalds 已提交
1332
 */
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
int dev_open(struct net_device *dev)
{
	int ret;

	if (dev->flags & IFF_UP)
		return 0;

	ret = __dev_open(dev);
	if (ret < 0)
		return ret;

1344
	rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING, GFP_KERNEL);
1345 1346 1347 1348 1349 1350
	call_netdevice_notifiers(NETDEV_UP, dev);

	return ret;
}
EXPORT_SYMBOL(dev_open);

1351
static int __dev_close_many(struct list_head *head)
L
Linus Torvalds 已提交
1352
{
1353
	struct net_device *dev;
1354

1355
	ASSERT_RTNL();
1356 1357
	might_sleep();

1358
	list_for_each_entry(dev, head, close_list) {
1359
		/* Temporarily disable netpoll until the interface is down */
1360
		netpoll_poll_disable(dev);
1361

1362
		call_netdevice_notifiers(NETDEV_GOING_DOWN, dev);
L
Linus Torvalds 已提交
1363

1364
		clear_bit(__LINK_STATE_START, &dev->state);
L
Linus Torvalds 已提交
1365

1366 1367 1368 1369 1370 1371
		/* Synchronize to scheduled poll. We cannot touch poll list, it
		 * can be even on different cpu. So just clear netif_running().
		 *
		 * dev->stop() will invoke napi_disable() on all of it's
		 * napi_struct instances on this device.
		 */
1372
		smp_mb__after_atomic(); /* Commit netif_running(). */
1373
	}
L
Linus Torvalds 已提交
1374

1375
	dev_deactivate_many(head);
1376

1377
	list_for_each_entry(dev, head, close_list) {
1378
		const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
1379

1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
		/*
		 *	Call the device specific close. This cannot fail.
		 *	Only if device is UP
		 *
		 *	We allow it to be called even after a DETACH hot-plug
		 *	event.
		 */
		if (ops->ndo_stop)
			ops->ndo_stop(dev);

		dev->flags &= ~IFF_UP;
1391
		netpoll_poll_enable(dev);
1392 1393 1394 1395 1396 1397 1398
	}

	return 0;
}

static int __dev_close(struct net_device *dev)
{
1399
	int retval;
1400 1401
	LIST_HEAD(single);

1402
	list_add(&dev->close_list, &single);
1403 1404
	retval = __dev_close_many(&single);
	list_del(&single);
1405

1406
	return retval;
1407 1408
}

1409
int dev_close_many(struct list_head *head, bool unlink)
1410 1411
{
	struct net_device *dev, *tmp;
L
Linus Torvalds 已提交
1412

1413 1414
	/* Remove the devices that don't need to be closed */
	list_for_each_entry_safe(dev, tmp, head, close_list)
1415
		if (!(dev->flags & IFF_UP))
1416
			list_del_init(&dev->close_list);
1417 1418

	__dev_close_many(head);
L
Linus Torvalds 已提交
1419

1420
	list_for_each_entry_safe(dev, tmp, head, close_list) {
1421
		rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING, GFP_KERNEL);
1422
		call_netdevice_notifiers(NETDEV_DOWN, dev);
1423 1424
		if (unlink)
			list_del_init(&dev->close_list);
1425
	}
1426 1427 1428

	return 0;
}
1429
EXPORT_SYMBOL(dev_close_many);
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441

/**
 *	dev_close - shutdown an interface.
 *	@dev: device to shutdown
 *
 *	This function moves an active device into down state. A
 *	%NETDEV_GOING_DOWN is sent to the netdev notifier chain. The device
 *	is then deactivated and finally a %NETDEV_DOWN is sent to the notifier
 *	chain.
 */
int dev_close(struct net_device *dev)
{
1442 1443
	if (dev->flags & IFF_UP) {
		LIST_HEAD(single);
L
Linus Torvalds 已提交
1444

1445
		list_add(&dev->close_list, &single);
1446
		dev_close_many(&single, true);
1447 1448
		list_del(&single);
	}
1449
	return 0;
L
Linus Torvalds 已提交
1450
}
E
Eric Dumazet 已提交
1451
EXPORT_SYMBOL(dev_close);
L
Linus Torvalds 已提交
1452 1453


1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
/**
 *	dev_disable_lro - disable Large Receive Offload on a device
 *	@dev: device
 *
 *	Disable Large Receive Offload (LRO) on a net device.  Must be
 *	called under RTNL.  This is needed if received packets may be
 *	forwarded to another interface.
 */
void dev_disable_lro(struct net_device *dev)
{
1464 1465
	struct net_device *lower_dev;
	struct list_head *iter;
1466

M
Michał Mirosław 已提交
1467 1468
	dev->wanted_features &= ~NETIF_F_LRO;
	netdev_update_features(dev);
1469

1470 1471
	if (unlikely(dev->features & NETIF_F_LRO))
		netdev_WARN(dev, "failed to disable LRO!\n");
1472 1473 1474

	netdev_for_each_lower_dev(dev, lower_dev, iter)
		dev_disable_lro(lower_dev);
1475 1476 1477
}
EXPORT_SYMBOL(dev_disable_lro);

1478 1479 1480 1481 1482 1483 1484 1485
static int call_netdevice_notifier(struct notifier_block *nb, unsigned long val,
				   struct net_device *dev)
{
	struct netdev_notifier_info info;

	netdev_notifier_info_init(&info, dev);
	return nb->notifier_call(nb, val, &info);
}
1486

1487 1488
static int dev_boot_phase = 1;

L
Linus Torvalds 已提交
1489 1490 1491 1492 1493 1494 1495 1496 1497 1498
/**
 *	register_netdevice_notifier - register a network notifier block
 *	@nb: notifier
 *
 *	Register a notifier to be called when network device events occur.
 *	The notifier passed is linked into the kernel structures and must
 *	not be reused until it has been unregistered. A negative errno code
 *	is returned on a failure.
 *
 * 	When registered all registration and up events are replayed
1499
 *	to the new notifier to allow device to have a race free
L
Linus Torvalds 已提交
1500 1501 1502 1503 1504 1505
 *	view of the network device list.
 */

int register_netdevice_notifier(struct notifier_block *nb)
{
	struct net_device *dev;
1506
	struct net_device *last;
1507
	struct net *net;
L
Linus Torvalds 已提交
1508 1509 1510
	int err;

	rtnl_lock();
1511
	err = raw_notifier_chain_register(&netdev_chain, nb);
1512 1513
	if (err)
		goto unlock;
1514 1515 1516 1517
	if (dev_boot_phase)
		goto unlock;
	for_each_net(net) {
		for_each_netdev(net, dev) {
1518
			err = call_netdevice_notifier(nb, NETDEV_REGISTER, dev);
1519 1520 1521 1522 1523 1524
			err = notifier_to_errno(err);
			if (err)
				goto rollback;

			if (!(dev->flags & IFF_UP))
				continue;
L
Linus Torvalds 已提交
1525

1526
			call_netdevice_notifier(nb, NETDEV_UP, dev);
1527
		}
L
Linus Torvalds 已提交
1528
	}
1529 1530

unlock:
L
Linus Torvalds 已提交
1531 1532
	rtnl_unlock();
	return err;
1533 1534 1535

rollback:
	last = dev;
1536 1537 1538
	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev == last)
1539
				goto outroll;
1540

1541
			if (dev->flags & IFF_UP) {
1542 1543 1544
				call_netdevice_notifier(nb, NETDEV_GOING_DOWN,
							dev);
				call_netdevice_notifier(nb, NETDEV_DOWN, dev);
1545
			}
1546
			call_netdevice_notifier(nb, NETDEV_UNREGISTER, dev);
1547 1548
		}
	}
1549

1550
outroll:
1551
	raw_notifier_chain_unregister(&netdev_chain, nb);
1552
	goto unlock;
L
Linus Torvalds 已提交
1553
}
E
Eric Dumazet 已提交
1554
EXPORT_SYMBOL(register_netdevice_notifier);
L
Linus Torvalds 已提交
1555 1556 1557 1558 1559 1560 1561 1562 1563

/**
 *	unregister_netdevice_notifier - unregister a network notifier block
 *	@nb: notifier
 *
 *	Unregister a notifier previously registered by
 *	register_netdevice_notifier(). The notifier is unlinked into the
 *	kernel structures and may then be reused. A negative errno code
 *	is returned on a failure.
1564 1565 1566 1567
 *
 * 	After unregistering unregister and down device events are synthesized
 *	for all devices on the device list to the removed notifier to remove
 *	the need for special case cleanup code.
L
Linus Torvalds 已提交
1568 1569 1570 1571
 */

int unregister_netdevice_notifier(struct notifier_block *nb)
{
1572 1573
	struct net_device *dev;
	struct net *net;
1574 1575 1576
	int err;

	rtnl_lock();
1577
	err = raw_notifier_chain_unregister(&netdev_chain, nb);
1578 1579 1580 1581 1582 1583
	if (err)
		goto unlock;

	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev->flags & IFF_UP) {
1584 1585 1586
				call_netdevice_notifier(nb, NETDEV_GOING_DOWN,
							dev);
				call_netdevice_notifier(nb, NETDEV_DOWN, dev);
1587
			}
1588
			call_netdevice_notifier(nb, NETDEV_UNREGISTER, dev);
1589 1590 1591
		}
	}
unlock:
1592 1593
	rtnl_unlock();
	return err;
L
Linus Torvalds 已提交
1594
}
E
Eric Dumazet 已提交
1595
EXPORT_SYMBOL(unregister_netdevice_notifier);
L
Linus Torvalds 已提交
1596

1597 1598 1599 1600 1601 1602 1603 1604 1605 1606
/**
 *	call_netdevice_notifiers_info - call all network notifier blocks
 *	@val: value passed unmodified to notifier function
 *	@dev: net_device pointer passed unmodified to notifier function
 *	@info: notifier information data
 *
 *	Call all network notifier blocks.  Parameters and return value
 *	are as for raw_notifier_call_chain().
 */

S
stephen hemminger 已提交
1607 1608 1609
static int call_netdevice_notifiers_info(unsigned long val,
					 struct net_device *dev,
					 struct netdev_notifier_info *info)
1610 1611 1612 1613 1614 1615
{
	ASSERT_RTNL();
	netdev_notifier_info_init(info, dev);
	return raw_notifier_call_chain(&netdev_chain, val, info);
}

L
Linus Torvalds 已提交
1616 1617 1618
/**
 *	call_netdevice_notifiers - call all network notifier blocks
 *      @val: value passed unmodified to notifier function
1619
 *      @dev: net_device pointer passed unmodified to notifier function
L
Linus Torvalds 已提交
1620 1621
 *
 *	Call all network notifier blocks.  Parameters and return value
1622
 *	are as for raw_notifier_call_chain().
L
Linus Torvalds 已提交
1623 1624
 */

1625
int call_netdevice_notifiers(unsigned long val, struct net_device *dev)
L
Linus Torvalds 已提交
1626
{
1627 1628 1629
	struct netdev_notifier_info info;

	return call_netdevice_notifiers_info(val, dev, &info);
L
Linus Torvalds 已提交
1630
}
1631
EXPORT_SYMBOL(call_netdevice_notifiers);
L
Linus Torvalds 已提交
1632

1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648
#ifdef CONFIG_NET_CLS_ACT
static struct static_key ingress_needed __read_mostly;

void net_inc_ingress_queue(void)
{
	static_key_slow_inc(&ingress_needed);
}
EXPORT_SYMBOL_GPL(net_inc_ingress_queue);

void net_dec_ingress_queue(void)
{
	static_key_slow_dec(&ingress_needed);
}
EXPORT_SYMBOL_GPL(net_dec_ingress_queue);
#endif

1649
static struct static_key netstamp_needed __read_mostly;
1650
#ifdef HAVE_JUMP_LABEL
1651
/* We are not allowed to call static_key_slow_dec() from irq context
1652
 * If net_disable_timestamp() is called from irq context, defer the
1653
 * static_key_slow_dec() calls.
1654 1655 1656
 */
static atomic_t netstamp_needed_deferred;
#endif
L
Linus Torvalds 已提交
1657 1658 1659

void net_enable_timestamp(void)
{
1660 1661 1662 1663 1664
#ifdef HAVE_JUMP_LABEL
	int deferred = atomic_xchg(&netstamp_needed_deferred, 0);

	if (deferred) {
		while (--deferred)
1665
			static_key_slow_dec(&netstamp_needed);
1666 1667 1668
		return;
	}
#endif
1669
	static_key_slow_inc(&netstamp_needed);
L
Linus Torvalds 已提交
1670
}
E
Eric Dumazet 已提交
1671
EXPORT_SYMBOL(net_enable_timestamp);
L
Linus Torvalds 已提交
1672 1673 1674

void net_disable_timestamp(void)
{
1675 1676 1677 1678 1679 1680
#ifdef HAVE_JUMP_LABEL
	if (in_interrupt()) {
		atomic_inc(&netstamp_needed_deferred);
		return;
	}
#endif
1681
	static_key_slow_dec(&netstamp_needed);
L
Linus Torvalds 已提交
1682
}
E
Eric Dumazet 已提交
1683
EXPORT_SYMBOL(net_disable_timestamp);
L
Linus Torvalds 已提交
1684

E
Eric Dumazet 已提交
1685
static inline void net_timestamp_set(struct sk_buff *skb)
L
Linus Torvalds 已提交
1686
{
1687
	skb->tstamp.tv64 = 0;
1688
	if (static_key_false(&netstamp_needed))
1689
		__net_timestamp(skb);
L
Linus Torvalds 已提交
1690 1691
}

1692
#define net_timestamp_check(COND, SKB)			\
1693
	if (static_key_false(&netstamp_needed)) {		\
1694 1695 1696
		if ((COND) && !(SKB)->tstamp.tv64)	\
			__net_timestamp(SKB);		\
	}						\
E
Eric Dumazet 已提交
1697

1698
bool is_skb_forwardable(struct net_device *dev, struct sk_buff *skb)
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716
{
	unsigned int len;

	if (!(dev->flags & IFF_UP))
		return false;

	len = dev->mtu + dev->hard_header_len + VLAN_HLEN;
	if (skb->len <= len)
		return true;

	/* if TSO is enabled, we don't care about the length as the packet
	 * could be forwarded without being segmented before
	 */
	if (skb_is_gso(skb))
		return true;

	return false;
}
1717
EXPORT_SYMBOL_GPL(is_skb_forwardable);
1718

H
Herbert Xu 已提交
1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735
int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb)
{
	if (skb_shinfo(skb)->tx_flags & SKBTX_DEV_ZEROCOPY) {
		if (skb_copy_ubufs(skb, GFP_ATOMIC)) {
			atomic_long_inc(&dev->rx_dropped);
			kfree_skb(skb);
			return NET_RX_DROP;
		}
	}

	if (unlikely(!is_skb_forwardable(dev, skb))) {
		atomic_long_inc(&dev->rx_dropped);
		kfree_skb(skb);
		return NET_RX_DROP;
	}

	skb_scrub_packet(skb, true);
1736
	skb->priority = 0;
H
Herbert Xu 已提交
1737
	skb->protocol = eth_type_trans(skb, dev);
1738
	skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
H
Herbert Xu 已提交
1739 1740 1741 1742 1743

	return 0;
}
EXPORT_SYMBOL_GPL(__dev_forward_skb);

1744 1745 1746 1747 1748 1749 1750 1751
/**
 * dev_forward_skb - loopback an skb to another netif
 *
 * @dev: destination network device
 * @skb: buffer to forward
 *
 * return values:
 *	NET_RX_SUCCESS	(no congestion)
1752
 *	NET_RX_DROP     (packet was dropped, but freed)
1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
 *
 * dev_forward_skb can be used for injecting an skb from the
 * start_xmit function of one device into the receive queue
 * of another device.
 *
 * The receiving device may be in another namespace, so
 * we have to clear all information in the skb that could
 * impact namespace isolation.
 */
int dev_forward_skb(struct net_device *dev, struct sk_buff *skb)
{
H
Herbert Xu 已提交
1764
	return __dev_forward_skb(dev, skb) ?: netif_rx_internal(skb);
1765 1766 1767
}
EXPORT_SYMBOL_GPL(dev_forward_skb);

1768 1769 1770 1771
static inline int deliver_skb(struct sk_buff *skb,
			      struct packet_type *pt_prev,
			      struct net_device *orig_dev)
{
1772 1773
	if (unlikely(skb_orphan_frags(skb, GFP_ATOMIC)))
		return -ENOMEM;
1774 1775 1776 1777
	atomic_inc(&skb->users);
	return pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
}

1778 1779
static inline void deliver_ptype_list_skb(struct sk_buff *skb,
					  struct packet_type **pt,
1780 1781
					  struct net_device *orig_dev,
					  __be16 type,
1782 1783 1784 1785 1786 1787 1788 1789
					  struct list_head *ptype_list)
{
	struct packet_type *ptype, *pt_prev = *pt;

	list_for_each_entry_rcu(ptype, ptype_list, list) {
		if (ptype->type != type)
			continue;
		if (pt_prev)
1790
			deliver_skb(skb, pt_prev, orig_dev);
1791 1792 1793 1794 1795
		pt_prev = ptype;
	}
	*pt = pt_prev;
}

1796 1797
static inline bool skb_loop_sk(struct packet_type *ptype, struct sk_buff *skb)
{
1798
	if (!ptype->af_packet_priv || !skb->sk)
1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
		return false;

	if (ptype->id_match)
		return ptype->id_match(ptype, skb->sk);
	else if ((struct sock *)ptype->af_packet_priv == skb->sk)
		return true;

	return false;
}

L
Linus Torvalds 已提交
1809 1810 1811 1812 1813
/*
 *	Support routine. Sends outgoing frames to any network
 *	taps currently in use.
 */

1814
static void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
L
Linus Torvalds 已提交
1815 1816
{
	struct packet_type *ptype;
1817 1818
	struct sk_buff *skb2 = NULL;
	struct packet_type *pt_prev = NULL;
1819
	struct list_head *ptype_list = &ptype_all;
1820

L
Linus Torvalds 已提交
1821
	rcu_read_lock();
1822 1823
again:
	list_for_each_entry_rcu(ptype, ptype_list, list) {
L
Linus Torvalds 已提交
1824 1825 1826
		/* Never send packets back to the socket
		 * they originated from - MvS (miquels@drinkel.ow.org)
		 */
1827 1828
		if (skb_loop_sk(ptype, skb))
			continue;
1829

1830 1831 1832 1833 1834
		if (pt_prev) {
			deliver_skb(skb2, pt_prev, skb->dev);
			pt_prev = ptype;
			continue;
		}
L
Linus Torvalds 已提交
1835

1836 1837 1838 1839
		/* need to clone skb, done only once */
		skb2 = skb_clone(skb, GFP_ATOMIC);
		if (!skb2)
			goto out_unlock;
1840

1841
		net_timestamp_set(skb2);
L
Linus Torvalds 已提交
1842

1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
		/* skb->nh should be correctly
		 * set by sender, so that the second statement is
		 * just protection against buggy protocols.
		 */
		skb_reset_mac_header(skb2);

		if (skb_network_header(skb2) < skb2->data ||
		    skb_network_header(skb2) > skb_tail_pointer(skb2)) {
			net_crit_ratelimited("protocol %04x is buggy, dev %s\n",
					     ntohs(skb2->protocol),
					     dev->name);
			skb_reset_network_header(skb2);
L
Linus Torvalds 已提交
1855
		}
1856 1857 1858 1859 1860 1861 1862 1863 1864

		skb2->transport_header = skb2->network_header;
		skb2->pkt_type = PACKET_OUTGOING;
		pt_prev = ptype;
	}

	if (ptype_list == &ptype_all) {
		ptype_list = &dev->ptype_all;
		goto again;
L
Linus Torvalds 已提交
1865
	}
1866
out_unlock:
1867 1868
	if (pt_prev)
		pt_prev->func(skb2, skb->dev, pt_prev, skb->dev);
L
Linus Torvalds 已提交
1869 1870 1871
	rcu_read_unlock();
}

1872 1873
/**
 * netif_setup_tc - Handle tc mappings on real_num_tx_queues change
1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884
 * @dev: Network device
 * @txq: number of queues available
 *
 * If real_num_tx_queues is changed the tc mappings may no longer be
 * valid. To resolve this verify the tc mapping remains valid and if
 * not NULL the mapping. With no priorities mapping to this
 * offset/count pair it will no longer be used. In the worst case TC0
 * is invalid nothing can be done so disable priority mappings. If is
 * expected that drivers will fix this mapping if they can before
 * calling netif_set_real_num_tx_queues.
 */
E
Eric Dumazet 已提交
1885
static void netif_setup_tc(struct net_device *dev, unsigned int txq)
1886 1887 1888 1889 1890 1891
{
	int i;
	struct netdev_tc_txq *tc = &dev->tc_to_txq[0];

	/* If TC0 is invalidated disable TC mapping */
	if (tc->offset + tc->count > txq) {
1892
		pr_warn("Number of in use tx queues changed invalidating tc mappings. Priority traffic classification disabled!\n");
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902
		dev->num_tc = 0;
		return;
	}

	/* Invalidated prio to tc mappings set to TC0 */
	for (i = 1; i < TC_BITMASK + 1; i++) {
		int q = netdev_get_prio_tc_map(dev, i);

		tc = &dev->tc_to_txq[q];
		if (tc->offset + tc->count > txq) {
1903 1904
			pr_warn("Number of in use tx queues changed. Priority %i to tc mapping %i is no longer valid. Setting map to 0\n",
				i, q);
1905 1906 1907 1908 1909
			netdev_set_prio_tc_map(dev, i, 0);
		}
	}
}

1910 1911 1912 1913 1914
#ifdef CONFIG_XPS
static DEFINE_MUTEX(xps_map_mutex);
#define xmap_dereference(P)		\
	rcu_dereference_protected((P), lockdep_is_held(&xps_map_mutex))

1915 1916
static struct xps_map *remove_xps_queue(struct xps_dev_maps *dev_maps,
					int cpu, u16 index)
1917
{
1918 1919
	struct xps_map *map = NULL;
	int pos;
1920

1921 1922
	if (dev_maps)
		map = xmap_dereference(dev_maps->cpu_map[cpu]);
1923

1924 1925
	for (pos = 0; map && pos < map->len; pos++) {
		if (map->queues[pos] == index) {
1926 1927 1928
			if (map->len > 1) {
				map->queues[pos] = map->queues[--map->len];
			} else {
1929
				RCU_INIT_POINTER(dev_maps->cpu_map[cpu], NULL);
1930 1931 1932
				kfree_rcu(map, rcu);
				map = NULL;
			}
1933
			break;
1934 1935 1936
		}
	}

1937 1938 1939
	return map;
}

1940
static void netif_reset_xps_queues_gt(struct net_device *dev, u16 index)
1941 1942
{
	struct xps_dev_maps *dev_maps;
1943
	int cpu, i;
1944 1945 1946 1947 1948 1949 1950 1951 1952
	bool active = false;

	mutex_lock(&xps_map_mutex);
	dev_maps = xmap_dereference(dev->xps_maps);

	if (!dev_maps)
		goto out_no_maps;

	for_each_possible_cpu(cpu) {
1953 1954 1955 1956 1957
		for (i = index; i < dev->num_tx_queues; i++) {
			if (!remove_xps_queue(dev_maps, cpu, i))
				break;
		}
		if (i == dev->num_tx_queues)
1958 1959 1960 1961
			active = true;
	}

	if (!active) {
1962 1963 1964 1965
		RCU_INIT_POINTER(dev->xps_maps, NULL);
		kfree_rcu(dev_maps, rcu);
	}

1966 1967 1968 1969
	for (i = index; i < dev->num_tx_queues; i++)
		netdev_queue_numa_node_write(netdev_get_tx_queue(dev, i),
					     NUMA_NO_NODE);

1970 1971 1972 1973
out_no_maps:
	mutex_unlock(&xps_map_mutex);
}

1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
static struct xps_map *expand_xps_map(struct xps_map *map,
				      int cpu, u16 index)
{
	struct xps_map *new_map;
	int alloc_len = XPS_MIN_MAP_ALLOC;
	int i, pos;

	for (pos = 0; map && pos < map->len; pos++) {
		if (map->queues[pos] != index)
			continue;
		return map;
	}

	/* Need to add queue to this CPU's existing map */
	if (map) {
		if (pos < map->alloc_len)
			return map;

		alloc_len = map->alloc_len * 2;
	}

	/* Need to allocate new map to store queue on this CPU's map */
	new_map = kzalloc_node(XPS_MAP_SIZE(alloc_len), GFP_KERNEL,
			       cpu_to_node(cpu));
	if (!new_map)
		return NULL;

	for (i = 0; i < pos; i++)
		new_map->queues[i] = map->queues[i];
	new_map->alloc_len = alloc_len;
	new_map->len = pos;

	return new_map;
}

2009 2010
int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
			u16 index)
2011
{
2012
	struct xps_dev_maps *dev_maps, *new_dev_maps = NULL;
2013 2014
	struct xps_map *map, *new_map;
	int maps_sz = max_t(unsigned int, XPS_DEV_MAPS_SIZE, L1_CACHE_BYTES);
2015 2016
	int cpu, numa_node_id = -2;
	bool active = false;
2017 2018 2019 2020 2021

	mutex_lock(&xps_map_mutex);

	dev_maps = xmap_dereference(dev->xps_maps);

2022 2023 2024 2025 2026 2027 2028
	/* allocate memory for queue storage */
	for_each_online_cpu(cpu) {
		if (!cpumask_test_cpu(cpu, mask))
			continue;

		if (!new_dev_maps)
			new_dev_maps = kzalloc(maps_sz, GFP_KERNEL);
2029 2030
		if (!new_dev_maps) {
			mutex_unlock(&xps_map_mutex);
2031
			return -ENOMEM;
2032
		}
2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046

		map = dev_maps ? xmap_dereference(dev_maps->cpu_map[cpu]) :
				 NULL;

		map = expand_xps_map(map, cpu, index);
		if (!map)
			goto error;

		RCU_INIT_POINTER(new_dev_maps->cpu_map[cpu], map);
	}

	if (!new_dev_maps)
		goto out_no_new_maps;

2047
	for_each_possible_cpu(cpu) {
2048 2049 2050 2051 2052 2053 2054 2055 2056 2057
		if (cpumask_test_cpu(cpu, mask) && cpu_online(cpu)) {
			/* add queue to CPU maps */
			int pos = 0;

			map = xmap_dereference(new_dev_maps->cpu_map[cpu]);
			while ((pos < map->len) && (map->queues[pos] != index))
				pos++;

			if (pos == map->len)
				map->queues[map->len++] = index;
2058 2059 2060 2061 2062 2063
#ifdef CONFIG_NUMA
			if (numa_node_id == -2)
				numa_node_id = cpu_to_node(cpu);
			else if (numa_node_id != cpu_to_node(cpu))
				numa_node_id = -1;
#endif
2064 2065 2066 2067
		} else if (dev_maps) {
			/* fill in the new device map from the old device map */
			map = xmap_dereference(dev_maps->cpu_map[cpu]);
			RCU_INIT_POINTER(new_dev_maps->cpu_map[cpu], map);
2068
		}
2069

2070 2071
	}

2072 2073
	rcu_assign_pointer(dev->xps_maps, new_dev_maps);

2074
	/* Cleanup old maps */
2075 2076 2077 2078 2079 2080 2081
	if (dev_maps) {
		for_each_possible_cpu(cpu) {
			new_map = xmap_dereference(new_dev_maps->cpu_map[cpu]);
			map = xmap_dereference(dev_maps->cpu_map[cpu]);
			if (map && map != new_map)
				kfree_rcu(map, rcu);
		}
2082

2083
		kfree_rcu(dev_maps, rcu);
2084 2085
	}

2086 2087
	dev_maps = new_dev_maps;
	active = true;
2088

2089 2090
out_no_new_maps:
	/* update Tx queue numa node */
2091 2092 2093 2094
	netdev_queue_numa_node_write(netdev_get_tx_queue(dev, index),
				     (numa_node_id >= 0) ? numa_node_id :
				     NUMA_NO_NODE);

2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113
	if (!dev_maps)
		goto out_no_maps;

	/* removes queue from unused CPUs */
	for_each_possible_cpu(cpu) {
		if (cpumask_test_cpu(cpu, mask) && cpu_online(cpu))
			continue;

		if (remove_xps_queue(dev_maps, cpu, index))
			active = true;
	}

	/* free map if not active */
	if (!active) {
		RCU_INIT_POINTER(dev->xps_maps, NULL);
		kfree_rcu(dev_maps, rcu);
	}

out_no_maps:
2114 2115 2116 2117
	mutex_unlock(&xps_map_mutex);

	return 0;
error:
2118 2119 2120 2121 2122 2123 2124 2125 2126
	/* remove any maps that we added */
	for_each_possible_cpu(cpu) {
		new_map = xmap_dereference(new_dev_maps->cpu_map[cpu]);
		map = dev_maps ? xmap_dereference(dev_maps->cpu_map[cpu]) :
				 NULL;
		if (new_map && new_map != map)
			kfree(new_map);
	}

2127 2128 2129 2130 2131 2132 2133 2134
	mutex_unlock(&xps_map_mutex);

	kfree(new_dev_maps);
	return -ENOMEM;
}
EXPORT_SYMBOL(netif_set_xps_queue);

#endif
2135 2136 2137 2138
/*
 * Routine to help set real_num_tx_queues. To avoid skbs mapped to queues
 * greater then real_num_tx_queues stale skbs on the qdisc must be flushed.
 */
2139
int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq)
2140
{
T
Tom Herbert 已提交
2141 2142
	int rc;

2143 2144
	if (txq < 1 || txq > dev->num_tx_queues)
		return -EINVAL;
2145

2146 2147
	if (dev->reg_state == NETREG_REGISTERED ||
	    dev->reg_state == NETREG_UNREGISTERING) {
2148 2149
		ASSERT_RTNL();

T
Tom Herbert 已提交
2150 2151
		rc = netdev_queue_update_kobjects(dev, dev->real_num_tx_queues,
						  txq);
T
Tom Herbert 已提交
2152 2153 2154
		if (rc)
			return rc;

2155 2156 2157
		if (dev->num_tc)
			netif_setup_tc(dev, txq);

2158
		if (txq < dev->real_num_tx_queues) {
2159
			qdisc_reset_all_tx_gt(dev, txq);
2160 2161 2162 2163
#ifdef CONFIG_XPS
			netif_reset_xps_queues_gt(dev, txq);
#endif
		}
2164
	}
2165 2166 2167

	dev->real_num_tx_queues = txq;
	return 0;
2168 2169
}
EXPORT_SYMBOL(netif_set_real_num_tx_queues);
2170

2171
#ifdef CONFIG_SYSFS
2172 2173 2174 2175 2176 2177 2178
/**
 *	netif_set_real_num_rx_queues - set actual number of RX queues used
 *	@dev: Network device
 *	@rxq: Actual number of RX queues
 *
 *	This must be called either with the rtnl_lock held or before
 *	registration of the net device.  Returns 0 on success, or a
2179 2180
 *	negative error code.  If called before registration, it always
 *	succeeds.
2181 2182 2183 2184 2185
 */
int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
{
	int rc;

T
Tom Herbert 已提交
2186 2187 2188
	if (rxq < 1 || rxq > dev->num_rx_queues)
		return -EINVAL;

2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203
	if (dev->reg_state == NETREG_REGISTERED) {
		ASSERT_RTNL();

		rc = net_rx_queue_update_kobjects(dev, dev->real_num_rx_queues,
						  rxq);
		if (rc)
			return rc;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}
EXPORT_SYMBOL(netif_set_real_num_rx_queues);
#endif

2204 2205
/**
 * netif_get_num_default_rss_queues - default number of RSS queues
2206 2207 2208 2209
 *
 * This routine should set an upper limit on the number of RSS queues
 * used by default by multiqueue devices.
 */
2210
int netif_get_num_default_rss_queues(void)
2211 2212 2213 2214 2215
{
	return min_t(int, DEFAULT_MAX_NUM_RSS_QUEUES, num_online_cpus());
}
EXPORT_SYMBOL(netif_get_num_default_rss_queues);

2216
static inline void __netif_reschedule(struct Qdisc *q)
2217
{
2218 2219
	struct softnet_data *sd;
	unsigned long flags;
2220

2221
	local_irq_save(flags);
2222
	sd = this_cpu_ptr(&softnet_data);
2223 2224 2225
	q->next_sched = NULL;
	*sd->output_queue_tailp = q;
	sd->output_queue_tailp = &q->next_sched;
2226 2227 2228 2229 2230 2231 2232 2233
	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_restore(flags);
}

void __netif_schedule(struct Qdisc *q)
{
	if (!test_and_set_bit(__QDISC_STATE_SCHED, &q->state))
		__netif_reschedule(q);
2234 2235 2236
}
EXPORT_SYMBOL(__netif_schedule);

2237 2238 2239 2240 2241
struct dev_kfree_skb_cb {
	enum skb_free_reason reason;
};

static struct dev_kfree_skb_cb *get_kfree_skb_cb(const struct sk_buff *skb)
2242
{
2243 2244 2245
	return (struct dev_kfree_skb_cb *)skb->cb;
}

2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
void netif_schedule_queue(struct netdev_queue *txq)
{
	rcu_read_lock();
	if (!(txq->state & QUEUE_STATE_ANY_XOFF)) {
		struct Qdisc *q = rcu_dereference(txq->qdisc);

		__netif_schedule(q);
	}
	rcu_read_unlock();
}
EXPORT_SYMBOL(netif_schedule_queue);

/**
 *	netif_wake_subqueue - allow sending packets on subqueue
 *	@dev: network device
 *	@queue_index: sub queue index
 *
 * Resume individual transmit queue of a device with multiple transmit queues.
 */
void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
{
	struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);

	if (test_and_clear_bit(__QUEUE_STATE_DRV_XOFF, &txq->state)) {
		struct Qdisc *q;

		rcu_read_lock();
		q = rcu_dereference(txq->qdisc);
		__netif_schedule(q);
		rcu_read_unlock();
	}
}
EXPORT_SYMBOL(netif_wake_subqueue);

void netif_tx_wake_queue(struct netdev_queue *dev_queue)
{
	if (test_and_clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state)) {
		struct Qdisc *q;

		rcu_read_lock();
		q = rcu_dereference(dev_queue->qdisc);
		__netif_schedule(q);
		rcu_read_unlock();
	}
}
EXPORT_SYMBOL(netif_tx_wake_queue);

2293
void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason)
2294
{
2295
	unsigned long flags;
2296

2297 2298 2299 2300 2301
	if (likely(atomic_read(&skb->users) == 1)) {
		smp_rmb();
		atomic_set(&skb->users, 0);
	} else if (likely(!atomic_dec_and_test(&skb->users))) {
		return;
2302
	}
2303 2304 2305 2306 2307 2308
	get_kfree_skb_cb(skb)->reason = reason;
	local_irq_save(flags);
	skb->next = __this_cpu_read(softnet_data.completion_queue);
	__this_cpu_write(softnet_data.completion_queue, skb);
	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_restore(flags);
2309
}
2310
EXPORT_SYMBOL(__dev_kfree_skb_irq);
2311

2312
void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason)
2313 2314
{
	if (in_irq() || irqs_disabled())
2315
		__dev_kfree_skb_irq(skb, reason);
2316 2317 2318
	else
		dev_kfree_skb(skb);
}
2319
EXPORT_SYMBOL(__dev_kfree_skb_any);
2320 2321


2322 2323 2324 2325 2326 2327
/**
 * netif_device_detach - mark device as removed
 * @dev: network device
 *
 * Mark device as removed from system and therefore no longer available.
 */
2328 2329 2330 2331
void netif_device_detach(struct net_device *dev)
{
	if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
2332
		netif_tx_stop_all_queues(dev);
2333 2334 2335 2336
	}
}
EXPORT_SYMBOL(netif_device_detach);

2337 2338 2339 2340 2341 2342
/**
 * netif_device_attach - mark device as attached
 * @dev: network device
 *
 * Mark device as attached from system and restart if needed.
 */
2343 2344 2345 2346
void netif_device_attach(struct net_device *dev)
{
	if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
2347
		netif_tx_wake_all_queues(dev);
2348
		__netdev_watchdog_up(dev);
2349 2350 2351 2352
	}
}
EXPORT_SYMBOL(netif_device_attach);

J
Jiri Pirko 已提交
2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380
/*
 * Returns a Tx hash based on the given packet descriptor a Tx queues' number
 * to be used as a distribution range.
 */
u16 __skb_tx_hash(const struct net_device *dev, struct sk_buff *skb,
		  unsigned int num_tx_queues)
{
	u32 hash;
	u16 qoffset = 0;
	u16 qcount = num_tx_queues;

	if (skb_rx_queue_recorded(skb)) {
		hash = skb_get_rx_queue(skb);
		while (unlikely(hash >= num_tx_queues))
			hash -= num_tx_queues;
		return hash;
	}

	if (dev->num_tc) {
		u8 tc = netdev_get_prio_tc_map(dev, skb->priority);
		qoffset = dev->tc_to_txq[tc].offset;
		qcount = dev->tc_to_txq[tc].count;
	}

	return (u16) reciprocal_scale(skb_get_hash(skb), qcount) + qoffset;
}
EXPORT_SYMBOL(__skb_tx_hash);

2381 2382
static void skb_warn_bad_offload(const struct sk_buff *skb)
{
2383
	static const netdev_features_t null_features = 0;
2384 2385 2386
	struct net_device *dev = skb->dev;
	const char *driver = "";

2387 2388 2389
	if (!net_ratelimit())
		return;

2390 2391 2392 2393 2394
	if (dev && dev->dev.parent)
		driver = dev_driver_string(dev->dev.parent);

	WARN(1, "%s: caps=(%pNF, %pNF) len=%d data_len=%d gso_size=%d "
	     "gso_type=%d ip_summed=%d\n",
2395 2396
	     driver, dev ? &dev->features : &null_features,
	     skb->sk ? &skb->sk->sk_route_caps : &null_features,
2397 2398 2399 2400
	     skb->len, skb->data_len, skb_shinfo(skb)->gso_size,
	     skb_shinfo(skb)->gso_type, skb->ip_summed);
}

L
Linus Torvalds 已提交
2401 2402 2403 2404
/*
 * Invalidate hardware checksum when packet is to be mangled, and
 * complete checksum manually on outgoing path.
 */
2405
int skb_checksum_help(struct sk_buff *skb)
L
Linus Torvalds 已提交
2406
{
2407
	__wsum csum;
2408
	int ret = 0, offset;
L
Linus Torvalds 已提交
2409

2410
	if (skb->ip_summed == CHECKSUM_COMPLETE)
2411 2412 2413
		goto out_set_summed;

	if (unlikely(skb_shinfo(skb)->gso_size)) {
2414 2415
		skb_warn_bad_offload(skb);
		return -EINVAL;
L
Linus Torvalds 已提交
2416 2417
	}

2418 2419 2420 2421 2422 2423 2424 2425 2426
	/* Before computing a checksum, we should make sure no frag could
	 * be modified by an external entity : checksum could be wrong.
	 */
	if (skb_has_shared_frag(skb)) {
		ret = __skb_linearize(skb);
		if (ret)
			goto out;
	}

2427
	offset = skb_checksum_start_offset(skb);
2428 2429 2430 2431 2432 2433 2434 2435
	BUG_ON(offset >= skb_headlen(skb));
	csum = skb_checksum(skb, offset, skb->len - offset, 0);

	offset += skb->csum_offset;
	BUG_ON(offset + sizeof(__sum16) > skb_headlen(skb));

	if (skb_cloned(skb) &&
	    !skb_clone_writable(skb, offset + sizeof(__sum16))) {
L
Linus Torvalds 已提交
2436 2437 2438 2439 2440
		ret = pskb_expand_head(skb, 0, 0, GFP_ATOMIC);
		if (ret)
			goto out;
	}

2441
	*(__sum16 *)(skb->data + offset) = csum_fold(csum);
2442
out_set_summed:
L
Linus Torvalds 已提交
2443
	skb->ip_summed = CHECKSUM_NONE;
2444
out:
L
Linus Torvalds 已提交
2445 2446
	return ret;
}
E
Eric Dumazet 已提交
2447
EXPORT_SYMBOL(skb_checksum_help);
L
Linus Torvalds 已提交
2448

2449
__be16 skb_network_protocol(struct sk_buff *skb, int *depth)
2450
{
A
Al Viro 已提交
2451
	__be16 type = skb->protocol;
2452

2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463
	/* Tunnel gso handlers can set protocol to ethernet. */
	if (type == htons(ETH_P_TEB)) {
		struct ethhdr *eth;

		if (unlikely(!pskb_may_pull(skb, sizeof(struct ethhdr))))
			return 0;

		eth = (struct ethhdr *)skb_mac_header(skb);
		type = eth->h_proto;
	}

2464
	return __vlan_get_protocol(skb, type, depth);
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476
}

/**
 *	skb_mac_gso_segment - mac layer segmentation handler.
 *	@skb: buffer to segment
 *	@features: features for the output path (see dev->features)
 */
struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
				    netdev_features_t features)
{
	struct sk_buff *segs = ERR_PTR(-EPROTONOSUPPORT);
	struct packet_offload *ptype;
2477 2478
	int vlan_depth = skb->mac_len;
	__be16 type = skb_network_protocol(skb, &vlan_depth);
2479 2480 2481 2482

	if (unlikely(!type))
		return ERR_PTR(-EINVAL);

2483
	__skb_pull(skb, vlan_depth);
2484 2485

	rcu_read_lock();
2486
	list_for_each_entry_rcu(ptype, &offload_base, list) {
2487 2488
		if (ptype->type == type && ptype->callbacks.gso_segment) {
			segs = ptype->callbacks.gso_segment(skb, features);
2489 2490 2491 2492 2493
			break;
		}
	}
	rcu_read_unlock();

2494
	__skb_push(skb, skb->data - skb_mac_header(skb));
2495

2496 2497
	return segs;
}
2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529
EXPORT_SYMBOL(skb_mac_gso_segment);


/* openvswitch calls this on rx path, so we need a different check.
 */
static inline bool skb_needs_check(struct sk_buff *skb, bool tx_path)
{
	if (tx_path)
		return skb->ip_summed != CHECKSUM_PARTIAL;
	else
		return skb->ip_summed == CHECKSUM_NONE;
}

/**
 *	__skb_gso_segment - Perform segmentation on skb.
 *	@skb: buffer to segment
 *	@features: features for the output path (see dev->features)
 *	@tx_path: whether it is called in TX path
 *
 *	This function segments the given skb and returns a list of segments.
 *
 *	It may return NULL if the skb requires no segmentation.  This is
 *	only possible when GSO is used for verifying header integrity.
 */
struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
				  netdev_features_t features, bool tx_path)
{
	if (unlikely(skb_needs_check(skb, tx_path))) {
		int err;

		skb_warn_bad_offload(skb);

2530 2531
		err = skb_cow_head(skb, 0);
		if (err < 0)
2532 2533 2534
			return ERR_PTR(err);
	}

2535
	SKB_GSO_CB(skb)->mac_offset = skb_headroom(skb);
2536 2537
	SKB_GSO_CB(skb)->encap_level = 0;

2538 2539 2540 2541 2542
	skb_reset_mac_header(skb);
	skb_reset_mac_len(skb);

	return skb_mac_gso_segment(skb, features);
}
2543
EXPORT_SYMBOL(__skb_gso_segment);
2544

2545 2546 2547 2548 2549
/* Take action when hardware reception checksum errors are detected. */
#ifdef CONFIG_BUG
void netdev_rx_csum_fault(struct net_device *dev)
{
	if (net_ratelimit()) {
2550
		pr_err("%s: hw csum failure\n", dev ? dev->name : "<unknown>");
2551 2552 2553 2554 2555 2556
		dump_stack();
	}
}
EXPORT_SYMBOL(netdev_rx_csum_fault);
#endif

L
Linus Torvalds 已提交
2557 2558 2559 2560 2561
/* Actually, we should eliminate this check as soon as we know, that:
 * 1. IOMMU is present and allows to map all the memory.
 * 2. No high memory really exists on this machine.
 */

2562
static int illegal_highdma(struct net_device *dev, struct sk_buff *skb)
L
Linus Torvalds 已提交
2563
{
2564
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
2565
	int i;
2566
	if (!(dev->features & NETIF_F_HIGHDMA)) {
2567 2568 2569
		for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			if (PageHighMem(skb_frag_page(frag)))
2570
				return 1;
2571
		}
2572
	}
L
Linus Torvalds 已提交
2573

2574 2575
	if (PCI_DMA_BUS_IS_PHYS) {
		struct device *pdev = dev->dev.parent;
L
Linus Torvalds 已提交
2576

E
Eric Dumazet 已提交
2577 2578
		if (!pdev)
			return 0;
2579
		for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
2580 2581
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			dma_addr_t addr = page_to_phys(skb_frag_page(frag));
2582 2583 2584 2585
			if (!pdev->dma_mask || addr + PAGE_SIZE - 1 > *pdev->dma_mask)
				return 1;
		}
	}
2586
#endif
L
Linus Torvalds 已提交
2587 2588 2589
	return 0;
}

2590 2591 2592
/* If MPLS offload request, verify we are testing hardware MPLS features
 * instead of standard features for the netdev.
 */
P
Pravin B Shelar 已提交
2593
#if IS_ENABLED(CONFIG_NET_MPLS_GSO)
2594 2595 2596 2597
static netdev_features_t net_mpls_features(struct sk_buff *skb,
					   netdev_features_t features,
					   __be16 type)
{
2598
	if (eth_p_mpls(type))
2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611
		features &= skb->dev->mpls_features;

	return features;
}
#else
static netdev_features_t net_mpls_features(struct sk_buff *skb,
					   netdev_features_t features,
					   __be16 type)
{
	return features;
}
#endif

2612
static netdev_features_t harmonize_features(struct sk_buff *skb,
2613
	netdev_features_t features)
2614
{
2615
	int tmp;
2616 2617 2618 2619
	__be16 type;

	type = skb_network_protocol(skb, &tmp);
	features = net_mpls_features(skb, features, type);
2620

2621
	if (skb->ip_summed != CHECKSUM_NONE &&
2622
	    !can_checksum_protocol(features, type)) {
2623
		features &= ~NETIF_F_ALL_CSUM;
2624
	} else if (illegal_highdma(skb->dev, skb)) {
2625 2626 2627 2628 2629 2630
		features &= ~NETIF_F_SG;
	}

	return features;
}

2631 2632 2633 2634 2635 2636 2637 2638
netdev_features_t passthru_features_check(struct sk_buff *skb,
					  struct net_device *dev,
					  netdev_features_t features)
{
	return features;
}
EXPORT_SYMBOL(passthru_features_check);

2639 2640 2641 2642 2643 2644 2645
static netdev_features_t dflt_features_check(const struct sk_buff *skb,
					     struct net_device *dev,
					     netdev_features_t features)
{
	return vlan_features_check(skb, features);
}

2646
netdev_features_t netif_skb_features(struct sk_buff *skb)
2647
{
2648
	struct net_device *dev = skb->dev;
2649 2650
	netdev_features_t features = dev->features;
	u16 gso_segs = skb_shinfo(skb)->gso_segs;
2651

2652
	if (gso_segs > dev->gso_max_segs || gso_segs < dev->gso_min_segs)
2653 2654
		features &= ~NETIF_F_GSO_MASK;

2655 2656 2657 2658 2659 2660 2661
	/* If encapsulation offload request, verify we are testing
	 * hardware encapsulation features instead of standard
	 * features for the netdev
	 */
	if (skb->encapsulation)
		features &= dev->hw_enc_features;

2662 2663 2664 2665 2666
	if (skb_vlan_tagged(skb))
		features = netdev_intersect_features(features,
						     dev->vlan_features |
						     NETIF_F_HW_VLAN_CTAG_TX |
						     NETIF_F_HW_VLAN_STAG_TX);
2667

2668 2669 2670
	if (dev->netdev_ops->ndo_features_check)
		features &= dev->netdev_ops->ndo_features_check(skb, dev,
								features);
2671 2672
	else
		features &= dflt_features_check(skb, dev, features);
2673

2674
	return harmonize_features(skb, features);
2675
}
2676
EXPORT_SYMBOL(netif_skb_features);
2677

2678
static int xmit_one(struct sk_buff *skb, struct net_device *dev,
2679
		    struct netdev_queue *txq, bool more)
2680
{
2681 2682
	unsigned int len;
	int rc;
2683

2684
	if (!list_empty(&ptype_all) || !list_empty(&dev->ptype_all))
2685
		dev_queue_xmit_nit(skb, dev);
2686

2687 2688
	len = skb->len;
	trace_net_dev_start_xmit(skb, dev);
2689
	rc = netdev_start_xmit(skb, dev, txq, more);
2690
	trace_net_dev_xmit(skb, rc, dev, len);
E
Eric Dumazet 已提交
2691

2692 2693
	return rc;
}
2694

2695 2696
struct sk_buff *dev_hard_start_xmit(struct sk_buff *first, struct net_device *dev,
				    struct netdev_queue *txq, int *ret)
2697 2698 2699
{
	struct sk_buff *skb = first;
	int rc = NETDEV_TX_OK;
2700

2701 2702
	while (skb) {
		struct sk_buff *next = skb->next;
2703

2704
		skb->next = NULL;
2705
		rc = xmit_one(skb, dev, txq, next != NULL);
2706 2707 2708 2709
		if (unlikely(!dev_xmit_complete(rc))) {
			skb->next = next;
			goto out;
		}
2710

2711 2712 2713 2714
		skb = next;
		if (netif_xmit_stopped(txq) && skb) {
			rc = NETDEV_TX_BUSY;
			break;
2715
		}
2716
	}
2717

2718 2719 2720 2721
out:
	*ret = rc;
	return skb;
}
2722

E
Eric Dumazet 已提交
2723 2724
static struct sk_buff *validate_xmit_vlan(struct sk_buff *skb,
					  netdev_features_t features)
2725
{
2726
	if (skb_vlan_tag_present(skb) &&
2727 2728
	    !vlan_hw_offload_capable(features, skb->vlan_proto))
		skb = __vlan_hwaccel_push_inside(skb);
2729 2730
	return skb;
}
2731

2732
static struct sk_buff *validate_xmit_skb(struct sk_buff *skb, struct net_device *dev)
2733 2734
{
	netdev_features_t features;
2735

2736 2737
	if (skb->next)
		return skb;
2738

2739 2740 2741 2742
	features = netif_skb_features(skb);
	skb = validate_xmit_vlan(skb, features);
	if (unlikely(!skb))
		goto out_null;
2743

2744
	if (netif_needs_gso(skb, features)) {
2745 2746 2747
		struct sk_buff *segs;

		segs = skb_gso_segment(skb, features);
2748
		if (IS_ERR(segs)) {
2749
			goto out_kfree_skb;
2750 2751 2752
		} else if (segs) {
			consume_skb(skb);
			skb = segs;
2753
		}
2754 2755 2756 2757
	} else {
		if (skb_needs_linearize(skb, features) &&
		    __skb_linearize(skb))
			goto out_kfree_skb;
2758

2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772
		/* If packet is not checksummed and device does not
		 * support checksumming for this protocol, complete
		 * checksumming here.
		 */
		if (skb->ip_summed == CHECKSUM_PARTIAL) {
			if (skb->encapsulation)
				skb_set_inner_transport_header(skb,
							       skb_checksum_start_offset(skb));
			else
				skb_set_transport_header(skb,
							 skb_checksum_start_offset(skb));
			if (!(features & NETIF_F_ALL_CSUM) &&
			    skb_checksum_help(skb))
				goto out_kfree_skb;
2773
		}
2774
	}
2775

2776
	return skb;
2777

2778 2779
out_kfree_skb:
	kfree_skb(skb);
2780 2781 2782
out_null:
	return NULL;
}
2783

2784 2785 2786 2787
struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev)
{
	struct sk_buff *next, *head = NULL, *tail;

2788
	for (; skb != NULL; skb = next) {
2789 2790
		next = skb->next;
		skb->next = NULL;
2791 2792 2793 2794

		/* in case skb wont be segmented, point to itself */
		skb->prev = skb;

2795
		skb = validate_xmit_skb(skb, dev);
2796 2797
		if (!skb)
			continue;
2798

2799 2800 2801 2802 2803 2804 2805 2806
		if (!head)
			head = skb;
		else
			tail->next = skb;
		/* If skb was segmented, skb->prev points to
		 * the last segment. If not, it still contains skb.
		 */
		tail = skb->prev;
2807 2808
	}
	return head;
2809 2810
}

2811 2812 2813 2814 2815 2816 2817 2818 2819 2820
static void qdisc_pkt_len_init(struct sk_buff *skb)
{
	const struct skb_shared_info *shinfo = skb_shinfo(skb);

	qdisc_skb_cb(skb)->pkt_len = skb->len;

	/* To get more precise estimation of bytes sent on wire,
	 * we add to pkt_len the headers size of all segments
	 */
	if (shinfo->gso_size)  {
2821
		unsigned int hdr_len;
2822
		u16 gso_segs = shinfo->gso_segs;
2823

2824 2825 2826 2827
		/* mac layer + network layer */
		hdr_len = skb_transport_header(skb) - skb_mac_header(skb);

		/* + transport layer */
2828 2829 2830 2831
		if (likely(shinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6)))
			hdr_len += tcp_hdrlen(skb);
		else
			hdr_len += sizeof(struct udphdr);
2832 2833 2834 2835 2836 2837

		if (shinfo->gso_type & SKB_GSO_DODGY)
			gso_segs = DIV_ROUND_UP(skb->len - hdr_len,
						shinfo->gso_size);

		qdisc_skb_cb(skb)->pkt_len += (gso_segs - 1) * hdr_len;
2838 2839 2840
	}
}

2841 2842 2843 2844 2845
static inline int __dev_xmit_skb(struct sk_buff *skb, struct Qdisc *q,
				 struct net_device *dev,
				 struct netdev_queue *txq)
{
	spinlock_t *root_lock = qdisc_lock(q);
E
Eric Dumazet 已提交
2846
	bool contended;
2847 2848
	int rc;

2849
	qdisc_pkt_len_init(skb);
E
Eric Dumazet 已提交
2850
	qdisc_calculate_pkt_len(skb, q);
2851 2852 2853
	/*
	 * Heuristic to force contended enqueues to serialize on a
	 * separate lock before trying to get qdisc main lock.
Y
Ying Xue 已提交
2854 2855
	 * This permits __QDISC___STATE_RUNNING owner to get the lock more
	 * often and dequeue packets faster.
2856
	 */
E
Eric Dumazet 已提交
2857
	contended = qdisc_is_running(q);
2858 2859 2860
	if (unlikely(contended))
		spin_lock(&q->busylock);

2861 2862 2863 2864 2865
	spin_lock(root_lock);
	if (unlikely(test_bit(__QDISC_STATE_DEACTIVATED, &q->state))) {
		kfree_skb(skb);
		rc = NET_XMIT_DROP;
	} else if ((q->flags & TCQ_F_CAN_BYPASS) && !qdisc_qlen(q) &&
2866
		   qdisc_run_begin(q)) {
2867 2868 2869 2870 2871
		/*
		 * This is a work-conserving queue; there are no old skbs
		 * waiting to be sent out; and the qdisc is not running -
		 * xmit the skb directly.
		 */
2872 2873 2874

		qdisc_bstats_update(q, skb);

2875
		if (sch_direct_xmit(skb, q, dev, txq, root_lock, true)) {
2876 2877 2878 2879
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
2880
			__qdisc_run(q);
2881
		} else
2882
			qdisc_run_end(q);
2883 2884 2885

		rc = NET_XMIT_SUCCESS;
	} else {
E
Eric Dumazet 已提交
2886
		rc = q->enqueue(skb, q) & NET_XMIT_MASK;
2887 2888 2889 2890 2891 2892 2893
		if (qdisc_run_begin(q)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
			__qdisc_run(q);
		}
2894 2895
	}
	spin_unlock(root_lock);
2896 2897
	if (unlikely(contended))
		spin_unlock(&q->busylock);
2898 2899 2900
	return rc;
}

2901
#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
2902 2903
static void skb_update_prio(struct sk_buff *skb)
{
I
Igor Maravic 已提交
2904
	struct netprio_map *map = rcu_dereference_bh(skb->dev->priomap);
2905

2906 2907 2908 2909 2910 2911
	if (!skb->priority && skb->sk && map) {
		unsigned int prioidx = skb->sk->sk_cgrp_prioidx;

		if (prioidx < map->priomap_len)
			skb->priority = map->priomap[prioidx];
	}
2912 2913 2914 2915 2916
}
#else
#define skb_update_prio(skb)
#endif

2917 2918 2919
DEFINE_PER_CPU(int, xmit_recursion);
EXPORT_SYMBOL(xmit_recursion);

2920
#define RECURSION_LIMIT 10
2921

2922 2923 2924 2925
/**
 *	dev_loopback_xmit - loop back @skb
 *	@skb: buffer to transmit
 */
2926
int dev_loopback_xmit(struct sock *sk, struct sk_buff *skb)
2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938
{
	skb_reset_mac_header(skb);
	__skb_pull(skb, skb_network_offset(skb));
	skb->pkt_type = PACKET_LOOPBACK;
	skb->ip_summed = CHECKSUM_UNNECESSARY;
	WARN_ON(!skb_dst(skb));
	skb_dst_force(skb);
	netif_rx_ni(skb);
	return 0;
}
EXPORT_SYMBOL(dev_loopback_xmit);

2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016
static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
{
#ifdef CONFIG_XPS
	struct xps_dev_maps *dev_maps;
	struct xps_map *map;
	int queue_index = -1;

	rcu_read_lock();
	dev_maps = rcu_dereference(dev->xps_maps);
	if (dev_maps) {
		map = rcu_dereference(
		    dev_maps->cpu_map[skb->sender_cpu - 1]);
		if (map) {
			if (map->len == 1)
				queue_index = map->queues[0];
			else
				queue_index = map->queues[reciprocal_scale(skb_get_hash(skb),
									   map->len)];
			if (unlikely(queue_index >= dev->real_num_tx_queues))
				queue_index = -1;
		}
	}
	rcu_read_unlock();

	return queue_index;
#else
	return -1;
#endif
}

static u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
	int queue_index = sk_tx_queue_get(sk);

	if (queue_index < 0 || skb->ooo_okay ||
	    queue_index >= dev->real_num_tx_queues) {
		int new_index = get_xps_queue(dev, skb);
		if (new_index < 0)
			new_index = skb_tx_hash(dev, skb);

		if (queue_index != new_index && sk &&
		    rcu_access_pointer(sk->sk_dst_cache))
			sk_tx_queue_set(sk, new_index);

		queue_index = new_index;
	}

	return queue_index;
}

struct netdev_queue *netdev_pick_tx(struct net_device *dev,
				    struct sk_buff *skb,
				    void *accel_priv)
{
	int queue_index = 0;

#ifdef CONFIG_XPS
	if (skb->sender_cpu == 0)
		skb->sender_cpu = raw_smp_processor_id() + 1;
#endif

	if (dev->real_num_tx_queues != 1) {
		const struct net_device_ops *ops = dev->netdev_ops;
		if (ops->ndo_select_queue)
			queue_index = ops->ndo_select_queue(dev, skb, accel_priv,
							    __netdev_pick_tx);
		else
			queue_index = __netdev_pick_tx(dev, skb);

		if (!accel_priv)
			queue_index = netdev_cap_txqueue(dev, queue_index);
	}

	skb_set_queue_mapping(skb, queue_index);
	return netdev_get_tx_queue(dev, queue_index);
}

3017
/**
3018
 *	__dev_queue_xmit - transmit a buffer
3019
 *	@skb: buffer to transmit
3020
 *	@accel_priv: private data used for L2 forwarding offload
3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042
 *
 *	Queue a buffer for transmission to a network device. The caller must
 *	have set the device and priority and built the buffer before calling
 *	this function. The function can be called from an interrupt.
 *
 *	A negative errno code is returned on a failure. A success does not
 *	guarantee the frame will be transmitted as it may be dropped due
 *	to congestion or traffic shaping.
 *
 * -----------------------------------------------------------------------------------
 *      I notice this method can also return errors from the queue disciplines,
 *      including NET_XMIT_DROP, which is a positive value.  So, errors can also
 *      be positive.
 *
 *      Regardless of the return value, the skb is consumed, so it is currently
 *      difficult to retry a send to this method.  (You can bump the ref count
 *      before sending to hold a reference for retry if you are careful.)
 *
 *      When calling this method, interrupts MUST be enabled.  This is because
 *      the BH enable code must have IRQs enabled so that it will not deadlock.
 *          --BLG
 */
3043
static int __dev_queue_xmit(struct sk_buff *skb, void *accel_priv)
L
Linus Torvalds 已提交
3044 3045
{
	struct net_device *dev = skb->dev;
3046
	struct netdev_queue *txq;
L
Linus Torvalds 已提交
3047 3048 3049
	struct Qdisc *q;
	int rc = -ENOMEM;

3050 3051
	skb_reset_mac_header(skb);

3052 3053 3054
	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_SCHED_TSTAMP))
		__skb_tstamp_tx(skb, NULL, skb->sk, SCM_TSTAMP_SCHED);

3055 3056
	/* Disable soft irqs for various locks below. Also
	 * stops preemption for RCU.
L
Linus Torvalds 已提交
3057
	 */
3058
	rcu_read_lock_bh();
L
Linus Torvalds 已提交
3059

3060 3061
	skb_update_prio(skb);

3062 3063 3064 3065 3066 3067 3068 3069
	/* If device/qdisc don't need skb->dst, release it right now while
	 * its hot in this cpu cache.
	 */
	if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
		skb_dst_drop(skb);
	else
		skb_dst_force(skb);

3070
	txq = netdev_pick_tx(dev, skb, accel_priv);
3071
	q = rcu_dereference_bh(txq->qdisc);
3072

L
Linus Torvalds 已提交
3073
#ifdef CONFIG_NET_CLS_ACT
E
Eric Dumazet 已提交
3074
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_EGRESS);
L
Linus Torvalds 已提交
3075
#endif
3076
	trace_net_dev_queue(skb);
L
Linus Torvalds 已提交
3077
	if (q->enqueue) {
3078
		rc = __dev_xmit_skb(skb, q, dev, txq);
3079
		goto out;
L
Linus Torvalds 已提交
3080 3081 3082 3083 3084
	}

	/* The device has no queue. Common case for software devices:
	   loopback, all the sorts of tunnels...

H
Herbert Xu 已提交
3085 3086
	   Really, it is unlikely that netif_tx_lock protection is necessary
	   here.  (f.e. loopback and IP tunnels are clean ignoring statistics
L
Linus Torvalds 已提交
3087 3088 3089 3090 3091 3092 3093 3094 3095 3096
	   counters.)
	   However, it is possible, that they rely on protection
	   made by us here.

	   Check this and shot the lock. It is not prone from deadlocks.
	   Either shot noqueue qdisc, it is even simpler 8)
	 */
	if (dev->flags & IFF_UP) {
		int cpu = smp_processor_id(); /* ok because BHs are off */

3097
		if (txq->xmit_lock_owner != cpu) {
L
Linus Torvalds 已提交
3098

3099 3100 3101
			if (__this_cpu_read(xmit_recursion) > RECURSION_LIMIT)
				goto recursion_alert;

3102 3103 3104 3105
			skb = validate_xmit_skb(skb, dev);
			if (!skb)
				goto drop;

3106
			HARD_TX_LOCK(dev, txq, cpu);
L
Linus Torvalds 已提交
3107

3108
			if (!netif_xmit_stopped(txq)) {
3109
				__this_cpu_inc(xmit_recursion);
3110
				skb = dev_hard_start_xmit(skb, dev, txq, &rc);
3111
				__this_cpu_dec(xmit_recursion);
3112
				if (dev_xmit_complete(rc)) {
3113
					HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
3114 3115 3116
					goto out;
				}
			}
3117
			HARD_TX_UNLOCK(dev, txq);
3118 3119
			net_crit_ratelimited("Virtual device %s asks to queue packet!\n",
					     dev->name);
L
Linus Torvalds 已提交
3120 3121
		} else {
			/* Recursion is detected! It is possible,
3122 3123 3124
			 * unfortunately
			 */
recursion_alert:
3125 3126
			net_crit_ratelimited("Dead loop on virtual device %s, fix it urgently!\n",
					     dev->name);
L
Linus Torvalds 已提交
3127 3128 3129 3130
		}
	}

	rc = -ENETDOWN;
3131
drop:
3132
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
3133

3134
	atomic_long_inc(&dev->tx_dropped);
3135
	kfree_skb_list(skb);
L
Linus Torvalds 已提交
3136 3137
	return rc;
out:
3138
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
3139 3140
	return rc;
}
3141

3142
int dev_queue_xmit_sk(struct sock *sk, struct sk_buff *skb)
3143 3144 3145
{
	return __dev_queue_xmit(skb, NULL);
}
3146
EXPORT_SYMBOL(dev_queue_xmit_sk);
L
Linus Torvalds 已提交
3147

3148 3149 3150 3151 3152 3153
int dev_queue_xmit_accel(struct sk_buff *skb, void *accel_priv)
{
	return __dev_queue_xmit(skb, accel_priv);
}
EXPORT_SYMBOL(dev_queue_xmit_accel);

L
Linus Torvalds 已提交
3154 3155 3156 3157 3158

/*=======================================================================
			Receiver routines
  =======================================================================*/

3159
int netdev_max_backlog __read_mostly = 1000;
E
Eric Dumazet 已提交
3160 3161
EXPORT_SYMBOL(netdev_max_backlog);

E
Eric Dumazet 已提交
3162
int netdev_tstamp_prequeue __read_mostly = 1;
3163 3164
int netdev_budget __read_mostly = 300;
int weight_p __read_mostly = 64;            /* old backlog weight */
L
Linus Torvalds 已提交
3165

E
Eric Dumazet 已提交
3166 3167 3168 3169 3170 3171 3172 3173
/* Called with irq disabled */
static inline void ____napi_schedule(struct softnet_data *sd,
				     struct napi_struct *napi)
{
	list_add_tail(&napi->poll_list, &sd->poll_list);
	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
}

3174 3175 3176
#ifdef CONFIG_RPS

/* One global table that all flow-based protocols share. */
E
Eric Dumazet 已提交
3177
struct rps_sock_flow_table __rcu *rps_sock_flow_table __read_mostly;
3178
EXPORT_SYMBOL(rps_sock_flow_table);
3179 3180
u32 rps_cpu_mask __read_mostly;
EXPORT_SYMBOL(rps_cpu_mask);
3181

3182
struct static_key rps_needed __read_mostly;
3183

3184 3185 3186 3187
static struct rps_dev_flow *
set_rps_cpu(struct net_device *dev, struct sk_buff *skb,
	    struct rps_dev_flow *rflow, u16 next_cpu)
{
3188
	if (next_cpu < nr_cpu_ids) {
3189 3190 3191 3192 3193 3194 3195 3196 3197
#ifdef CONFIG_RFS_ACCEL
		struct netdev_rx_queue *rxqueue;
		struct rps_dev_flow_table *flow_table;
		struct rps_dev_flow *old_rflow;
		u32 flow_id;
		u16 rxq_index;
		int rc;

		/* Should we steer this flow to a different hardware queue? */
3198 3199
		if (!skb_rx_queue_recorded(skb) || !dev->rx_cpu_rmap ||
		    !(dev->features & NETIF_F_NTUPLE))
3200 3201 3202 3203 3204 3205 3206 3207 3208
			goto out;
		rxq_index = cpu_rmap_lookup_index(dev->rx_cpu_rmap, next_cpu);
		if (rxq_index == skb_get_rx_queue(skb))
			goto out;

		rxqueue = dev->_rx + rxq_index;
		flow_table = rcu_dereference(rxqueue->rps_flow_table);
		if (!flow_table)
			goto out;
3209
		flow_id = skb_get_hash(skb) & flow_table->mask;
3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221
		rc = dev->netdev_ops->ndo_rx_flow_steer(dev, skb,
							rxq_index, flow_id);
		if (rc < 0)
			goto out;
		old_rflow = rflow;
		rflow = &flow_table->flows[flow_id];
		rflow->filter = rc;
		if (old_rflow->filter == rflow->filter)
			old_rflow->filter = RPS_NO_FILTER;
	out:
#endif
		rflow->last_qtail =
3222
			per_cpu(softnet_data, next_cpu).input_queue_head;
3223 3224
	}

3225
	rflow->cpu = next_cpu;
3226 3227 3228
	return rflow;
}

3229 3230 3231 3232 3233 3234 3235 3236
/*
 * get_rps_cpu is called from netif_receive_skb and returns the target
 * CPU from the RPS map of the receiving queue for a given skb.
 * rcu_read_lock must be held on entry.
 */
static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb,
		       struct rps_dev_flow **rflowp)
{
3237 3238
	const struct rps_sock_flow_table *sock_flow_table;
	struct netdev_rx_queue *rxqueue = dev->_rx;
3239
	struct rps_dev_flow_table *flow_table;
3240
	struct rps_map *map;
3241
	int cpu = -1;
3242
	u32 tcpu;
3243
	u32 hash;
3244 3245 3246

	if (skb_rx_queue_recorded(skb)) {
		u16 index = skb_get_rx_queue(skb);
3247

3248 3249 3250 3251 3252
		if (unlikely(index >= dev->real_num_rx_queues)) {
			WARN_ONCE(dev->real_num_rx_queues > 1,
				  "%s received packet on queue %u, but number "
				  "of RX queues is %u\n",
				  dev->name, index, dev->real_num_rx_queues);
3253 3254
			goto done;
		}
3255 3256
		rxqueue += index;
	}
3257

3258 3259 3260
	/* Avoid computing hash if RFS/RPS is not active for this rxqueue */

	flow_table = rcu_dereference(rxqueue->rps_flow_table);
E
Eric Dumazet 已提交
3261
	map = rcu_dereference(rxqueue->rps_map);
3262
	if (!flow_table && !map)
3263 3264
		goto done;

3265
	skb_reset_network_header(skb);
3266 3267
	hash = skb_get_hash(skb);
	if (!hash)
3268 3269
		goto done;

T
Tom Herbert 已提交
3270 3271 3272
	sock_flow_table = rcu_dereference(rps_sock_flow_table);
	if (flow_table && sock_flow_table) {
		struct rps_dev_flow *rflow;
3273 3274 3275 3276 3277 3278 3279
		u32 next_cpu;
		u32 ident;

		/* First check into global flow table if there is a match */
		ident = sock_flow_table->ents[hash & sock_flow_table->mask];
		if ((ident ^ hash) & ~rps_cpu_mask)
			goto try_rps;
T
Tom Herbert 已提交
3280

3281 3282 3283 3284 3285
		next_cpu = ident & rps_cpu_mask;

		/* OK, now we know there is a match,
		 * we can look at the local (per receive queue) flow table
		 */
3286
		rflow = &flow_table->flows[hash & flow_table->mask];
T
Tom Herbert 已提交
3287 3288 3289 3290 3291 3292
		tcpu = rflow->cpu;

		/*
		 * If the desired CPU (where last recvmsg was done) is
		 * different from current CPU (one in the rx-queue flow
		 * table entry), switch if one of the following holds:
3293
		 *   - Current CPU is unset (>= nr_cpu_ids).
T
Tom Herbert 已提交
3294 3295 3296 3297 3298 3299 3300
		 *   - Current CPU is offline.
		 *   - The current CPU's queue tail has advanced beyond the
		 *     last packet that was enqueued using this table entry.
		 *     This guarantees that all previous packets for the flow
		 *     have been dequeued, thus preserving in order delivery.
		 */
		if (unlikely(tcpu != next_cpu) &&
3301
		    (tcpu >= nr_cpu_ids || !cpu_online(tcpu) ||
T
Tom Herbert 已提交
3302
		     ((int)(per_cpu(softnet_data, tcpu).input_queue_head -
3303 3304
		      rflow->last_qtail)) >= 0)) {
			tcpu = next_cpu;
3305
			rflow = set_rps_cpu(dev, skb, rflow, next_cpu);
3306
		}
3307

3308
		if (tcpu < nr_cpu_ids && cpu_online(tcpu)) {
T
Tom Herbert 已提交
3309 3310 3311 3312 3313 3314
			*rflowp = rflow;
			cpu = tcpu;
			goto done;
		}
	}

3315 3316
try_rps:

T
Tom Herbert 已提交
3317
	if (map) {
3318
		tcpu = map->cpus[reciprocal_scale(hash, map->len)];
T
Tom Herbert 已提交
3319 3320 3321 3322 3323 3324 3325 3326 3327 3328
		if (cpu_online(tcpu)) {
			cpu = tcpu;
			goto done;
		}
	}

done:
	return cpu;
}

3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348
#ifdef CONFIG_RFS_ACCEL

/**
 * rps_may_expire_flow - check whether an RFS hardware filter may be removed
 * @dev: Device on which the filter was set
 * @rxq_index: RX queue index
 * @flow_id: Flow ID passed to ndo_rx_flow_steer()
 * @filter_id: Filter ID returned by ndo_rx_flow_steer()
 *
 * Drivers that implement ndo_rx_flow_steer() should periodically call
 * this function for each installed filter and remove the filters for
 * which it returns %true.
 */
bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index,
			 u32 flow_id, u16 filter_id)
{
	struct netdev_rx_queue *rxqueue = dev->_rx + rxq_index;
	struct rps_dev_flow_table *flow_table;
	struct rps_dev_flow *rflow;
	bool expire = true;
3349
	unsigned int cpu;
3350 3351 3352 3353 3354 3355

	rcu_read_lock();
	flow_table = rcu_dereference(rxqueue->rps_flow_table);
	if (flow_table && flow_id <= flow_table->mask) {
		rflow = &flow_table->flows[flow_id];
		cpu = ACCESS_ONCE(rflow->cpu);
3356
		if (rflow->filter == filter_id && cpu < nr_cpu_ids &&
3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368
		    ((int)(per_cpu(softnet_data, cpu).input_queue_head -
			   rflow->last_qtail) <
		     (int)(10 * flow_table->mask)))
			expire = false;
	}
	rcu_read_unlock();
	return expire;
}
EXPORT_SYMBOL(rps_may_expire_flow);

#endif /* CONFIG_RFS_ACCEL */

T
Tom Herbert 已提交
3369
/* Called from hardirq (IPI) context */
E
Eric Dumazet 已提交
3370
static void rps_trigger_softirq(void *data)
T
Tom Herbert 已提交
3371
{
E
Eric Dumazet 已提交
3372 3373
	struct softnet_data *sd = data;

E
Eric Dumazet 已提交
3374
	____napi_schedule(sd, &sd->backlog);
C
Changli Gao 已提交
3375
	sd->received_rps++;
T
Tom Herbert 已提交
3376
}
E
Eric Dumazet 已提交
3377

T
Tom Herbert 已提交
3378
#endif /* CONFIG_RPS */
T
Tom Herbert 已提交
3379

E
Eric Dumazet 已提交
3380 3381 3382 3383 3384 3385 3386 3387
/*
 * Check if this softnet_data structure is another cpu one
 * If yes, queue it to our IPI list and return 1
 * If no, return 0
 */
static int rps_ipi_queued(struct softnet_data *sd)
{
#ifdef CONFIG_RPS
3388
	struct softnet_data *mysd = this_cpu_ptr(&softnet_data);
E
Eric Dumazet 已提交
3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400

	if (sd != mysd) {
		sd->rps_ipi_next = mysd->rps_ipi_list;
		mysd->rps_ipi_list = sd;

		__raise_softirq_irqoff(NET_RX_SOFTIRQ);
		return 1;
	}
#endif /* CONFIG_RPS */
	return 0;
}

3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414
#ifdef CONFIG_NET_FLOW_LIMIT
int netdev_flow_limit_table_len __read_mostly = (1 << 12);
#endif

static bool skb_flow_limit(struct sk_buff *skb, unsigned int qlen)
{
#ifdef CONFIG_NET_FLOW_LIMIT
	struct sd_flow_limit *fl;
	struct softnet_data *sd;
	unsigned int old_flow, new_flow;

	if (qlen < (netdev_max_backlog >> 1))
		return false;

3415
	sd = this_cpu_ptr(&softnet_data);
3416 3417 3418 3419

	rcu_read_lock();
	fl = rcu_dereference(sd->flow_limit);
	if (fl) {
3420
		new_flow = skb_get_hash(skb) & (fl->num_buckets - 1);
3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440
		old_flow = fl->history[fl->history_head];
		fl->history[fl->history_head] = new_flow;

		fl->history_head++;
		fl->history_head &= FLOW_LIMIT_HISTORY - 1;

		if (likely(fl->buckets[old_flow]))
			fl->buckets[old_flow]--;

		if (++fl->buckets[new_flow] > (FLOW_LIMIT_HISTORY >> 1)) {
			fl->count++;
			rcu_read_unlock();
			return true;
		}
	}
	rcu_read_unlock();
#endif
	return false;
}

T
Tom Herbert 已提交
3441 3442 3443 3444
/*
 * enqueue_to_backlog is called to queue an skb to a per CPU backlog
 * queue (may be a remote CPU queue).
 */
T
Tom Herbert 已提交
3445 3446
static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
			      unsigned int *qtail)
T
Tom Herbert 已提交
3447
{
E
Eric Dumazet 已提交
3448
	struct softnet_data *sd;
T
Tom Herbert 已提交
3449
	unsigned long flags;
3450
	unsigned int qlen;
T
Tom Herbert 已提交
3451

E
Eric Dumazet 已提交
3452
	sd = &per_cpu(softnet_data, cpu);
T
Tom Herbert 已提交
3453 3454 3455

	local_irq_save(flags);

E
Eric Dumazet 已提交
3456
	rps_lock(sd);
3457 3458
	qlen = skb_queue_len(&sd->input_pkt_queue);
	if (qlen <= netdev_max_backlog && !skb_flow_limit(skb, qlen)) {
3459
		if (qlen) {
T
Tom Herbert 已提交
3460
enqueue:
E
Eric Dumazet 已提交
3461
			__skb_queue_tail(&sd->input_pkt_queue, skb);
3462
			input_queue_tail_incr_save(sd, qtail);
E
Eric Dumazet 已提交
3463
			rps_unlock(sd);
3464
			local_irq_restore(flags);
T
Tom Herbert 已提交
3465 3466 3467
			return NET_RX_SUCCESS;
		}

3468 3469 3470 3471
		/* Schedule NAPI for backlog device
		 * We can use non atomic operation since we own the queue lock
		 */
		if (!__test_and_set_bit(NAPI_STATE_SCHED, &sd->backlog.state)) {
E
Eric Dumazet 已提交
3472
			if (!rps_ipi_queued(sd))
E
Eric Dumazet 已提交
3473
				____napi_schedule(sd, &sd->backlog);
T
Tom Herbert 已提交
3474 3475 3476 3477
		}
		goto enqueue;
	}

C
Changli Gao 已提交
3478
	sd->dropped++;
E
Eric Dumazet 已提交
3479
	rps_unlock(sd);
T
Tom Herbert 已提交
3480 3481 3482

	local_irq_restore(flags);

3483
	atomic_long_inc(&skb->dev->rx_dropped);
T
Tom Herbert 已提交
3484 3485 3486
	kfree_skb(skb);
	return NET_RX_DROP;
}
L
Linus Torvalds 已提交
3487

3488
static int netif_rx_internal(struct sk_buff *skb)
L
Linus Torvalds 已提交
3489
{
3490
	int ret;
L
Linus Torvalds 已提交
3491

3492
	net_timestamp_check(netdev_tstamp_prequeue, skb);
L
Linus Torvalds 已提交
3493

3494
	trace_netif_rx(skb);
E
Eric Dumazet 已提交
3495
#ifdef CONFIG_RPS
3496
	if (static_key_false(&rps_needed)) {
T
Tom Herbert 已提交
3497
		struct rps_dev_flow voidflow, *rflow = &voidflow;
3498 3499
		int cpu;

3500
		preempt_disable();
3501
		rcu_read_lock();
T
Tom Herbert 已提交
3502 3503

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
3504 3505
		if (cpu < 0)
			cpu = smp_processor_id();
T
Tom Herbert 已提交
3506 3507 3508

		ret = enqueue_to_backlog(skb, cpu, &rflow->last_qtail);

3509
		rcu_read_unlock();
3510
		preempt_enable();
3511 3512
	} else
#endif
T
Tom Herbert 已提交
3513 3514 3515 3516 3517
	{
		unsigned int qtail;
		ret = enqueue_to_backlog(skb, get_cpu(), &qtail);
		put_cpu();
	}
3518
	return ret;
L
Linus Torvalds 已提交
3519
}
3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541

/**
 *	netif_rx	-	post buffer to the network code
 *	@skb: buffer to post
 *
 *	This function receives a packet from a device driver and queues it for
 *	the upper (protocol) levels to process.  It always succeeds. The buffer
 *	may be dropped during processing for congestion control or by the
 *	protocol layers.
 *
 *	return values:
 *	NET_RX_SUCCESS	(no congestion)
 *	NET_RX_DROP     (packet was dropped)
 *
 */

int netif_rx(struct sk_buff *skb)
{
	trace_netif_rx_entry(skb);

	return netif_rx_internal(skb);
}
E
Eric Dumazet 已提交
3542
EXPORT_SYMBOL(netif_rx);
L
Linus Torvalds 已提交
3543 3544 3545 3546 3547

int netif_rx_ni(struct sk_buff *skb)
{
	int err;

3548 3549
	trace_netif_rx_ni_entry(skb);

L
Linus Torvalds 已提交
3550
	preempt_disable();
3551
	err = netif_rx_internal(skb);
L
Linus Torvalds 已提交
3552 3553 3554 3555 3556 3557 3558 3559 3560 3561
	if (local_softirq_pending())
		do_softirq();
	preempt_enable();

	return err;
}
EXPORT_SYMBOL(netif_rx_ni);

static void net_tx_action(struct softirq_action *h)
{
3562
	struct softnet_data *sd = this_cpu_ptr(&softnet_data);
L
Linus Torvalds 已提交
3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575

	if (sd->completion_queue) {
		struct sk_buff *clist;

		local_irq_disable();
		clist = sd->completion_queue;
		sd->completion_queue = NULL;
		local_irq_enable();

		while (clist) {
			struct sk_buff *skb = clist;
			clist = clist->next;

3576
			WARN_ON(atomic_read(&skb->users));
3577 3578 3579 3580
			if (likely(get_kfree_skb_cb(skb)->reason == SKB_REASON_CONSUMED))
				trace_consume_skb(skb);
			else
				trace_kfree_skb(skb, net_tx_action);
L
Linus Torvalds 已提交
3581 3582 3583 3584 3585
			__kfree_skb(skb);
		}
	}

	if (sd->output_queue) {
3586
		struct Qdisc *head;
L
Linus Torvalds 已提交
3587 3588 3589 3590

		local_irq_disable();
		head = sd->output_queue;
		sd->output_queue = NULL;
3591
		sd->output_queue_tailp = &sd->output_queue;
L
Linus Torvalds 已提交
3592 3593 3594
		local_irq_enable();

		while (head) {
3595 3596 3597
			struct Qdisc *q = head;
			spinlock_t *root_lock;

L
Linus Torvalds 已提交
3598 3599
			head = head->next_sched;

3600
			root_lock = qdisc_lock(q);
3601
			if (spin_trylock(root_lock)) {
3602
				smp_mb__before_atomic();
3603 3604
				clear_bit(__QDISC_STATE_SCHED,
					  &q->state);
3605 3606
				qdisc_run(q);
				spin_unlock(root_lock);
L
Linus Torvalds 已提交
3607
			} else {
3608
				if (!test_bit(__QDISC_STATE_DEACTIVATED,
3609
					      &q->state)) {
3610
					__netif_reschedule(q);
3611
				} else {
3612
					smp_mb__before_atomic();
3613 3614 3615
					clear_bit(__QDISC_STATE_SCHED,
						  &q->state);
				}
L
Linus Torvalds 已提交
3616 3617 3618 3619 3620
			}
		}
	}
}

3621 3622
#if (defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE)) && \
    (defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE))
3623 3624 3625
/* This hook is defined here for ATM LANE */
int (*br_fdb_test_addr_hook)(struct net_device *dev,
			     unsigned char *addr) __read_mostly;
3626
EXPORT_SYMBOL_GPL(br_fdb_test_addr_hook);
3627
#endif
L
Linus Torvalds 已提交
3628 3629

#ifdef CONFIG_NET_CLS_ACT
3630 3631 3632 3633
static inline struct sk_buff *handle_ing(struct sk_buff *skb,
					 struct packet_type **pt_prev,
					 int *ret, struct net_device *orig_dev)
{
3634 3635
	struct tcf_proto *cl = rcu_dereference_bh(skb->dev->ingress_cl_list);
	struct tcf_result cl_res;
3636

3637 3638 3639
	/* If there's at least one ingress present somewhere (so
	 * we get here via enabled static key), remaining devices
	 * that are not configured with an ingress qdisc will bail
3640
	 * out here.
3641
	 */
3642
	if (!cl)
3643
		return skb;
3644 3645 3646
	if (*pt_prev) {
		*ret = deliver_skb(skb, *pt_prev, orig_dev);
		*pt_prev = NULL;
L
Linus Torvalds 已提交
3647 3648
	}

3649
	qdisc_bstats_update_cpu(cl->q, skb);
3650 3651
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_INGRESS);

3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664
	switch (tc_classify(skb, cl, &cl_res)) {
	case TC_ACT_OK:
	case TC_ACT_RECLASSIFY:
		skb->tc_index = TC_H_MIN(cl_res.classid);
		break;
	case TC_ACT_SHOT:
		qdisc_qstats_drop_cpu(cl->q);
	case TC_ACT_STOLEN:
	case TC_ACT_QUEUED:
		kfree_skb(skb);
		return NULL;
	default:
		break;
3665 3666 3667
	}

	return skb;
L
Linus Torvalds 已提交
3668 3669 3670
}
#endif

3671 3672 3673 3674
/**
 *	netdev_rx_handler_register - register receive handler
 *	@dev: device to register a handler for
 *	@rx_handler: receive handler to register
J
Jiri Pirko 已提交
3675
 *	@rx_handler_data: data pointer that is used by rx handler
3676
 *
3677
 *	Register a receive handler for a device. This handler will then be
3678 3679 3680 3681
 *	called from __netif_receive_skb. A negative errno code is returned
 *	on a failure.
 *
 *	The caller must hold the rtnl_mutex.
3682 3683
 *
 *	For a general description of rx_handler, see enum rx_handler_result.
3684 3685
 */
int netdev_rx_handler_register(struct net_device *dev,
J
Jiri Pirko 已提交
3686 3687
			       rx_handler_func_t *rx_handler,
			       void *rx_handler_data)
3688 3689 3690 3691 3692 3693
{
	ASSERT_RTNL();

	if (dev->rx_handler)
		return -EBUSY;

3694
	/* Note: rx_handler_data must be set before rx_handler */
J
Jiri Pirko 已提交
3695
	rcu_assign_pointer(dev->rx_handler_data, rx_handler_data);
3696 3697 3698 3699 3700 3701 3702 3703 3704 3705
	rcu_assign_pointer(dev->rx_handler, rx_handler);

	return 0;
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_register);

/**
 *	netdev_rx_handler_unregister - unregister receive handler
 *	@dev: device to unregister a handler from
 *
K
Kusanagi Kouichi 已提交
3706
 *	Unregister a receive handler from a device.
3707 3708 3709 3710 3711 3712 3713
 *
 *	The caller must hold the rtnl_mutex.
 */
void netdev_rx_handler_unregister(struct net_device *dev)
{

	ASSERT_RTNL();
3714
	RCU_INIT_POINTER(dev->rx_handler, NULL);
3715 3716 3717 3718 3719
	/* a reader seeing a non NULL rx_handler in a rcu_read_lock()
	 * section has a guarantee to see a non NULL rx_handler_data
	 * as well.
	 */
	synchronize_net();
3720
	RCU_INIT_POINTER(dev->rx_handler_data, NULL);
3721 3722 3723
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);

3724 3725 3726 3727 3728 3729 3730
/*
 * Limit the use of PFMEMALLOC reserves to those protocols that implement
 * the special handling of PFMEMALLOC skbs.
 */
static bool skb_pfmemalloc_protocol(struct sk_buff *skb)
{
	switch (skb->protocol) {
3731 3732 3733 3734 3735
	case htons(ETH_P_ARP):
	case htons(ETH_P_IP):
	case htons(ETH_P_IPV6):
	case htons(ETH_P_8021Q):
	case htons(ETH_P_8021AD):
3736 3737 3738 3739 3740 3741
		return true;
	default:
		return false;
	}
}

3742
static int __netif_receive_skb_core(struct sk_buff *skb, bool pfmemalloc)
L
Linus Torvalds 已提交
3743 3744
{
	struct packet_type *ptype, *pt_prev;
3745
	rx_handler_func_t *rx_handler;
D
David S. Miller 已提交
3746
	struct net_device *orig_dev;
3747
	bool deliver_exact = false;
L
Linus Torvalds 已提交
3748
	int ret = NET_RX_DROP;
A
Al Viro 已提交
3749
	__be16 type;
L
Linus Torvalds 已提交
3750

3751
	net_timestamp_check(!netdev_tstamp_prequeue, skb);
3752

3753
	trace_netif_receive_skb(skb);
3754

J
Joe Eykholt 已提交
3755
	orig_dev = skb->dev;
3756

3757
	skb_reset_network_header(skb);
3758 3759
	if (!skb_transport_header_was_set(skb))
		skb_reset_transport_header(skb);
3760
	skb_reset_mac_len(skb);
L
Linus Torvalds 已提交
3761 3762 3763 3764 3765

	pt_prev = NULL;

	rcu_read_lock();

3766
another_round:
3767
	skb->skb_iif = skb->dev->ifindex;
3768 3769 3770

	__this_cpu_inc(softnet_data.processed);

P
Patrick McHardy 已提交
3771 3772
	if (skb->protocol == cpu_to_be16(ETH_P_8021Q) ||
	    skb->protocol == cpu_to_be16(ETH_P_8021AD)) {
3773
		skb = skb_vlan_untag(skb);
3774
		if (unlikely(!skb))
3775
			goto unlock;
3776 3777
	}

L
Linus Torvalds 已提交
3778 3779 3780 3781 3782 3783 3784
#ifdef CONFIG_NET_CLS_ACT
	if (skb->tc_verd & TC_NCLS) {
		skb->tc_verd = CLR_TC_NCLS(skb->tc_verd);
		goto ncls;
	}
#endif

3785
	if (pfmemalloc)
3786 3787
		goto skip_taps;

L
Linus Torvalds 已提交
3788
	list_for_each_entry_rcu(ptype, &ptype_all, list) {
3789 3790 3791 3792 3793 3794 3795 3796 3797
		if (pt_prev)
			ret = deliver_skb(skb, pt_prev, orig_dev);
		pt_prev = ptype;
	}

	list_for_each_entry_rcu(ptype, &skb->dev->ptype_all, list) {
		if (pt_prev)
			ret = deliver_skb(skb, pt_prev, orig_dev);
		pt_prev = ptype;
L
Linus Torvalds 已提交
3798 3799
	}

3800
skip_taps:
L
Linus Torvalds 已提交
3801
#ifdef CONFIG_NET_CLS_ACT
3802 3803 3804 3805 3806 3807 3808
	if (static_key_false(&ingress_needed)) {
		skb = handle_ing(skb, &pt_prev, &ret, orig_dev);
		if (!skb)
			goto unlock;
	}

	skb->tc_verd = 0;
L
Linus Torvalds 已提交
3809 3810
ncls:
#endif
3811
	if (pfmemalloc && !skb_pfmemalloc_protocol(skb))
3812 3813
		goto drop;

3814
	if (skb_vlan_tag_present(skb)) {
3815 3816 3817 3818
		if (pt_prev) {
			ret = deliver_skb(skb, pt_prev, orig_dev);
			pt_prev = NULL;
		}
3819
		if (vlan_do_receive(&skb))
3820 3821
			goto another_round;
		else if (unlikely(!skb))
3822
			goto unlock;
3823 3824
	}

3825
	rx_handler = rcu_dereference(skb->dev->rx_handler);
3826 3827 3828 3829 3830
	if (rx_handler) {
		if (pt_prev) {
			ret = deliver_skb(skb, pt_prev, orig_dev);
			pt_prev = NULL;
		}
3831 3832
		switch (rx_handler(&skb)) {
		case RX_HANDLER_CONSUMED:
3833
			ret = NET_RX_SUCCESS;
3834
			goto unlock;
3835
		case RX_HANDLER_ANOTHER:
3836
			goto another_round;
3837 3838 3839 3840 3841 3842 3843
		case RX_HANDLER_EXACT:
			deliver_exact = true;
		case RX_HANDLER_PASS:
			break;
		default:
			BUG();
		}
3844
	}
L
Linus Torvalds 已提交
3845

3846 3847
	if (unlikely(skb_vlan_tag_present(skb))) {
		if (skb_vlan_tag_get_id(skb))
E
Eric Dumazet 已提交
3848 3849 3850 3851 3852 3853 3854
			skb->pkt_type = PACKET_OTHERHOST;
		/* Note: we might in the future use prio bits
		 * and set skb->priority like in vlan_do_receive()
		 * For the time being, just ignore Priority Code Point
		 */
		skb->vlan_tci = 0;
	}
3855

3856 3857
	type = skb->protocol;

3858
	/* deliver only exact match when indicated */
3859 3860 3861 3862 3863
	if (likely(!deliver_exact)) {
		deliver_ptype_list_skb(skb, &pt_prev, orig_dev, type,
				       &ptype_base[ntohs(type) &
						   PTYPE_HASH_MASK]);
	}
3864

3865 3866 3867 3868 3869 3870
	deliver_ptype_list_skb(skb, &pt_prev, orig_dev, type,
			       &orig_dev->ptype_specific);

	if (unlikely(skb->dev != orig_dev)) {
		deliver_ptype_list_skb(skb, &pt_prev, orig_dev, type,
				       &skb->dev->ptype_specific);
L
Linus Torvalds 已提交
3871 3872 3873
	}

	if (pt_prev) {
3874
		if (unlikely(skb_orphan_frags(skb, GFP_ATOMIC)))
3875
			goto drop;
3876 3877
		else
			ret = pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3878
	} else {
3879
drop:
3880
		atomic_long_inc(&skb->dev->rx_dropped);
L
Linus Torvalds 已提交
3881 3882 3883 3884 3885 3886 3887
		kfree_skb(skb);
		/* Jamal, now you will not able to escape explaining
		 * me how you were going to use this. :-)
		 */
		ret = NET_RX_DROP;
	}

3888
unlock:
L
Linus Torvalds 已提交
3889
	rcu_read_unlock();
3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914
	return ret;
}

static int __netif_receive_skb(struct sk_buff *skb)
{
	int ret;

	if (sk_memalloc_socks() && skb_pfmemalloc(skb)) {
		unsigned long pflags = current->flags;

		/*
		 * PFMEMALLOC skbs are special, they should
		 * - be delivered to SOCK_MEMALLOC sockets only
		 * - stay away from userspace
		 * - have bounded memory usage
		 *
		 * Use PF_MEMALLOC as this saves us from propagating the allocation
		 * context down to all allocation sites.
		 */
		current->flags |= PF_MEMALLOC;
		ret = __netif_receive_skb_core(skb, true);
		tsk_restore_flags(current, pflags, PF_MEMALLOC);
	} else
		ret = __netif_receive_skb_core(skb, false);

L
Linus Torvalds 已提交
3915 3916
	return ret;
}
T
Tom Herbert 已提交
3917

3918
static int netif_receive_skb_internal(struct sk_buff *skb)
T
Tom Herbert 已提交
3919
{
3920
	net_timestamp_check(netdev_tstamp_prequeue, skb);
E
Eric Dumazet 已提交
3921

3922 3923 3924
	if (skb_defer_rx_timestamp(skb))
		return NET_RX_SUCCESS;

E
Eric Dumazet 已提交
3925
#ifdef CONFIG_RPS
3926
	if (static_key_false(&rps_needed)) {
E
Eric Dumazet 已提交
3927 3928
		struct rps_dev_flow voidflow, *rflow = &voidflow;
		int cpu, ret;
T
Tom Herbert 已提交
3929

E
Eric Dumazet 已提交
3930 3931 3932
		rcu_read_lock();

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
T
Tom Herbert 已提交
3933

E
Eric Dumazet 已提交
3934 3935 3936
		if (cpu >= 0) {
			ret = enqueue_to_backlog(skb, cpu, &rflow->last_qtail);
			rcu_read_unlock();
3937
			return ret;
E
Eric Dumazet 已提交
3938
		}
3939
		rcu_read_unlock();
T
Tom Herbert 已提交
3940
	}
3941
#endif
3942
	return __netif_receive_skb(skb);
T
Tom Herbert 已提交
3943
}
3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959

/**
 *	netif_receive_skb - process receive buffer from network
 *	@skb: buffer to process
 *
 *	netif_receive_skb() is the main receive data processing function.
 *	It always succeeds. The buffer may be dropped during processing
 *	for congestion control or by the protocol layers.
 *
 *	This function may only be called from softirq context and interrupts
 *	should be enabled.
 *
 *	Return values (usually ignored):
 *	NET_RX_SUCCESS: no congestion
 *	NET_RX_DROP: packet was dropped
 */
3960
int netif_receive_skb_sk(struct sock *sk, struct sk_buff *skb)
3961 3962 3963 3964 3965
{
	trace_netif_receive_skb_entry(skb);

	return netif_receive_skb_internal(skb);
}
3966
EXPORT_SYMBOL(netif_receive_skb_sk);
L
Linus Torvalds 已提交
3967

E
Eric Dumazet 已提交
3968 3969 3970
/* Network device is going away, flush any packets still pending
 * Called with irqs disabled.
 */
3971
static void flush_backlog(void *arg)
3972
{
3973
	struct net_device *dev = arg;
3974
	struct softnet_data *sd = this_cpu_ptr(&softnet_data);
3975 3976
	struct sk_buff *skb, *tmp;

E
Eric Dumazet 已提交
3977
	rps_lock(sd);
3978
	skb_queue_walk_safe(&sd->input_pkt_queue, skb, tmp) {
3979
		if (skb->dev == dev) {
E
Eric Dumazet 已提交
3980
			__skb_unlink(skb, &sd->input_pkt_queue);
3981
			kfree_skb(skb);
3982
			input_queue_head_incr(sd);
3983
		}
3984
	}
E
Eric Dumazet 已提交
3985
	rps_unlock(sd);
3986 3987 3988 3989 3990

	skb_queue_walk_safe(&sd->process_queue, skb, tmp) {
		if (skb->dev == dev) {
			__skb_unlink(skb, &sd->process_queue);
			kfree_skb(skb);
3991
			input_queue_head_incr(sd);
3992 3993
		}
	}
3994 3995
}

3996 3997
static int napi_gro_complete(struct sk_buff *skb)
{
3998
	struct packet_offload *ptype;
3999
	__be16 type = skb->protocol;
4000
	struct list_head *head = &offload_base;
4001 4002
	int err = -ENOENT;

4003 4004
	BUILD_BUG_ON(sizeof(struct napi_gro_cb) > sizeof(skb->cb));

4005 4006
	if (NAPI_GRO_CB(skb)->count == 1) {
		skb_shinfo(skb)->gso_size = 0;
4007
		goto out;
4008
	}
4009 4010 4011

	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
4012
		if (ptype->type != type || !ptype->callbacks.gro_complete)
4013 4014
			continue;

4015
		err = ptype->callbacks.gro_complete(skb, 0);
4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026
		break;
	}
	rcu_read_unlock();

	if (err) {
		WARN_ON(&ptype->list == head);
		kfree_skb(skb);
		return NET_RX_SUCCESS;
	}

out:
4027
	return netif_receive_skb_internal(skb);
4028 4029
}

4030 4031 4032 4033 4034
/* napi->gro_list contains packets ordered by age.
 * youngest packets at the head of it.
 * Complete skbs in reverse order to reduce latencies.
 */
void napi_gro_flush(struct napi_struct *napi, bool flush_old)
4035
{
4036
	struct sk_buff *skb, *prev = NULL;
4037

4038 4039 4040 4041 4042 4043 4044
	/* scan list and build reverse chain */
	for (skb = napi->gro_list; skb != NULL; skb = skb->next) {
		skb->prev = prev;
		prev = skb;
	}

	for (skb = prev; skb; skb = prev) {
4045
		skb->next = NULL;
4046 4047 4048 4049 4050

		if (flush_old && NAPI_GRO_CB(skb)->age == jiffies)
			return;

		prev = skb->prev;
4051
		napi_gro_complete(skb);
4052
		napi->gro_count--;
4053 4054 4055 4056
	}

	napi->gro_list = NULL;
}
E
Eric Dumazet 已提交
4057
EXPORT_SYMBOL(napi_gro_flush);
4058

E
Eric Dumazet 已提交
4059 4060 4061 4062
static void gro_list_prepare(struct napi_struct *napi, struct sk_buff *skb)
{
	struct sk_buff *p;
	unsigned int maclen = skb->dev->hard_header_len;
4063
	u32 hash = skb_get_hash_raw(skb);
E
Eric Dumazet 已提交
4064 4065 4066 4067

	for (p = napi->gro_list; p; p = p->next) {
		unsigned long diffs;

4068 4069 4070 4071 4072 4073 4074
		NAPI_GRO_CB(p)->flush = 0;

		if (hash != skb_get_hash_raw(p)) {
			NAPI_GRO_CB(p)->same_flow = 0;
			continue;
		}

E
Eric Dumazet 已提交
4075 4076 4077 4078
		diffs = (unsigned long)p->dev ^ (unsigned long)skb->dev;
		diffs |= p->vlan_tci ^ skb->vlan_tci;
		if (maclen == ETH_HLEN)
			diffs |= compare_ether_header(skb_mac_header(p),
4079
						      skb_mac_header(skb));
E
Eric Dumazet 已提交
4080 4081
		else if (!diffs)
			diffs = memcmp(skb_mac_header(p),
4082
				       skb_mac_header(skb),
E
Eric Dumazet 已提交
4083 4084 4085 4086 4087
				       maclen);
		NAPI_GRO_CB(p)->same_flow = !diffs;
	}
}

4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101
static void skb_gro_reset_offset(struct sk_buff *skb)
{
	const struct skb_shared_info *pinfo = skb_shinfo(skb);
	const skb_frag_t *frag0 = &pinfo->frags[0];

	NAPI_GRO_CB(skb)->data_offset = 0;
	NAPI_GRO_CB(skb)->frag0 = NULL;
	NAPI_GRO_CB(skb)->frag0_len = 0;

	if (skb_mac_header(skb) == skb_tail_pointer(skb) &&
	    pinfo->nr_frags &&
	    !PageHighMem(skb_frag_page(frag0))) {
		NAPI_GRO_CB(skb)->frag0 = skb_frag_address(frag0);
		NAPI_GRO_CB(skb)->frag0_len = skb_frag_size(frag0);
E
Eric Dumazet 已提交
4102 4103 4104
	}
}

4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125
static void gro_pull_from_frag0(struct sk_buff *skb, int grow)
{
	struct skb_shared_info *pinfo = skb_shinfo(skb);

	BUG_ON(skb->end - skb->tail < grow);

	memcpy(skb_tail_pointer(skb), NAPI_GRO_CB(skb)->frag0, grow);

	skb->data_len -= grow;
	skb->tail += grow;

	pinfo->frags[0].page_offset += grow;
	skb_frag_size_sub(&pinfo->frags[0], grow);

	if (unlikely(!skb_frag_size(&pinfo->frags[0]))) {
		skb_frag_unref(skb, 0);
		memmove(pinfo->frags, pinfo->frags + 1,
			--pinfo->nr_frags * sizeof(pinfo->frags[0]));
	}
}

R
Rami Rosen 已提交
4126
static enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
4127 4128
{
	struct sk_buff **pp = NULL;
4129
	struct packet_offload *ptype;
4130
	__be16 type = skb->protocol;
4131
	struct list_head *head = &offload_base;
H
Herbert Xu 已提交
4132
	int same_flow;
4133
	enum gro_result ret;
4134
	int grow;
4135

4136
	if (!(skb->dev->features & NETIF_F_GRO))
4137 4138
		goto normal;

4139
	if (skb_is_gso(skb) || skb_has_frag_list(skb) || skb->csum_bad)
4140 4141
		goto normal;

E
Eric Dumazet 已提交
4142 4143
	gro_list_prepare(napi, skb);

4144 4145
	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
4146
		if (ptype->type != type || !ptype->callbacks.gro_receive)
4147 4148
			continue;

4149
		skb_set_network_header(skb, skb_gro_offset(skb));
4150
		skb_reset_mac_len(skb);
4151 4152
		NAPI_GRO_CB(skb)->same_flow = 0;
		NAPI_GRO_CB(skb)->flush = 0;
H
Herbert Xu 已提交
4153
		NAPI_GRO_CB(skb)->free = 0;
4154
		NAPI_GRO_CB(skb)->udp_mark = 0;
4155
		NAPI_GRO_CB(skb)->gro_remcsum_start = 0;
4156

4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171
		/* Setup for GRO checksum validation */
		switch (skb->ip_summed) {
		case CHECKSUM_COMPLETE:
			NAPI_GRO_CB(skb)->csum = skb->csum;
			NAPI_GRO_CB(skb)->csum_valid = 1;
			NAPI_GRO_CB(skb)->csum_cnt = 0;
			break;
		case CHECKSUM_UNNECESSARY:
			NAPI_GRO_CB(skb)->csum_cnt = skb->csum_level + 1;
			NAPI_GRO_CB(skb)->csum_valid = 0;
			break;
		default:
			NAPI_GRO_CB(skb)->csum_cnt = 0;
			NAPI_GRO_CB(skb)->csum_valid = 0;
		}
4172

4173
		pp = ptype->callbacks.gro_receive(&napi->gro_list, skb);
4174 4175 4176 4177 4178 4179 4180
		break;
	}
	rcu_read_unlock();

	if (&ptype->list == head)
		goto normal;

H
Herbert Xu 已提交
4181
	same_flow = NAPI_GRO_CB(skb)->same_flow;
4182
	ret = NAPI_GRO_CB(skb)->free ? GRO_MERGED_FREE : GRO_MERGED;
H
Herbert Xu 已提交
4183

4184 4185 4186 4187 4188 4189
	if (pp) {
		struct sk_buff *nskb = *pp;

		*pp = nskb->next;
		nskb->next = NULL;
		napi_gro_complete(nskb);
4190
		napi->gro_count--;
4191 4192
	}

H
Herbert Xu 已提交
4193
	if (same_flow)
4194 4195
		goto ok;

4196
	if (NAPI_GRO_CB(skb)->flush)
4197 4198
		goto normal;

4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212
	if (unlikely(napi->gro_count >= MAX_GRO_SKBS)) {
		struct sk_buff *nskb = napi->gro_list;

		/* locate the end of the list to select the 'oldest' flow */
		while (nskb->next) {
			pp = &nskb->next;
			nskb = *pp;
		}
		*pp = NULL;
		nskb->next = NULL;
		napi_gro_complete(nskb);
	} else {
		napi->gro_count++;
	}
4213
	NAPI_GRO_CB(skb)->count = 1;
4214
	NAPI_GRO_CB(skb)->age = jiffies;
4215
	NAPI_GRO_CB(skb)->last = skb;
4216
	skb_shinfo(skb)->gso_size = skb_gro_len(skb);
4217 4218
	skb->next = napi->gro_list;
	napi->gro_list = skb;
4219
	ret = GRO_HELD;
4220

4221
pull:
4222 4223 4224
	grow = skb_gro_offset(skb) - skb_headlen(skb);
	if (grow > 0)
		gro_pull_from_frag0(skb, grow);
4225
ok:
4226
	return ret;
4227 4228

normal:
4229 4230
	ret = GRO_NORMAL;
	goto pull;
H
Herbert Xu 已提交
4231
}
4232

4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244
struct packet_offload *gro_find_receive_by_type(__be16 type)
{
	struct list_head *offload_head = &offload_base;
	struct packet_offload *ptype;

	list_for_each_entry_rcu(ptype, offload_head, list) {
		if (ptype->type != type || !ptype->callbacks.gro_receive)
			continue;
		return ptype;
	}
	return NULL;
}
4245
EXPORT_SYMBOL(gro_find_receive_by_type);
4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258

struct packet_offload *gro_find_complete_by_type(__be16 type)
{
	struct list_head *offload_head = &offload_base;
	struct packet_offload *ptype;

	list_for_each_entry_rcu(ptype, offload_head, list) {
		if (ptype->type != type || !ptype->callbacks.gro_complete)
			continue;
		return ptype;
	}
	return NULL;
}
4259
EXPORT_SYMBOL(gro_find_complete_by_type);
H
Herbert Xu 已提交
4260

R
Rami Rosen 已提交
4261
static gro_result_t napi_skb_finish(gro_result_t ret, struct sk_buff *skb)
H
Herbert Xu 已提交
4262
{
4263 4264
	switch (ret) {
	case GRO_NORMAL:
4265
		if (netif_receive_skb_internal(skb))
4266 4267
			ret = GRO_DROP;
		break;
H
Herbert Xu 已提交
4268

4269
	case GRO_DROP:
H
Herbert Xu 已提交
4270 4271
		kfree_skb(skb);
		break;
4272

4273
	case GRO_MERGED_FREE:
4274 4275 4276 4277
		if (NAPI_GRO_CB(skb)->free == NAPI_GRO_FREE_STOLEN_HEAD)
			kmem_cache_free(skbuff_head_cache, skb);
		else
			__kfree_skb(skb);
4278 4279
		break;

4280 4281 4282
	case GRO_HELD:
	case GRO_MERGED:
		break;
H
Herbert Xu 已提交
4283 4284
	}

4285
	return ret;
4286 4287
}

4288
gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
4289
{
4290
	trace_napi_gro_receive_entry(skb);
4291

4292 4293
	skb_gro_reset_offset(skb);

E
Eric Dumazet 已提交
4294
	return napi_skb_finish(dev_gro_receive(napi, skb), skb);
4295 4296 4297
}
EXPORT_SYMBOL(napi_gro_receive);

S
stephen hemminger 已提交
4298
static void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
4299
{
4300 4301 4302 4303
	if (unlikely(skb->pfmemalloc)) {
		consume_skb(skb);
		return;
	}
4304
	__skb_pull(skb, skb_headlen(skb));
4305 4306
	/* restore the reserve we had after netdev_alloc_skb_ip_align() */
	skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN - skb_headroom(skb));
4307
	skb->vlan_tci = 0;
H
Herbert Xu 已提交
4308
	skb->dev = napi->dev;
A
Andy Gospodarek 已提交
4309
	skb->skb_iif = 0;
4310 4311
	skb->encapsulation = 0;
	skb_shinfo(skb)->gso_type = 0;
4312
	skb->truesize = SKB_TRUESIZE(skb_end_offset(skb));
4313 4314 4315 4316

	napi->skb = skb;
}

4317
struct sk_buff *napi_get_frags(struct napi_struct *napi)
H
Herbert Xu 已提交
4318 4319 4320 4321
{
	struct sk_buff *skb = napi->skb;

	if (!skb) {
4322
		skb = napi_alloc_skb(napi, GRO_MAX_HEAD);
E
Eric Dumazet 已提交
4323
		napi->skb = skb;
4324
	}
4325 4326
	return skb;
}
4327
EXPORT_SYMBOL(napi_get_frags);
4328

4329 4330 4331
static gro_result_t napi_frags_finish(struct napi_struct *napi,
				      struct sk_buff *skb,
				      gro_result_t ret)
4332
{
4333 4334
	switch (ret) {
	case GRO_NORMAL:
4335 4336 4337 4338
	case GRO_HELD:
		__skb_push(skb, ETH_HLEN);
		skb->protocol = eth_type_trans(skb, skb->dev);
		if (ret == GRO_NORMAL && netif_receive_skb_internal(skb))
4339
			ret = GRO_DROP;
4340
		break;
H
Herbert Xu 已提交
4341

4342 4343 4344 4345
	case GRO_DROP:
	case GRO_MERGED_FREE:
		napi_reuse_skb(napi, skb);
		break;
4346 4347 4348

	case GRO_MERGED:
		break;
4349
	}
H
Herbert Xu 已提交
4350

4351
	return ret;
H
Herbert Xu 已提交
4352
}
4353

4354 4355 4356 4357
/* Upper GRO stack assumes network header starts at gro_offset=0
 * Drivers could call both napi_gro_frags() and napi_gro_receive()
 * We copy ethernet header into skb->data to have a common layout.
 */
E
Eric Dumazet 已提交
4358
static struct sk_buff *napi_frags_skb(struct napi_struct *napi)
4359 4360
{
	struct sk_buff *skb = napi->skb;
4361 4362
	const struct ethhdr *eth;
	unsigned int hlen = sizeof(*eth);
4363 4364 4365

	napi->skb = NULL;

4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379
	skb_reset_mac_header(skb);
	skb_gro_reset_offset(skb);

	eth = skb_gro_header_fast(skb, 0);
	if (unlikely(skb_gro_header_hard(skb, hlen))) {
		eth = skb_gro_header_slow(skb, hlen, 0);
		if (unlikely(!eth)) {
			napi_reuse_skb(napi, skb);
			return NULL;
		}
	} else {
		gro_pull_from_frag0(skb, hlen);
		NAPI_GRO_CB(skb)->frag0 += hlen;
		NAPI_GRO_CB(skb)->frag0_len -= hlen;
4380
	}
4381 4382 4383 4384 4385 4386 4387 4388
	__skb_pull(skb, hlen);

	/*
	 * This works because the only protocols we care about don't require
	 * special handling.
	 * We'll fix it up properly in napi_frags_finish()
	 */
	skb->protocol = eth->h_proto;
4389 4390 4391 4392

	return skb;
}

4393
gro_result_t napi_gro_frags(struct napi_struct *napi)
4394
{
4395
	struct sk_buff *skb = napi_frags_skb(napi);
4396 4397

	if (!skb)
4398
		return GRO_DROP;
4399

4400 4401
	trace_napi_gro_frags_entry(skb);

E
Eric Dumazet 已提交
4402
	return napi_frags_finish(napi, skb, dev_gro_receive(napi, skb));
4403
}
H
Herbert Xu 已提交
4404 4405
EXPORT_SYMBOL(napi_gro_frags);

T
Tom Herbert 已提交
4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430
/* Compute the checksum from gro_offset and return the folded value
 * after adding in any pseudo checksum.
 */
__sum16 __skb_gro_checksum_complete(struct sk_buff *skb)
{
	__wsum wsum;
	__sum16 sum;

	wsum = skb_checksum(skb, skb_gro_offset(skb), skb_gro_len(skb), 0);

	/* NAPI_GRO_CB(skb)->csum holds pseudo checksum */
	sum = csum_fold(csum_add(NAPI_GRO_CB(skb)->csum, wsum));
	if (likely(!sum)) {
		if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE) &&
		    !skb->csum_complete_sw)
			netdev_rx_csum_fault(skb->dev);
	}

	NAPI_GRO_CB(skb)->csum = wsum;
	NAPI_GRO_CB(skb)->csum_valid = 1;

	return sum;
}
EXPORT_SYMBOL(__skb_gro_checksum_complete);

4431
/*
4432
 * net_rps_action_and_irq_enable sends any pending IPI's for rps.
4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449
 * Note: called with local irq disabled, but exits with local irq enabled.
 */
static void net_rps_action_and_irq_enable(struct softnet_data *sd)
{
#ifdef CONFIG_RPS
	struct softnet_data *remsd = sd->rps_ipi_list;

	if (remsd) {
		sd->rps_ipi_list = NULL;

		local_irq_enable();

		/* Send pending IPI's to kick RPS processing on remote cpus. */
		while (remsd) {
			struct softnet_data *next = remsd->rps_ipi_next;

			if (cpu_online(remsd->cpu))
4450
				smp_call_function_single_async(remsd->cpu,
4451
							   &remsd->csd);
4452 4453 4454 4455 4456 4457 4458
			remsd = next;
		}
	} else
#endif
		local_irq_enable();
}

E
Eric Dumazet 已提交
4459 4460 4461 4462 4463 4464 4465 4466 4467
static bool sd_has_rps_ipi_waiting(struct softnet_data *sd)
{
#ifdef CONFIG_RPS
	return sd->rps_ipi_list != NULL;
#else
	return false;
#endif
}

4468
static int process_backlog(struct napi_struct *napi, int quota)
L
Linus Torvalds 已提交
4469 4470
{
	int work = 0;
E
Eric Dumazet 已提交
4471
	struct softnet_data *sd = container_of(napi, struct softnet_data, backlog);
L
Linus Torvalds 已提交
4472

4473 4474 4475
	/* Check if we have pending ipi, its better to send them now,
	 * not waiting net_rx_action() end.
	 */
E
Eric Dumazet 已提交
4476
	if (sd_has_rps_ipi_waiting(sd)) {
4477 4478 4479
		local_irq_disable();
		net_rps_action_and_irq_enable(sd);
	}
E
Eric Dumazet 已提交
4480

4481
	napi->weight = weight_p;
4482
	local_irq_disable();
4483
	while (1) {
L
Linus Torvalds 已提交
4484
		struct sk_buff *skb;
4485 4486 4487 4488 4489

		while ((skb = __skb_dequeue(&sd->process_queue))) {
			local_irq_enable();
			__netif_receive_skb(skb);
			local_irq_disable();
4490 4491 4492 4493 4494
			input_queue_head_incr(sd);
			if (++work >= quota) {
				local_irq_enable();
				return work;
			}
4495
		}
L
Linus Torvalds 已提交
4496

E
Eric Dumazet 已提交
4497
		rps_lock(sd);
4498
		if (skb_queue_empty(&sd->input_pkt_queue)) {
E
Eric Dumazet 已提交
4499 4500 4501
			/*
			 * Inline a custom version of __napi_complete().
			 * only current cpu owns and manipulates this napi,
4502 4503 4504
			 * and NAPI_STATE_SCHED is the only possible flag set
			 * on backlog.
			 * We can use a plain write instead of clear_bit(),
E
Eric Dumazet 已提交
4505 4506 4507
			 * and we dont need an smp_mb() memory barrier.
			 */
			napi->state = 0;
4508
			rps_unlock(sd);
E
Eric Dumazet 已提交
4509

4510
			break;
4511
		}
4512 4513 4514

		skb_queue_splice_tail_init(&sd->input_pkt_queue,
					   &sd->process_queue);
E
Eric Dumazet 已提交
4515
		rps_unlock(sd);
4516 4517
	}
	local_irq_enable();
L
Linus Torvalds 已提交
4518

4519 4520
	return work;
}
L
Linus Torvalds 已提交
4521

4522 4523
/**
 * __napi_schedule - schedule for receive
4524
 * @n: entry to schedule
4525
 *
4526 4527
 * The entry's receive function will be scheduled to run.
 * Consider using __napi_schedule_irqoff() if hard irqs are masked.
4528
 */
H
Harvey Harrison 已提交
4529
void __napi_schedule(struct napi_struct *n)
4530 4531
{
	unsigned long flags;
L
Linus Torvalds 已提交
4532

4533
	local_irq_save(flags);
4534
	____napi_schedule(this_cpu_ptr(&softnet_data), n);
4535
	local_irq_restore(flags);
L
Linus Torvalds 已提交
4536
}
4537 4538
EXPORT_SYMBOL(__napi_schedule);

4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550
/**
 * __napi_schedule_irqoff - schedule for receive
 * @n: entry to schedule
 *
 * Variant of __napi_schedule() assuming hard irqs are masked
 */
void __napi_schedule_irqoff(struct napi_struct *n)
{
	____napi_schedule(this_cpu_ptr(&softnet_data), n);
}
EXPORT_SYMBOL(__napi_schedule_irqoff);

4551 4552 4553 4554
void __napi_complete(struct napi_struct *n)
{
	BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));

E
Eric Dumazet 已提交
4555
	list_del_init(&n->poll_list);
4556
	smp_mb__before_atomic();
4557 4558 4559 4560
	clear_bit(NAPI_STATE_SCHED, &n->state);
}
EXPORT_SYMBOL(__napi_complete);

4561
void napi_complete_done(struct napi_struct *n, int work_done)
4562 4563 4564 4565 4566 4567 4568 4569 4570 4571
{
	unsigned long flags;

	/*
	 * don't let napi dequeue from the cpu poll list
	 * just in case its running on a different cpu
	 */
	if (unlikely(test_bit(NAPI_STATE_NPSVC, &n->state)))
		return;

4572 4573
	if (n->gro_list) {
		unsigned long timeout = 0;
E
Eric Dumazet 已提交
4574

4575 4576 4577 4578 4579 4580 4581 4582 4583
		if (work_done)
			timeout = n->dev->gro_flush_timeout;

		if (timeout)
			hrtimer_start(&n->timer, ns_to_ktime(timeout),
				      HRTIMER_MODE_REL_PINNED);
		else
			napi_gro_flush(n, false);
	}
E
Eric Dumazet 已提交
4584 4585 4586 4587 4588 4589 4590 4591
	if (likely(list_empty(&n->poll_list))) {
		WARN_ON_ONCE(!test_and_clear_bit(NAPI_STATE_SCHED, &n->state));
	} else {
		/* If n->poll_list is not empty, we need to mask irqs */
		local_irq_save(flags);
		__napi_complete(n);
		local_irq_restore(flags);
	}
4592
}
4593
EXPORT_SYMBOL(napi_complete_done);
4594

E
Eliezer Tamir 已提交
4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646
/* must be called under rcu_read_lock(), as we dont take a reference */
struct napi_struct *napi_by_id(unsigned int napi_id)
{
	unsigned int hash = napi_id % HASH_SIZE(napi_hash);
	struct napi_struct *napi;

	hlist_for_each_entry_rcu(napi, &napi_hash[hash], napi_hash_node)
		if (napi->napi_id == napi_id)
			return napi;

	return NULL;
}
EXPORT_SYMBOL_GPL(napi_by_id);

void napi_hash_add(struct napi_struct *napi)
{
	if (!test_and_set_bit(NAPI_STATE_HASHED, &napi->state)) {

		spin_lock(&napi_hash_lock);

		/* 0 is not a valid id, we also skip an id that is taken
		 * we expect both events to be extremely rare
		 */
		napi->napi_id = 0;
		while (!napi->napi_id) {
			napi->napi_id = ++napi_gen_id;
			if (napi_by_id(napi->napi_id))
				napi->napi_id = 0;
		}

		hlist_add_head_rcu(&napi->napi_hash_node,
			&napi_hash[napi->napi_id % HASH_SIZE(napi_hash)]);

		spin_unlock(&napi_hash_lock);
	}
}
EXPORT_SYMBOL_GPL(napi_hash_add);

/* Warning : caller is responsible to make sure rcu grace period
 * is respected before freeing memory containing @napi
 */
void napi_hash_del(struct napi_struct *napi)
{
	spin_lock(&napi_hash_lock);

	if (test_and_clear_bit(NAPI_STATE_HASHED, &napi->state))
		hlist_del_rcu(&napi->napi_hash_node);

	spin_unlock(&napi_hash_lock);
}
EXPORT_SYMBOL_GPL(napi_hash_del);

4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657
static enum hrtimer_restart napi_watchdog(struct hrtimer *timer)
{
	struct napi_struct *napi;

	napi = container_of(timer, struct napi_struct, timer);
	if (napi->gro_list)
		napi_schedule(napi);

	return HRTIMER_NORESTART;
}

4658 4659 4660 4661
void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
		    int (*poll)(struct napi_struct *, int), int weight)
{
	INIT_LIST_HEAD(&napi->poll_list);
4662 4663
	hrtimer_init(&napi->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_PINNED);
	napi->timer.function = napi_watchdog;
4664
	napi->gro_count = 0;
4665
	napi->gro_list = NULL;
H
Herbert Xu 已提交
4666
	napi->skb = NULL;
4667
	napi->poll = poll;
E
Eric Dumazet 已提交
4668 4669 4670
	if (weight > NAPI_POLL_WEIGHT)
		pr_err_once("netif_napi_add() called with weight %d on device %s\n",
			    weight, dev->name);
4671 4672 4673
	napi->weight = weight;
	list_add(&napi->dev_list, &dev->napi_list);
	napi->dev = dev;
H
Herbert Xu 已提交
4674
#ifdef CONFIG_NETPOLL
4675 4676 4677 4678 4679 4680 4681
	spin_lock_init(&napi->poll_lock);
	napi->poll_owner = -1;
#endif
	set_bit(NAPI_STATE_SCHED, &napi->state);
}
EXPORT_SYMBOL(netif_napi_add);

4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695
void napi_disable(struct napi_struct *n)
{
	might_sleep();
	set_bit(NAPI_STATE_DISABLE, &n->state);

	while (test_and_set_bit(NAPI_STATE_SCHED, &n->state))
		msleep(1);

	hrtimer_cancel(&n->timer);

	clear_bit(NAPI_STATE_DISABLE, &n->state);
}
EXPORT_SYMBOL(napi_disable);

4696 4697
void netif_napi_del(struct napi_struct *napi)
{
4698
	list_del_init(&napi->dev_list);
4699
	napi_free_frags(napi);
4700

E
Eric Dumazet 已提交
4701
	kfree_skb_list(napi->gro_list);
4702
	napi->gro_list = NULL;
4703
	napi->gro_count = 0;
4704 4705 4706
}
EXPORT_SYMBOL(netif_napi_del);

4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751
static int napi_poll(struct napi_struct *n, struct list_head *repoll)
{
	void *have;
	int work, weight;

	list_del_init(&n->poll_list);

	have = netpoll_poll_lock(n);

	weight = n->weight;

	/* This NAPI_STATE_SCHED test is for avoiding a race
	 * with netpoll's poll_napi().  Only the entity which
	 * obtains the lock and sees NAPI_STATE_SCHED set will
	 * actually make the ->poll() call.  Therefore we avoid
	 * accidentally calling ->poll() when NAPI is not scheduled.
	 */
	work = 0;
	if (test_bit(NAPI_STATE_SCHED, &n->state)) {
		work = n->poll(n, weight);
		trace_napi_poll(n);
	}

	WARN_ON_ONCE(work > weight);

	if (likely(work < weight))
		goto out_unlock;

	/* Drivers must not modify the NAPI state if they
	 * consume the entire weight.  In such cases this code
	 * still "owns" the NAPI instance and therefore can
	 * move the instance around on the list at-will.
	 */
	if (unlikely(napi_disable_pending(n))) {
		napi_complete(n);
		goto out_unlock;
	}

	if (n->gro_list) {
		/* flush too old packets
		 * If HZ < 1000, flush all packets.
		 */
		napi_gro_flush(n, HZ >= 1000);
	}

4752 4753 4754 4755 4756 4757 4758 4759 4760
	/* Some drivers may have called napi_schedule
	 * prior to exhausting their budget.
	 */
	if (unlikely(!list_empty(&n->poll_list))) {
		pr_warn_once("%s: Budget exhausted after napi rescheduled\n",
			     n->dev ? n->dev->name : "backlog");
		goto out_unlock;
	}

4761 4762 4763 4764 4765 4766 4767 4768
	list_add_tail(&n->poll_list, repoll);

out_unlock:
	netpoll_poll_unlock(have);

	return work;
}

L
Linus Torvalds 已提交
4769 4770
static void net_rx_action(struct softirq_action *h)
{
4771
	struct softnet_data *sd = this_cpu_ptr(&softnet_data);
4772
	unsigned long time_limit = jiffies + 2;
4773
	int budget = netdev_budget;
E
Eric Dumazet 已提交
4774 4775
	LIST_HEAD(list);
	LIST_HEAD(repoll);
4776

L
Linus Torvalds 已提交
4777
	local_irq_disable();
E
Eric Dumazet 已提交
4778 4779
	list_splice_init(&sd->poll_list, &list);
	local_irq_enable();
L
Linus Torvalds 已提交
4780

H
Herbert Xu 已提交
4781
	for (;;) {
4782
		struct napi_struct *n;
L
Linus Torvalds 已提交
4783

H
Herbert Xu 已提交
4784 4785 4786 4787 4788 4789
		if (list_empty(&list)) {
			if (!sd_has_rps_ipi_waiting(sd) && list_empty(&repoll))
				return;
			break;
		}

4790 4791 4792
		n = list_first_entry(&list, struct napi_struct, poll_list);
		budget -= napi_poll(n, &repoll);

E
Eric Dumazet 已提交
4793
		/* If softirq window is exhausted then punt.
4794 4795
		 * Allow this to run for 2 jiffies since which will allow
		 * an average latency of 1.5/HZ.
4796
		 */
H
Herbert Xu 已提交
4797 4798 4799 4800 4801
		if (unlikely(budget <= 0 ||
			     time_after_eq(jiffies, time_limit))) {
			sd->time_squeeze++;
			break;
		}
L
Linus Torvalds 已提交
4802
	}
E
Eric Dumazet 已提交
4803 4804 4805 4806 4807 4808 4809 4810 4811

	local_irq_disable();

	list_splice_tail_init(&sd->poll_list, &list);
	list_splice_tail(&repoll, &list);
	list_splice(&list, &sd->poll_list);
	if (!list_empty(&sd->poll_list))
		__raise_softirq_irqoff(NET_RX_SOFTIRQ);

4812
	net_rps_action_and_irq_enable(sd);
L
Linus Torvalds 已提交
4813 4814
}

4815
struct netdev_adjacent {
J
Jiri Pirko 已提交
4816
	struct net_device *dev;
4817 4818

	/* upper master flag, there can only be one master device per list */
J
Jiri Pirko 已提交
4819
	bool master;
4820 4821 4822 4823

	/* counter for the number of times this device was added to us */
	u16 ref_nr;

4824 4825 4826
	/* private field for the users */
	void *private;

J
Jiri Pirko 已提交
4827 4828 4829 4830
	struct list_head list;
	struct rcu_head rcu;
};

4831 4832
static struct netdev_adjacent *__netdev_find_adj(struct net_device *dev,
						 struct net_device *adj_dev,
4833
						 struct list_head *adj_list)
J
Jiri Pirko 已提交
4834
{
4835 4836
	struct netdev_adjacent *adj;

4837
	list_for_each_entry(adj, adj_list, list) {
4838 4839
		if (adj->dev == adj_dev)
			return adj;
J
Jiri Pirko 已提交
4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857
	}
	return NULL;
}

/**
 * netdev_has_upper_dev - Check if device is linked to an upper device
 * @dev: device
 * @upper_dev: upper device to check
 *
 * Find out if a device is linked to specified upper device and return true
 * in case it is. Note that this checks only immediate upper device,
 * not through a complete stack of devices. The caller must hold the RTNL lock.
 */
bool netdev_has_upper_dev(struct net_device *dev,
			  struct net_device *upper_dev)
{
	ASSERT_RTNL();

4858
	return __netdev_find_adj(dev, upper_dev, &dev->all_adj_list.upper);
J
Jiri Pirko 已提交
4859 4860 4861 4862 4863 4864 4865 4866 4867 4868
}
EXPORT_SYMBOL(netdev_has_upper_dev);

/**
 * netdev_has_any_upper_dev - Check if device is linked to some device
 * @dev: device
 *
 * Find out if a device is linked to an upper device and return true in case
 * it is. The caller must hold the RTNL lock.
 */
S
stephen hemminger 已提交
4869
static bool netdev_has_any_upper_dev(struct net_device *dev)
J
Jiri Pirko 已提交
4870 4871 4872
{
	ASSERT_RTNL();

4873
	return !list_empty(&dev->all_adj_list.upper);
J
Jiri Pirko 已提交
4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884
}

/**
 * netdev_master_upper_dev_get - Get master upper device
 * @dev: device
 *
 * Find a master upper device and return pointer to it or NULL in case
 * it's not there. The caller must hold the RTNL lock.
 */
struct net_device *netdev_master_upper_dev_get(struct net_device *dev)
{
4885
	struct netdev_adjacent *upper;
J
Jiri Pirko 已提交
4886 4887 4888

	ASSERT_RTNL();

4889
	if (list_empty(&dev->adj_list.upper))
J
Jiri Pirko 已提交
4890 4891
		return NULL;

4892
	upper = list_first_entry(&dev->adj_list.upper,
4893
				 struct netdev_adjacent, list);
J
Jiri Pirko 已提交
4894 4895 4896 4897 4898 4899
	if (likely(upper->master))
		return upper->dev;
	return NULL;
}
EXPORT_SYMBOL(netdev_master_upper_dev_get);

4900 4901 4902 4903 4904 4905 4906 4907 4908 4909
void *netdev_adjacent_get_private(struct list_head *adj_list)
{
	struct netdev_adjacent *adj;

	adj = list_entry(adj_list, struct netdev_adjacent, list);

	return adj->private;
}
EXPORT_SYMBOL(netdev_adjacent_get_private);

4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935
/**
 * netdev_upper_get_next_dev_rcu - Get the next dev from upper list
 * @dev: device
 * @iter: list_head ** of the current position
 *
 * Gets the next device from the dev's upper list, starting from iter
 * position. The caller must hold RCU read lock.
 */
struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
						 struct list_head **iter)
{
	struct netdev_adjacent *upper;

	WARN_ON_ONCE(!rcu_read_lock_held() && !lockdep_rtnl_is_held());

	upper = list_entry_rcu((*iter)->next, struct netdev_adjacent, list);

	if (&upper->list == &dev->adj_list.upper)
		return NULL;

	*iter = &upper->list;

	return upper->dev;
}
EXPORT_SYMBOL(netdev_upper_get_next_dev_rcu);

4936 4937
/**
 * netdev_all_upper_get_next_dev_rcu - Get the next dev from upper list
4938 4939 4940 4941 4942 4943
 * @dev: device
 * @iter: list_head ** of the current position
 *
 * Gets the next device from the dev's upper list, starting from iter
 * position. The caller must hold RCU read lock.
 */
4944 4945
struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
						     struct list_head **iter)
4946 4947 4948
{
	struct netdev_adjacent *upper;

4949
	WARN_ON_ONCE(!rcu_read_lock_held() && !lockdep_rtnl_is_held());
4950 4951 4952

	upper = list_entry_rcu((*iter)->next, struct netdev_adjacent, list);

4953
	if (&upper->list == &dev->all_adj_list.upper)
4954 4955 4956 4957 4958 4959
		return NULL;

	*iter = &upper->list;

	return upper->dev;
}
4960
EXPORT_SYMBOL(netdev_all_upper_get_next_dev_rcu);
4961

4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982
/**
 * netdev_lower_get_next_private - Get the next ->private from the
 *				   lower neighbour list
 * @dev: device
 * @iter: list_head ** of the current position
 *
 * Gets the next netdev_adjacent->private from the dev's lower neighbour
 * list, starting from iter position. The caller must hold either hold the
 * RTNL lock or its own locking that guarantees that the neighbour lower
 * list will remain unchainged.
 */
void *netdev_lower_get_next_private(struct net_device *dev,
				    struct list_head **iter)
{
	struct netdev_adjacent *lower;

	lower = list_entry(*iter, struct netdev_adjacent, list);

	if (&lower->list == &dev->adj_list.lower)
		return NULL;

4983
	*iter = lower->list.next;
4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010

	return lower->private;
}
EXPORT_SYMBOL(netdev_lower_get_next_private);

/**
 * netdev_lower_get_next_private_rcu - Get the next ->private from the
 *				       lower neighbour list, RCU
 *				       variant
 * @dev: device
 * @iter: list_head ** of the current position
 *
 * Gets the next netdev_adjacent->private from the dev's lower neighbour
 * list, starting from iter position. The caller must hold RCU read lock.
 */
void *netdev_lower_get_next_private_rcu(struct net_device *dev,
					struct list_head **iter)
{
	struct netdev_adjacent *lower;

	WARN_ON_ONCE(!rcu_read_lock_held());

	lower = list_entry_rcu((*iter)->next, struct netdev_adjacent, list);

	if (&lower->list == &dev->adj_list.lower)
		return NULL;

5011
	*iter = &lower->list;
5012 5013 5014 5015 5016

	return lower->private;
}
EXPORT_SYMBOL(netdev_lower_get_next_private_rcu);

5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042
/**
 * netdev_lower_get_next - Get the next device from the lower neighbour
 *                         list
 * @dev: device
 * @iter: list_head ** of the current position
 *
 * Gets the next netdev_adjacent from the dev's lower neighbour
 * list, starting from iter position. The caller must hold RTNL lock or
 * its own locking that guarantees that the neighbour lower
 * list will remain unchainged.
 */
void *netdev_lower_get_next(struct net_device *dev, struct list_head **iter)
{
	struct netdev_adjacent *lower;

	lower = list_entry((*iter)->next, struct netdev_adjacent, list);

	if (&lower->list == &dev->adj_list.lower)
		return NULL;

	*iter = &lower->list;

	return lower->dev;
}
EXPORT_SYMBOL(netdev_lower_get_next);

5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063
/**
 * netdev_lower_get_first_private_rcu - Get the first ->private from the
 *				       lower neighbour list, RCU
 *				       variant
 * @dev: device
 *
 * Gets the first netdev_adjacent->private from the dev's lower neighbour
 * list. The caller must hold RCU read lock.
 */
void *netdev_lower_get_first_private_rcu(struct net_device *dev)
{
	struct netdev_adjacent *lower;

	lower = list_first_or_null_rcu(&dev->adj_list.lower,
			struct netdev_adjacent, list);
	if (lower)
		return lower->private;
	return NULL;
}
EXPORT_SYMBOL(netdev_lower_get_first_private_rcu);

J
Jiri Pirko 已提交
5064 5065 5066 5067 5068 5069 5070 5071 5072
/**
 * netdev_master_upper_dev_get_rcu - Get master upper device
 * @dev: device
 *
 * Find a master upper device and return pointer to it or NULL in case
 * it's not there. The caller must hold the RCU read lock.
 */
struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev)
{
5073
	struct netdev_adjacent *upper;
J
Jiri Pirko 已提交
5074

5075
	upper = list_first_or_null_rcu(&dev->adj_list.upper,
5076
				       struct netdev_adjacent, list);
J
Jiri Pirko 已提交
5077 5078 5079 5080 5081 5082
	if (upper && likely(upper->master))
		return upper->dev;
	return NULL;
}
EXPORT_SYMBOL(netdev_master_upper_dev_get_rcu);

5083
static int netdev_adjacent_sysfs_add(struct net_device *dev,
5084 5085 5086 5087 5088 5089 5090 5091 5092
			      struct net_device *adj_dev,
			      struct list_head *dev_list)
{
	char linkname[IFNAMSIZ+7];
	sprintf(linkname, dev_list == &dev->adj_list.upper ?
		"upper_%s" : "lower_%s", adj_dev->name);
	return sysfs_create_link(&(dev->dev.kobj), &(adj_dev->dev.kobj),
				 linkname);
}
5093
static void netdev_adjacent_sysfs_del(struct net_device *dev,
5094 5095 5096 5097 5098 5099 5100 5101 5102
			       char *name,
			       struct list_head *dev_list)
{
	char linkname[IFNAMSIZ+7];
	sprintf(linkname, dev_list == &dev->adj_list.upper ?
		"upper_%s" : "lower_%s", name);
	sysfs_remove_link(&(dev->dev.kobj), linkname);
}

5103 5104 5105 5106 5107 5108 5109 5110
static inline bool netdev_adjacent_is_neigh_list(struct net_device *dev,
						 struct net_device *adj_dev,
						 struct list_head *dev_list)
{
	return (dev_list == &dev->adj_list.upper ||
		dev_list == &dev->adj_list.lower) &&
		net_eq(dev_net(dev), dev_net(adj_dev));
}
5111

5112 5113
static int __netdev_adjacent_dev_insert(struct net_device *dev,
					struct net_device *adj_dev,
5114
					struct list_head *dev_list,
5115
					void *private, bool master)
5116 5117
{
	struct netdev_adjacent *adj;
5118
	int ret;
5119

5120
	adj = __netdev_find_adj(dev, adj_dev, dev_list);
5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133

	if (adj) {
		adj->ref_nr++;
		return 0;
	}

	adj = kmalloc(sizeof(*adj), GFP_KERNEL);
	if (!adj)
		return -ENOMEM;

	adj->dev = adj_dev;
	adj->master = master;
	adj->ref_nr = 1;
5134
	adj->private = private;
5135
	dev_hold(adj_dev);
5136 5137 5138

	pr_debug("dev_hold for %s, because of link added from %s to %s\n",
		 adj_dev->name, dev->name, adj_dev->name);
5139

5140
	if (netdev_adjacent_is_neigh_list(dev, adj_dev, dev_list)) {
5141
		ret = netdev_adjacent_sysfs_add(dev, adj_dev, dev_list);
5142 5143 5144 5145
		if (ret)
			goto free_adj;
	}

5146
	/* Ensure that master link is always the first item in list. */
5147 5148 5149 5150
	if (master) {
		ret = sysfs_create_link(&(dev->dev.kobj),
					&(adj_dev->dev.kobj), "master");
		if (ret)
5151
			goto remove_symlinks;
5152

5153
		list_add_rcu(&adj->list, dev_list);
5154
	} else {
5155
		list_add_tail_rcu(&adj->list, dev_list);
5156
	}
5157 5158

	return 0;
5159

5160
remove_symlinks:
5161
	if (netdev_adjacent_is_neigh_list(dev, adj_dev, dev_list))
5162
		netdev_adjacent_sysfs_del(dev, adj_dev->name, dev_list);
5163 5164
free_adj:
	kfree(adj);
5165
	dev_put(adj_dev);
5166 5167

	return ret;
5168 5169
}

S
stephen hemminger 已提交
5170 5171 5172
static void __netdev_adjacent_dev_remove(struct net_device *dev,
					 struct net_device *adj_dev,
					 struct list_head *dev_list)
5173 5174 5175
{
	struct netdev_adjacent *adj;

5176
	adj = __netdev_find_adj(dev, adj_dev, dev_list);
5177

5178 5179 5180
	if (!adj) {
		pr_err("tried to remove device %s from %s\n",
		       dev->name, adj_dev->name);
5181
		BUG();
5182
	}
5183 5184

	if (adj->ref_nr > 1) {
5185 5186
		pr_debug("%s to %s ref_nr-- = %d\n", dev->name, adj_dev->name,
			 adj->ref_nr-1);
5187 5188 5189 5190
		adj->ref_nr--;
		return;
	}

5191 5192 5193
	if (adj->master)
		sysfs_remove_link(&(dev->dev.kobj), "master");

5194
	if (netdev_adjacent_is_neigh_list(dev, adj_dev, dev_list))
5195
		netdev_adjacent_sysfs_del(dev, adj_dev->name, dev_list);
5196

5197
	list_del_rcu(&adj->list);
5198 5199
	pr_debug("dev_put for %s, because link removed from %s to %s\n",
		 adj_dev->name, dev->name, adj_dev->name);
5200 5201 5202 5203
	dev_put(adj_dev);
	kfree_rcu(adj, rcu);
}

S
stephen hemminger 已提交
5204 5205 5206 5207 5208
static int __netdev_adjacent_dev_link_lists(struct net_device *dev,
					    struct net_device *upper_dev,
					    struct list_head *up_list,
					    struct list_head *down_list,
					    void *private, bool master)
5209 5210 5211
{
	int ret;

5212 5213
	ret = __netdev_adjacent_dev_insert(dev, upper_dev, up_list, private,
					   master);
5214 5215 5216
	if (ret)
		return ret;

5217 5218
	ret = __netdev_adjacent_dev_insert(upper_dev, dev, down_list, private,
					   false);
5219
	if (ret) {
5220
		__netdev_adjacent_dev_remove(dev, upper_dev, up_list);
5221 5222 5223 5224 5225 5226
		return ret;
	}

	return 0;
}

S
stephen hemminger 已提交
5227 5228
static int __netdev_adjacent_dev_link(struct net_device *dev,
				      struct net_device *upper_dev)
5229
{
5230 5231 5232
	return __netdev_adjacent_dev_link_lists(dev, upper_dev,
						&dev->all_adj_list.upper,
						&upper_dev->all_adj_list.lower,
5233
						NULL, false);
5234 5235
}

S
stephen hemminger 已提交
5236 5237 5238 5239
static void __netdev_adjacent_dev_unlink_lists(struct net_device *dev,
					       struct net_device *upper_dev,
					       struct list_head *up_list,
					       struct list_head *down_list)
5240
{
5241 5242
	__netdev_adjacent_dev_remove(dev, upper_dev, up_list);
	__netdev_adjacent_dev_remove(upper_dev, dev, down_list);
5243 5244
}

S
stephen hemminger 已提交
5245 5246
static void __netdev_adjacent_dev_unlink(struct net_device *dev,
					 struct net_device *upper_dev)
5247
{
5248 5249 5250 5251 5252
	__netdev_adjacent_dev_unlink_lists(dev, upper_dev,
					   &dev->all_adj_list.upper,
					   &upper_dev->all_adj_list.lower);
}

S
stephen hemminger 已提交
5253 5254 5255
static int __netdev_adjacent_dev_link_neighbour(struct net_device *dev,
						struct net_device *upper_dev,
						void *private, bool master)
5256 5257 5258 5259 5260 5261 5262 5263 5264
{
	int ret = __netdev_adjacent_dev_link(dev, upper_dev);

	if (ret)
		return ret;

	ret = __netdev_adjacent_dev_link_lists(dev, upper_dev,
					       &dev->adj_list.upper,
					       &upper_dev->adj_list.lower,
5265
					       private, master);
5266 5267 5268 5269 5270 5271
	if (ret) {
		__netdev_adjacent_dev_unlink(dev, upper_dev);
		return ret;
	}

	return 0;
5272 5273
}

S
stephen hemminger 已提交
5274 5275
static void __netdev_adjacent_dev_unlink_neighbour(struct net_device *dev,
						   struct net_device *upper_dev)
5276 5277 5278 5279 5280 5281
{
	__netdev_adjacent_dev_unlink(dev, upper_dev);
	__netdev_adjacent_dev_unlink_lists(dev, upper_dev,
					   &dev->adj_list.upper,
					   &upper_dev->adj_list.lower);
}
5282

J
Jiri Pirko 已提交
5283
static int __netdev_upper_dev_link(struct net_device *dev,
5284 5285
				   struct net_device *upper_dev, bool master,
				   void *private)
J
Jiri Pirko 已提交
5286
{
5287 5288
	struct netdev_adjacent *i, *j, *to_i, *to_j;
	int ret = 0;
J
Jiri Pirko 已提交
5289 5290 5291 5292 5293 5294 5295

	ASSERT_RTNL();

	if (dev == upper_dev)
		return -EBUSY;

	/* To prevent loops, check if dev is not upper device to upper_dev. */
5296
	if (__netdev_find_adj(upper_dev, dev, &upper_dev->all_adj_list.upper))
J
Jiri Pirko 已提交
5297 5298
		return -EBUSY;

5299
	if (__netdev_find_adj(dev, upper_dev, &dev->adj_list.upper))
J
Jiri Pirko 已提交
5300 5301 5302 5303 5304
		return -EEXIST;

	if (master && netdev_master_upper_dev_get(dev))
		return -EBUSY;

5305 5306
	ret = __netdev_adjacent_dev_link_neighbour(dev, upper_dev, private,
						   master);
5307 5308
	if (ret)
		return ret;
J
Jiri Pirko 已提交
5309

5310
	/* Now that we linked these devs, make all the upper_dev's
5311
	 * all_adj_list.upper visible to every dev's all_adj_list.lower an
5312 5313 5314
	 * versa, and don't forget the devices itself. All of these
	 * links are non-neighbours.
	 */
5315 5316 5317 5318
	list_for_each_entry(i, &dev->all_adj_list.lower, list) {
		list_for_each_entry(j, &upper_dev->all_adj_list.upper, list) {
			pr_debug("Interlinking %s with %s, non-neighbour\n",
				 i->dev->name, j->dev->name);
5319 5320 5321 5322 5323 5324 5325
			ret = __netdev_adjacent_dev_link(i->dev, j->dev);
			if (ret)
				goto rollback_mesh;
		}
	}

	/* add dev to every upper_dev's upper device */
5326 5327 5328
	list_for_each_entry(i, &upper_dev->all_adj_list.upper, list) {
		pr_debug("linking %s's upper device %s with %s\n",
			 upper_dev->name, i->dev->name, dev->name);
5329 5330 5331 5332 5333 5334
		ret = __netdev_adjacent_dev_link(dev, i->dev);
		if (ret)
			goto rollback_upper_mesh;
	}

	/* add upper_dev to every dev's lower device */
5335 5336 5337
	list_for_each_entry(i, &dev->all_adj_list.lower, list) {
		pr_debug("linking %s's lower device %s with %s\n", dev->name,
			 i->dev->name, upper_dev->name);
5338 5339 5340 5341
		ret = __netdev_adjacent_dev_link(i->dev, upper_dev);
		if (ret)
			goto rollback_lower_mesh;
	}
J
Jiri Pirko 已提交
5342

5343
	call_netdevice_notifiers(NETDEV_CHANGEUPPER, dev);
J
Jiri Pirko 已提交
5344
	return 0;
5345 5346 5347

rollback_lower_mesh:
	to_i = i;
5348
	list_for_each_entry(i, &dev->all_adj_list.lower, list) {
5349 5350 5351 5352 5353 5354 5355 5356 5357
		if (i == to_i)
			break;
		__netdev_adjacent_dev_unlink(i->dev, upper_dev);
	}

	i = NULL;

rollback_upper_mesh:
	to_i = i;
5358
	list_for_each_entry(i, &upper_dev->all_adj_list.upper, list) {
5359 5360 5361 5362 5363 5364 5365 5366 5367 5368
		if (i == to_i)
			break;
		__netdev_adjacent_dev_unlink(dev, i->dev);
	}

	i = j = NULL;

rollback_mesh:
	to_i = i;
	to_j = j;
5369 5370
	list_for_each_entry(i, &dev->all_adj_list.lower, list) {
		list_for_each_entry(j, &upper_dev->all_adj_list.upper, list) {
5371 5372 5373 5374 5375 5376 5377 5378
			if (i == to_i && j == to_j)
				break;
			__netdev_adjacent_dev_unlink(i->dev, j->dev);
		}
		if (i == to_i)
			break;
	}

5379
	__netdev_adjacent_dev_unlink_neighbour(dev, upper_dev);
5380 5381

	return ret;
J
Jiri Pirko 已提交
5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396
}

/**
 * netdev_upper_dev_link - Add a link to the upper device
 * @dev: device
 * @upper_dev: new upper device
 *
 * Adds a link to device which is upper to this one. The caller must hold
 * the RTNL lock. On a failure a negative errno code is returned.
 * On success the reference counts are adjusted and the function
 * returns zero.
 */
int netdev_upper_dev_link(struct net_device *dev,
			  struct net_device *upper_dev)
{
5397
	return __netdev_upper_dev_link(dev, upper_dev, false, NULL);
J
Jiri Pirko 已提交
5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414
}
EXPORT_SYMBOL(netdev_upper_dev_link);

/**
 * netdev_master_upper_dev_link - Add a master link to the upper device
 * @dev: device
 * @upper_dev: new upper device
 *
 * Adds a link to device which is upper to this one. In this case, only
 * one master upper device can be linked, although other non-master devices
 * might be linked as well. The caller must hold the RTNL lock.
 * On a failure a negative errno code is returned. On success the reference
 * counts are adjusted and the function returns zero.
 */
int netdev_master_upper_dev_link(struct net_device *dev,
				 struct net_device *upper_dev)
{
5415
	return __netdev_upper_dev_link(dev, upper_dev, true, NULL);
J
Jiri Pirko 已提交
5416 5417 5418
}
EXPORT_SYMBOL(netdev_master_upper_dev_link);

5419 5420 5421 5422 5423 5424 5425 5426
int netdev_master_upper_dev_link_private(struct net_device *dev,
					 struct net_device *upper_dev,
					 void *private)
{
	return __netdev_upper_dev_link(dev, upper_dev, true, private);
}
EXPORT_SYMBOL(netdev_master_upper_dev_link_private);

J
Jiri Pirko 已提交
5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437
/**
 * netdev_upper_dev_unlink - Removes a link to upper device
 * @dev: device
 * @upper_dev: new upper device
 *
 * Removes a link to device which is upper to this one. The caller must hold
 * the RTNL lock.
 */
void netdev_upper_dev_unlink(struct net_device *dev,
			     struct net_device *upper_dev)
{
5438
	struct netdev_adjacent *i, *j;
J
Jiri Pirko 已提交
5439 5440
	ASSERT_RTNL();

5441
	__netdev_adjacent_dev_unlink_neighbour(dev, upper_dev);
5442 5443 5444 5445 5446

	/* Here is the tricky part. We must remove all dev's lower
	 * devices from all upper_dev's upper devices and vice
	 * versa, to maintain the graph relationship.
	 */
5447 5448
	list_for_each_entry(i, &dev->all_adj_list.lower, list)
		list_for_each_entry(j, &upper_dev->all_adj_list.upper, list)
5449 5450 5451 5452 5453
			__netdev_adjacent_dev_unlink(i->dev, j->dev);

	/* remove also the devices itself from lower/upper device
	 * list
	 */
5454
	list_for_each_entry(i, &dev->all_adj_list.lower, list)
5455 5456
		__netdev_adjacent_dev_unlink(i->dev, upper_dev);

5457
	list_for_each_entry(i, &upper_dev->all_adj_list.upper, list)
5458 5459
		__netdev_adjacent_dev_unlink(dev, i->dev);

5460
	call_netdevice_notifiers(NETDEV_CHANGEUPPER, dev);
J
Jiri Pirko 已提交
5461 5462 5463
}
EXPORT_SYMBOL(netdev_upper_dev_unlink);

5464 5465 5466
/**
 * netdev_bonding_info_change - Dispatch event about slave change
 * @dev: device
5467
 * @bonding_info: info to dispatch
5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483
 *
 * Send NETDEV_BONDING_INFO to netdev notifiers with info.
 * The caller must hold the RTNL lock.
 */
void netdev_bonding_info_change(struct net_device *dev,
				struct netdev_bonding_info *bonding_info)
{
	struct netdev_notifier_bonding_info	info;

	memcpy(&info.bonding_info, bonding_info,
	       sizeof(struct netdev_bonding_info));
	call_netdevice_notifiers_info(NETDEV_BONDING_INFO, dev,
				      &info.info);
}
EXPORT_SYMBOL(netdev_bonding_info_change);

E
Eric Dumazet 已提交
5484
static void netdev_adjacent_add_links(struct net_device *dev)
5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508
{
	struct netdev_adjacent *iter;

	struct net *net = dev_net(dev);

	list_for_each_entry(iter, &dev->adj_list.upper, list) {
		if (!net_eq(net,dev_net(iter->dev)))
			continue;
		netdev_adjacent_sysfs_add(iter->dev, dev,
					  &iter->dev->adj_list.lower);
		netdev_adjacent_sysfs_add(dev, iter->dev,
					  &dev->adj_list.upper);
	}

	list_for_each_entry(iter, &dev->adj_list.lower, list) {
		if (!net_eq(net,dev_net(iter->dev)))
			continue;
		netdev_adjacent_sysfs_add(iter->dev, dev,
					  &iter->dev->adj_list.upper);
		netdev_adjacent_sysfs_add(dev, iter->dev,
					  &dev->adj_list.lower);
	}
}

E
Eric Dumazet 已提交
5509
static void netdev_adjacent_del_links(struct net_device *dev)
5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533
{
	struct netdev_adjacent *iter;

	struct net *net = dev_net(dev);

	list_for_each_entry(iter, &dev->adj_list.upper, list) {
		if (!net_eq(net,dev_net(iter->dev)))
			continue;
		netdev_adjacent_sysfs_del(iter->dev, dev->name,
					  &iter->dev->adj_list.lower);
		netdev_adjacent_sysfs_del(dev, iter->dev->name,
					  &dev->adj_list.upper);
	}

	list_for_each_entry(iter, &dev->adj_list.lower, list) {
		if (!net_eq(net,dev_net(iter->dev)))
			continue;
		netdev_adjacent_sysfs_del(iter->dev, dev->name,
					  &iter->dev->adj_list.upper);
		netdev_adjacent_sysfs_del(dev, iter->dev->name,
					  &dev->adj_list.lower);
	}
}

5534
void netdev_adjacent_rename_links(struct net_device *dev, char *oldname)
5535
{
5536
	struct netdev_adjacent *iter;
5537

5538 5539
	struct net *net = dev_net(dev);

5540
	list_for_each_entry(iter, &dev->adj_list.upper, list) {
5541 5542
		if (!net_eq(net,dev_net(iter->dev)))
			continue;
5543 5544 5545 5546 5547
		netdev_adjacent_sysfs_del(iter->dev, oldname,
					  &iter->dev->adj_list.lower);
		netdev_adjacent_sysfs_add(iter->dev, dev,
					  &iter->dev->adj_list.lower);
	}
5548

5549
	list_for_each_entry(iter, &dev->adj_list.lower, list) {
5550 5551
		if (!net_eq(net,dev_net(iter->dev)))
			continue;
5552 5553 5554 5555 5556
		netdev_adjacent_sysfs_del(iter->dev, oldname,
					  &iter->dev->adj_list.upper);
		netdev_adjacent_sysfs_add(iter->dev, dev,
					  &iter->dev->adj_list.upper);
	}
5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573
}

void *netdev_lower_dev_get_private(struct net_device *dev,
				   struct net_device *lower_dev)
{
	struct netdev_adjacent *lower;

	if (!lower_dev)
		return NULL;
	lower = __netdev_find_adj(dev, lower_dev, &dev->adj_list.lower);
	if (!lower)
		return NULL;

	return lower->private;
}
EXPORT_SYMBOL(netdev_lower_dev_get_private);

5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597

int dev_get_nest_level(struct net_device *dev,
		       bool (*type_check)(struct net_device *dev))
{
	struct net_device *lower = NULL;
	struct list_head *iter;
	int max_nest = -1;
	int nest;

	ASSERT_RTNL();

	netdev_for_each_lower_dev(dev, lower, iter) {
		nest = dev_get_nest_level(lower, type_check);
		if (max_nest < nest)
			max_nest = nest;
	}

	if (type_check(dev))
		max_nest++;

	return max_nest;
}
EXPORT_SYMBOL(dev_get_nest_level);

5598 5599
static void dev_change_rx_flags(struct net_device *dev, int flags)
{
5600 5601
	const struct net_device_ops *ops = dev->netdev_ops;

5602
	if (ops->ndo_change_rx_flags)
5603
		ops->ndo_change_rx_flags(dev, flags);
5604 5605
}

5606
static int __dev_set_promiscuity(struct net_device *dev, int inc, bool notify)
L
Linus Torvalds 已提交
5607
{
5608
	unsigned int old_flags = dev->flags;
5609 5610
	kuid_t uid;
	kgid_t gid;
L
Linus Torvalds 已提交
5611

5612 5613
	ASSERT_RTNL();

5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624
	dev->flags |= IFF_PROMISC;
	dev->promiscuity += inc;
	if (dev->promiscuity == 0) {
		/*
		 * Avoid overflow.
		 * If inc causes overflow, untouch promisc and return error.
		 */
		if (inc < 0)
			dev->flags &= ~IFF_PROMISC;
		else {
			dev->promiscuity -= inc;
5625 5626
			pr_warn("%s: promiscuity touches roof, set promiscuity failed. promiscuity feature of device might be broken.\n",
				dev->name);
5627 5628 5629
			return -EOVERFLOW;
		}
	}
5630
	if (dev->flags != old_flags) {
5631 5632 5633
		pr_info("device %s %s promiscuous mode\n",
			dev->name,
			dev->flags & IFF_PROMISC ? "entered" : "left");
5634 5635
		if (audit_enabled) {
			current_uid_gid(&uid, &gid);
5636 5637 5638 5639 5640
			audit_log(current->audit_context, GFP_ATOMIC,
				AUDIT_ANOM_PROMISCUOUS,
				"dev=%s prom=%d old_prom=%d auid=%u uid=%u gid=%u ses=%u",
				dev->name, (dev->flags & IFF_PROMISC),
				(old_flags & IFF_PROMISC),
5641
				from_kuid(&init_user_ns, audit_get_loginuid(current)),
5642 5643
				from_kuid(&init_user_ns, uid),
				from_kgid(&init_user_ns, gid),
5644
				audit_get_sessionid(current));
5645
		}
5646

5647
		dev_change_rx_flags(dev, IFF_PROMISC);
L
Linus Torvalds 已提交
5648
	}
5649 5650
	if (notify)
		__dev_notify_flags(dev, old_flags, IFF_PROMISC);
5651
	return 0;
L
Linus Torvalds 已提交
5652 5653
}

5654 5655 5656 5657 5658 5659 5660 5661 5662
/**
 *	dev_set_promiscuity	- update promiscuity count on a device
 *	@dev: device
 *	@inc: modifier
 *
 *	Add or remove promiscuity from a device. While the count in the device
 *	remains above zero the interface remains promiscuous. Once it hits zero
 *	the device reverts back to normal filtering operation. A negative inc
 *	value is used to drop promiscuity on the device.
5663
 *	Return 0 if successful or a negative errno code on error.
5664
 */
5665
int dev_set_promiscuity(struct net_device *dev, int inc)
5666
{
5667
	unsigned int old_flags = dev->flags;
5668
	int err;
5669

5670
	err = __dev_set_promiscuity(dev, inc, true);
5671
	if (err < 0)
5672
		return err;
5673 5674
	if (dev->flags != old_flags)
		dev_set_rx_mode(dev);
5675
	return err;
5676
}
E
Eric Dumazet 已提交
5677
EXPORT_SYMBOL(dev_set_promiscuity);
5678

5679
static int __dev_set_allmulti(struct net_device *dev, int inc, bool notify)
L
Linus Torvalds 已提交
5680
{
5681
	unsigned int old_flags = dev->flags, old_gflags = dev->gflags;
L
Linus Torvalds 已提交
5682

5683 5684
	ASSERT_RTNL();

L
Linus Torvalds 已提交
5685
	dev->flags |= IFF_ALLMULTI;
5686 5687 5688 5689 5690 5691 5692 5693 5694 5695
	dev->allmulti += inc;
	if (dev->allmulti == 0) {
		/*
		 * Avoid overflow.
		 * If inc causes overflow, untouch allmulti and return error.
		 */
		if (inc < 0)
			dev->flags &= ~IFF_ALLMULTI;
		else {
			dev->allmulti -= inc;
5696 5697
			pr_warn("%s: allmulti touches roof, set allmulti failed. allmulti feature of device might be broken.\n",
				dev->name);
5698 5699 5700
			return -EOVERFLOW;
		}
	}
5701
	if (dev->flags ^ old_flags) {
5702
		dev_change_rx_flags(dev, IFF_ALLMULTI);
5703
		dev_set_rx_mode(dev);
5704 5705 5706
		if (notify)
			__dev_notify_flags(dev, old_flags,
					   dev->gflags ^ old_gflags);
5707
	}
5708
	return 0;
5709
}
5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727

/**
 *	dev_set_allmulti	- update allmulti count on a device
 *	@dev: device
 *	@inc: modifier
 *
 *	Add or remove reception of all multicast frames to a device. While the
 *	count in the device remains above zero the interface remains listening
 *	to all interfaces. Once it hits zero the device reverts back to normal
 *	filtering operation. A negative @inc value is used to drop the counter
 *	when releasing a resource needing all multicasts.
 *	Return 0 if successful or a negative errno code on error.
 */

int dev_set_allmulti(struct net_device *dev, int inc)
{
	return __dev_set_allmulti(dev, inc, true);
}
E
Eric Dumazet 已提交
5728
EXPORT_SYMBOL(dev_set_allmulti);
5729 5730 5731 5732

/*
 *	Upload unicast and multicast address lists to device and
 *	configure RX filtering. When the device doesn't support unicast
J
Joe Perches 已提交
5733
 *	filtering it is put in promiscuous mode while unicast addresses
5734 5735 5736 5737
 *	are present.
 */
void __dev_set_rx_mode(struct net_device *dev)
{
5738 5739
	const struct net_device_ops *ops = dev->netdev_ops;

5740 5741 5742 5743 5744
	/* dev_open will call this function so the list will stay sane. */
	if (!(dev->flags&IFF_UP))
		return;

	if (!netif_device_present(dev))
5745
		return;
5746

5747
	if (!(dev->priv_flags & IFF_UNICAST_FLT)) {
5748 5749 5750
		/* Unicast addresses changes may only happen under the rtnl,
		 * therefore calling __dev_set_promiscuity here is safe.
		 */
5751
		if (!netdev_uc_empty(dev) && !dev->uc_promisc) {
5752
			__dev_set_promiscuity(dev, 1, false);
5753
			dev->uc_promisc = true;
5754
		} else if (netdev_uc_empty(dev) && dev->uc_promisc) {
5755
			__dev_set_promiscuity(dev, -1, false);
5756
			dev->uc_promisc = false;
5757 5758
		}
	}
5759 5760 5761

	if (ops->ndo_set_rx_mode)
		ops->ndo_set_rx_mode(dev);
5762 5763 5764 5765
}

void dev_set_rx_mode(struct net_device *dev)
{
5766
	netif_addr_lock_bh(dev);
5767
	__dev_set_rx_mode(dev);
5768
	netif_addr_unlock_bh(dev);
L
Linus Torvalds 已提交
5769 5770
}

5771 5772 5773 5774 5775 5776
/**
 *	dev_get_flags - get flags reported to userspace
 *	@dev: device
 *
 *	Get the combination of flag bits exported through APIs to userspace.
 */
5777
unsigned int dev_get_flags(const struct net_device *dev)
L
Linus Torvalds 已提交
5778
{
5779
	unsigned int flags;
L
Linus Torvalds 已提交
5780 5781 5782

	flags = (dev->flags & ~(IFF_PROMISC |
				IFF_ALLMULTI |
S
Stefan Rompf 已提交
5783 5784 5785
				IFF_RUNNING |
				IFF_LOWER_UP |
				IFF_DORMANT)) |
L
Linus Torvalds 已提交
5786 5787 5788
		(dev->gflags & (IFF_PROMISC |
				IFF_ALLMULTI));

S
Stefan Rompf 已提交
5789 5790 5791 5792 5793 5794 5795 5796
	if (netif_running(dev)) {
		if (netif_oper_up(dev))
			flags |= IFF_RUNNING;
		if (netif_carrier_ok(dev))
			flags |= IFF_LOWER_UP;
		if (netif_dormant(dev))
			flags |= IFF_DORMANT;
	}
L
Linus Torvalds 已提交
5797 5798 5799

	return flags;
}
E
Eric Dumazet 已提交
5800
EXPORT_SYMBOL(dev_get_flags);
L
Linus Torvalds 已提交
5801

5802
int __dev_change_flags(struct net_device *dev, unsigned int flags)
L
Linus Torvalds 已提交
5803
{
5804
	unsigned int old_flags = dev->flags;
5805
	int ret;
L
Linus Torvalds 已提交
5806

5807 5808
	ASSERT_RTNL();

L
Linus Torvalds 已提交
5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822
	/*
	 *	Set the flags on our device.
	 */

	dev->flags = (flags & (IFF_DEBUG | IFF_NOTRAILERS | IFF_NOARP |
			       IFF_DYNAMIC | IFF_MULTICAST | IFF_PORTSEL |
			       IFF_AUTOMEDIA)) |
		     (dev->flags & (IFF_UP | IFF_VOLATILE | IFF_PROMISC |
				    IFF_ALLMULTI));

	/*
	 *	Load in the correct multicast list now the flags have changed.
	 */

5823 5824
	if ((old_flags ^ flags) & IFF_MULTICAST)
		dev_change_rx_flags(dev, IFF_MULTICAST);
5825

5826
	dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
5827 5828 5829 5830 5831 5832 5833 5834

	/*
	 *	Have we downed the interface. We handle IFF_UP ourselves
	 *	according to user attempts to set it, rather than blindly
	 *	setting it.
	 */

	ret = 0;
5835
	if ((old_flags ^ flags) & IFF_UP)
5836
		ret = ((old_flags & IFF_UP) ? __dev_close : __dev_open)(dev);
L
Linus Torvalds 已提交
5837 5838

	if ((flags ^ dev->gflags) & IFF_PROMISC) {
E
Eric Dumazet 已提交
5839
		int inc = (flags & IFF_PROMISC) ? 1 : -1;
5840
		unsigned int old_flags = dev->flags;
E
Eric Dumazet 已提交
5841

L
Linus Torvalds 已提交
5842
		dev->gflags ^= IFF_PROMISC;
5843 5844 5845 5846

		if (__dev_set_promiscuity(dev, inc, false) >= 0)
			if (dev->flags != old_flags)
				dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
5847 5848 5849 5850 5851 5852 5853
	}

	/* NOTE: order of synchronization of IFF_PROMISC and IFF_ALLMULTI
	   is important. Some (broken) drivers set IFF_PROMISC, when
	   IFF_ALLMULTI is requested not asking us and not reporting.
	 */
	if ((flags ^ dev->gflags) & IFF_ALLMULTI) {
E
Eric Dumazet 已提交
5854 5855
		int inc = (flags & IFF_ALLMULTI) ? 1 : -1;

L
Linus Torvalds 已提交
5856
		dev->gflags ^= IFF_ALLMULTI;
5857
		__dev_set_allmulti(dev, inc, false);
L
Linus Torvalds 已提交
5858 5859
	}

5860 5861 5862
	return ret;
}

5863 5864
void __dev_notify_flags(struct net_device *dev, unsigned int old_flags,
			unsigned int gchanges)
5865 5866 5867
{
	unsigned int changes = dev->flags ^ old_flags;

5868
	if (gchanges)
5869
		rtmsg_ifinfo(RTM_NEWLINK, dev, gchanges, GFP_ATOMIC);
5870

5871 5872 5873 5874 5875 5876 5877 5878
	if (changes & IFF_UP) {
		if (dev->flags & IFF_UP)
			call_netdevice_notifiers(NETDEV_UP, dev);
		else
			call_netdevice_notifiers(NETDEV_DOWN, dev);
	}

	if (dev->flags & IFF_UP &&
5879 5880 5881 5882 5883 5884 5885
	    (changes & ~(IFF_UP | IFF_PROMISC | IFF_ALLMULTI | IFF_VOLATILE))) {
		struct netdev_notifier_change_info change_info;

		change_info.flags_changed = changes;
		call_netdevice_notifiers_info(NETDEV_CHANGE, dev,
					      &change_info.info);
	}
5886 5887 5888 5889 5890 5891 5892 5893 5894 5895
}

/**
 *	dev_change_flags - change device settings
 *	@dev: device
 *	@flags: device state flags
 *
 *	Change settings on device based state flags. The flags are
 *	in the userspace exported format.
 */
5896
int dev_change_flags(struct net_device *dev, unsigned int flags)
5897
{
5898
	int ret;
5899
	unsigned int changes, old_flags = dev->flags, old_gflags = dev->gflags;
5900 5901 5902 5903 5904

	ret = __dev_change_flags(dev, flags);
	if (ret < 0)
		return ret;

5905
	changes = (old_flags ^ dev->flags) | (old_gflags ^ dev->gflags);
5906
	__dev_notify_flags(dev, old_flags, changes);
L
Linus Torvalds 已提交
5907 5908
	return ret;
}
E
Eric Dumazet 已提交
5909
EXPORT_SYMBOL(dev_change_flags);
L
Linus Torvalds 已提交
5910

5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921
static int __dev_set_mtu(struct net_device *dev, int new_mtu)
{
	const struct net_device_ops *ops = dev->netdev_ops;

	if (ops->ndo_change_mtu)
		return ops->ndo_change_mtu(dev, new_mtu);

	dev->mtu = new_mtu;
	return 0;
}

5922 5923 5924 5925 5926 5927 5928
/**
 *	dev_set_mtu - Change maximum transfer unit
 *	@dev: device
 *	@new_mtu: new transfer unit
 *
 *	Change the maximum transfer size of the network device.
 */
L
Linus Torvalds 已提交
5929 5930
int dev_set_mtu(struct net_device *dev, int new_mtu)
{
5931
	int err, orig_mtu;
L
Linus Torvalds 已提交
5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942

	if (new_mtu == dev->mtu)
		return 0;

	/*	MTU must be positive.	 */
	if (new_mtu < 0)
		return -EINVAL;

	if (!netif_device_present(dev))
		return -ENODEV;

5943 5944 5945 5946
	err = call_netdevice_notifiers(NETDEV_PRECHANGEMTU, dev);
	err = notifier_to_errno(err);
	if (err)
		return err;
5947

5948 5949
	orig_mtu = dev->mtu;
	err = __dev_set_mtu(dev, new_mtu);
5950

5951 5952 5953 5954 5955 5956 5957 5958 5959 5960 5961
	if (!err) {
		err = call_netdevice_notifiers(NETDEV_CHANGEMTU, dev);
		err = notifier_to_errno(err);
		if (err) {
			/* setting mtu back and notifying everyone again,
			 * so that they have a chance to revert changes.
			 */
			__dev_set_mtu(dev, orig_mtu);
			call_netdevice_notifiers(NETDEV_CHANGEMTU, dev);
		}
	}
L
Linus Torvalds 已提交
5962 5963
	return err;
}
E
Eric Dumazet 已提交
5964
EXPORT_SYMBOL(dev_set_mtu);
L
Linus Torvalds 已提交
5965

5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976
/**
 *	dev_set_group - Change group this device belongs to
 *	@dev: device
 *	@new_group: group this device should belong to
 */
void dev_set_group(struct net_device *dev, int new_group)
{
	dev->group = new_group;
}
EXPORT_SYMBOL(dev_set_group);

5977 5978 5979 5980 5981 5982 5983
/**
 *	dev_set_mac_address - Change Media Access Control Address
 *	@dev: device
 *	@sa: new address
 *
 *	Change the hardware (MAC) address of the device
 */
L
Linus Torvalds 已提交
5984 5985
int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa)
{
5986
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
5987 5988
	int err;

5989
	if (!ops->ndo_set_mac_address)
L
Linus Torvalds 已提交
5990 5991 5992 5993 5994
		return -EOPNOTSUPP;
	if (sa->sa_family != dev->type)
		return -EINVAL;
	if (!netif_device_present(dev))
		return -ENODEV;
5995
	err = ops->ndo_set_mac_address(dev, sa);
5996 5997
	if (err)
		return err;
J
Jiri Pirko 已提交
5998
	dev->addr_assign_type = NET_ADDR_SET;
5999
	call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
6000
	add_device_randomness(dev->dev_addr, dev->addr_len);
6001
	return 0;
L
Linus Torvalds 已提交
6002
}
E
Eric Dumazet 已提交
6003
EXPORT_SYMBOL(dev_set_mac_address);
L
Linus Torvalds 已提交
6004

J
Jiri Pirko 已提交
6005 6006 6007
/**
 *	dev_change_carrier - Change device carrier
 *	@dev: device
6008
 *	@new_carrier: new value
J
Jiri Pirko 已提交
6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023
 *
 *	Change device carrier
 */
int dev_change_carrier(struct net_device *dev, bool new_carrier)
{
	const struct net_device_ops *ops = dev->netdev_ops;

	if (!ops->ndo_change_carrier)
		return -EOPNOTSUPP;
	if (!netif_device_present(dev))
		return -ENODEV;
	return ops->ndo_change_carrier(dev, new_carrier);
}
EXPORT_SYMBOL(dev_change_carrier);

6024 6025 6026 6027 6028 6029 6030 6031
/**
 *	dev_get_phys_port_id - Get device physical port ID
 *	@dev: device
 *	@ppid: port ID
 *
 *	Get device physical port ID
 */
int dev_get_phys_port_id(struct net_device *dev,
6032
			 struct netdev_phys_item_id *ppid)
6033 6034 6035 6036 6037 6038 6039 6040 6041
{
	const struct net_device_ops *ops = dev->netdev_ops;

	if (!ops->ndo_get_phys_port_id)
		return -EOPNOTSUPP;
	return ops->ndo_get_phys_port_id(dev, ppid);
}
EXPORT_SYMBOL(dev_get_phys_port_id);

6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059
/**
 *	dev_get_phys_port_name - Get device physical port name
 *	@dev: device
 *	@name: port name
 *
 *	Get device physical port name
 */
int dev_get_phys_port_name(struct net_device *dev,
			   char *name, size_t len)
{
	const struct net_device_ops *ops = dev->netdev_ops;

	if (!ops->ndo_get_phys_port_name)
		return -EOPNOTSUPP;
	return ops->ndo_get_phys_port_name(dev, name, len);
}
EXPORT_SYMBOL(dev_get_phys_port_name);

L
Linus Torvalds 已提交
6060 6061
/**
 *	dev_new_index	-	allocate an ifindex
6062
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
6063 6064 6065 6066 6067
 *
 *	Returns a suitable unique value for a new device interface
 *	number.  The caller must hold the rtnl semaphore or the
 *	dev_base_lock to be sure it remains unique.
 */
6068
static int dev_new_index(struct net *net)
L
Linus Torvalds 已提交
6069
{
6070
	int ifindex = net->ifindex;
L
Linus Torvalds 已提交
6071 6072 6073
	for (;;) {
		if (++ifindex <= 0)
			ifindex = 1;
6074
		if (!__dev_get_by_index(net, ifindex))
6075
			return net->ifindex = ifindex;
L
Linus Torvalds 已提交
6076 6077 6078 6079
	}
}

/* Delayed registration/unregisteration */
6080
static LIST_HEAD(net_todo_list);
6081
DECLARE_WAIT_QUEUE_HEAD(netdev_unregistering_wq);
L
Linus Torvalds 已提交
6082

6083
static void net_set_todo(struct net_device *dev)
L
Linus Torvalds 已提交
6084 6085
{
	list_add_tail(&dev->todo_list, &net_todo_list);
6086
	dev_net(dev)->dev_unreg_count++;
L
Linus Torvalds 已提交
6087 6088
}

6089
static void rollback_registered_many(struct list_head *head)
6090
{
6091
	struct net_device *dev, *tmp;
6092
	LIST_HEAD(close_head);
6093

6094 6095 6096
	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

6097
	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
6098
		/* Some devices call without registering
6099 6100
		 * for initialization unwind. Remove those
		 * devices and proceed with the remaining.
6101 6102
		 */
		if (dev->reg_state == NETREG_UNINITIALIZED) {
6103 6104
			pr_debug("unregister_netdevice: device %s/%p never was registered\n",
				 dev->name, dev);
6105

6106
			WARN_ON(1);
6107 6108
			list_del(&dev->unreg_list);
			continue;
6109
		}
6110
		dev->dismantle = true;
6111
		BUG_ON(dev->reg_state != NETREG_REGISTERED);
6112
	}
6113

6114
	/* If device is running, close it first. */
6115 6116
	list_for_each_entry(dev, head, unreg_list)
		list_add_tail(&dev->close_list, &close_head);
6117
	dev_close_many(&close_head, true);
6118

6119
	list_for_each_entry(dev, head, unreg_list) {
6120 6121
		/* And unlink it from device chain. */
		unlist_netdevice(dev);
6122

6123 6124
		dev->reg_state = NETREG_UNREGISTERING;
	}
6125 6126 6127

	synchronize_net();

6128
	list_for_each_entry(dev, head, unreg_list) {
6129 6130
		struct sk_buff *skb = NULL;

6131 6132
		/* Shutdown queueing discipline. */
		dev_shutdown(dev);
6133 6134


6135 6136 6137 6138
		/* Notify protocols, that we are about to destroy
		   this device. They should clean all the things.
		*/
		call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
6139

6140 6141 6142 6143 6144
		if (!dev->rtnl_link_ops ||
		    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
			skb = rtmsg_ifinfo_build_skb(RTM_DELLINK, dev, ~0U,
						     GFP_KERNEL);

6145 6146 6147
		/*
		 *	Flush the unicast and multicast chains
		 */
6148
		dev_uc_flush(dev);
6149
		dev_mc_flush(dev);
6150

6151 6152
		if (dev->netdev_ops->ndo_uninit)
			dev->netdev_ops->ndo_uninit(dev);
6153

6154 6155
		if (skb)
			rtmsg_ifinfo_send(skb, dev, GFP_KERNEL);
6156

J
Jiri Pirko 已提交
6157 6158
		/* Notifier chain MUST detach us all upper devices. */
		WARN_ON(netdev_has_any_upper_dev(dev));
6159

6160 6161
		/* Remove entries from kobject tree */
		netdev_unregister_kobject(dev);
6162 6163 6164 6165
#ifdef CONFIG_XPS
		/* Remove XPS queueing entries */
		netif_reset_xps_queues_gt(dev, 0);
#endif
6166
	}
6167

6168
	synchronize_net();
6169

6170
	list_for_each_entry(dev, head, unreg_list)
6171 6172 6173 6174 6175 6176 6177 6178 6179
		dev_put(dev);
}

static void rollback_registered(struct net_device *dev)
{
	LIST_HEAD(single);

	list_add(&dev->unreg_list, &single);
	rollback_registered_many(&single);
E
Eric Dumazet 已提交
6180
	list_del(&single);
6181 6182
}

6183 6184
static netdev_features_t netdev_fix_features(struct net_device *dev,
	netdev_features_t features)
6185
{
6186 6187 6188
	/* Fix illegal checksum combinations */
	if ((features & NETIF_F_HW_CSUM) &&
	    (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
6189
		netdev_warn(dev, "mixed HW and IP checksum settings.\n");
6190 6191 6192
		features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
	}

6193
	/* TSO requires that SG is present as well. */
6194
	if ((features & NETIF_F_ALL_TSO) && !(features & NETIF_F_SG)) {
6195
		netdev_dbg(dev, "Dropping TSO features since no SG feature.\n");
6196
		features &= ~NETIF_F_ALL_TSO;
6197 6198
	}

6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211
	if ((features & NETIF_F_TSO) && !(features & NETIF_F_HW_CSUM) &&
					!(features & NETIF_F_IP_CSUM)) {
		netdev_dbg(dev, "Dropping TSO features since no CSUM feature.\n");
		features &= ~NETIF_F_TSO;
		features &= ~NETIF_F_TSO_ECN;
	}

	if ((features & NETIF_F_TSO6) && !(features & NETIF_F_HW_CSUM) &&
					 !(features & NETIF_F_IPV6_CSUM)) {
		netdev_dbg(dev, "Dropping TSO6 features since no CSUM feature.\n");
		features &= ~NETIF_F_TSO6;
	}

6212 6213 6214 6215
	/* TSO ECN requires that TSO is present as well. */
	if ((features & NETIF_F_ALL_TSO) == NETIF_F_TSO_ECN)
		features &= ~NETIF_F_TSO_ECN;

6216 6217
	/* Software GSO depends on SG. */
	if ((features & NETIF_F_GSO) && !(features & NETIF_F_SG)) {
6218
		netdev_dbg(dev, "Dropping NETIF_F_GSO since no SG feature.\n");
6219 6220 6221
		features &= ~NETIF_F_GSO;
	}

6222
	/* UFO needs SG and checksumming */
6223
	if (features & NETIF_F_UFO) {
6224 6225 6226 6227
		/* maybe split UFO into V4 and V6? */
		if (!((features & NETIF_F_GEN_CSUM) ||
		    (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))
			    == (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
6228
			netdev_dbg(dev,
6229
				"Dropping NETIF_F_UFO since no checksum offload features.\n");
6230 6231 6232 6233
			features &= ~NETIF_F_UFO;
		}

		if (!(features & NETIF_F_SG)) {
6234
			netdev_dbg(dev,
6235
				"Dropping NETIF_F_UFO since no NETIF_F_SG feature.\n");
6236 6237 6238 6239
			features &= ~NETIF_F_UFO;
		}
	}

J
Jiri Pirko 已提交
6240 6241 6242 6243 6244 6245 6246
#ifdef CONFIG_NET_RX_BUSY_POLL
	if (dev->netdev_ops->ndo_busy_poll)
		features |= NETIF_F_BUSY_POLL;
	else
#endif
		features &= ~NETIF_F_BUSY_POLL;

6247 6248 6249
	return features;
}

6250
int __netdev_update_features(struct net_device *dev)
6251
{
6252
	netdev_features_t features;
6253 6254
	int err = 0;

6255 6256
	ASSERT_RTNL();

6257 6258 6259 6260 6261 6262 6263 6264 6265
	features = netdev_get_wanted_features(dev);

	if (dev->netdev_ops->ndo_fix_features)
		features = dev->netdev_ops->ndo_fix_features(dev, features);

	/* driver might be less strict about feature dependencies */
	features = netdev_fix_features(dev, features);

	if (dev->features == features)
6266
		return 0;
6267

6268 6269
	netdev_dbg(dev, "Features changed: %pNF -> %pNF\n",
		&dev->features, &features);
6270 6271 6272 6273

	if (dev->netdev_ops->ndo_set_features)
		err = dev->netdev_ops->ndo_set_features(dev, features);

6274
	if (unlikely(err < 0)) {
6275
		netdev_err(dev,
6276 6277
			"set_features() failed (%d); wanted %pNF, left %pNF\n",
			err, &features, &dev->features);
6278 6279 6280 6281 6282 6283 6284 6285 6286
		return -1;
	}

	if (!err)
		dev->features = features;

	return 1;
}

6287 6288 6289 6290 6291 6292 6293 6294
/**
 *	netdev_update_features - recalculate device features
 *	@dev: the device to check
 *
 *	Recalculate dev->features set and send notifications if it
 *	has changed. Should be called after driver or hardware dependent
 *	conditions might have changed that influence the features.
 */
6295 6296 6297 6298
void netdev_update_features(struct net_device *dev)
{
	if (__netdev_update_features(dev))
		netdev_features_change(dev);
6299 6300 6301
}
EXPORT_SYMBOL(netdev_update_features);

6302 6303 6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318
/**
 *	netdev_change_features - recalculate device features
 *	@dev: the device to check
 *
 *	Recalculate dev->features set and send notifications even
 *	if they have not changed. Should be called instead of
 *	netdev_update_features() if also dev->vlan_features might
 *	have changed to allow the changes to be propagated to stacked
 *	VLAN devices.
 */
void netdev_change_features(struct net_device *dev)
{
	__netdev_update_features(dev);
	netdev_features_change(dev);
}
EXPORT_SYMBOL(netdev_change_features);

6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345
/**
 *	netif_stacked_transfer_operstate -	transfer operstate
 *	@rootdev: the root or lower level device to transfer state from
 *	@dev: the device to transfer operstate to
 *
 *	Transfer operational state from root to device. This is normally
 *	called when a stacking relationship exists between the root
 *	device and the device(a leaf device).
 */
void netif_stacked_transfer_operstate(const struct net_device *rootdev,
					struct net_device *dev)
{
	if (rootdev->operstate == IF_OPER_DORMANT)
		netif_dormant_on(dev);
	else
		netif_dormant_off(dev);

	if (netif_carrier_ok(rootdev)) {
		if (!netif_carrier_ok(dev))
			netif_carrier_on(dev);
	} else {
		if (netif_carrier_ok(dev))
			netif_carrier_off(dev);
	}
}
EXPORT_SYMBOL(netif_stacked_transfer_operstate);

6346
#ifdef CONFIG_SYSFS
6347 6348 6349
static int netif_alloc_rx_queues(struct net_device *dev)
{
	unsigned int i, count = dev->num_rx_queues;
T
Tom Herbert 已提交
6350
	struct netdev_rx_queue *rx;
6351
	size_t sz = count * sizeof(*rx);
6352

T
Tom Herbert 已提交
6353
	BUG_ON(count < 1);
6354

6355 6356 6357 6358 6359 6360
	rx = kzalloc(sz, GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);
	if (!rx) {
		rx = vzalloc(sz);
		if (!rx)
			return -ENOMEM;
	}
T
Tom Herbert 已提交
6361 6362 6363
	dev->_rx = rx;

	for (i = 0; i < count; i++)
T
Tom Herbert 已提交
6364
		rx[i].dev = dev;
6365 6366
	return 0;
}
T
Tom Herbert 已提交
6367
#endif
6368

C
Changli Gao 已提交
6369 6370 6371 6372 6373 6374 6375
static void netdev_init_one_queue(struct net_device *dev,
				  struct netdev_queue *queue, void *_unused)
{
	/* Initialize queue lock */
	spin_lock_init(&queue->_xmit_lock);
	netdev_set_xmit_lockdep_class(&queue->_xmit_lock, dev->type);
	queue->xmit_lock_owner = -1;
6376
	netdev_queue_numa_node_write(queue, NUMA_NO_NODE);
C
Changli Gao 已提交
6377
	queue->dev = dev;
T
Tom Herbert 已提交
6378 6379 6380
#ifdef CONFIG_BQL
	dql_init(&queue->dql, HZ);
#endif
C
Changli Gao 已提交
6381 6382
}

6383 6384
static void netif_free_tx_queues(struct net_device *dev)
{
W
WANG Cong 已提交
6385
	kvfree(dev->_tx);
6386 6387
}

6388 6389 6390 6391
static int netif_alloc_netdev_queues(struct net_device *dev)
{
	unsigned int count = dev->num_tx_queues;
	struct netdev_queue *tx;
6392
	size_t sz = count * sizeof(*tx);
6393

6394
	BUG_ON(count < 1 || count > 0xffff);
6395

6396 6397 6398 6399 6400 6401
	tx = kzalloc(sz, GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);
	if (!tx) {
		tx = vzalloc(sz);
		if (!tx)
			return -ENOMEM;
	}
6402
	dev->_tx = tx;
T
Tom Herbert 已提交
6403

6404 6405
	netdev_for_each_tx_queue(dev, netdev_init_one_queue, NULL);
	spin_lock_init(&dev->tx_global_lock);
C
Changli Gao 已提交
6406 6407

	return 0;
6408 6409
}

6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420
void netif_tx_stop_all_queues(struct net_device *dev)
{
	unsigned int i;

	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
		netif_tx_stop_queue(txq);
	}
}
EXPORT_SYMBOL(netif_tx_stop_all_queues);

L
Linus Torvalds 已提交
6421 6422 6423 6424 6425 6426 6427 6428 6429 6430 6431 6432 6433 6434 6435 6436 6437 6438 6439 6440
/**
 *	register_netdevice	- register a network device
 *	@dev: device to register
 *
 *	Take a completed network device structure and add it to the kernel
 *	interfaces. A %NETDEV_REGISTER message is sent to the netdev notifier
 *	chain. 0 is returned on success. A negative errno code is returned
 *	on a failure to set up the device, or if the name is a duplicate.
 *
 *	Callers must hold the rtnl semaphore. You may want
 *	register_netdev() instead of this.
 *
 *	BUGS:
 *	The locking appears insufficient to guarantee two parallel registers
 *	will not get the same name.
 */

int register_netdevice(struct net_device *dev)
{
	int ret;
6441
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
6442 6443 6444 6445

	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

6446 6447
	might_sleep();

L
Linus Torvalds 已提交
6448 6449
	/* When net_device's are persistent, this will be fatal. */
	BUG_ON(dev->reg_state != NETREG_UNINITIALIZED);
6450
	BUG_ON(!net);
L
Linus Torvalds 已提交
6451

6452
	spin_lock_init(&dev->addr_list_lock);
6453
	netdev_set_addr_lockdep_class(dev);
L
Linus Torvalds 已提交
6454

6455
	ret = dev_get_valid_name(net, dev, dev->name);
6456 6457 6458
	if (ret < 0)
		goto out;

L
Linus Torvalds 已提交
6459
	/* Init, if this function is available */
6460 6461
	if (dev->netdev_ops->ndo_init) {
		ret = dev->netdev_ops->ndo_init(dev);
L
Linus Torvalds 已提交
6462 6463 6464
		if (ret) {
			if (ret > 0)
				ret = -EIO;
6465
			goto out;
L
Linus Torvalds 已提交
6466 6467
		}
	}
6468

6469 6470
	if (((dev->hw_features | dev->features) &
	     NETIF_F_HW_VLAN_CTAG_FILTER) &&
6471 6472 6473 6474 6475 6476 6477
	    (!dev->netdev_ops->ndo_vlan_rx_add_vid ||
	     !dev->netdev_ops->ndo_vlan_rx_kill_vid)) {
		netdev_WARN(dev, "Buggy VLAN acceleration in driver!\n");
		ret = -EINVAL;
		goto err_uninit;
	}

6478 6479 6480 6481 6482 6483
	ret = -EBUSY;
	if (!dev->ifindex)
		dev->ifindex = dev_new_index(net);
	else if (__dev_get_by_index(net, dev->ifindex))
		goto err_uninit;

6484 6485 6486 6487
	/* Transfer changeable features to wanted_features and enable
	 * software offloads (GSO and GRO).
	 */
	dev->hw_features |= NETIF_F_SOFT_FEATURES;
6488 6489
	dev->features |= NETIF_F_SOFT_FEATURES;
	dev->wanted_features = dev->features & dev->hw_features;
L
Linus Torvalds 已提交
6490

6491 6492
	if (!(dev->flags & IFF_LOOPBACK)) {
		dev->hw_features |= NETIF_F_NOCACHE_COPY;
6493 6494
	}

6495
	/* Make NETIF_F_HIGHDMA inheritable to VLAN devices.
6496
	 */
6497
	dev->vlan_features |= NETIF_F_HIGHDMA;
6498

6499 6500 6501 6502
	/* Make NETIF_F_SG inheritable to tunnel devices.
	 */
	dev->hw_enc_features |= NETIF_F_SG;

S
Simon Horman 已提交
6503 6504 6505 6506
	/* Make NETIF_F_SG inheritable to MPLS.
	 */
	dev->mpls_features |= NETIF_F_SG;

6507 6508 6509 6510 6511
	ret = call_netdevice_notifiers(NETDEV_POST_INIT, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		goto err_uninit;

6512
	ret = netdev_register_kobject(dev);
6513
	if (ret)
6514
		goto err_uninit;
6515 6516
	dev->reg_state = NETREG_REGISTERED;

6517
	__netdev_update_features(dev);
6518

L
Linus Torvalds 已提交
6519 6520 6521 6522 6523 6524 6525
	/*
	 *	Default initial state at registry is that the
	 *	device is present.
	 */

	set_bit(__LINK_STATE_PRESENT, &dev->state);

6526 6527
	linkwatch_init_dev(dev);

L
Linus Torvalds 已提交
6528 6529
	dev_init_scheduler(dev);
	dev_hold(dev);
6530
	list_netdevice(dev);
6531
	add_device_randomness(dev->dev_addr, dev->addr_len);
L
Linus Torvalds 已提交
6532

6533 6534 6535 6536 6537 6538 6539
	/* If the device has permanent device address, driver should
	 * set dev_addr and also addr_assign_type should be set to
	 * NET_ADDR_PERM (default value).
	 */
	if (dev->addr_assign_type == NET_ADDR_PERM)
		memcpy(dev->perm_addr, dev->dev_addr, dev->addr_len);

L
Linus Torvalds 已提交
6540
	/* Notify protocols, that a new device appeared. */
6541
	ret = call_netdevice_notifiers(NETDEV_REGISTER, dev);
6542
	ret = notifier_to_errno(ret);
6543 6544 6545 6546
	if (ret) {
		rollback_registered(dev);
		dev->reg_state = NETREG_UNREGISTERED;
	}
6547 6548 6549 6550
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
6551 6552
	if (!dev->rtnl_link_ops ||
	    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
6553
		rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U, GFP_KERNEL);
L
Linus Torvalds 已提交
6554 6555 6556

out:
	return ret;
6557 6558

err_uninit:
6559 6560
	if (dev->netdev_ops->ndo_uninit)
		dev->netdev_ops->ndo_uninit(dev);
6561
	goto out;
L
Linus Torvalds 已提交
6562
}
E
Eric Dumazet 已提交
6563
EXPORT_SYMBOL(register_netdevice);
L
Linus Torvalds 已提交
6564

6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595
/**
 *	init_dummy_netdev	- init a dummy network device for NAPI
 *	@dev: device to init
 *
 *	This takes a network device structure and initialize the minimum
 *	amount of fields so it can be used to schedule NAPI polls without
 *	registering a full blown interface. This is to be used by drivers
 *	that need to tie several hardware interfaces to a single NAPI
 *	poll scheduler due to HW limitations.
 */
int init_dummy_netdev(struct net_device *dev)
{
	/* Clear everything. Note we don't initialize spinlocks
	 * are they aren't supposed to be taken by any of the
	 * NAPI code and this dummy netdev is supposed to be
	 * only ever used for NAPI polls
	 */
	memset(dev, 0, sizeof(struct net_device));

	/* make sure we BUG if trying to hit standard
	 * register/unregister code path
	 */
	dev->reg_state = NETREG_DUMMY;

	/* NAPI wants this */
	INIT_LIST_HEAD(&dev->napi_list);

	/* a dummy interface is started by default */
	set_bit(__LINK_STATE_PRESENT, &dev->state);
	set_bit(__LINK_STATE_START, &dev->state);

E
Eric Dumazet 已提交
6596 6597 6598 6599 6600
	/* Note : We dont allocate pcpu_refcnt for dummy devices,
	 * because users of this 'device' dont need to change
	 * its refcount.
	 */

6601 6602 6603 6604 6605
	return 0;
}
EXPORT_SYMBOL_GPL(init_dummy_netdev);


L
Linus Torvalds 已提交
6606 6607 6608 6609 6610 6611 6612 6613 6614
/**
 *	register_netdev	- register a network device
 *	@dev: device to register
 *
 *	Take a completed network device structure and add it to the kernel
 *	interfaces. A %NETDEV_REGISTER message is sent to the netdev notifier
 *	chain. 0 is returned on success. A negative errno code is returned
 *	on a failure to set up the device, or if the name is a duplicate.
 *
6615
 *	This is a wrapper around register_netdevice that takes the rtnl semaphore
L
Linus Torvalds 已提交
6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629
 *	and expands the device name if you passed a format string to
 *	alloc_netdev.
 */
int register_netdev(struct net_device *dev)
{
	int err;

	rtnl_lock();
	err = register_netdevice(dev);
	rtnl_unlock();
	return err;
}
EXPORT_SYMBOL(register_netdev);

E
Eric Dumazet 已提交
6630 6631 6632 6633 6634 6635 6636 6637 6638 6639
int netdev_refcnt_read(const struct net_device *dev)
{
	int i, refcnt = 0;

	for_each_possible_cpu(i)
		refcnt += *per_cpu_ptr(dev->pcpu_refcnt, i);
	return refcnt;
}
EXPORT_SYMBOL(netdev_refcnt_read);

6640
/**
L
Linus Torvalds 已提交
6641
 * netdev_wait_allrefs - wait until all references are gone.
6642
 * @dev: target net_device
L
Linus Torvalds 已提交
6643 6644 6645 6646 6647 6648 6649
 *
 * This is called when unregistering network devices.
 *
 * Any protocol or device that holds a reference should register
 * for netdevice notification, and cleanup and put back the
 * reference if they receive an UNREGISTER event.
 * We can get stuck here if buggy protocols don't correctly
6650
 * call dev_put.
L
Linus Torvalds 已提交
6651 6652 6653 6654
 */
static void netdev_wait_allrefs(struct net_device *dev)
{
	unsigned long rebroadcast_time, warning_time;
E
Eric Dumazet 已提交
6655
	int refcnt;
L
Linus Torvalds 已提交
6656

6657 6658
	linkwatch_forget_dev(dev);

L
Linus Torvalds 已提交
6659
	rebroadcast_time = warning_time = jiffies;
E
Eric Dumazet 已提交
6660 6661 6662
	refcnt = netdev_refcnt_read(dev);

	while (refcnt != 0) {
L
Linus Torvalds 已提交
6663
		if (time_after(jiffies, rebroadcast_time + 1 * HZ)) {
6664
			rtnl_lock();
L
Linus Torvalds 已提交
6665 6666

			/* Rebroadcast unregister notification */
6667
			call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
L
Linus Torvalds 已提交
6668

6669
			__rtnl_unlock();
6670
			rcu_barrier();
6671 6672
			rtnl_lock();

6673
			call_netdevice_notifiers(NETDEV_UNREGISTER_FINAL, dev);
L
Linus Torvalds 已提交
6674 6675 6676 6677 6678 6679 6680 6681 6682 6683 6684
			if (test_bit(__LINK_STATE_LINKWATCH_PENDING,
				     &dev->state)) {
				/* We must not have linkwatch events
				 * pending on unregister. If this
				 * happens, we simply run the queue
				 * unscheduled, resulting in a noop
				 * for this device.
				 */
				linkwatch_run_queue();
			}

6685
			__rtnl_unlock();
L
Linus Torvalds 已提交
6686 6687 6688 6689 6690 6691

			rebroadcast_time = jiffies;
		}

		msleep(250);

E
Eric Dumazet 已提交
6692 6693
		refcnt = netdev_refcnt_read(dev);

L
Linus Torvalds 已提交
6694
		if (time_after(jiffies, warning_time + 10 * HZ)) {
6695 6696
			pr_emerg("unregister_netdevice: waiting for %s to become free. Usage count = %d\n",
				 dev->name, refcnt);
L
Linus Torvalds 已提交
6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711 6712 6713 6714 6715
			warning_time = jiffies;
		}
	}
}

/* The sequence is:
 *
 *	rtnl_lock();
 *	...
 *	register_netdevice(x1);
 *	register_netdevice(x2);
 *	...
 *	unregister_netdevice(y1);
 *	unregister_netdevice(y2);
 *      ...
 *	rtnl_unlock();
 *	free_netdev(y1);
 *	free_netdev(y2);
 *
H
Herbert Xu 已提交
6716
 * We are invoked by rtnl_unlock().
L
Linus Torvalds 已提交
6717
 * This allows us to deal with problems:
6718
 * 1) We can delete sysfs objects which invoke hotplug
L
Linus Torvalds 已提交
6719 6720 6721
 *    without deadlocking with linkwatch via keventd.
 * 2) Since we run with the RTNL semaphore not held, we can sleep
 *    safely in order to wait for the netdev refcnt to drop to zero.
H
Herbert Xu 已提交
6722 6723 6724
 *
 * We must not return until all unregister events added during
 * the interval the lock was held have been completed.
L
Linus Torvalds 已提交
6725 6726 6727
 */
void netdev_run_todo(void)
{
6728
	struct list_head list;
L
Linus Torvalds 已提交
6729 6730

	/* Snapshot list, allow later requests */
6731
	list_replace_init(&net_todo_list, &list);
H
Herbert Xu 已提交
6732 6733

	__rtnl_unlock();
6734

6735 6736

	/* Wait for rcu callbacks to finish before next phase */
6737 6738 6739
	if (!list_empty(&list))
		rcu_barrier();

L
Linus Torvalds 已提交
6740 6741
	while (!list_empty(&list)) {
		struct net_device *dev
6742
			= list_first_entry(&list, struct net_device, todo_list);
L
Linus Torvalds 已提交
6743 6744
		list_del(&dev->todo_list);

6745
		rtnl_lock();
6746
		call_netdevice_notifiers(NETDEV_UNREGISTER_FINAL, dev);
6747
		__rtnl_unlock();
6748

6749
		if (unlikely(dev->reg_state != NETREG_UNREGISTERING)) {
6750
			pr_err("network todo '%s' but state %d\n",
6751 6752 6753 6754
			       dev->name, dev->reg_state);
			dump_stack();
			continue;
		}
L
Linus Torvalds 已提交
6755

6756
		dev->reg_state = NETREG_UNREGISTERED;
L
Linus Torvalds 已提交
6757

6758
		on_each_cpu(flush_backlog, dev, 1);
6759

6760
		netdev_wait_allrefs(dev);
L
Linus Torvalds 已提交
6761

6762
		/* paranoia */
E
Eric Dumazet 已提交
6763
		BUG_ON(netdev_refcnt_read(dev));
6764 6765
		BUG_ON(!list_empty(&dev->ptype_all));
		BUG_ON(!list_empty(&dev->ptype_specific));
6766 6767
		WARN_ON(rcu_access_pointer(dev->ip_ptr));
		WARN_ON(rcu_access_pointer(dev->ip6_ptr));
6768
		WARN_ON(dev->dn_ptr);
L
Linus Torvalds 已提交
6769

6770 6771
		if (dev->destructor)
			dev->destructor(dev);
6772

6773 6774 6775 6776 6777 6778
		/* Report a network device has been unregistered */
		rtnl_lock();
		dev_net(dev)->dev_unreg_count--;
		__rtnl_unlock();
		wake_up(&netdev_unregistering_wq);

6779 6780
		/* Free network device */
		kobject_put(&dev->dev.kobj);
L
Linus Torvalds 已提交
6781 6782 6783
	}
}

6784 6785 6786
/* Convert net_device_stats to rtnl_link_stats64.  They have the same
 * fields in the same order, with only the type differing.
 */
6787 6788
void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
			     const struct net_device_stats *netdev_stats)
6789 6790
{
#if BITS_PER_LONG == 64
6791 6792
	BUILD_BUG_ON(sizeof(*stats64) != sizeof(*netdev_stats));
	memcpy(stats64, netdev_stats, sizeof(*stats64));
6793 6794 6795 6796 6797 6798 6799 6800 6801 6802 6803
#else
	size_t i, n = sizeof(*stats64) / sizeof(u64);
	const unsigned long *src = (const unsigned long *)netdev_stats;
	u64 *dst = (u64 *)stats64;

	BUILD_BUG_ON(sizeof(*netdev_stats) / sizeof(unsigned long) !=
		     sizeof(*stats64) / sizeof(u64));
	for (i = 0; i < n; i++)
		dst[i] = src[i];
#endif
}
6804
EXPORT_SYMBOL(netdev_stats_to_stats64);
6805

6806 6807 6808
/**
 *	dev_get_stats	- get network device statistics
 *	@dev: device to get statistics from
6809
 *	@storage: place to store stats
6810
 *
6811 6812 6813 6814
 *	Get network statistics from device. Return @storage.
 *	The device driver may provide its own method by setting
 *	dev->netdev_ops->get_stats64 or dev->netdev_ops->get_stats;
 *	otherwise the internal statistics structure is used.
6815
 */
6816 6817
struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
					struct rtnl_link_stats64 *storage)
6818
{
6819 6820
	const struct net_device_ops *ops = dev->netdev_ops;

6821 6822
	if (ops->ndo_get_stats64) {
		memset(storage, 0, sizeof(*storage));
6823 6824
		ops->ndo_get_stats64(dev, storage);
	} else if (ops->ndo_get_stats) {
6825
		netdev_stats_to_stats64(storage, ops->ndo_get_stats(dev));
6826 6827
	} else {
		netdev_stats_to_stats64(storage, &dev->stats);
6828
	}
6829
	storage->rx_dropped += atomic_long_read(&dev->rx_dropped);
6830
	storage->tx_dropped += atomic_long_read(&dev->tx_dropped);
6831
	return storage;
R
Rusty Russell 已提交
6832
}
6833
EXPORT_SYMBOL(dev_get_stats);
R
Rusty Russell 已提交
6834

6835
struct netdev_queue *dev_ingress_queue_create(struct net_device *dev)
6836
{
6837
	struct netdev_queue *queue = dev_ingress_queue(dev);
6838

6839 6840 6841 6842 6843 6844 6845
#ifdef CONFIG_NET_CLS_ACT
	if (queue)
		return queue;
	queue = kzalloc(sizeof(*queue), GFP_KERNEL);
	if (!queue)
		return NULL;
	netdev_init_one_queue(dev, queue, NULL);
E
Eric Dumazet 已提交
6846
	RCU_INIT_POINTER(queue->qdisc, &noop_qdisc);
6847 6848 6849 6850
	queue->qdisc_sleeping = &noop_qdisc;
	rcu_assign_pointer(dev->ingress_queue, queue);
#endif
	return queue;
6851 6852
}

6853 6854
static const struct ethtool_ops default_ethtool_ops;

6855 6856 6857 6858 6859 6860 6861 6862
void netdev_set_default_ethtool_ops(struct net_device *dev,
				    const struct ethtool_ops *ops)
{
	if (dev->ethtool_ops == &default_ethtool_ops)
		dev->ethtool_ops = ops;
}
EXPORT_SYMBOL_GPL(netdev_set_default_ethtool_ops);

6863 6864 6865 6866
void netdev_freemem(struct net_device *dev)
{
	char *addr = (char *)dev - dev->padded;

W
WANG Cong 已提交
6867
	kvfree(addr);
6868 6869
}

L
Linus Torvalds 已提交
6870
/**
T
Tom Herbert 已提交
6871
 *	alloc_netdev_mqs - allocate network device
6872 6873 6874 6875 6876 6877
 *	@sizeof_priv:		size of private data to allocate space for
 *	@name:			device name format string
 *	@name_assign_type: 	origin of device name
 *	@setup:			callback to initialize device
 *	@txqs:			the number of TX subqueues to allocate
 *	@rxqs:			the number of RX subqueues to allocate
L
Linus Torvalds 已提交
6878 6879
 *
 *	Allocates a struct net_device with private data area for driver use
L
Li Zhong 已提交
6880
 *	and performs basic initialization.  Also allocates subqueue structs
T
Tom Herbert 已提交
6881
 *	for each queue on the device.
L
Linus Torvalds 已提交
6882
 */
T
Tom Herbert 已提交
6883
struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
6884
		unsigned char name_assign_type,
T
Tom Herbert 已提交
6885 6886
		void (*setup)(struct net_device *),
		unsigned int txqs, unsigned int rxqs)
L
Linus Torvalds 已提交
6887 6888
{
	struct net_device *dev;
6889
	size_t alloc_size;
6890
	struct net_device *p;
L
Linus Torvalds 已提交
6891

6892 6893
	BUG_ON(strlen(name) >= sizeof(dev->name));

T
Tom Herbert 已提交
6894
	if (txqs < 1) {
6895
		pr_err("alloc_netdev: Unable to allocate device with zero queues\n");
6896 6897 6898
		return NULL;
	}

6899
#ifdef CONFIG_SYSFS
T
Tom Herbert 已提交
6900
	if (rxqs < 1) {
6901
		pr_err("alloc_netdev: Unable to allocate device with zero RX queues\n");
T
Tom Herbert 已提交
6902 6903 6904 6905
		return NULL;
	}
#endif

6906
	alloc_size = sizeof(struct net_device);
6907 6908
	if (sizeof_priv) {
		/* ensure 32-byte alignment of private area */
6909
		alloc_size = ALIGN(alloc_size, NETDEV_ALIGN);
6910 6911 6912
		alloc_size += sizeof_priv;
	}
	/* ensure 32-byte alignment of whole construct */
6913
	alloc_size += NETDEV_ALIGN - 1;
L
Linus Torvalds 已提交
6914

6915 6916 6917
	p = kzalloc(alloc_size, GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);
	if (!p)
		p = vzalloc(alloc_size);
6918
	if (!p)
L
Linus Torvalds 已提交
6919 6920
		return NULL;

6921
	dev = PTR_ALIGN(p, NETDEV_ALIGN);
L
Linus Torvalds 已提交
6922
	dev->padded = (char *)dev - (char *)p;
6923

E
Eric Dumazet 已提交
6924 6925
	dev->pcpu_refcnt = alloc_percpu(int);
	if (!dev->pcpu_refcnt)
6926
		goto free_dev;
6927 6928

	if (dev_addr_init(dev))
E
Eric Dumazet 已提交
6929
		goto free_pcpu;
6930

6931
	dev_mc_init(dev);
6932
	dev_uc_init(dev);
J
Jiri Pirko 已提交
6933

6934
	dev_net_set(dev, &init_net);
L
Linus Torvalds 已提交
6935

6936
	dev->gso_max_size = GSO_MAX_SIZE;
6937
	dev->gso_max_segs = GSO_MAX_SEGS;
6938
	dev->gso_min_segs = 0;
6939 6940 6941

	INIT_LIST_HEAD(&dev->napi_list);
	INIT_LIST_HEAD(&dev->unreg_list);
6942
	INIT_LIST_HEAD(&dev->close_list);
6943
	INIT_LIST_HEAD(&dev->link_watch_list);
6944 6945 6946 6947
	INIT_LIST_HEAD(&dev->adj_list.upper);
	INIT_LIST_HEAD(&dev->adj_list.lower);
	INIT_LIST_HEAD(&dev->all_adj_list.upper);
	INIT_LIST_HEAD(&dev->all_adj_list.lower);
6948 6949
	INIT_LIST_HEAD(&dev->ptype_all);
	INIT_LIST_HEAD(&dev->ptype_specific);
6950
	dev->priv_flags = IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM;
6951 6952
	setup(dev);

T
Tom Herbert 已提交
6953 6954
	dev->num_tx_queues = txqs;
	dev->real_num_tx_queues = txqs;
6955
	if (netif_alloc_netdev_queues(dev))
6956
		goto free_all;
6957

6958
#ifdef CONFIG_SYSFS
T
Tom Herbert 已提交
6959 6960
	dev->num_rx_queues = rxqs;
	dev->real_num_rx_queues = rxqs;
T
Tom Herbert 已提交
6961
	if (netif_alloc_rx_queues(dev))
6962
		goto free_all;
E
Eric Dumazet 已提交
6963
#endif
T
Tom Herbert 已提交
6964

L
Linus Torvalds 已提交
6965
	strcpy(dev->name, name);
6966
	dev->name_assign_type = name_assign_type;
6967
	dev->group = INIT_NETDEV_GROUP;
6968 6969
	if (!dev->ethtool_ops)
		dev->ethtool_ops = &default_ethtool_ops;
L
Linus Torvalds 已提交
6970
	return dev;
6971

6972 6973 6974 6975
free_all:
	free_netdev(dev);
	return NULL;

E
Eric Dumazet 已提交
6976 6977
free_pcpu:
	free_percpu(dev->pcpu_refcnt);
6978 6979
free_dev:
	netdev_freemem(dev);
6980
	return NULL;
L
Linus Torvalds 已提交
6981
}
T
Tom Herbert 已提交
6982
EXPORT_SYMBOL(alloc_netdev_mqs);
L
Linus Torvalds 已提交
6983 6984 6985 6986 6987

/**
 *	free_netdev - free network device
 *	@dev: device
 *
6988 6989
 *	This function does the last stage of destroying an allocated device
 * 	interface. The reference to the device object is released.
L
Linus Torvalds 已提交
6990 6991 6992 6993
 *	If this is the last reference then it will be freed.
 */
void free_netdev(struct net_device *dev)
{
6994 6995
	struct napi_struct *p, *n;

6996
	netif_free_tx_queues(dev);
6997
#ifdef CONFIG_SYSFS
6998
	kvfree(dev->_rx);
T
Tom Herbert 已提交
6999
#endif
7000

7001
	kfree(rcu_dereference_protected(dev->ingress_queue, 1));
7002

7003 7004 7005
	/* Flush device addresses */
	dev_addr_flush(dev);

7006 7007 7008
	list_for_each_entry_safe(p, n, &dev->napi_list, dev_list)
		netif_napi_del(p);

E
Eric Dumazet 已提交
7009 7010 7011
	free_percpu(dev->pcpu_refcnt);
	dev->pcpu_refcnt = NULL;

S
Stephen Hemminger 已提交
7012
	/*  Compatibility with error handling in drivers */
L
Linus Torvalds 已提交
7013
	if (dev->reg_state == NETREG_UNINITIALIZED) {
7014
		netdev_freemem(dev);
L
Linus Torvalds 已提交
7015 7016 7017 7018 7019 7020
		return;
	}

	BUG_ON(dev->reg_state != NETREG_UNREGISTERED);
	dev->reg_state = NETREG_RELEASED;

7021 7022
	/* will free via device release */
	put_device(&dev->dev);
L
Linus Torvalds 已提交
7023
}
E
Eric Dumazet 已提交
7024
EXPORT_SYMBOL(free_netdev);
7025

7026 7027 7028 7029 7030 7031
/**
 *	synchronize_net -  Synchronize with packet receive processing
 *
 *	Wait for packets currently being received to be done.
 *	Does not block later packets from starting.
 */
7032
void synchronize_net(void)
L
Linus Torvalds 已提交
7033 7034
{
	might_sleep();
7035 7036 7037 7038
	if (rtnl_is_locked())
		synchronize_rcu_expedited();
	else
		synchronize_rcu();
L
Linus Torvalds 已提交
7039
}
E
Eric Dumazet 已提交
7040
EXPORT_SYMBOL(synchronize_net);
L
Linus Torvalds 已提交
7041 7042

/**
7043
 *	unregister_netdevice_queue - remove device from the kernel
L
Linus Torvalds 已提交
7044
 *	@dev: device
7045
 *	@head: list
7046
 *
L
Linus Torvalds 已提交
7047
 *	This function shuts down a device interface and removes it
7048
 *	from the kernel tables.
7049
 *	If head not NULL, device is queued to be unregistered later.
L
Linus Torvalds 已提交
7050 7051 7052 7053 7054
 *
 *	Callers must hold the rtnl semaphore.  You may want
 *	unregister_netdev() instead of this.
 */

7055
void unregister_netdevice_queue(struct net_device *dev, struct list_head *head)
L
Linus Torvalds 已提交
7056
{
7057 7058
	ASSERT_RTNL();

7059
	if (head) {
7060
		list_move_tail(&dev->unreg_list, head);
7061 7062 7063 7064 7065
	} else {
		rollback_registered(dev);
		/* Finish processing unregister after unlock */
		net_set_todo(dev);
	}
L
Linus Torvalds 已提交
7066
}
7067
EXPORT_SYMBOL(unregister_netdevice_queue);
L
Linus Torvalds 已提交
7068

7069 7070 7071
/**
 *	unregister_netdevice_many - unregister many devices
 *	@head: list of devices
7072 7073 7074
 *
 *  Note: As most callers use a stack allocated list_head,
 *  we force a list_del() to make sure stack wont be corrupted later.
7075 7076 7077 7078 7079 7080 7081 7082 7083
 */
void unregister_netdevice_many(struct list_head *head)
{
	struct net_device *dev;

	if (!list_empty(head)) {
		rollback_registered_many(head);
		list_for_each_entry(dev, head, unreg_list)
			net_set_todo(dev);
7084
		list_del(head);
7085 7086
	}
}
7087
EXPORT_SYMBOL(unregister_netdevice_many);
7088

L
Linus Torvalds 已提交
7089 7090 7091 7092 7093
/**
 *	unregister_netdev - remove device from the kernel
 *	@dev: device
 *
 *	This function shuts down a device interface and removes it
7094
 *	from the kernel tables.
L
Linus Torvalds 已提交
7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106 7107
 *
 *	This is just a wrapper for unregister_netdevice that takes
 *	the rtnl semaphore.  In general you want to use this and not
 *	unregister_netdevice.
 */
void unregister_netdev(struct net_device *dev)
{
	rtnl_lock();
	unregister_netdevice(dev);
	rtnl_unlock();
}
EXPORT_SYMBOL(unregister_netdev);

7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138
/**
 *	dev_change_net_namespace - move device to different nethost namespace
 *	@dev: device
 *	@net: network namespace
 *	@pat: If not NULL name pattern to try if the current device name
 *	      is already taken in the destination network namespace.
 *
 *	This function shuts down a device interface and moves it
 *	to a new network namespace. On success 0 is returned, on
 *	a failure a netagive errno code is returned.
 *
 *	Callers must hold the rtnl semaphore.
 */

int dev_change_net_namespace(struct net_device *dev, struct net *net, const char *pat)
{
	int err;

	ASSERT_RTNL();

	/* Don't allow namespace local devices to be moved. */
	err = -EINVAL;
	if (dev->features & NETIF_F_NETNS_LOCAL)
		goto out;

	/* Ensure the device has been registrered */
	if (dev->reg_state != NETREG_REGISTERED)
		goto out;

	/* Get out if there is nothing todo */
	err = 0;
7139
	if (net_eq(dev_net(dev), net))
7140 7141 7142 7143 7144 7145
		goto out;

	/* Pick the destination device name, and ensure
	 * we can use it in the destination network namespace.
	 */
	err = -EEXIST;
7146
	if (__dev_get_by_name(net, dev->name)) {
7147 7148 7149
		/* We get here if we can't use the current device name */
		if (!pat)
			goto out;
7150
		if (dev_get_valid_name(net, dev, pat) < 0)
7151 7152 7153 7154 7155 7156 7157 7158
			goto out;
	}

	/*
	 * And now a mini version of register_netdevice unregister_netdevice.
	 */

	/* If device is running close it first. */
7159
	dev_close(dev);
7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171

	/* And unlink it from device chain */
	err = -ENODEV;
	unlist_netdevice(dev);

	synchronize_net();

	/* Shutdown queueing discipline. */
	dev_shutdown(dev);

	/* Notify protocols, that we are about to destroy
	   this device. They should clean all the things.
7172 7173 7174 7175

	   Note that dev->reg_state stays at NETREG_REGISTERED.
	   This is wanted because this way 8021q and macvlan know
	   the device is just moving and can keep their slaves up.
7176 7177
	*/
	call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
7178 7179
	rcu_barrier();
	call_netdevice_notifiers(NETDEV_UNREGISTER_FINAL, dev);
7180
	rtmsg_ifinfo(RTM_DELLINK, dev, ~0U, GFP_KERNEL);
7181 7182 7183 7184

	/*
	 *	Flush the unicast and multicast chains
	 */
7185
	dev_uc_flush(dev);
7186
	dev_mc_flush(dev);
7187

7188 7189
	/* Send a netdev-removed uevent to the old namespace */
	kobject_uevent(&dev->dev.kobj, KOBJ_REMOVE);
7190
	netdev_adjacent_del_links(dev);
7191

7192
	/* Actually switch the network namespace */
7193
	dev_net_set(dev, net);
7194 7195

	/* If there is an ifindex conflict assign a new one */
7196
	if (__dev_get_by_index(net, dev->ifindex))
7197 7198
		dev->ifindex = dev_new_index(net);

7199 7200
	/* Send a netdev-add uevent to the new namespace */
	kobject_uevent(&dev->dev.kobj, KOBJ_ADD);
7201
	netdev_adjacent_add_links(dev);
7202

7203
	/* Fixup kobjects */
7204
	err = device_rename(&dev->dev, dev->name);
7205
	WARN_ON(err);
7206 7207 7208 7209 7210 7211 7212

	/* Add the device back in the hashes */
	list_netdevice(dev);

	/* Notify protocols, that a new device appeared. */
	call_netdevice_notifiers(NETDEV_REGISTER, dev);

7213 7214 7215 7216
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
7217
	rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U, GFP_KERNEL);
7218

7219 7220 7221 7222 7223
	synchronize_net();
	err = 0;
out:
	return err;
}
7224
EXPORT_SYMBOL_GPL(dev_change_net_namespace);
7225

L
Linus Torvalds 已提交
7226 7227 7228 7229 7230 7231 7232 7233 7234
static int dev_cpu_callback(struct notifier_block *nfb,
			    unsigned long action,
			    void *ocpu)
{
	struct sk_buff **list_skb;
	struct sk_buff *skb;
	unsigned int cpu, oldcpu = (unsigned long)ocpu;
	struct softnet_data *sd, *oldsd;

7235
	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN)
L
Linus Torvalds 已提交
7236 7237 7238 7239 7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251
		return NOTIFY_OK;

	local_irq_disable();
	cpu = smp_processor_id();
	sd = &per_cpu(softnet_data, cpu);
	oldsd = &per_cpu(softnet_data, oldcpu);

	/* Find end of our completion_queue. */
	list_skb = &sd->completion_queue;
	while (*list_skb)
		list_skb = &(*list_skb)->next;
	/* Append completion queue from offline CPU. */
	*list_skb = oldsd->completion_queue;
	oldsd->completion_queue = NULL;

	/* Append output queue from offline CPU. */
7252 7253 7254 7255 7256 7257
	if (oldsd->output_queue) {
		*sd->output_queue_tailp = oldsd->output_queue;
		sd->output_queue_tailp = oldsd->output_queue_tailp;
		oldsd->output_queue = NULL;
		oldsd->output_queue_tailp = &oldsd->output_queue;
	}
E
Eric Dumazet 已提交
7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271
	/* Append NAPI poll list from offline CPU, with one exception :
	 * process_backlog() must be called by cpu owning percpu backlog.
	 * We properly handle process_queue & input_pkt_queue later.
	 */
	while (!list_empty(&oldsd->poll_list)) {
		struct napi_struct *napi = list_first_entry(&oldsd->poll_list,
							    struct napi_struct,
							    poll_list);

		list_del_init(&napi->poll_list);
		if (napi->poll == process_backlog)
			napi->state = 0;
		else
			____napi_schedule(sd, napi);
7272
	}
L
Linus Torvalds 已提交
7273 7274 7275 7276 7277

	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_enable();

	/* Process offline CPU's input_pkt_queue */
7278
	while ((skb = __skb_dequeue(&oldsd->process_queue))) {
7279
		netif_rx_ni(skb);
7280
		input_queue_head_incr(oldsd);
T
Tom Herbert 已提交
7281
	}
E
Eric Dumazet 已提交
7282
	while ((skb = skb_dequeue(&oldsd->input_pkt_queue))) {
7283
		netif_rx_ni(skb);
7284 7285
		input_queue_head_incr(oldsd);
	}
L
Linus Torvalds 已提交
7286 7287 7288 7289 7290

	return NOTIFY_OK;
}


7291
/**
7292 7293 7294 7295
 *	netdev_increment_features - increment feature set by one
 *	@all: current feature set
 *	@one: new feature set
 *	@mask: mask feature set
7296 7297
 *
 *	Computes a new feature set after adding a device with feature set
7298 7299
 *	@one to the master device with current feature set @all.  Will not
 *	enable anything that is off in @mask. Returns the new feature set.
7300
 */
7301 7302
netdev_features_t netdev_increment_features(netdev_features_t all,
	netdev_features_t one, netdev_features_t mask)
7303
{
7304 7305 7306
	if (mask & NETIF_F_GEN_CSUM)
		mask |= NETIF_F_ALL_CSUM;
	mask |= NETIF_F_VLAN_CHALLENGED;
7307

7308 7309
	all |= one & (NETIF_F_ONE_FOR_ALL|NETIF_F_ALL_CSUM) & mask;
	all &= one | ~NETIF_F_ALL_FOR_ALL;
7310

7311 7312 7313
	/* If one device supports hw checksumming, set for all. */
	if (all & NETIF_F_GEN_CSUM)
		all &= ~(NETIF_F_ALL_CSUM & ~NETIF_F_GEN_CSUM);
7314 7315 7316

	return all;
}
7317
EXPORT_SYMBOL(netdev_increment_features);
7318

7319
static struct hlist_head * __net_init netdev_create_hash(void)
7320 7321 7322 7323 7324 7325 7326 7327 7328 7329 7330 7331
{
	int i;
	struct hlist_head *hash;

	hash = kmalloc(sizeof(*hash) * NETDEV_HASHENTRIES, GFP_KERNEL);
	if (hash != NULL)
		for (i = 0; i < NETDEV_HASHENTRIES; i++)
			INIT_HLIST_HEAD(&hash[i]);

	return hash;
}

7332
/* Initialize per network namespace state */
7333
static int __net_init netdev_init(struct net *net)
7334
{
7335 7336
	if (net != &init_net)
		INIT_LIST_HEAD(&net->dev_base_head);
7337

7338 7339 7340
	net->dev_name_head = netdev_create_hash();
	if (net->dev_name_head == NULL)
		goto err_name;
7341

7342 7343 7344
	net->dev_index_head = netdev_create_hash();
	if (net->dev_index_head == NULL)
		goto err_idx;
7345 7346

	return 0;
7347 7348 7349 7350 7351

err_idx:
	kfree(net->dev_name_head);
err_name:
	return -ENOMEM;
7352 7353
}

7354 7355 7356 7357 7358 7359
/**
 *	netdev_drivername - network driver for the device
 *	@dev: network device
 *
 *	Determine network driver for device.
 */
7360
const char *netdev_drivername(const struct net_device *dev)
7361
{
7362 7363
	const struct device_driver *driver;
	const struct device *parent;
7364
	const char *empty = "";
7365 7366 7367

	parent = dev->dev.parent;
	if (!parent)
7368
		return empty;
7369 7370 7371

	driver = parent->driver;
	if (driver && driver->name)
7372 7373
		return driver->name;
	return empty;
7374 7375
}

7376 7377
static void __netdev_printk(const char *level, const struct net_device *dev,
			    struct va_format *vaf)
7378
{
7379
	if (dev && dev->dev.parent) {
7380 7381 7382 7383 7384 7385 7386
		dev_printk_emit(level[1] - '0',
				dev->dev.parent,
				"%s %s %s%s: %pV",
				dev_driver_string(dev->dev.parent),
				dev_name(dev->dev.parent),
				netdev_name(dev), netdev_reg_state(dev),
				vaf);
7387
	} else if (dev) {
7388 7389
		printk("%s%s%s: %pV",
		       level, netdev_name(dev), netdev_reg_state(dev), vaf);
7390
	} else {
7391
		printk("%s(NULL net_device): %pV", level, vaf);
7392
	}
7393 7394
}

7395 7396
void netdev_printk(const char *level, const struct net_device *dev,
		   const char *format, ...)
7397 7398 7399 7400 7401 7402 7403 7404 7405
{
	struct va_format vaf;
	va_list args;

	va_start(args, format);

	vaf.fmt = format;
	vaf.va = &args;

7406
	__netdev_printk(level, dev, &vaf);
7407

7408 7409 7410 7411 7412
	va_end(args);
}
EXPORT_SYMBOL(netdev_printk);

#define define_netdev_printk_level(func, level)			\
7413
void func(const struct net_device *dev, const char *fmt, ...)	\
7414 7415 7416 7417 7418 7419 7420 7421 7422
{								\
	struct va_format vaf;					\
	va_list args;						\
								\
	va_start(args, fmt);					\
								\
	vaf.fmt = fmt;						\
	vaf.va = &args;						\
								\
7423
	__netdev_printk(level, dev, &vaf);			\
7424
								\
7425 7426 7427 7428 7429 7430 7431 7432 7433 7434 7435 7436
	va_end(args);						\
}								\
EXPORT_SYMBOL(func);

define_netdev_printk_level(netdev_emerg, KERN_EMERG);
define_netdev_printk_level(netdev_alert, KERN_ALERT);
define_netdev_printk_level(netdev_crit, KERN_CRIT);
define_netdev_printk_level(netdev_err, KERN_ERR);
define_netdev_printk_level(netdev_warn, KERN_WARNING);
define_netdev_printk_level(netdev_notice, KERN_NOTICE);
define_netdev_printk_level(netdev_info, KERN_INFO);

7437
static void __net_exit netdev_exit(struct net *net)
7438 7439 7440 7441 7442
{
	kfree(net->dev_name_head);
	kfree(net->dev_index_head);
}

7443
static struct pernet_operations __net_initdata netdev_net_ops = {
7444 7445 7446 7447
	.init = netdev_init,
	.exit = netdev_exit,
};

7448
static void __net_exit default_device_exit(struct net *net)
7449
{
7450
	struct net_device *dev, *aux;
7451
	/*
7452
	 * Push all migratable network devices back to the
7453 7454 7455
	 * initial network namespace
	 */
	rtnl_lock();
7456
	for_each_netdev_safe(net, dev, aux) {
7457
		int err;
7458
		char fb_name[IFNAMSIZ];
7459 7460 7461 7462 7463

		/* Ignore unmoveable devices (i.e. loopback) */
		if (dev->features & NETIF_F_NETNS_LOCAL)
			continue;

7464 7465 7466
		/* Leave virtual devices for the generic cleanup */
		if (dev->rtnl_link_ops)
			continue;
7467

L
Lucas De Marchi 已提交
7468
		/* Push remaining network devices to init_net */
7469 7470
		snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
		err = dev_change_net_namespace(dev, &init_net, fb_name);
7471
		if (err) {
7472 7473
			pr_emerg("%s: failed to move %s to init_net: %d\n",
				 __func__, dev->name, err);
7474
			BUG();
7475 7476 7477 7478 7479
		}
	}
	rtnl_unlock();
}

7480 7481 7482 7483 7484 7485 7486
static void __net_exit rtnl_lock_unregistering(struct list_head *net_list)
{
	/* Return with the rtnl_lock held when there are no network
	 * devices unregistering in any network namespace in net_list.
	 */
	struct net *net;
	bool unregistering;
7487
	DEFINE_WAIT_FUNC(wait, woken_wake_function);
7488

7489
	add_wait_queue(&netdev_unregistering_wq, &wait);
7490 7491 7492 7493 7494 7495 7496 7497 7498 7499 7500 7501
	for (;;) {
		unregistering = false;
		rtnl_lock();
		list_for_each_entry(net, net_list, exit_list) {
			if (net->dev_unreg_count > 0) {
				unregistering = true;
				break;
			}
		}
		if (!unregistering)
			break;
		__rtnl_unlock();
7502 7503

		wait_woken(&wait, TASK_UNINTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT);
7504
	}
7505
	remove_wait_queue(&netdev_unregistering_wq, &wait);
7506 7507
}

7508 7509 7510
static void __net_exit default_device_exit_batch(struct list_head *net_list)
{
	/* At exit all network devices most be removed from a network
7511
	 * namespace.  Do this in the reverse order of registration.
7512 7513 7514 7515 7516 7517 7518
	 * Do this across as many network namespaces as possible to
	 * improve batching efficiency.
	 */
	struct net_device *dev;
	struct net *net;
	LIST_HEAD(dev_kill_list);

7519 7520 7521 7522 7523 7524 7525 7526 7527 7528 7529 7530
	/* To prevent network device cleanup code from dereferencing
	 * loopback devices or network devices that have been freed
	 * wait here for all pending unregistrations to complete,
	 * before unregistring the loopback device and allowing the
	 * network namespace be freed.
	 *
	 * The netdev todo list containing all network devices
	 * unregistrations that happen in default_device_exit_batch
	 * will run in the rtnl_unlock() at the end of
	 * default_device_exit_batch.
	 */
	rtnl_lock_unregistering(net_list);
7531 7532
	list_for_each_entry(net, net_list, exit_list) {
		for_each_netdev_reverse(net, dev) {
7533
			if (dev->rtnl_link_ops && dev->rtnl_link_ops->dellink)
7534 7535 7536 7537 7538 7539 7540 7541 7542
				dev->rtnl_link_ops->dellink(dev, &dev_kill_list);
			else
				unregister_netdevice_queue(dev, &dev_kill_list);
		}
	}
	unregister_netdevice_many(&dev_kill_list);
	rtnl_unlock();
}

7543
static struct pernet_operations __net_initdata default_device_ops = {
7544
	.exit = default_device_exit,
7545
	.exit_batch = default_device_exit_batch,
7546 7547
};

L
Linus Torvalds 已提交
7548 7549 7550 7551 7552 7553 7554 7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567
/*
 *	Initialize the DEV module. At boot time this walks the device list and
 *	unhooks any devices that fail to initialise (normally hardware not
 *	present) and leaves us with a valid list of present and active devices.
 *
 */

/*
 *       This is called single threaded during boot, so no need
 *       to take the rtnl semaphore.
 */
static int __init net_dev_init(void)
{
	int i, rc = -ENOMEM;

	BUG_ON(!dev_boot_phase);

	if (dev_proc_init())
		goto out;

7568
	if (netdev_kobject_init())
L
Linus Torvalds 已提交
7569 7570 7571
		goto out;

	INIT_LIST_HEAD(&ptype_all);
7572
	for (i = 0; i < PTYPE_HASH_SIZE; i++)
L
Linus Torvalds 已提交
7573 7574
		INIT_LIST_HEAD(&ptype_base[i]);

7575 7576
	INIT_LIST_HEAD(&offload_base);

7577 7578
	if (register_pernet_subsys(&netdev_net_ops))
		goto out;
L
Linus Torvalds 已提交
7579 7580 7581 7582 7583

	/*
	 *	Initialise the packet receive queues.
	 */

7584
	for_each_possible_cpu(i) {
E
Eric Dumazet 已提交
7585
		struct softnet_data *sd = &per_cpu(softnet_data, i);
L
Linus Torvalds 已提交
7586

E
Eric Dumazet 已提交
7587
		skb_queue_head_init(&sd->input_pkt_queue);
7588
		skb_queue_head_init(&sd->process_queue);
E
Eric Dumazet 已提交
7589
		INIT_LIST_HEAD(&sd->poll_list);
7590
		sd->output_queue_tailp = &sd->output_queue;
E
Eric Dumazet 已提交
7591
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
7592 7593 7594
		sd->csd.func = rps_trigger_softirq;
		sd->csd.info = sd;
		sd->cpu = i;
7595
#endif
T
Tom Herbert 已提交
7596

E
Eric Dumazet 已提交
7597 7598
		sd->backlog.poll = process_backlog;
		sd->backlog.weight = weight_p;
L
Linus Torvalds 已提交
7599 7600 7601 7602
	}

	dev_boot_phase = 0;

7603 7604 7605 7606 7607 7608 7609 7610 7611 7612 7613 7614 7615 7616 7617
	/* The loopback device is special if any other network devices
	 * is present in a network namespace the loopback device must
	 * be present. Since we now dynamically allocate and free the
	 * loopback device ensure this invariant is maintained by
	 * keeping the loopback device as the first device on the
	 * list of network devices.  Ensuring the loopback devices
	 * is the first device that appears and the last network device
	 * that disappears.
	 */
	if (register_pernet_device(&loopback_net_ops))
		goto out;

	if (register_pernet_device(&default_device_ops))
		goto out;

7618 7619
	open_softirq(NET_TX_SOFTIRQ, net_tx_action);
	open_softirq(NET_RX_SOFTIRQ, net_rx_action);
L
Linus Torvalds 已提交
7620 7621 7622 7623 7624 7625 7626 7627 7628

	hotcpu_notifier(dev_cpu_callback, 0);
	dst_init();
	rc = 0;
out:
	return rc;
}

subsys_initcall(net_dev_init);