dev.c 162.6 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 99 100 101 102 103 104 105
#include <net/sock.h>
#include <linux/rtnetlink.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/stat.h>
#include <net/dst.h>
#include <net/pkt_sched.h>
#include <net/checksum.h>
106
#include <net/xfrm.h>
L
Linus Torvalds 已提交
107 108 109 110 111 112 113
#include <linux/highmem.h>
#include <linux/init.h>
#include <linux/kmod.h>
#include <linux/module.h>
#include <linux/netpoll.h>
#include <linux/rcupdate.h>
#include <linux/delay.h>
114
#include <net/wext.h>
L
Linus Torvalds 已提交
115 116
#include <net/iw_handler.h>
#include <asm/current.h>
S
Steve Grubb 已提交
117
#include <linux/audit.h>
118
#include <linux/dmaengine.h>
119
#include <linux/err.h>
120
#include <linux/ctype.h>
121
#include <linux/if_arp.h>
122
#include <linux/if_vlan.h>
123
#include <linux/ip.h>
124
#include <net/ip.h>
125 126
#include <linux/ipv6.h>
#include <linux/in.h>
D
David S. Miller 已提交
127 128
#include <linux/jhash.h>
#include <linux/random.h>
129
#include <trace/events/napi.h>
130
#include <trace/events/net.h>
131
#include <trace/events/skb.h>
132
#include <linux/pci.h>
133
#include <linux/inetdevice.h>
134
#include <linux/cpu_rmap.h>
135
#include <linux/net_tstamp.h>
136
#include <linux/static_key.h>
137
#include <net/flow_keys.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
147 148 149 150 151 152 153 154 155 156 157
/*
 *	The list of packet types we will receive (as opposed to discard)
 *	and the routines to invoke.
 *
 *	Why 16. Because with 16 the only overlap we get on a hash of the
 *	low nibble of the protocol value is RARP/SNAP/X.25.
 *
 *      NOTE:  That is no longer true with the addition of VLAN tags.  Not
 *             sure which should go first, but I bet it won't make much
 *             difference if we are running VLANs.  The good news is that
 *             this protocol won't be in the list unless compiled in, so
S
Stephen Hemminger 已提交
158
 *             the average user (w/out VLANs) will not be adversely affected.
L
Linus Torvalds 已提交
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
 *             --BLG
 *
 *		0800	IP
 *		8100    802.1Q VLAN
 *		0001	802.3
 *		0002	AX.25
 *		0004	802.2
 *		8035	RARP
 *		0005	SNAP
 *		0805	X.25
 *		0806	ARP
 *		8137	IPX
 *		0009	Localtalk
 *		86DD	IPv6
 */

175 176 177
#define PTYPE_HASH_SIZE	(16)
#define PTYPE_HASH_MASK	(PTYPE_HASH_SIZE - 1)

L
Linus Torvalds 已提交
178
static DEFINE_SPINLOCK(ptype_lock);
179
static struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
180
static struct list_head ptype_all __read_mostly;	/* Taps */
L
Linus Torvalds 已提交
181 182

/*
183
 * The @dev_base_head list is protected by @dev_base_lock and the rtnl
L
Linus Torvalds 已提交
184 185
 * semaphore.
 *
186
 * Pure readers hold dev_base_lock for reading, or rcu_read_lock()
L
Linus Torvalds 已提交
187 188
 *
 * Writers must hold the rtnl semaphore while they loop through the
189
 * dev_base_head list, and hold dev_base_lock for writing when they do the
L
Linus Torvalds 已提交
190 191 192 193 194 195 196 197 198 199 200 201 202 203
 * 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);

204 205 206 207 208
static inline void dev_base_seq_inc(struct net *net)
{
	while (++net->dev_base_seq == 0);
}

209
static inline struct hlist_head *dev_name_hash(struct net *net, const char *name)
L
Linus Torvalds 已提交
210
{
211 212
	unsigned int hash = full_name_hash(name, strnlen(name, IFNAMSIZ));

213
	return &net->dev_name_head[hash_32(hash, NETDEV_HASHBITS)];
L
Linus Torvalds 已提交
214 215
}

216
static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex)
L
Linus Torvalds 已提交
217
{
218
	return &net->dev_index_head[ifindex & (NETDEV_HASHENTRIES - 1)];
L
Linus Torvalds 已提交
219 220
}

E
Eric Dumazet 已提交
221
static inline void rps_lock(struct softnet_data *sd)
222 223
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
224
	spin_lock(&sd->input_pkt_queue.lock);
225 226 227
#endif
}

E
Eric Dumazet 已提交
228
static inline void rps_unlock(struct softnet_data *sd)
229 230
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
231
	spin_unlock(&sd->input_pkt_queue.lock);
232 233 234
#endif
}

235 236 237
/* Device list insertion */
static int list_netdevice(struct net_device *dev)
{
238
	struct net *net = dev_net(dev);
239 240 241 242

	ASSERT_RTNL();

	write_lock_bh(&dev_base_lock);
243
	list_add_tail_rcu(&dev->dev_list, &net->dev_base_head);
244
	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net, dev->name));
245 246
	hlist_add_head_rcu(&dev->index_hlist,
			   dev_index_hash(net, dev->ifindex));
247
	write_unlock_bh(&dev_base_lock);
248 249 250

	dev_base_seq_inc(net);

251 252 253
	return 0;
}

254 255 256
/* Device list removal
 * caller must respect a RCU grace period before freeing/reusing dev
 */
257 258 259 260 261 262
static void unlist_netdevice(struct net_device *dev)
{
	ASSERT_RTNL();

	/* Unlink dev from the device chain */
	write_lock_bh(&dev_base_lock);
263
	list_del_rcu(&dev->dev_list);
264
	hlist_del_rcu(&dev->name_hlist);
265
	hlist_del_rcu(&dev->index_hlist);
266
	write_unlock_bh(&dev_base_lock);
267 268

	dev_base_seq_inc(dev_net(dev));
269 270
}

L
Linus Torvalds 已提交
271 272 273 274
/*
 *	Our notifier list
 */

275
static RAW_NOTIFIER_HEAD(netdev_chain);
L
Linus Torvalds 已提交
276 277 278 279 280

/*
 *	Device drivers call our routines to queue packets here. We empty the
 *	queue in the local softnet handler.
 */
281

282
DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
E
Eric Dumazet 已提交
283
EXPORT_PER_CPU_SYMBOL(softnet_data);
L
Linus Torvalds 已提交
284

285
#ifdef CONFIG_LOCKDEP
286
/*
287
 * register_netdevice() inits txq->_xmit_lock and sets lockdep class
288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
 * 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,
303 304 305
	 ARPHRD_FCFABRIC, ARPHRD_IEEE80211, ARPHRD_IEEE80211_PRISM,
	 ARPHRD_IEEE80211_RADIOTAP, ARPHRD_PHONET, ARPHRD_PHONET_PIPE,
	 ARPHRD_IEEE802154, ARPHRD_VOID, ARPHRD_NONE};
306

307
static const char *const netdev_lock_name[] =
308 309 310 311 312 313 314 315 316 317 318 319
	{"_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",
320 321 322
	 "_xmit_FCFABRIC", "_xmit_IEEE80211", "_xmit_IEEE80211_PRISM",
	 "_xmit_IEEE80211_RADIOTAP", "_xmit_PHONET", "_xmit_PHONET_PIPE",
	 "_xmit_IEEE802154", "_xmit_VOID", "_xmit_NONE"};
323 324

static struct lock_class_key netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)];
325
static struct lock_class_key netdev_addr_lock_key[ARRAY_SIZE(netdev_lock_type)];
326 327 328 329 330 331 332 333 334 335 336 337

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;
}

338 339
static inline void netdev_set_xmit_lockdep_class(spinlock_t *lock,
						 unsigned short dev_type)
340 341 342 343 344 345 346
{
	int i;

	i = netdev_lock_pos(dev_type);
	lockdep_set_class_and_name(lock, &netdev_xmit_lock_key[i],
				   netdev_lock_name[i]);
}
347 348 349 350 351 352 353 354 355 356

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]);
}
357
#else
358 359 360 361 362
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)
363 364 365
{
}
#endif
L
Linus Torvalds 已提交
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388

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

		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)
 */

389 390 391 392 393 394 395 396
static inline struct list_head *ptype_head(const struct packet_type *pt)
{
	if (pt->type == htons(ETH_P_ALL))
		return &ptype_all;
	else
		return &ptype_base[ntohs(pt->type) & PTYPE_HASH_MASK];
}

L
Linus Torvalds 已提交
397 398 399 400 401 402 403 404
/**
 *	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.
 *
405
 *	This call does not sleep therefore it can not
L
Linus Torvalds 已提交
406 407 408 409 410 411
 *	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)
{
412
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
413

414 415 416
	spin_lock(&ptype_lock);
	list_add_rcu(&pt->list, head);
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
417
}
E
Eric Dumazet 已提交
418
EXPORT_SYMBOL(dev_add_pack);
L
Linus Torvalds 已提交
419 420 421 422 423 424 425 426

/**
 *	__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
427
 *	returns.
L
Linus Torvalds 已提交
428 429 430 431 432 433 434
 *
 *      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)
{
435
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
436 437
	struct packet_type *pt1;

438
	spin_lock(&ptype_lock);
L
Linus Torvalds 已提交
439 440 441 442 443 444 445 446

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

447
	pr_warn("dev_remove_pack: %p not found\n", pt);
L
Linus Torvalds 已提交
448
out:
449
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
450
}
E
Eric Dumazet 已提交
451 452
EXPORT_SYMBOL(__dev_remove_pack);

L
Linus Torvalds 已提交
453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
/**
 *	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);
468

L
Linus Torvalds 已提交
469 470
	synchronize_net();
}
E
Eric Dumazet 已提交
471
EXPORT_SYMBOL(dev_remove_pack);
L
Linus Torvalds 已提交
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499

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

		      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));
500
			strlcpy(s[i].name, name, IFNAMSIZ);
L
Linus Torvalds 已提交
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524
			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] != ' ' &&
525
		    !strcmp(dev->name, s[i].name)) {
L
Linus Torvalds 已提交
526 527 528 529 530 531 532 533 534
			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
535
EXPORT_SYMBOL(netdev_boot_setup_check);
L
Linus Torvalds 已提交
536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559


/**
 *	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
	 */
560
	if (__dev_get_by_name(&init_net, name))
L
Linus Torvalds 已提交
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605
		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

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

/**
 *	__dev_get_by_name	- find a device by its name
606
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
607 608 609 610 611 612 613 614 615
 *	@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.
 */

616
struct net_device *__dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
617 618
{
	struct hlist_node *p;
E
Eric Dumazet 已提交
619 620
	struct net_device *dev;
	struct hlist_head *head = dev_name_hash(net, name);
L
Linus Torvalds 已提交
621

E
Eric Dumazet 已提交
622
	hlist_for_each_entry(dev, p, head, name_hlist)
L
Linus Torvalds 已提交
623 624
		if (!strncmp(dev->name, name, IFNAMSIZ))
			return dev;
E
Eric Dumazet 已提交
625

L
Linus Torvalds 已提交
626 627
	return NULL;
}
E
Eric Dumazet 已提交
628
EXPORT_SYMBOL(__dev_get_by_name);
L
Linus Torvalds 已提交
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
/**
 *	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 hlist_node *p;
	struct net_device *dev;
	struct hlist_head *head = dev_name_hash(net, name);

	hlist_for_each_entry_rcu(dev, p, head, name_hlist)
		if (!strncmp(dev->name, name, IFNAMSIZ))
			return dev;

	return NULL;
}
EXPORT_SYMBOL(dev_get_by_name_rcu);

L
Linus Torvalds 已提交
656 657
/**
 *	dev_get_by_name		- find a device by its name
658
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
659 660 661 662 663 664 665 666 667
 *	@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.
 */

668
struct net_device *dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
669 670 671
{
	struct net_device *dev;

672 673
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
L
Linus Torvalds 已提交
674 675
	if (dev)
		dev_hold(dev);
676
	rcu_read_unlock();
L
Linus Torvalds 已提交
677 678
	return dev;
}
E
Eric Dumazet 已提交
679
EXPORT_SYMBOL(dev_get_by_name);
L
Linus Torvalds 已提交
680 681 682

/**
 *	__dev_get_by_index - find a device by its ifindex
683
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
684 685 686 687 688 689 690 691 692
 *	@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.
 */

693
struct net_device *__dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
694 695
{
	struct hlist_node *p;
E
Eric Dumazet 已提交
696 697
	struct net_device *dev;
	struct hlist_head *head = dev_index_hash(net, ifindex);
L
Linus Torvalds 已提交
698

E
Eric Dumazet 已提交
699
	hlist_for_each_entry(dev, p, head, index_hlist)
L
Linus Torvalds 已提交
700 701
		if (dev->ifindex == ifindex)
			return dev;
E
Eric Dumazet 已提交
702

L
Linus Torvalds 已提交
703 704
	return NULL;
}
E
Eric Dumazet 已提交
705
EXPORT_SYMBOL(__dev_get_by_index);
L
Linus Torvalds 已提交
706

707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
/**
 *	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 hlist_node *p;
	struct net_device *dev;
	struct hlist_head *head = dev_index_hash(net, ifindex);

	hlist_for_each_entry_rcu(dev, p, head, index_hlist)
		if (dev->ifindex == ifindex)
			return dev;

	return NULL;
}
EXPORT_SYMBOL(dev_get_by_index_rcu);

L
Linus Torvalds 已提交
732 733 734

/**
 *	dev_get_by_index - find a device by its ifindex
735
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
736 737 738 739 740 741 742 743
 *	@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.
 */

744
struct net_device *dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
745 746 747
{
	struct net_device *dev;

748 749
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifindex);
L
Linus Torvalds 已提交
750 751
	if (dev)
		dev_hold(dev);
752
	rcu_read_unlock();
L
Linus Torvalds 已提交
753 754
	return dev;
}
E
Eric Dumazet 已提交
755
EXPORT_SYMBOL(dev_get_by_index);
L
Linus Torvalds 已提交
756 757

/**
758
 *	dev_getbyhwaddr_rcu - find a device by its hardware address
759
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
760 761 762 763
 *	@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
764 765
 *	is not found or a pointer to the device.
 *	The caller must hold RCU or RTNL.
766
 *	The returned device has not had its ref count increased
L
Linus Torvalds 已提交
767 768 769 770
 *	and the caller must therefore be careful about locking
 *
 */

771 772
struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
				       const char *ha)
L
Linus Torvalds 已提交
773 774 775
{
	struct net_device *dev;

776
	for_each_netdev_rcu(net, dev)
L
Linus Torvalds 已提交
777 778
		if (dev->type == type &&
		    !memcmp(dev->dev_addr, ha, dev->addr_len))
779 780 781
			return dev;

	return NULL;
L
Linus Torvalds 已提交
782
}
783
EXPORT_SYMBOL(dev_getbyhwaddr_rcu);
784

785
struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type)
L
Linus Torvalds 已提交
786 787 788
{
	struct net_device *dev;

789
	ASSERT_RTNL();
790
	for_each_netdev(net, dev)
791
		if (dev->type == type)
792 793 794
			return dev;

	return NULL;
795 796 797
}
EXPORT_SYMBOL(__dev_getfirstbyhwtype);

798
struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
799
{
800
	struct net_device *dev, *ret = NULL;
801

802 803 804 805 806 807 808 809 810
	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
811 812 813 814
}
EXPORT_SYMBOL(dev_getfirstbyhwtype);

/**
E
Eric Dumazet 已提交
815
 *	dev_get_by_flags_rcu - find any device with given flags
816
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
817 818 819 820
 *	@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
821 822
 *	is not found or a pointer to the device. Must be called inside
 *	rcu_read_lock(), and result refcount is unchanged.
L
Linus Torvalds 已提交
823 824
 */

E
Eric Dumazet 已提交
825
struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
E
Eric Dumazet 已提交
826
				    unsigned short mask)
L
Linus Torvalds 已提交
827
{
828
	struct net_device *dev, *ret;
L
Linus Torvalds 已提交
829

830
	ret = NULL;
831
	for_each_netdev_rcu(net, dev) {
L
Linus Torvalds 已提交
832
		if (((dev->flags ^ if_flags) & mask) == 0) {
833
			ret = dev;
L
Linus Torvalds 已提交
834 835 836
			break;
		}
	}
837
	return ret;
L
Linus Torvalds 已提交
838
}
E
Eric Dumazet 已提交
839
EXPORT_SYMBOL(dev_get_by_flags_rcu);
L
Linus Torvalds 已提交
840 841 842 843 844 845

/**
 *	dev_valid_name - check if name is okay for network device
 *	@name: name string
 *
 *	Network device names need to be valid file names to
846 847
 *	to allow sysfs to work.  We also disallow any kind of
 *	whitespace.
L
Linus Torvalds 已提交
848
 */
849
bool dev_valid_name(const char *name)
L
Linus Torvalds 已提交
850
{
851
	if (*name == '\0')
852
		return false;
853
	if (strlen(name) >= IFNAMSIZ)
854
		return false;
855
	if (!strcmp(name, ".") || !strcmp(name, ".."))
856
		return false;
857 858 859

	while (*name) {
		if (*name == '/' || isspace(*name))
860
			return false;
861 862
		name++;
	}
863
	return true;
L
Linus Torvalds 已提交
864
}
E
Eric Dumazet 已提交
865
EXPORT_SYMBOL(dev_valid_name);
L
Linus Torvalds 已提交
866 867

/**
868 869
 *	__dev_alloc_name - allocate a name for a device
 *	@net: network namespace to allocate the device name in
L
Linus Torvalds 已提交
870
 *	@name: name format string
871
 *	@buf:  scratch buffer and result name string
L
Linus Torvalds 已提交
872 873
 *
 *	Passed a format string - eg "lt%d" it will try and find a suitable
S
Stephen Hemminger 已提交
874 875 876 877 878 879
 *	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
880 881
 */

882
static int __dev_alloc_name(struct net *net, const char *name, char *buf)
L
Linus Torvalds 已提交
883 884 885 886
{
	int i = 0;
	const char *p;
	const int max_netdevices = 8*PAGE_SIZE;
S
Stephen Hemminger 已提交
887
	unsigned long *inuse;
L
Linus Torvalds 已提交
888 889 890 891 892 893 894 895 896 897 898 899 900
	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
901
		inuse = (unsigned long *) get_zeroed_page(GFP_ATOMIC);
L
Linus Torvalds 已提交
902 903 904
		if (!inuse)
			return -ENOMEM;

905
		for_each_netdev(net, d) {
L
Linus Torvalds 已提交
906 907 908 909 910 911
			if (!sscanf(d->name, name, &i))
				continue;
			if (i < 0 || i >= max_netdevices)
				continue;

			/*  avoid cases where sscanf is not exact inverse of printf */
912
			snprintf(buf, IFNAMSIZ, name, i);
L
Linus Torvalds 已提交
913 914 915 916 917 918 919 920
			if (!strncmp(buf, d->name, IFNAMSIZ))
				set_bit(i, inuse);
		}

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

921 922
	if (buf != name)
		snprintf(buf, IFNAMSIZ, name, i);
923
	if (!__dev_get_by_name(net, buf))
L
Linus Torvalds 已提交
924 925 926 927 928 929 930 931 932
		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;
}

933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952
/**
 *	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;

953 954
	BUG_ON(!dev_net(dev));
	net = dev_net(dev);
955 956 957 958 959
	ret = __dev_alloc_name(net, name, buf);
	if (ret >= 0)
		strlcpy(dev->name, buf, IFNAMSIZ);
	return ret;
}
E
Eric Dumazet 已提交
960
EXPORT_SYMBOL(dev_alloc_name);
961

962 963 964
static int dev_alloc_name_ns(struct net *net,
			     struct net_device *dev,
			     const char *name)
965
{
966 967
	char buf[IFNAMSIZ];
	int ret;
968

969 970 971 972 973 974 975 976 977 978 979
	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);
980

981 982 983
	if (!dev_valid_name(name))
		return -EINVAL;

984
	if (strchr(name, '%'))
985
		return dev_alloc_name_ns(net, dev, name);
986 987
	else if (__dev_get_by_name(net, name))
		return -EEXIST;
988 989
	else if (dev->name != name)
		strlcpy(dev->name, name, IFNAMSIZ);
990 991 992

	return 0;
}
L
Linus Torvalds 已提交
993 994 995 996 997 998 999 1000 1001

/**
 *	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.
 */
1002
int dev_change_name(struct net_device *dev, const char *newname)
L
Linus Torvalds 已提交
1003
{
1004
	char oldname[IFNAMSIZ];
L
Linus Torvalds 已提交
1005
	int err = 0;
1006
	int ret;
1007
	struct net *net;
L
Linus Torvalds 已提交
1008 1009

	ASSERT_RTNL();
1010
	BUG_ON(!dev_net(dev));
L
Linus Torvalds 已提交
1011

1012
	net = dev_net(dev);
L
Linus Torvalds 已提交
1013 1014 1015
	if (dev->flags & IFF_UP)
		return -EBUSY;

1016 1017 1018
	if (strncmp(newname, dev->name, IFNAMSIZ) == 0)
		return 0;

1019 1020
	memcpy(oldname, dev->name, IFNAMSIZ);

1021
	err = dev_get_valid_name(net, dev, newname);
1022 1023
	if (err < 0)
		return err;
L
Linus Torvalds 已提交
1024

1025
rollback:
1026 1027 1028 1029
	ret = device_rename(&dev->dev, dev->name);
	if (ret) {
		memcpy(dev->name, oldname, IFNAMSIZ);
		return ret;
1030
	}
1031 1032

	write_lock_bh(&dev_base_lock);
1033
	hlist_del_rcu(&dev->name_hlist);
1034 1035 1036 1037 1038 1039
	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));
1040 1041
	write_unlock_bh(&dev_base_lock);

1042
	ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev);
1043 1044 1045
	ret = notifier_to_errno(ret);

	if (ret) {
1046 1047
		/* err >= 0 after dev_alloc_name() or stores the first errno */
		if (err >= 0) {
1048 1049 1050
			err = ret;
			memcpy(dev->name, oldname, IFNAMSIZ);
			goto rollback;
1051
		} else {
1052
			pr_err("%s: name change rollback failed: %d\n",
1053
			       dev->name, ret);
1054 1055
		}
	}
L
Linus Torvalds 已提交
1056 1057 1058 1059

	return err;
}

1060 1061 1062 1063
/**
 *	dev_set_alias - change ifalias of a device
 *	@dev: device
 *	@alias: name up to IFALIASZ
1064
 *	@len: limit of bytes to copy from info
1065 1066 1067 1068 1069
 *
 *	Set ifalias for a device,
 */
int dev_set_alias(struct net_device *dev, const char *alias, size_t len)
{
1070 1071
	char *new_ifalias;

1072 1073 1074 1075 1076
	ASSERT_RTNL();

	if (len >= IFALIASZ)
		return -EINVAL;

1077 1078 1079 1080 1081 1082 1083 1084
	if (!len) {
		if (dev->ifalias) {
			kfree(dev->ifalias);
			dev->ifalias = NULL;
		}
		return 0;
	}

1085 1086
	new_ifalias = krealloc(dev->ifalias, len + 1, GFP_KERNEL);
	if (!new_ifalias)
1087
		return -ENOMEM;
1088
	dev->ifalias = new_ifalias;
1089 1090 1091 1092 1093 1094

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


1095
/**
S
Stephen Hemminger 已提交
1096
 *	netdev_features_change - device changes features
1097 1098 1099 1100 1101 1102
 *	@dev: device to cause notification
 *
 *	Called to indicate a device has changed features.
 */
void netdev_features_change(struct net_device *dev)
{
1103
	call_netdevice_notifiers(NETDEV_FEAT_CHANGE, dev);
1104 1105 1106
}
EXPORT_SYMBOL(netdev_features_change);

L
Linus Torvalds 已提交
1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
/**
 *	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) {
1118
		call_netdevice_notifiers(NETDEV_CHANGE, dev);
L
Linus Torvalds 已提交
1119 1120 1121
		rtmsg_ifinfo(RTM_NEWLINK, dev, 0);
	}
}
E
Eric Dumazet 已提交
1122
EXPORT_SYMBOL(netdev_state_change);
L
Linus Torvalds 已提交
1123

1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
/**
 * 	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)
{
	rtnl_lock();
	call_netdevice_notifiers(NETDEV_NOTIFY_PEERS, dev);
	rtnl_unlock();
}
EXPORT_SYMBOL(netdev_notify_peers);

L
Linus Torvalds 已提交
1142 1143
/**
 *	dev_load 	- load a network module
1144
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
1145 1146 1147 1148 1149 1150 1151
 *	@name: name of interface
 *
 *	If a network interface is not present and the process has suitable
 *	privileges this function loads the module. If module loading is not
 *	available in this kernel then it becomes a nop.
 */

1152
void dev_load(struct net *net, const char *name)
L
Linus Torvalds 已提交
1153
{
1154
	struct net_device *dev;
1155
	int no_module;
L
Linus Torvalds 已提交
1156

1157 1158 1159
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
	rcu_read_unlock();
L
Linus Torvalds 已提交
1160

1161 1162 1163 1164 1165
	no_module = !dev;
	if (no_module && capable(CAP_NET_ADMIN))
		no_module = request_module("netdev-%s", name);
	if (no_module && capable(CAP_SYS_MODULE)) {
		if (!request_module("%s", name))
1166 1167
			pr_warn("Loading kernel module for a network device with CAP_SYS_MODULE (deprecated).  Use CAP_NET_ADMIN and alias netdev-%s instead.\n",
				name);
1168
	}
L
Linus Torvalds 已提交
1169
}
E
Eric Dumazet 已提交
1170
EXPORT_SYMBOL(dev_load);
L
Linus Torvalds 已提交
1171

1172
static int __dev_open(struct net_device *dev)
L
Linus Torvalds 已提交
1173
{
1174
	const struct net_device_ops *ops = dev->netdev_ops;
1175
	int ret;
L
Linus Torvalds 已提交
1176

1177 1178
	ASSERT_RTNL();

L
Linus Torvalds 已提交
1179 1180 1181
	if (!netif_device_present(dev))
		return -ENODEV;

1182 1183 1184 1185 1186
	ret = call_netdevice_notifiers(NETDEV_PRE_UP, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		return ret;

L
Linus Torvalds 已提交
1187
	set_bit(__LINK_STATE_START, &dev->state);
1188

1189 1190
	if (ops->ndo_validate_addr)
		ret = ops->ndo_validate_addr(dev);
1191

1192 1193
	if (!ret && ops->ndo_open)
		ret = ops->ndo_open(dev);
L
Linus Torvalds 已提交
1194

1195 1196 1197
	if (ret)
		clear_bit(__LINK_STATE_START, &dev->state);
	else {
L
Linus Torvalds 已提交
1198
		dev->flags |= IFF_UP;
1199
		net_dmaengine_get();
1200
		dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
1201
		dev_activate(dev);
1202
		add_device_randomness(dev->dev_addr, dev->addr_len);
L
Linus Torvalds 已提交
1203
	}
1204

L
Linus Torvalds 已提交
1205 1206 1207 1208
	return ret;
}

/**
1209 1210
 *	dev_open	- prepare an interface for use.
 *	@dev:	device to open
L
Linus Torvalds 已提交
1211
 *
1212 1213 1214 1215 1216 1217 1218
 *	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
1219
 */
1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
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;

	rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING);
	call_netdevice_notifiers(NETDEV_UP, dev);

	return ret;
}
EXPORT_SYMBOL(dev_open);

1238
static int __dev_close_many(struct list_head *head)
L
Linus Torvalds 已提交
1239
{
1240
	struct net_device *dev;
1241

1242
	ASSERT_RTNL();
1243 1244
	might_sleep();

1245 1246
	list_for_each_entry(dev, head, unreg_list) {
		call_netdevice_notifiers(NETDEV_GOING_DOWN, dev);
L
Linus Torvalds 已提交
1247

1248
		clear_bit(__LINK_STATE_START, &dev->state);
L
Linus Torvalds 已提交
1249

1250 1251 1252 1253 1254 1255 1256 1257
		/* 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.
		 */
		smp_mb__after_clear_bit(); /* Commit netif_running(). */
	}
L
Linus Torvalds 已提交
1258

1259
	dev_deactivate_many(head);
1260

1261 1262
	list_for_each_entry(dev, head, unreg_list) {
		const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
1263

1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282
		/*
		 *	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;
		net_dmaengine_put();
	}

	return 0;
}

static int __dev_close(struct net_device *dev)
{
1283
	int retval;
1284 1285 1286
	LIST_HEAD(single);

	list_add(&dev->unreg_list, &single);
1287 1288 1289
	retval = __dev_close_many(&single);
	list_del(&single);
	return retval;
1290 1291
}

E
Eric Dumazet 已提交
1292
static int dev_close_many(struct list_head *head)
1293 1294 1295
{
	struct net_device *dev, *tmp;
	LIST_HEAD(tmp_list);
L
Linus Torvalds 已提交
1296

1297 1298 1299 1300 1301
	list_for_each_entry_safe(dev, tmp, head, unreg_list)
		if (!(dev->flags & IFF_UP))
			list_move(&dev->unreg_list, &tmp_list);

	__dev_close_many(head);
L
Linus Torvalds 已提交
1302

1303 1304 1305 1306
	list_for_each_entry(dev, head, unreg_list) {
		rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING);
		call_netdevice_notifiers(NETDEV_DOWN, dev);
	}
1307

1308 1309
	/* rollback_registered_many needs the complete original list */
	list_splice(&tmp_list, head);
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323
	return 0;
}

/**
 *	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)
{
1324 1325
	if (dev->flags & IFF_UP) {
		LIST_HEAD(single);
L
Linus Torvalds 已提交
1326

1327 1328 1329 1330
		list_add(&dev->unreg_list, &single);
		dev_close_many(&single);
		list_del(&single);
	}
L
Linus Torvalds 已提交
1331 1332
	return 0;
}
E
Eric Dumazet 已提交
1333
EXPORT_SYMBOL(dev_close);
L
Linus Torvalds 已提交
1334 1335


1336 1337 1338 1339 1340 1341 1342 1343 1344 1345
/**
 *	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)
{
1346 1347 1348 1349 1350 1351 1352
	/*
	 * If we're trying to disable lro on a vlan device
	 * use the underlying physical device instead
	 */
	if (is_vlan_dev(dev))
		dev = vlan_dev_real_dev(dev);

M
Michał Mirosław 已提交
1353 1354
	dev->wanted_features &= ~NETIF_F_LRO;
	netdev_update_features(dev);
1355

1356 1357
	if (unlikely(dev->features & NETIF_F_LRO))
		netdev_WARN(dev, "failed to disable LRO!\n");
1358 1359 1360 1361
}
EXPORT_SYMBOL(dev_disable_lro);


1362 1363
static int dev_boot_phase = 1;

L
Linus Torvalds 已提交
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
/**
 *	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
1374
 *	to the new notifier to allow device to have a race free
L
Linus Torvalds 已提交
1375 1376 1377 1378 1379 1380
 *	view of the network device list.
 */

int register_netdevice_notifier(struct notifier_block *nb)
{
	struct net_device *dev;
1381
	struct net_device *last;
1382
	struct net *net;
L
Linus Torvalds 已提交
1383 1384 1385
	int err;

	rtnl_lock();
1386
	err = raw_notifier_chain_register(&netdev_chain, nb);
1387 1388
	if (err)
		goto unlock;
1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399
	if (dev_boot_phase)
		goto unlock;
	for_each_net(net) {
		for_each_netdev(net, dev) {
			err = nb->notifier_call(nb, NETDEV_REGISTER, dev);
			err = notifier_to_errno(err);
			if (err)
				goto rollback;

			if (!(dev->flags & IFF_UP))
				continue;
L
Linus Torvalds 已提交
1400

1401 1402
			nb->notifier_call(nb, NETDEV_UP, dev);
		}
L
Linus Torvalds 已提交
1403
	}
1404 1405

unlock:
L
Linus Torvalds 已提交
1406 1407
	rtnl_unlock();
	return err;
1408 1409 1410

rollback:
	last = dev;
1411 1412 1413
	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev == last)
1414
				goto outroll;
1415

1416 1417 1418 1419 1420
			if (dev->flags & IFF_UP) {
				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
				nb->notifier_call(nb, NETDEV_DOWN, dev);
			}
			nb->notifier_call(nb, NETDEV_UNREGISTER, dev);
1421 1422
		}
	}
1423

1424
outroll:
1425
	raw_notifier_chain_unregister(&netdev_chain, nb);
1426
	goto unlock;
L
Linus Torvalds 已提交
1427
}
E
Eric Dumazet 已提交
1428
EXPORT_SYMBOL(register_netdevice_notifier);
L
Linus Torvalds 已提交
1429 1430 1431 1432 1433 1434 1435 1436 1437

/**
 *	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.
1438 1439 1440 1441
 *
 * 	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
1442 1443 1444 1445
 */

int unregister_netdevice_notifier(struct notifier_block *nb)
{
1446 1447
	struct net_device *dev;
	struct net *net;
1448 1449 1450
	int err;

	rtnl_lock();
1451
	err = raw_notifier_chain_unregister(&netdev_chain, nb);
1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464
	if (err)
		goto unlock;

	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev->flags & IFF_UP) {
				nb->notifier_call(nb, NETDEV_GOING_DOWN, dev);
				nb->notifier_call(nb, NETDEV_DOWN, dev);
			}
			nb->notifier_call(nb, NETDEV_UNREGISTER, dev);
		}
	}
unlock:
1465 1466
	rtnl_unlock();
	return err;
L
Linus Torvalds 已提交
1467
}
E
Eric Dumazet 已提交
1468
EXPORT_SYMBOL(unregister_netdevice_notifier);
L
Linus Torvalds 已提交
1469 1470 1471 1472

/**
 *	call_netdevice_notifiers - call all network notifier blocks
 *      @val: value passed unmodified to notifier function
1473
 *      @dev: net_device pointer passed unmodified to notifier function
L
Linus Torvalds 已提交
1474 1475
 *
 *	Call all network notifier blocks.  Parameters and return value
1476
 *	are as for raw_notifier_call_chain().
L
Linus Torvalds 已提交
1477 1478
 */

1479
int call_netdevice_notifiers(unsigned long val, struct net_device *dev)
L
Linus Torvalds 已提交
1480
{
1481
	ASSERT_RTNL();
1482
	return raw_notifier_call_chain(&netdev_chain, val, dev);
L
Linus Torvalds 已提交
1483
}
1484
EXPORT_SYMBOL(call_netdevice_notifiers);
L
Linus Torvalds 已提交
1485

1486
static struct static_key netstamp_needed __read_mostly;
1487
#ifdef HAVE_JUMP_LABEL
1488
/* We are not allowed to call static_key_slow_dec() from irq context
1489
 * If net_disable_timestamp() is called from irq context, defer the
1490
 * static_key_slow_dec() calls.
1491 1492 1493
 */
static atomic_t netstamp_needed_deferred;
#endif
L
Linus Torvalds 已提交
1494 1495 1496

void net_enable_timestamp(void)
{
1497 1498 1499 1500 1501
#ifdef HAVE_JUMP_LABEL
	int deferred = atomic_xchg(&netstamp_needed_deferred, 0);

	if (deferred) {
		while (--deferred)
1502
			static_key_slow_dec(&netstamp_needed);
1503 1504 1505 1506
		return;
	}
#endif
	WARN_ON(in_interrupt());
1507
	static_key_slow_inc(&netstamp_needed);
L
Linus Torvalds 已提交
1508
}
E
Eric Dumazet 已提交
1509
EXPORT_SYMBOL(net_enable_timestamp);
L
Linus Torvalds 已提交
1510 1511 1512

void net_disable_timestamp(void)
{
1513 1514 1515 1516 1517 1518
#ifdef HAVE_JUMP_LABEL
	if (in_interrupt()) {
		atomic_inc(&netstamp_needed_deferred);
		return;
	}
#endif
1519
	static_key_slow_dec(&netstamp_needed);
L
Linus Torvalds 已提交
1520
}
E
Eric Dumazet 已提交
1521
EXPORT_SYMBOL(net_disable_timestamp);
L
Linus Torvalds 已提交
1522

E
Eric Dumazet 已提交
1523
static inline void net_timestamp_set(struct sk_buff *skb)
L
Linus Torvalds 已提交
1524
{
1525
	skb->tstamp.tv64 = 0;
1526
	if (static_key_false(&netstamp_needed))
1527
		__net_timestamp(skb);
L
Linus Torvalds 已提交
1528 1529
}

1530
#define net_timestamp_check(COND, SKB)			\
1531
	if (static_key_false(&netstamp_needed)) {		\
1532 1533 1534
		if ((COND) && !(SKB)->tstamp.tv64)	\
			__net_timestamp(SKB);		\
	}						\
E
Eric Dumazet 已提交
1535

1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
static int net_hwtstamp_validate(struct ifreq *ifr)
{
	struct hwtstamp_config cfg;
	enum hwtstamp_tx_types tx_type;
	enum hwtstamp_rx_filters rx_filter;
	int tx_type_valid = 0;
	int rx_filter_valid = 0;

	if (copy_from_user(&cfg, ifr->ifr_data, sizeof(cfg)))
		return -EFAULT;

	if (cfg.flags) /* reserved for future extensions */
		return -EINVAL;

	tx_type = cfg.tx_type;
	rx_filter = cfg.rx_filter;

	switch (tx_type) {
	case HWTSTAMP_TX_OFF:
	case HWTSTAMP_TX_ON:
	case HWTSTAMP_TX_ONESTEP_SYNC:
		tx_type_valid = 1;
		break;
	}

	switch (rx_filter) {
	case HWTSTAMP_FILTER_NONE:
	case HWTSTAMP_FILTER_ALL:
	case HWTSTAMP_FILTER_SOME:
	case HWTSTAMP_FILTER_PTP_V1_L4_EVENT:
	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
	case HWTSTAMP_FILTER_PTP_V2_L4_EVENT:
	case HWTSTAMP_FILTER_PTP_V2_L4_SYNC:
	case HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ:
	case HWTSTAMP_FILTER_PTP_V2_L2_EVENT:
	case HWTSTAMP_FILTER_PTP_V2_L2_SYNC:
	case HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ:
	case HWTSTAMP_FILTER_PTP_V2_EVENT:
	case HWTSTAMP_FILTER_PTP_V2_SYNC:
	case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
		rx_filter_valid = 1;
		break;
	}

	if (!tx_type_valid || !rx_filter_valid)
		return -ERANGE;

	return 0;
}

1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
static inline bool is_skb_forwardable(struct net_device *dev,
				      struct sk_buff *skb)
{
	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;
}

1608 1609 1610 1611 1612 1613 1614 1615
/**
 * dev_forward_skb - loopback an skb to another netif
 *
 * @dev: destination network device
 * @skb: buffer to forward
 *
 * return values:
 *	NET_RX_SUCCESS	(no congestion)
1616
 *	NET_RX_DROP     (packet was dropped, but freed)
1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627
 *
 * 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)
{
1628 1629 1630 1631 1632 1633 1634 1635
	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;
		}
	}

1636
	skb_orphan(skb);
1637
	nf_reset(skb);
1638

1639
	if (unlikely(!is_skb_forwardable(dev, skb))) {
1640
		atomic_long_inc(&dev->rx_dropped);
1641
		kfree_skb(skb);
1642
		return NET_RX_DROP;
1643
	}
1644
	skb->skb_iif = 0;
1645 1646
	skb->dev = dev;
	skb_dst_drop(skb);
1647 1648 1649
	skb->tstamp.tv64 = 0;
	skb->pkt_type = PACKET_HOST;
	skb->protocol = eth_type_trans(skb, dev);
1650 1651 1652
	skb->mark = 0;
	secpath_reset(skb);
	nf_reset(skb);
1653 1654 1655 1656
	return netif_rx(skb);
}
EXPORT_SYMBOL_GPL(dev_forward_skb);

1657 1658 1659 1660
static inline int deliver_skb(struct sk_buff *skb,
			      struct packet_type *pt_prev,
			      struct net_device *orig_dev)
{
1661 1662
	if (unlikely(skb_orphan_frags(skb, GFP_ATOMIC)))
		return -ENOMEM;
1663 1664 1665 1666
	atomic_inc(&skb->users);
	return pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
}

1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679
static inline bool skb_loop_sk(struct packet_type *ptype, struct sk_buff *skb)
{
	if (ptype->af_packet_priv == NULL)
		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
1680 1681 1682 1683 1684
/*
 *	Support routine. Sends outgoing frames to any network
 *	taps currently in use.
 */

1685
static void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
L
Linus Torvalds 已提交
1686 1687
{
	struct packet_type *ptype;
1688 1689
	struct sk_buff *skb2 = NULL;
	struct packet_type *pt_prev = NULL;
1690

L
Linus Torvalds 已提交
1691 1692 1693 1694 1695 1696
	rcu_read_lock();
	list_for_each_entry_rcu(ptype, &ptype_all, list) {
		/* Never send packets back to the socket
		 * they originated from - MvS (miquels@drinkel.ow.org)
		 */
		if ((ptype->dev == dev || !ptype->dev) &&
1697
		    (!skb_loop_sk(ptype, skb))) {
1698 1699 1700 1701 1702 1703 1704
			if (pt_prev) {
				deliver_skb(skb2, pt_prev, skb->dev);
				pt_prev = ptype;
				continue;
			}

			skb2 = skb_clone(skb, GFP_ATOMIC);
L
Linus Torvalds 已提交
1705 1706 1707
			if (!skb2)
				break;

1708 1709
			net_timestamp_set(skb2);

L
Linus Torvalds 已提交
1710 1711 1712 1713
			/* skb->nh should be correctly
			   set by sender, so that the second statement is
			   just protection against buggy protocols.
			 */
1714
			skb_reset_mac_header(skb2);
L
Linus Torvalds 已提交
1715

1716
			if (skb_network_header(skb2) < skb2->data ||
1717
			    skb2->network_header > skb2->tail) {
1718 1719 1720
				net_crit_ratelimited("protocol %04x is buggy, dev %s\n",
						     ntohs(skb2->protocol),
						     dev->name);
1721
				skb_reset_network_header(skb2);
L
Linus Torvalds 已提交
1722 1723
			}

1724
			skb2->transport_header = skb2->network_header;
L
Linus Torvalds 已提交
1725
			skb2->pkt_type = PACKET_OUTGOING;
1726
			pt_prev = ptype;
L
Linus Torvalds 已提交
1727 1728
		}
	}
1729 1730
	if (pt_prev)
		pt_prev->func(skb2, skb->dev, pt_prev, skb->dev);
L
Linus Torvalds 已提交
1731 1732 1733
	rcu_read_unlock();
}

1734 1735
/**
 * netif_setup_tc - Handle tc mappings on real_num_tx_queues change
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
 * @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
1747
static void netif_setup_tc(struct net_device *dev, unsigned int txq)
1748 1749 1750 1751 1752 1753
{
	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) {
1754
		pr_warn("Number of in use tx queues changed invalidating tc mappings. Priority traffic classification disabled!\n");
1755 1756 1757 1758 1759 1760 1761 1762 1763 1764
		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) {
1765 1766
			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);
1767 1768 1769 1770 1771
			netdev_set_prio_tc_map(dev, i, 0);
		}
	}
}

1772 1773 1774 1775
/*
 * 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.
 */
1776
int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq)
1777
{
T
Tom Herbert 已提交
1778 1779
	int rc;

1780 1781
	if (txq < 1 || txq > dev->num_tx_queues)
		return -EINVAL;
1782

1783 1784
	if (dev->reg_state == NETREG_REGISTERED ||
	    dev->reg_state == NETREG_UNREGISTERING) {
1785 1786
		ASSERT_RTNL();

T
Tom Herbert 已提交
1787 1788
		rc = netdev_queue_update_kobjects(dev, dev->real_num_tx_queues,
						  txq);
T
Tom Herbert 已提交
1789 1790 1791
		if (rc)
			return rc;

1792 1793 1794
		if (dev->num_tc)
			netif_setup_tc(dev, txq);

1795 1796
		if (txq < dev->real_num_tx_queues)
			qdisc_reset_all_tx_gt(dev, txq);
1797
	}
1798 1799 1800

	dev->real_num_tx_queues = txq;
	return 0;
1801 1802
}
EXPORT_SYMBOL(netif_set_real_num_tx_queues);
1803

1804 1805 1806 1807 1808 1809 1810 1811
#ifdef CONFIG_RPS
/**
 *	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
1812 1813
 *	negative error code.  If called before registration, it always
 *	succeeds.
1814 1815 1816 1817 1818
 */
int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
{
	int rc;

T
Tom Herbert 已提交
1819 1820 1821
	if (rxq < 1 || rxq > dev->num_rx_queues)
		return -EINVAL;

1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836
	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

1837 1838
/**
 * netif_get_num_default_rss_queues - default number of RSS queues
1839 1840 1841 1842
 *
 * This routine should set an upper limit on the number of RSS queues
 * used by default by multiqueue devices.
 */
1843
int netif_get_num_default_rss_queues(void)
1844 1845 1846 1847 1848
{
	return min_t(int, DEFAULT_MAX_NUM_RSS_QUEUES, num_online_cpus());
}
EXPORT_SYMBOL(netif_get_num_default_rss_queues);

1849
static inline void __netif_reschedule(struct Qdisc *q)
1850
{
1851 1852
	struct softnet_data *sd;
	unsigned long flags;
1853

1854 1855
	local_irq_save(flags);
	sd = &__get_cpu_var(softnet_data);
1856 1857 1858
	q->next_sched = NULL;
	*sd->output_queue_tailp = q;
	sd->output_queue_tailp = &q->next_sched;
1859 1860 1861 1862 1863 1864 1865 1866
	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);
1867 1868 1869
}
EXPORT_SYMBOL(__netif_schedule);

1870
void dev_kfree_skb_irq(struct sk_buff *skb)
1871
{
1872
	if (atomic_dec_and_test(&skb->users)) {
1873 1874
		struct softnet_data *sd;
		unsigned long flags;
1875

1876 1877 1878 1879 1880 1881 1882
		local_irq_save(flags);
		sd = &__get_cpu_var(softnet_data);
		skb->next = sd->completion_queue;
		sd->completion_queue = skb;
		raise_softirq_irqoff(NET_TX_SOFTIRQ);
		local_irq_restore(flags);
	}
1883
}
1884
EXPORT_SYMBOL(dev_kfree_skb_irq);
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895

void dev_kfree_skb_any(struct sk_buff *skb)
{
	if (in_irq() || irqs_disabled())
		dev_kfree_skb_irq(skb);
	else
		dev_kfree_skb(skb);
}
EXPORT_SYMBOL(dev_kfree_skb_any);


1896 1897 1898 1899 1900 1901
/**
 * netif_device_detach - mark device as removed
 * @dev: network device
 *
 * Mark device as removed from system and therefore no longer available.
 */
1902 1903 1904 1905
void netif_device_detach(struct net_device *dev)
{
	if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1906
		netif_tx_stop_all_queues(dev);
1907 1908 1909 1910
	}
}
EXPORT_SYMBOL(netif_device_detach);

1911 1912 1913 1914 1915 1916
/**
 * netif_device_attach - mark device as attached
 * @dev: network device
 *
 * Mark device as attached from system and restart if needed.
 */
1917 1918 1919 1920
void netif_device_attach(struct net_device *dev)
{
	if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1921
		netif_tx_wake_all_queues(dev);
1922
		__netdev_watchdog_up(dev);
1923 1924 1925 1926
	}
}
EXPORT_SYMBOL(netif_device_attach);

1927 1928
static void skb_warn_bad_offload(const struct sk_buff *skb)
{
1929
	static const netdev_features_t null_features = 0;
1930 1931 1932 1933 1934 1935 1936 1937
	struct net_device *dev = skb->dev;
	const char *driver = "";

	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",
1938 1939
	     driver, dev ? &dev->features : &null_features,
	     skb->sk ? &skb->sk->sk_route_caps : &null_features,
1940 1941 1942 1943
	     skb->len, skb->data_len, skb_shinfo(skb)->gso_size,
	     skb_shinfo(skb)->gso_type, skb->ip_summed);
}

L
Linus Torvalds 已提交
1944 1945 1946 1947
/*
 * Invalidate hardware checksum when packet is to be mangled, and
 * complete checksum manually on outgoing path.
 */
1948
int skb_checksum_help(struct sk_buff *skb)
L
Linus Torvalds 已提交
1949
{
1950
	__wsum csum;
1951
	int ret = 0, offset;
L
Linus Torvalds 已提交
1952

1953
	if (skb->ip_summed == CHECKSUM_COMPLETE)
1954 1955 1956
		goto out_set_summed;

	if (unlikely(skb_shinfo(skb)->gso_size)) {
1957 1958
		skb_warn_bad_offload(skb);
		return -EINVAL;
L
Linus Torvalds 已提交
1959 1960
	}

1961
	offset = skb_checksum_start_offset(skb);
1962 1963 1964 1965 1966 1967 1968 1969
	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
1970 1971 1972 1973 1974
		ret = pskb_expand_head(skb, 0, 0, GFP_ATOMIC);
		if (ret)
			goto out;
	}

1975
	*(__sum16 *)(skb->data + offset) = csum_fold(csum);
1976
out_set_summed:
L
Linus Torvalds 已提交
1977
	skb->ip_summed = CHECKSUM_NONE;
1978
out:
L
Linus Torvalds 已提交
1979 1980
	return ret;
}
E
Eric Dumazet 已提交
1981
EXPORT_SYMBOL(skb_checksum_help);
L
Linus Torvalds 已提交
1982

1983 1984 1985
/**
 *	skb_gso_segment - Perform segmentation on skb.
 *	@skb: buffer to segment
1986
 *	@features: features for the output path (see dev->features)
1987 1988
 *
 *	This function segments the given skb and returns a list of segments.
1989 1990 1991
 *
 *	It may return NULL if the skb requires no segmentation.  This is
 *	only possible when GSO is used for verifying header integrity.
1992
 */
1993 1994
struct sk_buff *skb_gso_segment(struct sk_buff *skb,
	netdev_features_t features)
1995 1996 1997
{
	struct sk_buff *segs = ERR_PTR(-EPROTONOSUPPORT);
	struct packet_type *ptype;
A
Al Viro 已提交
1998
	__be16 type = skb->protocol;
1999
	int vlan_depth = ETH_HLEN;
2000
	int err;
2001

2002 2003
	while (type == htons(ETH_P_8021Q)) {
		struct vlan_hdr *vh;
2004

2005
		if (unlikely(!pskb_may_pull(skb, vlan_depth + VLAN_HLEN)))
2006 2007
			return ERR_PTR(-EINVAL);

2008 2009 2010
		vh = (struct vlan_hdr *)(skb->data + vlan_depth);
		type = vh->h_vlan_encapsulated_proto;
		vlan_depth += VLAN_HLEN;
2011 2012
	}

2013
	skb_reset_mac_header(skb);
2014
	skb->mac_len = skb->network_header - skb->mac_header;
2015 2016
	__skb_pull(skb, skb->mac_len);

2017
	if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
2018
		skb_warn_bad_offload(skb);
2019

2020 2021 2022 2023 2024
		if (skb_header_cloned(skb) &&
		    (err = pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
			return ERR_PTR(err);
	}

2025
	rcu_read_lock();
2026 2027
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
2028
		if (ptype->type == type && !ptype->dev && ptype->gso_segment) {
2029
			if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
2030 2031 2032 2033
				err = ptype->gso_send_check(skb);
				segs = ERR_PTR(err);
				if (err || skb_gso_ok(skb, features))
					break;
2034 2035
				__skb_push(skb, (skb->data -
						 skb_network_header(skb)));
2036
			}
2037
			segs = ptype->gso_segment(skb, features);
2038 2039 2040 2041 2042
			break;
		}
	}
	rcu_read_unlock();

2043
	__skb_push(skb, skb->data - skb_mac_header(skb));
2044

2045 2046 2047 2048
	return segs;
}
EXPORT_SYMBOL(skb_gso_segment);

2049 2050 2051 2052 2053
/* Take action when hardware reception checksum errors are detected. */
#ifdef CONFIG_BUG
void netdev_rx_csum_fault(struct net_device *dev)
{
	if (net_ratelimit()) {
2054
		pr_err("%s: hw csum failure\n", dev ? dev->name : "<unknown>");
2055 2056 2057 2058 2059 2060
		dump_stack();
	}
}
EXPORT_SYMBOL(netdev_rx_csum_fault);
#endif

L
Linus Torvalds 已提交
2061 2062 2063 2064 2065
/* 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.
 */

E
Eric Dumazet 已提交
2066
static int illegal_highdma(struct net_device *dev, struct sk_buff *skb)
L
Linus Torvalds 已提交
2067
{
2068
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
2069
	int i;
2070
	if (!(dev->features & NETIF_F_HIGHDMA)) {
2071 2072 2073
		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)))
2074
				return 1;
2075
		}
2076
	}
L
Linus Torvalds 已提交
2077

2078 2079
	if (PCI_DMA_BUS_IS_PHYS) {
		struct device *pdev = dev->dev.parent;
L
Linus Torvalds 已提交
2080

E
Eric Dumazet 已提交
2081 2082
		if (!pdev)
			return 0;
2083
		for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
2084 2085
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			dma_addr_t addr = page_to_phys(skb_frag_page(frag));
2086 2087 2088 2089
			if (!pdev->dma_mask || addr + PAGE_SIZE - 1 > *pdev->dma_mask)
				return 1;
		}
	}
2090
#endif
L
Linus Torvalds 已提交
2091 2092 2093
	return 0;
}

2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119
struct dev_gso_cb {
	void (*destructor)(struct sk_buff *skb);
};

#define DEV_GSO_CB(skb) ((struct dev_gso_cb *)(skb)->cb)

static void dev_gso_skb_destructor(struct sk_buff *skb)
{
	struct dev_gso_cb *cb;

	do {
		struct sk_buff *nskb = skb->next;

		skb->next = nskb->next;
		nskb->next = NULL;
		kfree_skb(nskb);
	} while (skb->next);

	cb = DEV_GSO_CB(skb);
	if (cb->destructor)
		cb->destructor(skb);
}

/**
 *	dev_gso_segment - Perform emulated hardware segmentation on skb.
 *	@skb: buffer to segment
2120
 *	@features: device features as applicable to this skb
2121 2122 2123 2124
 *
 *	This function segments the given skb and stores the list of segments
 *	in skb->next.
 */
2125
static int dev_gso_segment(struct sk_buff *skb, netdev_features_t features)
2126 2127
{
	struct sk_buff *segs;
2128 2129 2130 2131 2132 2133

	segs = skb_gso_segment(skb, features);

	/* Verifying header integrity only. */
	if (!segs)
		return 0;
2134

2135
	if (IS_ERR(segs))
2136 2137 2138 2139 2140 2141 2142 2143 2144
		return PTR_ERR(segs);

	skb->next = segs;
	DEV_GSO_CB(skb)->destructor = skb->destructor;
	skb->destructor = dev_gso_skb_destructor;

	return 0;
}

2145
static bool can_checksum_protocol(netdev_features_t features, __be16 protocol)
2146 2147 2148 2149 2150 2151 2152 2153 2154 2155
{
	return ((features & NETIF_F_GEN_CSUM) ||
		((features & NETIF_F_V4_CSUM) &&
		 protocol == htons(ETH_P_IP)) ||
		((features & NETIF_F_V6_CSUM) &&
		 protocol == htons(ETH_P_IPV6)) ||
		((features & NETIF_F_FCOE_CRC) &&
		 protocol == htons(ETH_P_FCOE)));
}

2156 2157
static netdev_features_t harmonize_features(struct sk_buff *skb,
	__be16 protocol, netdev_features_t features)
2158
{
2159 2160
	if (skb->ip_summed != CHECKSUM_NONE &&
	    !can_checksum_protocol(features, protocol)) {
2161 2162 2163 2164 2165 2166 2167 2168 2169
		features &= ~NETIF_F_ALL_CSUM;
		features &= ~NETIF_F_SG;
	} else if (illegal_highdma(skb->dev, skb)) {
		features &= ~NETIF_F_SG;
	}

	return features;
}

2170
netdev_features_t netif_skb_features(struct sk_buff *skb)
2171 2172
{
	__be16 protocol = skb->protocol;
2173
	netdev_features_t features = skb->dev->features;
2174

2175 2176 2177
	if (skb_shinfo(skb)->gso_segs > skb->dev->gso_max_segs)
		features &= ~NETIF_F_GSO_MASK;

2178 2179 2180
	if (protocol == htons(ETH_P_8021Q)) {
		struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data;
		protocol = veh->h_vlan_encapsulated_proto;
2181 2182 2183
	} else if (!vlan_tx_tag_present(skb)) {
		return harmonize_features(skb, protocol, features);
	}
2184

2185
	features &= (skb->dev->vlan_features | NETIF_F_HW_VLAN_TX);
2186 2187 2188 2189 2190

	if (protocol != htons(ETH_P_8021Q)) {
		return harmonize_features(skb, protocol, features);
	} else {
		features &= NETIF_F_SG | NETIF_F_HIGHDMA | NETIF_F_FRAGLIST |
2191
				NETIF_F_GEN_CSUM | NETIF_F_HW_VLAN_TX;
2192 2193
		return harmonize_features(skb, protocol, features);
	}
2194
}
2195
EXPORT_SYMBOL(netif_skb_features);
2196

2197 2198 2199
/*
 * Returns true if either:
 *	1. skb has frag_list and the device doesn't support FRAGLIST, or
2200
 *	2. skb is fragmented and the device does not support SG.
2201 2202
 */
static inline int skb_needs_linearize(struct sk_buff *skb,
2203
				      int features)
2204
{
2205 2206 2207
	return skb_is_nonlinear(skb) &&
			((skb_has_frag_list(skb) &&
				!(features & NETIF_F_FRAGLIST)) ||
2208
			(skb_shinfo(skb)->nr_frags &&
2209
				!(features & NETIF_F_SG)));
2210 2211
}

2212 2213
int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
			struct netdev_queue *txq)
2214
{
2215
	const struct net_device_ops *ops = dev->netdev_ops;
2216
	int rc = NETDEV_TX_OK;
2217
	unsigned int skb_len;
2218

2219
	if (likely(!skb->next)) {
2220
		netdev_features_t features;
2221

2222
		/*
L
Lucas De Marchi 已提交
2223
		 * If device doesn't need skb->dst, release it right now while
2224 2225
		 * its hot in this cpu cache
		 */
E
Eric Dumazet 已提交
2226 2227 2228
		if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
			skb_dst_drop(skb);

2229 2230
		features = netif_skb_features(skb);

2231
		if (vlan_tx_tag_present(skb) &&
2232
		    !(features & NETIF_F_HW_VLAN_TX)) {
2233 2234 2235 2236 2237 2238 2239
			skb = __vlan_put_tag(skb, vlan_tx_tag_get(skb));
			if (unlikely(!skb))
				goto out;

			skb->vlan_tci = 0;
		}

2240
		if (netif_needs_gso(skb, features)) {
2241
			if (unlikely(dev_gso_segment(skb, features)))
2242 2243 2244
				goto out_kfree_skb;
			if (skb->next)
				goto gso;
2245
		} else {
2246
			if (skb_needs_linearize(skb, features) &&
2247 2248 2249 2250 2251 2252 2253 2254
			    __skb_linearize(skb))
				goto out_kfree_skb;

			/* If packet is not checksummed and device does not
			 * support checksumming for this protocol, complete
			 * checksumming here.
			 */
			if (skb->ip_summed == CHECKSUM_PARTIAL) {
2255 2256
				skb_set_transport_header(skb,
					skb_checksum_start_offset(skb));
2257
				if (!(features & NETIF_F_ALL_CSUM) &&
2258 2259 2260
				     skb_checksum_help(skb))
					goto out_kfree_skb;
			}
2261 2262
		}

2263 2264 2265
		if (!list_empty(&ptype_all))
			dev_queue_xmit_nit(skb, dev);

2266
		skb_len = skb->len;
2267
		rc = ops->ndo_start_xmit(skb, dev);
2268
		trace_net_dev_xmit(skb, rc, dev, skb_len);
2269
		if (rc == NETDEV_TX_OK)
E
Eric Dumazet 已提交
2270
			txq_trans_update(txq);
2271
		return rc;
2272 2273
	}

2274
gso:
2275 2276 2277 2278 2279
	do {
		struct sk_buff *nskb = skb->next;

		skb->next = nskb->next;
		nskb->next = NULL;
2280 2281

		/*
L
Lucas De Marchi 已提交
2282
		 * If device doesn't need nskb->dst, release it right now while
2283 2284 2285 2286 2287
		 * its hot in this cpu cache
		 */
		if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
			skb_dst_drop(nskb);

2288 2289 2290
		if (!list_empty(&ptype_all))
			dev_queue_xmit_nit(nskb, dev);

2291
		skb_len = nskb->len;
2292
		rc = ops->ndo_start_xmit(nskb, dev);
2293
		trace_net_dev_xmit(nskb, rc, dev, skb_len);
2294
		if (unlikely(rc != NETDEV_TX_OK)) {
2295 2296
			if (rc & ~NETDEV_TX_MASK)
				goto out_kfree_gso_skb;
2297
			nskb->next = skb->next;
2298 2299 2300
			skb->next = nskb;
			return rc;
		}
E
Eric Dumazet 已提交
2301
		txq_trans_update(txq);
2302
		if (unlikely(netif_xmit_stopped(txq) && skb->next))
2303
			return NETDEV_TX_BUSY;
2304
	} while (skb->next);
2305

2306 2307 2308
out_kfree_gso_skb:
	if (likely(skb->next == NULL))
		skb->destructor = DEV_GSO_CB(skb)->destructor;
2309 2310
out_kfree_skb:
	kfree_skb(skb);
2311
out:
2312
	return rc;
2313 2314
}

T
Tom Herbert 已提交
2315
static u32 hashrnd __read_mostly;
D
David S. Miller 已提交
2316

2317 2318 2319 2320 2321 2322
/*
 * 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, const struct sk_buff *skb,
		  unsigned int num_tx_queues)
2323
{
2324
	u32 hash;
2325 2326
	u16 qoffset = 0;
	u16 qcount = num_tx_queues;
D
David S. Miller 已提交
2327

2328 2329
	if (skb_rx_queue_recorded(skb)) {
		hash = skb_get_rx_queue(skb);
2330 2331
		while (unlikely(hash >= num_tx_queues))
			hash -= num_tx_queues;
2332 2333
		return hash;
	}
2334

2335 2336 2337 2338 2339 2340
	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;
	}

2341
	if (skb->sk && skb->sk->sk_hash)
2342
		hash = skb->sk->sk_hash;
2343
	else
E
Eric Dumazet 已提交
2344
		hash = (__force u16) skb->protocol;
T
Tom Herbert 已提交
2345
	hash = jhash_1word(hash, hashrnd);
D
David S. Miller 已提交
2346

2347
	return (u16) (((u64) hash * qcount) >> 32) + qoffset;
2348
}
2349
EXPORT_SYMBOL(__skb_tx_hash);
2350

2351 2352 2353
static inline u16 dev_cap_txqueue(struct net_device *dev, u16 queue_index)
{
	if (unlikely(queue_index >= dev->real_num_tx_queues)) {
2354 2355 2356
		net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
				     dev->name, queue_index,
				     dev->real_num_tx_queues);
2357 2358 2359 2360 2361
		return 0;
	}
	return queue_index;
}

T
Tom Herbert 已提交
2362 2363
static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
{
T
Tom Herbert 已提交
2364
#ifdef CONFIG_XPS
T
Tom Herbert 已提交
2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399
	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[raw_smp_processor_id()]);
		if (map) {
			if (map->len == 1)
				queue_index = map->queues[0];
			else {
				u32 hash;
				if (skb->sk && skb->sk->sk_hash)
					hash = skb->sk->sk_hash;
				else
					hash = (__force u16) skb->protocol ^
					    skb->rxhash;
				hash = jhash_1word(hash, hashrnd);
				queue_index = map->queues[
				    ((u64)hash * map->len) >> 32];
			}
			if (unlikely(queue_index >= dev->real_num_tx_queues))
				queue_index = -1;
		}
	}
	rcu_read_unlock();

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

2400 2401
struct netdev_queue *netdev_pick_tx(struct net_device *dev,
				    struct sk_buff *skb)
2402
{
2403
	int queue_index;
2404
	const struct net_device_ops *ops = dev->netdev_ops;
2405

2406 2407 2408
	if (dev->real_num_tx_queues == 1)
		queue_index = 0;
	else if (ops->ndo_select_queue) {
2409 2410 2411 2412 2413
		queue_index = ops->ndo_select_queue(dev, skb);
		queue_index = dev_cap_txqueue(dev, queue_index);
	} else {
		struct sock *sk = skb->sk;
		queue_index = sk_tx_queue_get(sk);
2414

2415 2416 2417
		if (queue_index < 0 || skb->ooo_okay ||
		    queue_index >= dev->real_num_tx_queues) {
			int old_index = queue_index;
2418

T
Tom Herbert 已提交
2419 2420 2421
			queue_index = get_xps_queue(dev, skb);
			if (queue_index < 0)
				queue_index = skb_tx_hash(dev, skb);
2422 2423 2424 2425

			if (queue_index != old_index && sk) {
				struct dst_entry *dst =
				    rcu_dereference_check(sk->sk_dst_cache, 1);
E
Eric Dumazet 已提交
2426 2427 2428 2429

				if (dst && skb_dst(skb) == dst)
					sk_tx_queue_set(sk, queue_index);
			}
2430 2431
		}
	}
2432

2433 2434
	skb_set_queue_mapping(skb, queue_index);
	return netdev_get_tx_queue(dev, queue_index);
2435 2436
}

2437 2438 2439 2440 2441
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
2442
	bool contended;
2443 2444
	int rc;

E
Eric Dumazet 已提交
2445 2446
	qdisc_skb_cb(skb)->pkt_len = skb->len;
	qdisc_calculate_pkt_len(skb, q);
2447 2448 2449 2450 2451 2452
	/*
	 * Heuristic to force contended enqueues to serialize on a
	 * separate lock before trying to get qdisc main lock.
	 * This permits __QDISC_STATE_RUNNING owner to get the lock more often
	 * and dequeue packets faster.
	 */
E
Eric Dumazet 已提交
2453
	contended = qdisc_is_running(q);
2454 2455 2456
	if (unlikely(contended))
		spin_lock(&q->busylock);

2457 2458 2459 2460 2461
	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) &&
2462
		   qdisc_run_begin(q)) {
2463 2464 2465 2466 2467
		/*
		 * 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.
		 */
E
Eric Dumazet 已提交
2468 2469
		if (!(dev->priv_flags & IFF_XMIT_DST_RELEASE))
			skb_dst_force(skb);
2470 2471 2472

		qdisc_bstats_update(q, skb);

2473 2474 2475 2476 2477
		if (sch_direct_xmit(skb, q, dev, txq, root_lock)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
2478
			__qdisc_run(q);
2479
		} else
2480
			qdisc_run_end(q);
2481 2482 2483

		rc = NET_XMIT_SUCCESS;
	} else {
E
Eric Dumazet 已提交
2484
		skb_dst_force(skb);
E
Eric Dumazet 已提交
2485
		rc = q->enqueue(skb, q) & NET_XMIT_MASK;
2486 2487 2488 2489 2490 2491 2492
		if (qdisc_run_begin(q)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
			__qdisc_run(q);
		}
2493 2494
	}
	spin_unlock(root_lock);
2495 2496
	if (unlikely(contended))
		spin_unlock(&q->busylock);
2497 2498 2499
	return rc;
}

2500 2501 2502
#if IS_ENABLED(CONFIG_NETPRIO_CGROUP)
static void skb_update_prio(struct sk_buff *skb)
{
I
Igor Maravic 已提交
2503
	struct netprio_map *map = rcu_dereference_bh(skb->dev->priomap);
2504

2505 2506 2507 2508 2509 2510
	if (!skb->priority && skb->sk && map) {
		unsigned int prioidx = skb->sk->sk_cgrp_prioidx;

		if (prioidx < map->priomap_len)
			skb->priority = map->priomap[prioidx];
	}
2511 2512 2513 2514 2515
}
#else
#define skb_update_prio(skb)
#endif

2516
static DEFINE_PER_CPU(int, xmit_recursion);
2517
#define RECURSION_LIMIT 10
2518

2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535
/**
 *	dev_loopback_xmit - loop back @skb
 *	@skb: buffer to transmit
 */
int dev_loopback_xmit(struct sk_buff *skb)
{
	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);

2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
/**
 *	dev_queue_xmit - transmit a buffer
 *	@skb: buffer to transmit
 *
 *	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
 */
L
Linus Torvalds 已提交
2561 2562 2563
int dev_queue_xmit(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
2564
	struct netdev_queue *txq;
L
Linus Torvalds 已提交
2565 2566 2567
	struct Qdisc *q;
	int rc = -ENOMEM;

2568 2569
	/* Disable soft irqs for various locks below. Also
	 * stops preemption for RCU.
L
Linus Torvalds 已提交
2570
	 */
2571
	rcu_read_lock_bh();
L
Linus Torvalds 已提交
2572

2573 2574
	skb_update_prio(skb);

2575
	txq = netdev_pick_tx(dev, skb);
2576
	q = rcu_dereference_bh(txq->qdisc);
2577

L
Linus Torvalds 已提交
2578
#ifdef CONFIG_NET_CLS_ACT
E
Eric Dumazet 已提交
2579
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_EGRESS);
L
Linus Torvalds 已提交
2580
#endif
2581
	trace_net_dev_queue(skb);
L
Linus Torvalds 已提交
2582
	if (q->enqueue) {
2583
		rc = __dev_xmit_skb(skb, q, dev, txq);
2584
		goto out;
L
Linus Torvalds 已提交
2585 2586 2587 2588 2589
	}

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

H
Herbert Xu 已提交
2590 2591
	   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
2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
	   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 */

2602
		if (txq->xmit_lock_owner != cpu) {
L
Linus Torvalds 已提交
2603

2604 2605 2606
			if (__this_cpu_read(xmit_recursion) > RECURSION_LIMIT)
				goto recursion_alert;

2607
			HARD_TX_LOCK(dev, txq, cpu);
L
Linus Torvalds 已提交
2608

2609
			if (!netif_xmit_stopped(txq)) {
2610
				__this_cpu_inc(xmit_recursion);
2611
				rc = dev_hard_start_xmit(skb, dev, txq);
2612
				__this_cpu_dec(xmit_recursion);
2613
				if (dev_xmit_complete(rc)) {
2614
					HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2615 2616 2617
					goto out;
				}
			}
2618
			HARD_TX_UNLOCK(dev, txq);
2619 2620
			net_crit_ratelimited("Virtual device %s asks to queue packet!\n",
					     dev->name);
L
Linus Torvalds 已提交
2621 2622
		} else {
			/* Recursion is detected! It is possible,
2623 2624 2625
			 * unfortunately
			 */
recursion_alert:
2626 2627
			net_crit_ratelimited("Dead loop on virtual device %s, fix it urgently!\n",
					     dev->name);
L
Linus Torvalds 已提交
2628 2629 2630 2631
		}
	}

	rc = -ENETDOWN;
2632
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2633 2634 2635 2636

	kfree_skb(skb);
	return rc;
out:
2637
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2638 2639
	return rc;
}
E
Eric Dumazet 已提交
2640
EXPORT_SYMBOL(dev_queue_xmit);
L
Linus Torvalds 已提交
2641 2642 2643 2644 2645 2646


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

2647
int netdev_max_backlog __read_mostly = 1000;
E
Eric Dumazet 已提交
2648
int netdev_tstamp_prequeue __read_mostly = 1;
2649 2650
int netdev_budget __read_mostly = 300;
int weight_p __read_mostly = 64;            /* old backlog weight */
L
Linus Torvalds 已提交
2651

E
Eric Dumazet 已提交
2652 2653 2654 2655 2656 2657 2658 2659
/* 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);
}

T
Tom Herbert 已提交
2660
/*
2661
 * __skb_get_rxhash: calculate a flow hash based on src/dst addresses
2662 2663 2664
 * and src/dst port numbers.  Sets rxhash in skb to non-zero hash value
 * on success, zero indicates no valid hash.  Also, sets l4_rxhash in skb
 * if hash is a canonical 4-tuple hash over transport ports.
T
Tom Herbert 已提交
2665
 */
2666
void __skb_get_rxhash(struct sk_buff *skb)
T
Tom Herbert 已提交
2667
{
2668 2669
	struct flow_keys keys;
	u32 hash;
2670

2671 2672
	if (!skb_flow_dissect(skb, &keys))
		return;
2673

2674
	if (keys.ports)
2675
		skb->l4_rxhash = 1;
T
Tom Herbert 已提交
2676

E
Eric Dumazet 已提交
2677
	/* get a consistent hash (same value on both flow directions) */
2678 2679 2680
	if (((__force u32)keys.dst < (__force u32)keys.src) ||
	    (((__force u32)keys.dst == (__force u32)keys.src) &&
	     ((__force u16)keys.port16[1] < (__force u16)keys.port16[0]))) {
2681
		swap(keys.dst, keys.src);
2682 2683
		swap(keys.port16[0], keys.port16[1]);
	}
T
Tom Herbert 已提交
2684

2685 2686 2687
	hash = jhash_3words((__force u32)keys.dst,
			    (__force u32)keys.src,
			    (__force u32)keys.ports, hashrnd);
2688 2689 2690
	if (!hash)
		hash = 1;

2691
	skb->rxhash = hash;
2692 2693 2694 2695 2696 2697
}
EXPORT_SYMBOL(__skb_get_rxhash);

#ifdef CONFIG_RPS

/* One global table that all flow-based protocols share. */
E
Eric Dumazet 已提交
2698
struct rps_sock_flow_table __rcu *rps_sock_flow_table __read_mostly;
2699 2700
EXPORT_SYMBOL(rps_sock_flow_table);

2701
struct static_key rps_needed __read_mostly;
2702

2703 2704 2705 2706
static struct rps_dev_flow *
set_rps_cpu(struct net_device *dev, struct sk_buff *skb,
	    struct rps_dev_flow *rflow, u16 next_cpu)
{
2707
	if (next_cpu != RPS_NO_CPU) {
2708 2709 2710 2711 2712 2713 2714 2715 2716
#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? */
2717 2718
		if (!skb_rx_queue_recorded(skb) || !dev->rx_cpu_rmap ||
		    !(dev->features & NETIF_F_NTUPLE))
2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740
			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;
		flow_id = skb->rxhash & flow_table->mask;
		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 =
2741
			per_cpu(softnet_data, next_cpu).input_queue_head;
2742 2743
	}

2744
	rflow->cpu = next_cpu;
2745 2746 2747
	return rflow;
}

2748 2749 2750 2751 2752 2753 2754 2755 2756
/*
 * 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)
{
	struct netdev_rx_queue *rxqueue;
E
Eric Dumazet 已提交
2757
	struct rps_map *map;
2758 2759 2760 2761 2762 2763 2764
	struct rps_dev_flow_table *flow_table;
	struct rps_sock_flow_table *sock_flow_table;
	int cpu = -1;
	u16 tcpu;

	if (skb_rx_queue_recorded(skb)) {
		u16 index = skb_get_rx_queue(skb);
2765 2766 2767 2768 2769
		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);
2770 2771 2772 2773 2774 2775
			goto done;
		}
		rxqueue = dev->_rx + index;
	} else
		rxqueue = dev->_rx;

E
Eric Dumazet 已提交
2776 2777
	map = rcu_dereference(rxqueue->rps_map);
	if (map) {
2778
		if (map->len == 1 &&
2779
		    !rcu_access_pointer(rxqueue->rps_flow_table)) {
2780 2781 2782 2783 2784
			tcpu = map->cpus[0];
			if (cpu_online(tcpu))
				cpu = tcpu;
			goto done;
		}
2785
	} else if (!rcu_access_pointer(rxqueue->rps_flow_table)) {
2786
		goto done;
2787
	}
2788

2789
	skb_reset_network_header(skb);
2790 2791 2792
	if (!skb_get_rxhash(skb))
		goto done;

T
Tom Herbert 已提交
2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818
	flow_table = rcu_dereference(rxqueue->rps_flow_table);
	sock_flow_table = rcu_dereference(rps_sock_flow_table);
	if (flow_table && sock_flow_table) {
		u16 next_cpu;
		struct rps_dev_flow *rflow;

		rflow = &flow_table->flows[skb->rxhash & flow_table->mask];
		tcpu = rflow->cpu;

		next_cpu = sock_flow_table->ents[skb->rxhash &
		    sock_flow_table->mask];

		/*
		 * 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:
		 *   - Current CPU is unset (equal to RPS_NO_CPU).
		 *   - 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) &&
		    (tcpu == RPS_NO_CPU || !cpu_online(tcpu) ||
		     ((int)(per_cpu(softnet_data, tcpu).input_queue_head -
2819 2820 2821
		      rflow->last_qtail)) >= 0))
			rflow = set_rps_cpu(dev, skb, rflow, next_cpu);

T
Tom Herbert 已提交
2822 2823 2824 2825 2826 2827 2828
		if (tcpu != RPS_NO_CPU && cpu_online(tcpu)) {
			*rflowp = rflow;
			cpu = tcpu;
			goto done;
		}
	}

T
Tom Herbert 已提交
2829
	if (map) {
T
Tom Herbert 已提交
2830
		tcpu = map->cpus[((u64) skb->rxhash * map->len) >> 32];
T
Tom Herbert 已提交
2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841

		if (cpu_online(tcpu)) {
			cpu = tcpu;
			goto done;
		}
	}

done:
	return cpu;
}

2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881
#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;
	int cpu;

	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);
		if (rflow->filter == filter_id && cpu != RPS_NO_CPU &&
		    ((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
2882
/* Called from hardirq (IPI) context */
E
Eric Dumazet 已提交
2883
static void rps_trigger_softirq(void *data)
T
Tom Herbert 已提交
2884
{
E
Eric Dumazet 已提交
2885 2886
	struct softnet_data *sd = data;

E
Eric Dumazet 已提交
2887
	____napi_schedule(sd, &sd->backlog);
C
Changli Gao 已提交
2888
	sd->received_rps++;
T
Tom Herbert 已提交
2889
}
E
Eric Dumazet 已提交
2890

T
Tom Herbert 已提交
2891
#endif /* CONFIG_RPS */
T
Tom Herbert 已提交
2892

E
Eric Dumazet 已提交
2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913
/*
 * 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
	struct softnet_data *mysd = &__get_cpu_var(softnet_data);

	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;
}

T
Tom Herbert 已提交
2914 2915 2916 2917
/*
 * 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
2918 2919
static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
			      unsigned int *qtail)
T
Tom Herbert 已提交
2920
{
E
Eric Dumazet 已提交
2921
	struct softnet_data *sd;
T
Tom Herbert 已提交
2922 2923
	unsigned long flags;

E
Eric Dumazet 已提交
2924
	sd = &per_cpu(softnet_data, cpu);
T
Tom Herbert 已提交
2925 2926 2927

	local_irq_save(flags);

E
Eric Dumazet 已提交
2928
	rps_lock(sd);
2929 2930
	if (skb_queue_len(&sd->input_pkt_queue) <= netdev_max_backlog) {
		if (skb_queue_len(&sd->input_pkt_queue)) {
T
Tom Herbert 已提交
2931
enqueue:
E
Eric Dumazet 已提交
2932
			__skb_queue_tail(&sd->input_pkt_queue, skb);
2933
			input_queue_tail_incr_save(sd, qtail);
E
Eric Dumazet 已提交
2934
			rps_unlock(sd);
2935
			local_irq_restore(flags);
T
Tom Herbert 已提交
2936 2937 2938
			return NET_RX_SUCCESS;
		}

2939 2940 2941 2942
		/* 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
2943
			if (!rps_ipi_queued(sd))
E
Eric Dumazet 已提交
2944
				____napi_schedule(sd, &sd->backlog);
T
Tom Herbert 已提交
2945 2946 2947 2948
		}
		goto enqueue;
	}

C
Changli Gao 已提交
2949
	sd->dropped++;
E
Eric Dumazet 已提交
2950
	rps_unlock(sd);
T
Tom Herbert 已提交
2951 2952 2953

	local_irq_restore(flags);

2954
	atomic_long_inc(&skb->dev->rx_dropped);
T
Tom Herbert 已提交
2955 2956 2957
	kfree_skb(skb);
	return NET_RX_DROP;
}
L
Linus Torvalds 已提交
2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975

/**
 *	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)
{
2976
	int ret;
L
Linus Torvalds 已提交
2977 2978 2979 2980 2981

	/* if netpoll wants it, pretend we never saw it */
	if (netpoll_rx(skb))
		return NET_RX_DROP;

2982
	net_timestamp_check(netdev_tstamp_prequeue, skb);
L
Linus Torvalds 已提交
2983

2984
	trace_netif_rx(skb);
E
Eric Dumazet 已提交
2985
#ifdef CONFIG_RPS
2986
	if (static_key_false(&rps_needed)) {
T
Tom Herbert 已提交
2987
		struct rps_dev_flow voidflow, *rflow = &voidflow;
2988 2989
		int cpu;

2990
		preempt_disable();
2991
		rcu_read_lock();
T
Tom Herbert 已提交
2992 2993

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
2994 2995
		if (cpu < 0)
			cpu = smp_processor_id();
T
Tom Herbert 已提交
2996 2997 2998

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

2999
		rcu_read_unlock();
3000
		preempt_enable();
3001 3002
	} else
#endif
T
Tom Herbert 已提交
3003 3004 3005 3006 3007
	{
		unsigned int qtail;
		ret = enqueue_to_backlog(skb, get_cpu(), &qtail);
		put_cpu();
	}
3008
	return ret;
L
Linus Torvalds 已提交
3009
}
E
Eric Dumazet 已提交
3010
EXPORT_SYMBOL(netif_rx);
L
Linus Torvalds 已提交
3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041

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

	preempt_disable();
	err = netif_rx(skb);
	if (local_softirq_pending())
		do_softirq();
	preempt_enable();

	return err;
}
EXPORT_SYMBOL(netif_rx_ni);

static void net_tx_action(struct softirq_action *h)
{
	struct softnet_data *sd = &__get_cpu_var(softnet_data);

	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;

3042
			WARN_ON(atomic_read(&skb->users));
3043
			trace_kfree_skb(skb, net_tx_action);
L
Linus Torvalds 已提交
3044 3045 3046 3047 3048
			__kfree_skb(skb);
		}
	}

	if (sd->output_queue) {
3049
		struct Qdisc *head;
L
Linus Torvalds 已提交
3050 3051 3052 3053

		local_irq_disable();
		head = sd->output_queue;
		sd->output_queue = NULL;
3054
		sd->output_queue_tailp = &sd->output_queue;
L
Linus Torvalds 已提交
3055 3056 3057
		local_irq_enable();

		while (head) {
3058 3059 3060
			struct Qdisc *q = head;
			spinlock_t *root_lock;

L
Linus Torvalds 已提交
3061 3062
			head = head->next_sched;

3063
			root_lock = qdisc_lock(q);
3064
			if (spin_trylock(root_lock)) {
3065 3066 3067
				smp_mb__before_clear_bit();
				clear_bit(__QDISC_STATE_SCHED,
					  &q->state);
3068 3069
				qdisc_run(q);
				spin_unlock(root_lock);
L
Linus Torvalds 已提交
3070
			} else {
3071
				if (!test_bit(__QDISC_STATE_DEACTIVATED,
3072
					      &q->state)) {
3073
					__netif_reschedule(q);
3074 3075 3076 3077 3078
				} else {
					smp_mb__before_clear_bit();
					clear_bit(__QDISC_STATE_SCHED,
						  &q->state);
				}
L
Linus Torvalds 已提交
3079 3080 3081 3082 3083
			}
		}
	}
}

3084 3085
#if (defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE)) && \
    (defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE))
3086 3087 3088
/* This hook is defined here for ATM LANE */
int (*br_fdb_test_addr_hook)(struct net_device *dev,
			     unsigned char *addr) __read_mostly;
3089
EXPORT_SYMBOL_GPL(br_fdb_test_addr_hook);
3090
#endif
L
Linus Torvalds 已提交
3091 3092 3093 3094 3095 3096

#ifdef CONFIG_NET_CLS_ACT
/* TODO: Maybe we should just force sch_ingress to be compiled in
 * when CONFIG_NET_CLS_ACT is? otherwise some useless instructions
 * a compare and 2 stores extra right now if we dont have it on
 * but have CONFIG_NET_CLS_ACT
L
Lucas De Marchi 已提交
3097 3098
 * NOTE: This doesn't stop any functionality; if you dont have
 * the ingress scheduler, you just can't add policies on ingress.
L
Linus Torvalds 已提交
3099 3100
 *
 */
3101
static int ing_filter(struct sk_buff *skb, struct netdev_queue *rxq)
L
Linus Torvalds 已提交
3102 3103
{
	struct net_device *dev = skb->dev;
3104
	u32 ttl = G_TC_RTTL(skb->tc_verd);
3105 3106
	int result = TC_ACT_OK;
	struct Qdisc *q;
3107

3108
	if (unlikely(MAX_RED_LOOP < ttl++)) {
3109 3110
		net_warn_ratelimited("Redir loop detected Dropping packet (%d->%d)\n",
				     skb->skb_iif, dev->ifindex);
3111 3112
		return TC_ACT_SHOT;
	}
L
Linus Torvalds 已提交
3113

3114 3115
	skb->tc_verd = SET_TC_RTTL(skb->tc_verd, ttl);
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_INGRESS);
L
Linus Torvalds 已提交
3116

3117
	q = rxq->qdisc;
3118
	if (q != &noop_qdisc) {
3119
		spin_lock(qdisc_lock(q));
3120 3121
		if (likely(!test_bit(__QDISC_STATE_DEACTIVATED, &q->state)))
			result = qdisc_enqueue_root(skb, q);
3122 3123
		spin_unlock(qdisc_lock(q));
	}
3124 3125 3126

	return result;
}
3127

3128 3129 3130 3131
static inline struct sk_buff *handle_ing(struct sk_buff *skb,
					 struct packet_type **pt_prev,
					 int *ret, struct net_device *orig_dev)
{
3132 3133 3134
	struct netdev_queue *rxq = rcu_dereference(skb->dev->ingress_queue);

	if (!rxq || rxq->qdisc == &noop_qdisc)
3135
		goto out;
L
Linus Torvalds 已提交
3136

3137 3138 3139
	if (*pt_prev) {
		*ret = deliver_skb(skb, *pt_prev, orig_dev);
		*pt_prev = NULL;
L
Linus Torvalds 已提交
3140 3141
	}

3142
	switch (ing_filter(skb, rxq)) {
3143 3144 3145 3146 3147 3148 3149 3150 3151
	case TC_ACT_SHOT:
	case TC_ACT_STOLEN:
		kfree_skb(skb);
		return NULL;
	}

out:
	skb->tc_verd = 0;
	return skb;
L
Linus Torvalds 已提交
3152 3153 3154
}
#endif

3155 3156 3157 3158
/**
 *	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
3159
 *	@rx_handler_data: data pointer that is used by rx handler
3160 3161 3162 3163 3164 3165
 *
 *	Register a receive hander for a device. This handler will then be
 *	called from __netif_receive_skb. A negative errno code is returned
 *	on a failure.
 *
 *	The caller must hold the rtnl_mutex.
3166 3167
 *
 *	For a general description of rx_handler, see enum rx_handler_result.
3168 3169
 */
int netdev_rx_handler_register(struct net_device *dev,
J
Jiri Pirko 已提交
3170 3171
			       rx_handler_func_t *rx_handler,
			       void *rx_handler_data)
3172 3173 3174 3175 3176 3177
{
	ASSERT_RTNL();

	if (dev->rx_handler)
		return -EBUSY;

J
Jiri Pirko 已提交
3178
	rcu_assign_pointer(dev->rx_handler_data, rx_handler_data);
3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196
	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
 *
 *	Unregister a receive hander from a device.
 *
 *	The caller must hold the rtnl_mutex.
 */
void netdev_rx_handler_unregister(struct net_device *dev)
{

	ASSERT_RTNL();
3197 3198
	RCU_INIT_POINTER(dev->rx_handler, NULL);
	RCU_INIT_POINTER(dev->rx_handler_data, NULL);
3199 3200 3201
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);

3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218
/*
 * 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) {
	case __constant_htons(ETH_P_ARP):
	case __constant_htons(ETH_P_IP):
	case __constant_htons(ETH_P_IPV6):
	case __constant_htons(ETH_P_8021Q):
		return true;
	default:
		return false;
	}
}

3219
static int __netif_receive_skb(struct sk_buff *skb)
L
Linus Torvalds 已提交
3220 3221
{
	struct packet_type *ptype, *pt_prev;
3222
	rx_handler_func_t *rx_handler;
D
David S. Miller 已提交
3223
	struct net_device *orig_dev;
3224
	struct net_device *null_or_dev;
3225
	bool deliver_exact = false;
L
Linus Torvalds 已提交
3226
	int ret = NET_RX_DROP;
A
Al Viro 已提交
3227
	__be16 type;
3228
	unsigned long pflags = current->flags;
L
Linus Torvalds 已提交
3229

3230
	net_timestamp_check(!netdev_tstamp_prequeue, skb);
3231

3232
	trace_netif_receive_skb(skb);
3233

3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245
	/*
	 * 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.
	 */
	if (sk_memalloc_socks() && skb_pfmemalloc(skb))
		current->flags |= PF_MEMALLOC;

L
Linus Torvalds 已提交
3246
	/* if we've gotten here through NAPI, check netpoll */
3247
	if (netpoll_receive_skb(skb))
3248
		goto out;
L
Linus Torvalds 已提交
3249

J
Joe Eykholt 已提交
3250
	orig_dev = skb->dev;
3251

3252
	skb_reset_network_header(skb);
3253
	skb_reset_transport_header(skb);
3254
	skb_reset_mac_len(skb);
L
Linus Torvalds 已提交
3255 3256 3257 3258 3259

	pt_prev = NULL;

	rcu_read_lock();

3260
another_round:
3261
	skb->skb_iif = skb->dev->ifindex;
3262 3263 3264

	__this_cpu_inc(softnet_data.processed);

3265 3266 3267
	if (skb->protocol == cpu_to_be16(ETH_P_8021Q)) {
		skb = vlan_untag(skb);
		if (unlikely(!skb))
3268
			goto unlock;
3269 3270
	}

L
Linus Torvalds 已提交
3271 3272 3273 3274 3275 3276 3277
#ifdef CONFIG_NET_CLS_ACT
	if (skb->tc_verd & TC_NCLS) {
		skb->tc_verd = CLR_TC_NCLS(skb->tc_verd);
		goto ncls;
	}
#endif

3278 3279 3280
	if (sk_memalloc_socks() && skb_pfmemalloc(skb))
		goto skip_taps;

L
Linus Torvalds 已提交
3281
	list_for_each_entry_rcu(ptype, &ptype_all, list) {
3282
		if (!ptype->dev || ptype->dev == skb->dev) {
3283
			if (pt_prev)
D
David S. Miller 已提交
3284
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3285 3286 3287 3288
			pt_prev = ptype;
		}
	}

3289
skip_taps:
L
Linus Torvalds 已提交
3290
#ifdef CONFIG_NET_CLS_ACT
3291 3292
	skb = handle_ing(skb, &pt_prev, &ret, orig_dev);
	if (!skb)
3293
		goto unlock;
L
Linus Torvalds 已提交
3294 3295 3296
ncls:
#endif

3297 3298 3299 3300
	if (sk_memalloc_socks() && skb_pfmemalloc(skb)
				&& !skb_pfmemalloc_protocol(skb))
		goto drop;

3301
	rx_handler = rcu_dereference(skb->dev->rx_handler);
3302 3303 3304 3305 3306
	if (vlan_tx_tag_present(skb)) {
		if (pt_prev) {
			ret = deliver_skb(skb, pt_prev, orig_dev);
			pt_prev = NULL;
		}
3307
		if (vlan_do_receive(&skb, !rx_handler))
3308 3309
			goto another_round;
		else if (unlikely(!skb))
3310
			goto unlock;
3311 3312
	}

3313 3314 3315 3316 3317
	if (rx_handler) {
		if (pt_prev) {
			ret = deliver_skb(skb, pt_prev, orig_dev);
			pt_prev = NULL;
		}
3318 3319
		switch (rx_handler(&skb)) {
		case RX_HANDLER_CONSUMED:
3320
			goto unlock;
3321
		case RX_HANDLER_ANOTHER:
3322
			goto another_round;
3323 3324 3325 3326 3327 3328 3329
		case RX_HANDLER_EXACT:
			deliver_exact = true;
		case RX_HANDLER_PASS:
			break;
		default:
			BUG();
		}
3330
	}
L
Linus Torvalds 已提交
3331

3332
	/* deliver only exact match when indicated */
3333
	null_or_dev = deliver_exact ? skb->dev : NULL;
3334

L
Linus Torvalds 已提交
3335
	type = skb->protocol;
3336 3337
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
3338
		if (ptype->type == type &&
3339 3340
		    (ptype->dev == null_or_dev || ptype->dev == skb->dev ||
		     ptype->dev == orig_dev)) {
3341
			if (pt_prev)
D
David S. Miller 已提交
3342
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3343 3344 3345 3346 3347
			pt_prev = ptype;
		}
	}

	if (pt_prev) {
3348
		if (unlikely(skb_orphan_frags(skb, GFP_ATOMIC)))
3349
			goto drop;
3350 3351
		else
			ret = pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3352
	} else {
3353
drop:
3354
		atomic_long_inc(&skb->dev->rx_dropped);
L
Linus Torvalds 已提交
3355 3356 3357 3358 3359 3360 3361
		kfree_skb(skb);
		/* Jamal, now you will not able to escape explaining
		 * me how you were going to use this. :-)
		 */
		ret = NET_RX_DROP;
	}

3362
unlock:
L
Linus Torvalds 已提交
3363
	rcu_read_unlock();
3364 3365
out:
	tsk_restore_flags(current, pflags, PF_MEMALLOC);
L
Linus Torvalds 已提交
3366 3367
	return ret;
}
T
Tom Herbert 已提交
3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385

/**
 *	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
 */
int netif_receive_skb(struct sk_buff *skb)
{
3386
	net_timestamp_check(netdev_tstamp_prequeue, skb);
E
Eric Dumazet 已提交
3387

3388 3389 3390
	if (skb_defer_rx_timestamp(skb))
		return NET_RX_SUCCESS;

E
Eric Dumazet 已提交
3391
#ifdef CONFIG_RPS
3392
	if (static_key_false(&rps_needed)) {
E
Eric Dumazet 已提交
3393 3394
		struct rps_dev_flow voidflow, *rflow = &voidflow;
		int cpu, ret;
T
Tom Herbert 已提交
3395

E
Eric Dumazet 已提交
3396 3397 3398
		rcu_read_lock();

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
T
Tom Herbert 已提交
3399

E
Eric Dumazet 已提交
3400 3401 3402
		if (cpu >= 0) {
			ret = enqueue_to_backlog(skb, cpu, &rflow->last_qtail);
			rcu_read_unlock();
3403
			return ret;
E
Eric Dumazet 已提交
3404
		}
3405
		rcu_read_unlock();
T
Tom Herbert 已提交
3406
	}
3407
#endif
3408
	return __netif_receive_skb(skb);
T
Tom Herbert 已提交
3409
}
E
Eric Dumazet 已提交
3410
EXPORT_SYMBOL(netif_receive_skb);
L
Linus Torvalds 已提交
3411

E
Eric Dumazet 已提交
3412 3413 3414
/* Network device is going away, flush any packets still pending
 * Called with irqs disabled.
 */
3415
static void flush_backlog(void *arg)
3416
{
3417
	struct net_device *dev = arg;
E
Eric Dumazet 已提交
3418
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3419 3420
	struct sk_buff *skb, *tmp;

E
Eric Dumazet 已提交
3421
	rps_lock(sd);
3422
	skb_queue_walk_safe(&sd->input_pkt_queue, skb, tmp) {
3423
		if (skb->dev == dev) {
E
Eric Dumazet 已提交
3424
			__skb_unlink(skb, &sd->input_pkt_queue);
3425
			kfree_skb(skb);
3426
			input_queue_head_incr(sd);
3427
		}
3428
	}
E
Eric Dumazet 已提交
3429
	rps_unlock(sd);
3430 3431 3432 3433 3434

	skb_queue_walk_safe(&sd->process_queue, skb, tmp) {
		if (skb->dev == dev) {
			__skb_unlink(skb, &sd->process_queue);
			kfree_skb(skb);
3435
			input_queue_head_incr(sd);
3436 3437
		}
	}
3438 3439
}

3440 3441 3442 3443 3444 3445 3446
static int napi_gro_complete(struct sk_buff *skb)
{
	struct packet_type *ptype;
	__be16 type = skb->protocol;
	struct list_head *head = &ptype_base[ntohs(type) & PTYPE_HASH_MASK];
	int err = -ENOENT;

3447 3448
	if (NAPI_GRO_CB(skb)->count == 1) {
		skb_shinfo(skb)->gso_size = 0;
3449
		goto out;
3450
	}
3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471

	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
		if (ptype->type != type || ptype->dev || !ptype->gro_complete)
			continue;

		err = ptype->gro_complete(skb);
		break;
	}
	rcu_read_unlock();

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

out:
	return netif_receive_skb(skb);
}

E
Eric Dumazet 已提交
3472
inline void napi_gro_flush(struct napi_struct *napi)
3473 3474 3475 3476 3477 3478 3479 3480 3481
{
	struct sk_buff *skb, *next;

	for (skb = napi->gro_list; skb; skb = next) {
		next = skb->next;
		skb->next = NULL;
		napi_gro_complete(skb);
	}

3482
	napi->gro_count = 0;
3483 3484
	napi->gro_list = NULL;
}
E
Eric Dumazet 已提交
3485
EXPORT_SYMBOL(napi_gro_flush);
3486

3487
enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3488 3489 3490 3491 3492
{
	struct sk_buff **pp = NULL;
	struct packet_type *ptype;
	__be16 type = skb->protocol;
	struct list_head *head = &ptype_base[ntohs(type) & PTYPE_HASH_MASK];
H
Herbert Xu 已提交
3493
	int same_flow;
3494
	int mac_len;
3495
	enum gro_result ret;
3496

3497
	if (!(skb->dev->features & NETIF_F_GRO) || netpoll_rx_on(skb))
3498 3499
		goto normal;

3500
	if (skb_is_gso(skb) || skb_has_frag_list(skb))
3501 3502
		goto normal;

3503 3504 3505 3506 3507
	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
		if (ptype->type != type || ptype->dev || !ptype->gro_receive)
			continue;

3508
		skb_set_network_header(skb, skb_gro_offset(skb));
3509 3510 3511 3512
		mac_len = skb->network_header - skb->mac_header;
		skb->mac_len = mac_len;
		NAPI_GRO_CB(skb)->same_flow = 0;
		NAPI_GRO_CB(skb)->flush = 0;
H
Herbert Xu 已提交
3513
		NAPI_GRO_CB(skb)->free = 0;
3514 3515 3516 3517 3518 3519 3520 3521 3522

		pp = ptype->gro_receive(&napi->gro_list, skb);
		break;
	}
	rcu_read_unlock();

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

H
Herbert Xu 已提交
3523
	same_flow = NAPI_GRO_CB(skb)->same_flow;
3524
	ret = NAPI_GRO_CB(skb)->free ? GRO_MERGED_FREE : GRO_MERGED;
H
Herbert Xu 已提交
3525

3526 3527 3528 3529 3530 3531
	if (pp) {
		struct sk_buff *nskb = *pp;

		*pp = nskb->next;
		nskb->next = NULL;
		napi_gro_complete(nskb);
3532
		napi->gro_count--;
3533 3534
	}

H
Herbert Xu 已提交
3535
	if (same_flow)
3536 3537
		goto ok;

3538
	if (NAPI_GRO_CB(skb)->flush || napi->gro_count >= MAX_GRO_SKBS)
3539 3540
		goto normal;

3541
	napi->gro_count++;
3542
	NAPI_GRO_CB(skb)->count = 1;
3543
	skb_shinfo(skb)->gso_size = skb_gro_len(skb);
3544 3545
	skb->next = napi->gro_list;
	napi->gro_list = skb;
3546
	ret = GRO_HELD;
3547

3548
pull:
H
Herbert Xu 已提交
3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559
	if (skb_headlen(skb) < skb_gro_offset(skb)) {
		int grow = skb_gro_offset(skb) - skb_headlen(skb);

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

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

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

		skb_shinfo(skb)->frags[0].page_offset += grow;
E
Eric Dumazet 已提交
3560
		skb_frag_size_sub(&skb_shinfo(skb)->frags[0], grow);
H
Herbert Xu 已提交
3561

E
Eric Dumazet 已提交
3562
		if (unlikely(!skb_frag_size(&skb_shinfo(skb)->frags[0]))) {
3563
			skb_frag_unref(skb, 0);
H
Herbert Xu 已提交
3564 3565
			memmove(skb_shinfo(skb)->frags,
				skb_shinfo(skb)->frags + 1,
3566
				--skb_shinfo(skb)->nr_frags * sizeof(skb_frag_t));
H
Herbert Xu 已提交
3567
		}
3568 3569
	}

3570
ok:
3571
	return ret;
3572 3573

normal:
3574 3575
	ret = GRO_NORMAL;
	goto pull;
H
Herbert Xu 已提交
3576
}
3577 3578
EXPORT_SYMBOL(dev_gro_receive);

3579
static inline gro_result_t
3580
__napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3581 3582
{
	struct sk_buff *p;
E
Eric Dumazet 已提交
3583
	unsigned int maclen = skb->dev->hard_header_len;
3584 3585

	for (p = napi->gro_list; p; p = p->next) {
3586 3587 3588
		unsigned long diffs;

		diffs = (unsigned long)p->dev ^ (unsigned long)skb->dev;
3589
		diffs |= p->vlan_tci ^ skb->vlan_tci;
E
Eric Dumazet 已提交
3590 3591 3592 3593 3594 3595 3596
		if (maclen == ETH_HLEN)
			diffs |= compare_ether_header(skb_mac_header(p),
						      skb_gro_mac_header(skb));
		else if (!diffs)
			diffs = memcmp(skb_mac_header(p),
				       skb_gro_mac_header(skb),
				       maclen);
3597
		NAPI_GRO_CB(p)->same_flow = !diffs;
3598 3599 3600 3601 3602
		NAPI_GRO_CB(p)->flush = 0;
	}

	return dev_gro_receive(napi, skb);
}
H
Herbert Xu 已提交
3603

3604
gro_result_t napi_skb_finish(gro_result_t ret, struct sk_buff *skb)
H
Herbert Xu 已提交
3605
{
3606 3607
	switch (ret) {
	case GRO_NORMAL:
3608 3609 3610
		if (netif_receive_skb(skb))
			ret = GRO_DROP;
		break;
H
Herbert Xu 已提交
3611

3612
	case GRO_DROP:
H
Herbert Xu 已提交
3613 3614
		kfree_skb(skb);
		break;
3615

3616
	case GRO_MERGED_FREE:
3617 3618 3619 3620
		if (NAPI_GRO_CB(skb)->free == NAPI_GRO_FREE_STOLEN_HEAD)
			kmem_cache_free(skbuff_head_cache, skb);
		else
			__kfree_skb(skb);
3621 3622
		break;

3623 3624 3625
	case GRO_HELD:
	case GRO_MERGED:
		break;
H
Herbert Xu 已提交
3626 3627
	}

3628
	return ret;
3629 3630 3631
}
EXPORT_SYMBOL(napi_skb_finish);

3632 3633 3634 3635
void skb_gro_reset_offset(struct sk_buff *skb)
{
	NAPI_GRO_CB(skb)->data_offset = 0;
	NAPI_GRO_CB(skb)->frag0 = NULL;
3636
	NAPI_GRO_CB(skb)->frag0_len = 0;
3637

3638
	if (skb->mac_header == skb->tail &&
3639
	    !PageHighMem(skb_frag_page(&skb_shinfo(skb)->frags[0]))) {
3640
		NAPI_GRO_CB(skb)->frag0 =
3641
			skb_frag_address(&skb_shinfo(skb)->frags[0]);
E
Eric Dumazet 已提交
3642
		NAPI_GRO_CB(skb)->frag0_len = skb_frag_size(&skb_shinfo(skb)->frags[0]);
3643
	}
3644 3645 3646
}
EXPORT_SYMBOL(skb_gro_reset_offset);

3647
gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3648
{
3649 3650
	skb_gro_reset_offset(skb);

3651
	return napi_skb_finish(__napi_gro_receive(napi, skb), skb);
3652 3653 3654
}
EXPORT_SYMBOL(napi_gro_receive);

S
stephen hemminger 已提交
3655
static void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
3656 3657
{
	__skb_pull(skb, skb_headlen(skb));
3658 3659
	/* restore the reserve we had after netdev_alloc_skb_ip_align() */
	skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN - skb_headroom(skb));
3660
	skb->vlan_tci = 0;
H
Herbert Xu 已提交
3661
	skb->dev = napi->dev;
A
Andy Gospodarek 已提交
3662
	skb->skb_iif = 0;
3663 3664 3665 3666

	napi->skb = skb;
}

3667
struct sk_buff *napi_get_frags(struct napi_struct *napi)
H
Herbert Xu 已提交
3668 3669 3670 3671
{
	struct sk_buff *skb = napi->skb;

	if (!skb) {
3672 3673 3674
		skb = netdev_alloc_skb_ip_align(napi->dev, GRO_MAX_HEAD);
		if (skb)
			napi->skb = skb;
3675
	}
3676 3677
	return skb;
}
3678
EXPORT_SYMBOL(napi_get_frags);
3679

3680 3681
gro_result_t napi_frags_finish(struct napi_struct *napi, struct sk_buff *skb,
			       gro_result_t ret)
3682
{
3683 3684
	switch (ret) {
	case GRO_NORMAL:
3685
	case GRO_HELD:
3686
		skb->protocol = eth_type_trans(skb, skb->dev);
3687

3688 3689 3690 3691
		if (ret == GRO_HELD)
			skb_gro_pull(skb, -ETH_HLEN);
		else if (netif_receive_skb(skb))
			ret = GRO_DROP;
3692
		break;
H
Herbert Xu 已提交
3693

3694 3695 3696 3697
	case GRO_DROP:
	case GRO_MERGED_FREE:
		napi_reuse_skb(napi, skb);
		break;
3698 3699 3700

	case GRO_MERGED:
		break;
3701
	}
H
Herbert Xu 已提交
3702

3703
	return ret;
H
Herbert Xu 已提交
3704
}
3705 3706
EXPORT_SYMBOL(napi_frags_finish);

E
Eric Dumazet 已提交
3707
static struct sk_buff *napi_frags_skb(struct napi_struct *napi)
3708 3709 3710
{
	struct sk_buff *skb = napi->skb;
	struct ethhdr *eth;
3711 3712
	unsigned int hlen;
	unsigned int off;
3713 3714 3715 3716 3717 3718

	napi->skb = NULL;

	skb_reset_mac_header(skb);
	skb_gro_reset_offset(skb);

3719 3720 3721 3722 3723 3724 3725 3726 3727 3728
	off = skb_gro_offset(skb);
	hlen = off + sizeof(*eth);
	eth = skb_gro_header_fast(skb, off);
	if (skb_gro_header_hard(skb, hlen)) {
		eth = skb_gro_header_slow(skb, hlen, off);
		if (unlikely(!eth)) {
			napi_reuse_skb(napi, skb);
			skb = NULL;
			goto out;
		}
3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742
	}

	skb_gro_pull(skb, sizeof(*eth));

	/*
	 * This works because the only protocols we care about don't require
	 * special handling.  We'll fix it up properly at the end.
	 */
	skb->protocol = eth->h_proto;

out:
	return skb;
}

3743
gro_result_t napi_gro_frags(struct napi_struct *napi)
3744
{
3745
	struct sk_buff *skb = napi_frags_skb(napi);
3746 3747

	if (!skb)
3748
		return GRO_DROP;
3749 3750 3751

	return napi_frags_finish(napi, skb, __napi_gro_receive(napi, skb));
}
H
Herbert Xu 已提交
3752 3753
EXPORT_SYMBOL(napi_gro_frags);

3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781
/*
 * net_rps_action sends any pending IPI's for rps.
 * 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))
				__smp_call_function_single(remsd->cpu,
							   &remsd->csd, 0);
			remsd = next;
		}
	} else
#endif
		local_irq_enable();
}

3782
static int process_backlog(struct napi_struct *napi, int quota)
L
Linus Torvalds 已提交
3783 3784
{
	int work = 0;
E
Eric Dumazet 已提交
3785
	struct softnet_data *sd = container_of(napi, struct softnet_data, backlog);
L
Linus Torvalds 已提交
3786

3787 3788 3789 3790 3791 3792 3793 3794 3795
#ifdef CONFIG_RPS
	/* Check if we have pending ipi, its better to send them now,
	 * not waiting net_rx_action() end.
	 */
	if (sd->rps_ipi_list) {
		local_irq_disable();
		net_rps_action_and_irq_enable(sd);
	}
#endif
3796
	napi->weight = weight_p;
3797 3798
	local_irq_disable();
	while (work < quota) {
L
Linus Torvalds 已提交
3799
		struct sk_buff *skb;
3800 3801 3802 3803 3804 3805
		unsigned int qlen;

		while ((skb = __skb_dequeue(&sd->process_queue))) {
			local_irq_enable();
			__netif_receive_skb(skb);
			local_irq_disable();
3806 3807 3808 3809 3810
			input_queue_head_incr(sd);
			if (++work >= quota) {
				local_irq_enable();
				return work;
			}
3811
		}
L
Linus Torvalds 已提交
3812

E
Eric Dumazet 已提交
3813
		rps_lock(sd);
3814
		qlen = skb_queue_len(&sd->input_pkt_queue);
3815
		if (qlen)
3816 3817
			skb_queue_splice_tail_init(&sd->input_pkt_queue,
						   &sd->process_queue);
3818

3819
		if (qlen < quota - work) {
E
Eric Dumazet 已提交
3820 3821 3822 3823 3824 3825 3826 3827 3828 3829
			/*
			 * Inline a custom version of __napi_complete().
			 * only current cpu owns and manipulates this napi,
			 * and NAPI_STATE_SCHED is the only possible flag set on backlog.
			 * we can use a plain write instead of clear_bit(),
			 * and we dont need an smp_mb() memory barrier.
			 */
			list_del(&napi->poll_list);
			napi->state = 0;

3830
			quota = work + qlen;
3831
		}
E
Eric Dumazet 已提交
3832
		rps_unlock(sd);
3833 3834
	}
	local_irq_enable();
L
Linus Torvalds 已提交
3835

3836 3837
	return work;
}
L
Linus Torvalds 已提交
3838

3839 3840
/**
 * __napi_schedule - schedule for receive
3841
 * @n: entry to schedule
3842 3843 3844
 *
 * The entry's receive function will be scheduled to run
 */
H
Harvey Harrison 已提交
3845
void __napi_schedule(struct napi_struct *n)
3846 3847
{
	unsigned long flags;
L
Linus Torvalds 已提交
3848

3849
	local_irq_save(flags);
E
Eric Dumazet 已提交
3850
	____napi_schedule(&__get_cpu_var(softnet_data), n);
3851
	local_irq_restore(flags);
L
Linus Torvalds 已提交
3852
}
3853 3854
EXPORT_SYMBOL(__napi_schedule);

3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887
void __napi_complete(struct napi_struct *n)
{
	BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
	BUG_ON(n->gro_list);

	list_del(&n->poll_list);
	smp_mb__before_clear_bit();
	clear_bit(NAPI_STATE_SCHED, &n->state);
}
EXPORT_SYMBOL(__napi_complete);

void napi_complete(struct napi_struct *n)
{
	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;

	napi_gro_flush(n);
	local_irq_save(flags);
	__napi_complete(n);
	local_irq_restore(flags);
}
EXPORT_SYMBOL(napi_complete);

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);
3888
	napi->gro_count = 0;
3889
	napi->gro_list = NULL;
H
Herbert Xu 已提交
3890
	napi->skb = NULL;
3891 3892 3893 3894
	napi->poll = poll;
	napi->weight = weight;
	list_add(&napi->dev_list, &dev->napi_list);
	napi->dev = dev;
H
Herbert Xu 已提交
3895
#ifdef CONFIG_NETPOLL
3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906
	spin_lock_init(&napi->poll_lock);
	napi->poll_owner = -1;
#endif
	set_bit(NAPI_STATE_SCHED, &napi->state);
}
EXPORT_SYMBOL(netif_napi_add);

void netif_napi_del(struct napi_struct *napi)
{
	struct sk_buff *skb, *next;

3907
	list_del_init(&napi->dev_list);
3908
	napi_free_frags(napi);
3909 3910 3911 3912 3913 3914 3915 3916

	for (skb = napi->gro_list; skb; skb = next) {
		next = skb->next;
		skb->next = NULL;
		kfree_skb(skb);
	}

	napi->gro_list = NULL;
3917
	napi->gro_count = 0;
3918 3919 3920
}
EXPORT_SYMBOL(netif_napi_del);

L
Linus Torvalds 已提交
3921 3922
static void net_rx_action(struct softirq_action *h)
{
3923
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3924
	unsigned long time_limit = jiffies + 2;
3925
	int budget = netdev_budget;
3926 3927
	void *have;

L
Linus Torvalds 已提交
3928 3929
	local_irq_disable();

3930
	while (!list_empty(&sd->poll_list)) {
3931 3932
		struct napi_struct *n;
		int work, weight;
L
Linus Torvalds 已提交
3933

3934
		/* If softirq window is exhuasted then punt.
3935 3936
		 * Allow this to run for 2 jiffies since which will allow
		 * an average latency of 1.5/HZ.
3937
		 */
3938
		if (unlikely(budget <= 0 || time_after(jiffies, time_limit)))
L
Linus Torvalds 已提交
3939 3940 3941 3942
			goto softnet_break;

		local_irq_enable();

3943 3944 3945 3946 3947
		/* Even though interrupts have been re-enabled, this
		 * access is safe because interrupts can only add new
		 * entries to the tail of this list, and only ->poll()
		 * calls can remove this head entry from the list.
		 */
3948
		n = list_first_entry(&sd->poll_list, struct napi_struct, poll_list);
L
Linus Torvalds 已提交
3949

3950 3951 3952 3953
		have = netpoll_poll_lock(n);

		weight = n->weight;

3954 3955 3956 3957
		/* 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
L
Lucas De Marchi 已提交
3958
		 * accidentally calling ->poll() when NAPI is not scheduled.
3959 3960
		 */
		work = 0;
3961
		if (test_bit(NAPI_STATE_SCHED, &n->state)) {
3962
			work = n->poll(n, weight);
3963 3964
			trace_napi_poll(n);
		}
3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976

		WARN_ON_ONCE(work > weight);

		budget -= work;

		local_irq_disable();

		/* 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.
		 */
3977
		if (unlikely(work == weight)) {
3978 3979 3980 3981 3982
			if (unlikely(napi_disable_pending(n))) {
				local_irq_enable();
				napi_complete(n);
				local_irq_disable();
			} else
3983
				list_move_tail(&n->poll_list, &sd->poll_list);
3984
		}
3985 3986

		netpoll_poll_unlock(have);
L
Linus Torvalds 已提交
3987 3988
	}
out:
3989
	net_rps_action_and_irq_enable(sd);
T
Tom Herbert 已提交
3990

3991 3992 3993 3994 3995
#ifdef CONFIG_NET_DMA
	/*
	 * There may not be any more sk_buffs coming right now, so push
	 * any pending DMA copies to hardware
	 */
3996
	dma_issue_pending_all();
3997
#endif
3998

L
Linus Torvalds 已提交
3999 4000 4001
	return;

softnet_break:
C
Changli Gao 已提交
4002
	sd->time_squeeze++;
L
Linus Torvalds 已提交
4003 4004 4005 4006
	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
	goto out;
}

E
Eric Dumazet 已提交
4007
static gifconf_func_t *gifconf_list[NPROTO];
L
Linus Torvalds 已提交
4008 4009 4010 4011 4012 4013 4014 4015 4016 4017

/**
 *	register_gifconf	-	register a SIOCGIF handler
 *	@family: Address family
 *	@gifconf: Function handler
 *
 *	Register protocol dependent address dumping routines. The handler
 *	that is passed must not be freed or reused until it has been replaced
 *	by another handler.
 */
E
Eric Dumazet 已提交
4018
int register_gifconf(unsigned int family, gifconf_func_t *gifconf)
L
Linus Torvalds 已提交
4019 4020 4021 4022 4023 4024
{
	if (family >= NPROTO)
		return -EINVAL;
	gifconf_list[family] = gifconf;
	return 0;
}
E
Eric Dumazet 已提交
4025
EXPORT_SYMBOL(register_gifconf);
L
Linus Torvalds 已提交
4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038


/*
 *	Map an interface index to its name (SIOCGIFNAME)
 */

/*
 *	We need this ioctl for efficient implementation of the
 *	if_indextoname() function required by the IPv6 API.  Without
 *	it, we would have to search all the interfaces to find a
 *	match.  --pb
 */

4039
static int dev_ifname(struct net *net, struct ifreq __user *arg)
L
Linus Torvalds 已提交
4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050
{
	struct net_device *dev;
	struct ifreq ifr;

	/*
	 *	Fetch the caller's info block.
	 */

	if (copy_from_user(&ifr, arg, sizeof(struct ifreq)))
		return -EFAULT;

4051 4052
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifr.ifr_ifindex);
L
Linus Torvalds 已提交
4053
	if (!dev) {
4054
		rcu_read_unlock();
L
Linus Torvalds 已提交
4055 4056 4057 4058
		return -ENODEV;
	}

	strcpy(ifr.ifr_name, dev->name);
4059
	rcu_read_unlock();
L
Linus Torvalds 已提交
4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071

	if (copy_to_user(arg, &ifr, sizeof(struct ifreq)))
		return -EFAULT;
	return 0;
}

/*
 *	Perform a SIOCGIFCONF call. This structure will change
 *	size eventually, and there is nothing I can do about it.
 *	Thus we will need a 'compatibility mode'.
 */

4072
static int dev_ifconf(struct net *net, char __user *arg)
L
Linus Torvalds 已提交
4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095
{
	struct ifconf ifc;
	struct net_device *dev;
	char __user *pos;
	int len;
	int total;
	int i;

	/*
	 *	Fetch the caller's info block.
	 */

	if (copy_from_user(&ifc, arg, sizeof(struct ifconf)))
		return -EFAULT;

	pos = ifc.ifc_buf;
	len = ifc.ifc_len;

	/*
	 *	Loop over the interfaces, and write an info block for each.
	 */

	total = 0;
4096
	for_each_netdev(net, dev) {
L
Linus Torvalds 已提交
4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109
		for (i = 0; i < NPROTO; i++) {
			if (gifconf_list[i]) {
				int done;
				if (!pos)
					done = gifconf_list[i](dev, NULL, 0);
				else
					done = gifconf_list[i](dev, pos + total,
							       len - total);
				if (done < 0)
					return -EFAULT;
				total += done;
			}
		}
4110
	}
L
Linus Torvalds 已提交
4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123

	/*
	 *	All done.  Write the updated control block back to the caller.
	 */
	ifc.ifc_len = total;

	/*
	 * 	Both BSD and Solaris return 0 here, so we do too.
	 */
	return copy_to_user(arg, &ifc, sizeof(struct ifconf)) ? -EFAULT : 0;
}

#ifdef CONFIG_PROC_FS
4124

E
Eric Dumazet 已提交
4125
#define BUCKET_SPACE (32 - NETDEV_HASHBITS - 1)
4126 4127 4128 4129 4130

#define get_bucket(x) ((x) >> BUCKET_SPACE)
#define get_offset(x) ((x) & ((1 << BUCKET_SPACE) - 1))
#define set_bucket_offset(b, o) ((b) << BUCKET_SPACE | (o))

E
Eric Dumazet 已提交
4131
static inline struct net_device *dev_from_same_bucket(struct seq_file *seq, loff_t *pos)
4132 4133 4134 4135 4136
{
	struct net *net = seq_file_net(seq);
	struct net_device *dev;
	struct hlist_node *p;
	struct hlist_head *h;
E
Eric Dumazet 已提交
4137
	unsigned int count = 0, offset = get_offset(*pos);
4138

E
Eric Dumazet 已提交
4139
	h = &net->dev_name_head[get_bucket(*pos)];
4140
	hlist_for_each_entry_rcu(dev, p, h, name_hlist) {
E
Eric Dumazet 已提交
4141
		if (++count == offset)
4142 4143 4144 4145 4146 4147
			return dev;
	}

	return NULL;
}

E
Eric Dumazet 已提交
4148
static inline struct net_device *dev_from_bucket(struct seq_file *seq, loff_t *pos)
4149 4150 4151 4152 4153
{
	struct net_device *dev;
	unsigned int bucket;

	do {
E
Eric Dumazet 已提交
4154
		dev = dev_from_same_bucket(seq, pos);
4155 4156 4157
		if (dev)
			return dev;

E
Eric Dumazet 已提交
4158 4159
		bucket = get_bucket(*pos) + 1;
		*pos = set_bucket_offset(bucket, 1);
4160 4161 4162 4163 4164
	} while (bucket < NETDEV_HASHENTRIES);

	return NULL;
}

L
Linus Torvalds 已提交
4165 4166 4167 4168
/*
 *	This is invoked by the /proc filesystem handler to display a device
 *	in detail.
 */
4169
void *dev_seq_start(struct seq_file *seq, loff_t *pos)
4170
	__acquires(RCU)
L
Linus Torvalds 已提交
4171
{
4172
	rcu_read_lock();
4173 4174
	if (!*pos)
		return SEQ_START_TOKEN;
L
Linus Torvalds 已提交
4175

E
Eric Dumazet 已提交
4176
	if (get_bucket(*pos) >= NETDEV_HASHENTRIES)
4177
		return NULL;
L
Linus Torvalds 已提交
4178

E
Eric Dumazet 已提交
4179
	return dev_from_bucket(seq, pos);
L
Linus Torvalds 已提交
4180 4181 4182 4183
}

void *dev_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
4184
	++*pos;
E
Eric Dumazet 已提交
4185
	return dev_from_bucket(seq, pos);
L
Linus Torvalds 已提交
4186 4187 4188
}

void dev_seq_stop(struct seq_file *seq, void *v)
4189
	__releases(RCU)
L
Linus Torvalds 已提交
4190
{
4191
	rcu_read_unlock();
L
Linus Torvalds 已提交
4192 4193 4194 4195
}

static void dev_seq_printf_stats(struct seq_file *seq, struct net_device *dev)
{
4196 4197
	struct rtnl_link_stats64 temp;
	const struct rtnl_link_stats64 *stats = dev_get_stats(dev, &temp);
L
Linus Torvalds 已提交
4198

4199 4200
	seq_printf(seq, "%6s: %7llu %7llu %4llu %4llu %4llu %5llu %10llu %9llu "
		   "%8llu %7llu %4llu %4llu %4llu %5llu %7llu %10llu\n",
4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215
		   dev->name, stats->rx_bytes, stats->rx_packets,
		   stats->rx_errors,
		   stats->rx_dropped + stats->rx_missed_errors,
		   stats->rx_fifo_errors,
		   stats->rx_length_errors + stats->rx_over_errors +
		    stats->rx_crc_errors + stats->rx_frame_errors,
		   stats->rx_compressed, stats->multicast,
		   stats->tx_bytes, stats->tx_packets,
		   stats->tx_errors, stats->tx_dropped,
		   stats->tx_fifo_errors, stats->collisions,
		   stats->tx_carrier_errors +
		    stats->tx_aborted_errors +
		    stats->tx_window_errors +
		    stats->tx_heartbeat_errors,
		   stats->tx_compressed);
L
Linus Torvalds 已提交
4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234
}

/*
 *	Called from the PROCfs module. This now uses the new arbitrary sized
 *	/proc/net interface to create /proc/net/dev
 */
static int dev_seq_show(struct seq_file *seq, void *v)
{
	if (v == SEQ_START_TOKEN)
		seq_puts(seq, "Inter-|   Receive                            "
			      "                    |  Transmit\n"
			      " face |bytes    packets errs drop fifo frame "
			      "compressed multicast|bytes    packets errs "
			      "drop fifo colls carrier compressed\n");
	else
		dev_seq_printf_stats(seq, v);
	return 0;
}

C
Changli Gao 已提交
4235
static struct softnet_data *softnet_get_online(loff_t *pos)
L
Linus Torvalds 已提交
4236
{
C
Changli Gao 已提交
4237
	struct softnet_data *sd = NULL;
L
Linus Torvalds 已提交
4238

4239
	while (*pos < nr_cpu_ids)
4240
		if (cpu_online(*pos)) {
C
Changli Gao 已提交
4241
			sd = &per_cpu(softnet_data, *pos);
L
Linus Torvalds 已提交
4242 4243 4244
			break;
		} else
			++*pos;
C
Changli Gao 已提交
4245
	return sd;
L
Linus Torvalds 已提交
4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264
}

static void *softnet_seq_start(struct seq_file *seq, loff_t *pos)
{
	return softnet_get_online(pos);
}

static void *softnet_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	++*pos;
	return softnet_get_online(pos);
}

static void softnet_seq_stop(struct seq_file *seq, void *v)
{
}

static int softnet_seq_show(struct seq_file *seq, void *v)
{
C
Changli Gao 已提交
4265
	struct softnet_data *sd = v;
L
Linus Torvalds 已提交
4266

T
Tom Herbert 已提交
4267
	seq_printf(seq, "%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
C
Changli Gao 已提交
4268
		   sd->processed, sd->dropped, sd->time_squeeze, 0,
4269
		   0, 0, 0, 0, /* was fastroute */
C
Changli Gao 已提交
4270
		   sd->cpu_collision, sd->received_rps);
L
Linus Torvalds 已提交
4271 4272 4273
	return 0;
}

4274
static const struct seq_operations dev_seq_ops = {
L
Linus Torvalds 已提交
4275 4276 4277 4278 4279 4280 4281 4282
	.start = dev_seq_start,
	.next  = dev_seq_next,
	.stop  = dev_seq_stop,
	.show  = dev_seq_show,
};

static int dev_seq_open(struct inode *inode, struct file *file)
{
4283
	return seq_open_net(inode, file, &dev_seq_ops,
E
Eric Dumazet 已提交
4284
			    sizeof(struct seq_net_private));
4285 4286
}

4287
static const struct file_operations dev_seq_fops = {
L
Linus Torvalds 已提交
4288 4289 4290 4291
	.owner	 = THIS_MODULE,
	.open    = dev_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
4292
	.release = seq_release_net,
L
Linus Torvalds 已提交
4293 4294
};

4295
static const struct seq_operations softnet_seq_ops = {
L
Linus Torvalds 已提交
4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306
	.start = softnet_seq_start,
	.next  = softnet_seq_next,
	.stop  = softnet_seq_stop,
	.show  = softnet_seq_show,
};

static int softnet_seq_open(struct inode *inode, struct file *file)
{
	return seq_open(file, &softnet_seq_ops);
}

4307
static const struct file_operations softnet_seq_fops = {
L
Linus Torvalds 已提交
4308 4309 4310 4311 4312 4313 4314
	.owner	 = THIS_MODULE,
	.open    = softnet_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release,
};

4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326
static void *ptype_get_idx(loff_t pos)
{
	struct packet_type *pt = NULL;
	loff_t i = 0;
	int t;

	list_for_each_entry_rcu(pt, &ptype_all, list) {
		if (i == pos)
			return pt;
		++i;
	}

4327
	for (t = 0; t < PTYPE_HASH_SIZE; t++) {
4328 4329 4330 4331 4332 4333 4334 4335 4336 4337
		list_for_each_entry_rcu(pt, &ptype_base[t], list) {
			if (i == pos)
				return pt;
			++i;
		}
	}
	return NULL;
}

static void *ptype_seq_start(struct seq_file *seq, loff_t *pos)
4338
	__acquires(RCU)
4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361
{
	rcu_read_lock();
	return *pos ? ptype_get_idx(*pos - 1) : SEQ_START_TOKEN;
}

static void *ptype_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	struct packet_type *pt;
	struct list_head *nxt;
	int hash;

	++*pos;
	if (v == SEQ_START_TOKEN)
		return ptype_get_idx(0);

	pt = v;
	nxt = pt->list.next;
	if (pt->type == htons(ETH_P_ALL)) {
		if (nxt != &ptype_all)
			goto found;
		hash = 0;
		nxt = ptype_base[0].next;
	} else
4362
		hash = ntohs(pt->type) & PTYPE_HASH_MASK;
4363 4364

	while (nxt == &ptype_base[hash]) {
4365
		if (++hash >= PTYPE_HASH_SIZE)
4366 4367 4368 4369 4370 4371 4372 4373
			return NULL;
		nxt = ptype_base[hash].next;
	}
found:
	return list_entry(nxt, struct packet_type, list);
}

static void ptype_seq_stop(struct seq_file *seq, void *v)
4374
	__releases(RCU)
4375 4376 4377 4378 4379 4380 4381 4382 4383 4384
{
	rcu_read_unlock();
}

static int ptype_seq_show(struct seq_file *seq, void *v)
{
	struct packet_type *pt = v;

	if (v == SEQ_START_TOKEN)
		seq_puts(seq, "Type Device      Function\n");
4385
	else if (pt->dev == NULL || dev_net(pt->dev) == seq_file_net(seq)) {
4386 4387 4388 4389 4390
		if (pt->type == htons(ETH_P_ALL))
			seq_puts(seq, "ALL ");
		else
			seq_printf(seq, "%04x", ntohs(pt->type));

A
Alexey Dobriyan 已提交
4391 4392
		seq_printf(seq, " %-8s %pF\n",
			   pt->dev ? pt->dev->name : "", pt->func);
4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406
	}

	return 0;
}

static const struct seq_operations ptype_seq_ops = {
	.start = ptype_seq_start,
	.next  = ptype_seq_next,
	.stop  = ptype_seq_stop,
	.show  = ptype_seq_show,
};

static int ptype_seq_open(struct inode *inode, struct file *file)
{
4407 4408
	return seq_open_net(inode, file, &ptype_seq_ops,
			sizeof(struct seq_net_private));
4409 4410 4411 4412 4413 4414 4415
}

static const struct file_operations ptype_seq_fops = {
	.owner	 = THIS_MODULE,
	.open    = ptype_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
4416
	.release = seq_release_net,
4417 4418 4419
};


4420
static int __net_init dev_proc_net_init(struct net *net)
L
Linus Torvalds 已提交
4421 4422 4423
{
	int rc = -ENOMEM;

4424
	if (!proc_net_fops_create(net, "dev", S_IRUGO, &dev_seq_fops))
L
Linus Torvalds 已提交
4425
		goto out;
4426
	if (!proc_net_fops_create(net, "softnet_stat", S_IRUGO, &softnet_seq_fops))
L
Linus Torvalds 已提交
4427
		goto out_dev;
4428
	if (!proc_net_fops_create(net, "ptype", S_IRUGO, &ptype_seq_fops))
4429
		goto out_softnet;
4430

4431
	if (wext_proc_init(net))
4432
		goto out_ptype;
L
Linus Torvalds 已提交
4433 4434 4435
	rc = 0;
out:
	return rc;
4436
out_ptype:
4437
	proc_net_remove(net, "ptype");
L
Linus Torvalds 已提交
4438
out_softnet:
4439
	proc_net_remove(net, "softnet_stat");
L
Linus Torvalds 已提交
4440
out_dev:
4441
	proc_net_remove(net, "dev");
L
Linus Torvalds 已提交
4442 4443
	goto out;
}
4444

4445
static void __net_exit dev_proc_net_exit(struct net *net)
4446 4447 4448 4449 4450 4451 4452 4453
{
	wext_proc_exit(net);

	proc_net_remove(net, "ptype");
	proc_net_remove(net, "softnet_stat");
	proc_net_remove(net, "dev");
}

4454
static struct pernet_operations __net_initdata dev_proc_ops = {
4455 4456 4457 4458 4459 4460 4461 4462
	.init = dev_proc_net_init,
	.exit = dev_proc_net_exit,
};

static int __init dev_proc_init(void)
{
	return register_pernet_subsys(&dev_proc_ops);
}
L
Linus Torvalds 已提交
4463 4464 4465 4466 4467 4468
#else
#define dev_proc_init() 0
#endif	/* CONFIG_PROC_FS */


/**
J
Jiri Pirko 已提交
4469
 *	netdev_set_master	-	set up master pointer
L
Linus Torvalds 已提交
4470 4471 4472 4473 4474 4475
 *	@slave: slave device
 *	@master: new master device
 *
 *	Changes the master device of the slave. Pass %NULL to break the
 *	bonding. The caller must hold the RTNL semaphore. On a failure
 *	a negative errno code is returned. On success the reference counts
J
Jiri Pirko 已提交
4476
 *	are adjusted and the function returns zero.
L
Linus Torvalds 已提交
4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490
 */
int netdev_set_master(struct net_device *slave, struct net_device *master)
{
	struct net_device *old = slave->master;

	ASSERT_RTNL();

	if (master) {
		if (old)
			return -EBUSY;
		dev_hold(master);
	}

	slave->master = master;
4491

4492
	if (old)
L
Linus Torvalds 已提交
4493
		dev_put(old);
J
Jiri Pirko 已提交
4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516
	return 0;
}
EXPORT_SYMBOL(netdev_set_master);

/**
 *	netdev_set_bond_master	-	set up bonding master/slave pair
 *	@slave: slave device
 *	@master: new master device
 *
 *	Changes the master device of the slave. Pass %NULL to break the
 *	bonding. The caller must hold the RTNL semaphore. On a failure
 *	a negative errno code is returned. On success %RTM_NEWLINK is sent
 *	to the routing socket and the function returns zero.
 */
int netdev_set_bond_master(struct net_device *slave, struct net_device *master)
{
	int err;

	ASSERT_RTNL();

	err = netdev_set_master(slave, master);
	if (err)
		return err;
L
Linus Torvalds 已提交
4517 4518 4519 4520 4521 4522 4523 4524
	if (master)
		slave->flags |= IFF_SLAVE;
	else
		slave->flags &= ~IFF_SLAVE;

	rtmsg_ifinfo(RTM_NEWLINK, slave, IFF_SLAVE);
	return 0;
}
J
Jiri Pirko 已提交
4525
EXPORT_SYMBOL(netdev_set_bond_master);
L
Linus Torvalds 已提交
4526

4527 4528
static void dev_change_rx_flags(struct net_device *dev, int flags)
{
4529 4530 4531 4532
	const struct net_device_ops *ops = dev->netdev_ops;

	if ((dev->flags & IFF_UP) && ops->ndo_change_rx_flags)
		ops->ndo_change_rx_flags(dev, flags);
4533 4534
}

4535
static int __dev_set_promiscuity(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4536
{
4537
	unsigned int old_flags = dev->flags;
4538 4539
	kuid_t uid;
	kgid_t gid;
L
Linus Torvalds 已提交
4540

4541 4542
	ASSERT_RTNL();

4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553
	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;
4554 4555
			pr_warn("%s: promiscuity touches roof, set promiscuity failed. promiscuity feature of device might be broken.\n",
				dev->name);
4556 4557 4558
			return -EOVERFLOW;
		}
	}
4559
	if (dev->flags != old_flags) {
4560 4561 4562
		pr_info("device %s %s promiscuous mode\n",
			dev->name,
			dev->flags & IFF_PROMISC ? "entered" : "left");
4563 4564
		if (audit_enabled) {
			current_uid_gid(&uid, &gid);
4565 4566 4567 4568 4569 4570
			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),
				audit_get_loginuid(current),
4571 4572
				from_kuid(&init_user_ns, uid),
				from_kgid(&init_user_ns, gid),
4573
				audit_get_sessionid(current));
4574
		}
4575

4576
		dev_change_rx_flags(dev, IFF_PROMISC);
L
Linus Torvalds 已提交
4577
	}
4578
	return 0;
L
Linus Torvalds 已提交
4579 4580
}

4581 4582 4583 4584 4585 4586 4587 4588 4589
/**
 *	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.
4590
 *	Return 0 if successful or a negative errno code on error.
4591
 */
4592
int dev_set_promiscuity(struct net_device *dev, int inc)
4593
{
4594
	unsigned int old_flags = dev->flags;
4595
	int err;
4596

4597
	err = __dev_set_promiscuity(dev, inc);
4598
	if (err < 0)
4599
		return err;
4600 4601
	if (dev->flags != old_flags)
		dev_set_rx_mode(dev);
4602
	return err;
4603
}
E
Eric Dumazet 已提交
4604
EXPORT_SYMBOL(dev_set_promiscuity);
4605

L
Linus Torvalds 已提交
4606 4607 4608 4609 4610 4611 4612 4613 4614 4615
/**
 *	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.
4616
 *	Return 0 if successful or a negative errno code on error.
L
Linus Torvalds 已提交
4617 4618
 */

4619
int dev_set_allmulti(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4620
{
4621
	unsigned int old_flags = dev->flags;
L
Linus Torvalds 已提交
4622

4623 4624
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4625
	dev->flags |= IFF_ALLMULTI;
4626 4627 4628 4629 4630 4631 4632 4633 4634 4635
	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;
4636 4637
			pr_warn("%s: allmulti touches roof, set allmulti failed. allmulti feature of device might be broken.\n",
				dev->name);
4638 4639 4640
			return -EOVERFLOW;
		}
	}
4641
	if (dev->flags ^ old_flags) {
4642
		dev_change_rx_flags(dev, IFF_ALLMULTI);
4643
		dev_set_rx_mode(dev);
4644
	}
4645
	return 0;
4646
}
E
Eric Dumazet 已提交
4647
EXPORT_SYMBOL(dev_set_allmulti);
4648 4649 4650 4651

/*
 *	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
4652
 *	filtering it is put in promiscuous mode while unicast addresses
4653 4654 4655 4656
 *	are present.
 */
void __dev_set_rx_mode(struct net_device *dev)
{
4657 4658
	const struct net_device_ops *ops = dev->netdev_ops;

4659 4660 4661 4662 4663
	/* dev_open will call this function so the list will stay sane. */
	if (!(dev->flags&IFF_UP))
		return;

	if (!netif_device_present(dev))
4664
		return;
4665

4666
	if (!(dev->priv_flags & IFF_UNICAST_FLT)) {
4667 4668 4669
		/* Unicast addresses changes may only happen under the rtnl,
		 * therefore calling __dev_set_promiscuity here is safe.
		 */
4670
		if (!netdev_uc_empty(dev) && !dev->uc_promisc) {
4671
			__dev_set_promiscuity(dev, 1);
4672
			dev->uc_promisc = true;
4673
		} else if (netdev_uc_empty(dev) && dev->uc_promisc) {
4674
			__dev_set_promiscuity(dev, -1);
4675
			dev->uc_promisc = false;
4676 4677
		}
	}
4678 4679 4680

	if (ops->ndo_set_rx_mode)
		ops->ndo_set_rx_mode(dev);
4681 4682 4683 4684
}

void dev_set_rx_mode(struct net_device *dev)
{
4685
	netif_addr_lock_bh(dev);
4686
	__dev_set_rx_mode(dev);
4687
	netif_addr_unlock_bh(dev);
L
Linus Torvalds 已提交
4688 4689
}

4690 4691 4692 4693 4694 4695
/**
 *	dev_get_flags - get flags reported to userspace
 *	@dev: device
 *
 *	Get the combination of flag bits exported through APIs to userspace.
 */
4696
unsigned int dev_get_flags(const struct net_device *dev)
L
Linus Torvalds 已提交
4697
{
4698
	unsigned int flags;
L
Linus Torvalds 已提交
4699 4700 4701

	flags = (dev->flags & ~(IFF_PROMISC |
				IFF_ALLMULTI |
S
Stefan Rompf 已提交
4702 4703 4704
				IFF_RUNNING |
				IFF_LOWER_UP |
				IFF_DORMANT)) |
L
Linus Torvalds 已提交
4705 4706 4707
		(dev->gflags & (IFF_PROMISC |
				IFF_ALLMULTI));

S
Stefan Rompf 已提交
4708 4709 4710 4711 4712 4713 4714 4715
	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
4716 4717 4718

	return flags;
}
E
Eric Dumazet 已提交
4719
EXPORT_SYMBOL(dev_get_flags);
L
Linus Torvalds 已提交
4720

4721
int __dev_change_flags(struct net_device *dev, unsigned int flags)
L
Linus Torvalds 已提交
4722
{
4723
	unsigned int old_flags = dev->flags;
4724
	int ret;
L
Linus Torvalds 已提交
4725

4726 4727
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741
	/*
	 *	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.
	 */

4742 4743
	if ((old_flags ^ flags) & IFF_MULTICAST)
		dev_change_rx_flags(dev, IFF_MULTICAST);
4744

4745
	dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4746 4747 4748 4749 4750 4751 4752 4753 4754

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

	ret = 0;
	if ((old_flags ^ flags) & IFF_UP) {	/* Bit is different  ? */
4755
		ret = ((old_flags & IFF_UP) ? __dev_close : __dev_open)(dev);
L
Linus Torvalds 已提交
4756 4757

		if (!ret)
4758
			dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4759 4760 4761
	}

	if ((flags ^ dev->gflags) & IFF_PROMISC) {
E
Eric Dumazet 已提交
4762 4763
		int inc = (flags & IFF_PROMISC) ? 1 : -1;

L
Linus Torvalds 已提交
4764 4765 4766 4767 4768 4769 4770 4771 4772
		dev->gflags ^= IFF_PROMISC;
		dev_set_promiscuity(dev, inc);
	}

	/* 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
4773 4774
		int inc = (flags & IFF_ALLMULTI) ? 1 : -1;

L
Linus Torvalds 已提交
4775 4776 4777 4778
		dev->gflags ^= IFF_ALLMULTI;
		dev_set_allmulti(dev, inc);
	}

4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805
	return ret;
}

void __dev_notify_flags(struct net_device *dev, unsigned int old_flags)
{
	unsigned int changes = dev->flags ^ old_flags;

	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 &&
	    (changes & ~(IFF_UP | IFF_PROMISC | IFF_ALLMULTI | IFF_VOLATILE)))
		call_netdevice_notifiers(NETDEV_CHANGE, dev);
}

/**
 *	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.
 */
4806
int dev_change_flags(struct net_device *dev, unsigned int flags)
4807
{
4808 4809
	int ret;
	unsigned int changes, old_flags = dev->flags;
4810 4811 4812 4813 4814 4815

	ret = __dev_change_flags(dev, flags);
	if (ret < 0)
		return ret;

	changes = old_flags ^ dev->flags;
4816 4817
	if (changes)
		rtmsg_ifinfo(RTM_NEWLINK, dev, changes);
L
Linus Torvalds 已提交
4818

4819
	__dev_notify_flags(dev, old_flags);
L
Linus Torvalds 已提交
4820 4821
	return ret;
}
E
Eric Dumazet 已提交
4822
EXPORT_SYMBOL(dev_change_flags);
L
Linus Torvalds 已提交
4823

4824 4825 4826 4827 4828 4829 4830
/**
 *	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
4831 4832
int dev_set_mtu(struct net_device *dev, int new_mtu)
{
4833
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846
	int err;

	if (new_mtu == dev->mtu)
		return 0;

	/*	MTU must be positive.	 */
	if (new_mtu < 0)
		return -EINVAL;

	if (!netif_device_present(dev))
		return -ENODEV;

	err = 0;
4847 4848
	if (ops->ndo_change_mtu)
		err = ops->ndo_change_mtu(dev, new_mtu);
L
Linus Torvalds 已提交
4849 4850
	else
		dev->mtu = new_mtu;
4851

L
Linus Torvalds 已提交
4852
	if (!err && dev->flags & IFF_UP)
4853
		call_netdevice_notifiers(NETDEV_CHANGEMTU, dev);
L
Linus Torvalds 已提交
4854 4855
	return err;
}
E
Eric Dumazet 已提交
4856
EXPORT_SYMBOL(dev_set_mtu);
L
Linus Torvalds 已提交
4857

4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868
/**
 *	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);

4869 4870 4871 4872 4873 4874 4875
/**
 *	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
4876 4877
int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa)
{
4878
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4879 4880
	int err;

4881
	if (!ops->ndo_set_mac_address)
L
Linus Torvalds 已提交
4882 4883 4884 4885 4886
		return -EOPNOTSUPP;
	if (sa->sa_family != dev->type)
		return -EINVAL;
	if (!netif_device_present(dev))
		return -ENODEV;
4887
	err = ops->ndo_set_mac_address(dev, sa);
L
Linus Torvalds 已提交
4888
	if (!err)
4889
		call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
4890
	add_device_randomness(dev->dev_addr, dev->addr_len);
L
Linus Torvalds 已提交
4891 4892
	return err;
}
E
Eric Dumazet 已提交
4893
EXPORT_SYMBOL(dev_set_mac_address);
L
Linus Torvalds 已提交
4894 4895

/*
E
Eric Dumazet 已提交
4896
 *	Perform the SIOCxIFxxx calls, inside rcu_read_lock()
L
Linus Torvalds 已提交
4897
 */
4898
static int dev_ifsioc_locked(struct net *net, struct ifreq *ifr, unsigned int cmd)
L
Linus Torvalds 已提交
4899 4900
{
	int err;
E
Eric Dumazet 已提交
4901
	struct net_device *dev = dev_get_by_name_rcu(net, ifr->ifr_name);
L
Linus Torvalds 已提交
4902 4903 4904 4905 4906

	if (!dev)
		return -ENODEV;

	switch (cmd) {
E
Eric Dumazet 已提交
4907 4908 4909
	case SIOCGIFFLAGS:	/* Get interface flags */
		ifr->ifr_flags = (short) dev_get_flags(dev);
		return 0;
L
Linus Torvalds 已提交
4910

E
Eric Dumazet 已提交
4911 4912 4913 4914
	case SIOCGIFMETRIC:	/* Get the metric on the interface
				   (currently unused) */
		ifr->ifr_metric = 0;
		return 0;
L
Linus Torvalds 已提交
4915

E
Eric Dumazet 已提交
4916 4917 4918
	case SIOCGIFMTU:	/* Get the MTU of a device */
		ifr->ifr_mtu = dev->mtu;
		return 0;
L
Linus Torvalds 已提交
4919

E
Eric Dumazet 已提交
4920 4921 4922 4923 4924 4925 4926 4927
	case SIOCGIFHWADDR:
		if (!dev->addr_len)
			memset(ifr->ifr_hwaddr.sa_data, 0, sizeof ifr->ifr_hwaddr.sa_data);
		else
			memcpy(ifr->ifr_hwaddr.sa_data, dev->dev_addr,
			       min(sizeof ifr->ifr_hwaddr.sa_data, (size_t) dev->addr_len));
		ifr->ifr_hwaddr.sa_family = dev->type;
		return 0;
L
Linus Torvalds 已提交
4928

E
Eric Dumazet 已提交
4929 4930 4931
	case SIOCGIFSLAVE:
		err = -EINVAL;
		break;
4932

E
Eric Dumazet 已提交
4933 4934 4935 4936 4937 4938 4939 4940
	case SIOCGIFMAP:
		ifr->ifr_map.mem_start = dev->mem_start;
		ifr->ifr_map.mem_end   = dev->mem_end;
		ifr->ifr_map.base_addr = dev->base_addr;
		ifr->ifr_map.irq       = dev->irq;
		ifr->ifr_map.dma       = dev->dma;
		ifr->ifr_map.port      = dev->if_port;
		return 0;
4941

E
Eric Dumazet 已提交
4942 4943 4944
	case SIOCGIFINDEX:
		ifr->ifr_ifindex = dev->ifindex;
		return 0;
4945

E
Eric Dumazet 已提交
4946 4947 4948
	case SIOCGIFTXQLEN:
		ifr->ifr_qlen = dev->tx_queue_len;
		return 0;
4949

E
Eric Dumazet 已提交
4950 4951 4952 4953 4954
	default:
		/* dev_ioctl() should ensure this case
		 * is never reached
		 */
		WARN_ON(1);
4955
		err = -ENOTTY;
E
Eric Dumazet 已提交
4956
		break;
4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968

	}
	return err;
}

/*
 *	Perform the SIOCxIFxxx calls, inside rtnl_lock()
 */
static int dev_ifsioc(struct net *net, struct ifreq *ifr, unsigned int cmd)
{
	int err;
	struct net_device *dev = __dev_get_by_name(net, ifr->ifr_name);
J
Jarek Poplawski 已提交
4969
	const struct net_device_ops *ops;
4970 4971 4972 4973

	if (!dev)
		return -ENODEV;

J
Jarek Poplawski 已提交
4974 4975
	ops = dev->netdev_ops;

4976
	switch (cmd) {
E
Eric Dumazet 已提交
4977 4978
	case SIOCSIFFLAGS:	/* Set interface flags */
		return dev_change_flags(dev, ifr->ifr_flags);
4979

E
Eric Dumazet 已提交
4980 4981 4982
	case SIOCSIFMETRIC:	/* Set the metric on the interface
				   (currently unused) */
		return -EOPNOTSUPP;
4983

E
Eric Dumazet 已提交
4984 4985
	case SIOCSIFMTU:	/* Set the MTU of a device */
		return dev_set_mtu(dev, ifr->ifr_mtu);
L
Linus Torvalds 已提交
4986

E
Eric Dumazet 已提交
4987 4988
	case SIOCSIFHWADDR:
		return dev_set_mac_address(dev, &ifr->ifr_hwaddr);
L
Linus Torvalds 已提交
4989

E
Eric Dumazet 已提交
4990 4991 4992 4993 4994 4995 4996
	case SIOCSIFHWBROADCAST:
		if (ifr->ifr_hwaddr.sa_family != dev->type)
			return -EINVAL;
		memcpy(dev->broadcast, ifr->ifr_hwaddr.sa_data,
		       min(sizeof ifr->ifr_hwaddr.sa_data, (size_t) dev->addr_len));
		call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
		return 0;
L
Linus Torvalds 已提交
4997

E
Eric Dumazet 已提交
4998 4999
	case SIOCSIFMAP:
		if (ops->ndo_set_config) {
L
Linus Torvalds 已提交
5000 5001
			if (!netif_device_present(dev))
				return -ENODEV;
E
Eric Dumazet 已提交
5002 5003 5004
			return ops->ndo_set_config(dev, &ifr->ifr_map);
		}
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
5005

E
Eric Dumazet 已提交
5006
	case SIOCADDMULTI:
5007
		if (!ops->ndo_set_rx_mode ||
E
Eric Dumazet 已提交
5008 5009 5010 5011
		    ifr->ifr_hwaddr.sa_family != AF_UNSPEC)
			return -EINVAL;
		if (!netif_device_present(dev))
			return -ENODEV;
5012
		return dev_mc_add_global(dev, ifr->ifr_hwaddr.sa_data);
E
Eric Dumazet 已提交
5013 5014

	case SIOCDELMULTI:
5015
		if (!ops->ndo_set_rx_mode ||
E
Eric Dumazet 已提交
5016 5017 5018 5019
		    ifr->ifr_hwaddr.sa_family != AF_UNSPEC)
			return -EINVAL;
		if (!netif_device_present(dev))
			return -ENODEV;
5020
		return dev_mc_del_global(dev, ifr->ifr_hwaddr.sa_data);
L
Linus Torvalds 已提交
5021

E
Eric Dumazet 已提交
5022 5023 5024 5025 5026
	case SIOCSIFTXQLEN:
		if (ifr->ifr_qlen < 0)
			return -EINVAL;
		dev->tx_queue_len = ifr->ifr_qlen;
		return 0;
L
Linus Torvalds 已提交
5027

E
Eric Dumazet 已提交
5028 5029 5030
	case SIOCSIFNAME:
		ifr->ifr_newname[IFNAMSIZ-1] = '\0';
		return dev_change_name(dev, ifr->ifr_newname);
L
Linus Torvalds 已提交
5031

5032 5033 5034 5035 5036 5037
	case SIOCSHWTSTAMP:
		err = net_hwtstamp_validate(ifr);
		if (err)
			return err;
		/* fall through */

E
Eric Dumazet 已提交
5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065
	/*
	 *	Unknown or private ioctl
	 */
	default:
		if ((cmd >= SIOCDEVPRIVATE &&
		    cmd <= SIOCDEVPRIVATE + 15) ||
		    cmd == SIOCBONDENSLAVE ||
		    cmd == SIOCBONDRELEASE ||
		    cmd == SIOCBONDSETHWADDR ||
		    cmd == SIOCBONDSLAVEINFOQUERY ||
		    cmd == SIOCBONDINFOQUERY ||
		    cmd == SIOCBONDCHANGEACTIVE ||
		    cmd == SIOCGMIIPHY ||
		    cmd == SIOCGMIIREG ||
		    cmd == SIOCSMIIREG ||
		    cmd == SIOCBRADDIF ||
		    cmd == SIOCBRDELIF ||
		    cmd == SIOCSHWTSTAMP ||
		    cmd == SIOCWANDEV) {
			err = -EOPNOTSUPP;
			if (ops->ndo_do_ioctl) {
				if (netif_device_present(dev))
					err = ops->ndo_do_ioctl(dev, ifr, cmd);
				else
					err = -ENODEV;
			}
		} else
			err = -EINVAL;
L
Linus Torvalds 已提交
5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077

	}
	return err;
}

/*
 *	This function handles all "interface"-type I/O control requests. The actual
 *	'doing' part of this is dev_ifsioc above.
 */

/**
 *	dev_ioctl	-	network device ioctl
5078
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
5079 5080 5081 5082 5083 5084 5085 5086 5087
 *	@cmd: command to issue
 *	@arg: pointer to a struct ifreq in user space
 *
 *	Issue ioctl functions to devices. This is normally called by the
 *	user space syscall interfaces but can sometimes be useful for
 *	other purposes. The return value is the return from the syscall if
 *	positive or a negative errno code on error.
 */

5088
int dev_ioctl(struct net *net, unsigned int cmd, void __user *arg)
L
Linus Torvalds 已提交
5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099
{
	struct ifreq ifr;
	int ret;
	char *colon;

	/* One special case: SIOCGIFCONF takes ifconf argument
	   and requires shared lock, because it sleeps writing
	   to user space.
	 */

	if (cmd == SIOCGIFCONF) {
5100
		rtnl_lock();
5101
		ret = dev_ifconf(net, (char __user *) arg);
5102
		rtnl_unlock();
L
Linus Torvalds 已提交
5103 5104 5105
		return ret;
	}
	if (cmd == SIOCGIFNAME)
5106
		return dev_ifname(net, (struct ifreq __user *)arg);
L
Linus Torvalds 已提交
5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121

	if (copy_from_user(&ifr, arg, sizeof(struct ifreq)))
		return -EFAULT;

	ifr.ifr_name[IFNAMSIZ-1] = 0;

	colon = strchr(ifr.ifr_name, ':');
	if (colon)
		*colon = 0;

	/*
	 *	See which interface the caller is talking about.
	 */

	switch (cmd) {
E
Eric Dumazet 已提交
5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136
	/*
	 *	These ioctl calls:
	 *	- can be done by all.
	 *	- atomic and do not require locking.
	 *	- return a value
	 */
	case SIOCGIFFLAGS:
	case SIOCGIFMETRIC:
	case SIOCGIFMTU:
	case SIOCGIFHWADDR:
	case SIOCGIFSLAVE:
	case SIOCGIFMAP:
	case SIOCGIFINDEX:
	case SIOCGIFTXQLEN:
		dev_load(net, ifr.ifr_name);
E
Eric Dumazet 已提交
5137
		rcu_read_lock();
E
Eric Dumazet 已提交
5138
		ret = dev_ifsioc_locked(net, &ifr, cmd);
E
Eric Dumazet 已提交
5139
		rcu_read_unlock();
E
Eric Dumazet 已提交
5140 5141 5142 5143 5144 5145 5146 5147
		if (!ret) {
			if (colon)
				*colon = ':';
			if (copy_to_user(arg, &ifr,
					 sizeof(struct ifreq)))
				ret = -EFAULT;
		}
		return ret;
L
Linus Torvalds 已提交
5148

E
Eric Dumazet 已提交
5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161
	case SIOCETHTOOL:
		dev_load(net, ifr.ifr_name);
		rtnl_lock();
		ret = dev_ethtool(net, &ifr);
		rtnl_unlock();
		if (!ret) {
			if (colon)
				*colon = ':';
			if (copy_to_user(arg, &ifr,
					 sizeof(struct ifreq)))
				ret = -EFAULT;
		}
		return ret;
L
Linus Torvalds 已提交
5162

E
Eric Dumazet 已提交
5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185
	/*
	 *	These ioctl calls:
	 *	- require superuser power.
	 *	- require strict serialization.
	 *	- return a value
	 */
	case SIOCGMIIPHY:
	case SIOCGMIIREG:
	case SIOCSIFNAME:
		if (!capable(CAP_NET_ADMIN))
			return -EPERM;
		dev_load(net, ifr.ifr_name);
		rtnl_lock();
		ret = dev_ifsioc(net, &ifr, cmd);
		rtnl_unlock();
		if (!ret) {
			if (colon)
				*colon = ':';
			if (copy_to_user(arg, &ifr,
					 sizeof(struct ifreq)))
				ret = -EFAULT;
		}
		return ret;
L
Linus Torvalds 已提交
5186

E
Eric Dumazet 已提交
5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228
	/*
	 *	These ioctl calls:
	 *	- require superuser power.
	 *	- require strict serialization.
	 *	- do not return a value
	 */
	case SIOCSIFFLAGS:
	case SIOCSIFMETRIC:
	case SIOCSIFMTU:
	case SIOCSIFMAP:
	case SIOCSIFHWADDR:
	case SIOCSIFSLAVE:
	case SIOCADDMULTI:
	case SIOCDELMULTI:
	case SIOCSIFHWBROADCAST:
	case SIOCSIFTXQLEN:
	case SIOCSMIIREG:
	case SIOCBONDENSLAVE:
	case SIOCBONDRELEASE:
	case SIOCBONDSETHWADDR:
	case SIOCBONDCHANGEACTIVE:
	case SIOCBRADDIF:
	case SIOCBRDELIF:
	case SIOCSHWTSTAMP:
		if (!capable(CAP_NET_ADMIN))
			return -EPERM;
		/* fall through */
	case SIOCBONDSLAVEINFOQUERY:
	case SIOCBONDINFOQUERY:
		dev_load(net, ifr.ifr_name);
		rtnl_lock();
		ret = dev_ifsioc(net, &ifr, cmd);
		rtnl_unlock();
		return ret;

	case SIOCGIFMEM:
		/* Get the per device memory space. We can add this but
		 * currently do not support it */
	case SIOCSIFMEM:
		/* Set the per device memory buffer space.
		 * Not applicable in our case */
	case SIOCSIFLINK:
5229
		return -ENOTTY;
E
Eric Dumazet 已提交
5230 5231 5232 5233 5234 5235 5236 5237

	/*
	 *	Unknown or private ioctl.
	 */
	default:
		if (cmd == SIOCWANDEV ||
		    (cmd >= SIOCDEVPRIVATE &&
		     cmd <= SIOCDEVPRIVATE + 15)) {
5238
			dev_load(net, ifr.ifr_name);
L
Linus Torvalds 已提交
5239
			rtnl_lock();
5240
			ret = dev_ifsioc(net, &ifr, cmd);
L
Linus Torvalds 已提交
5241
			rtnl_unlock();
E
Eric Dumazet 已提交
5242 5243 5244
			if (!ret && copy_to_user(arg, &ifr,
						 sizeof(struct ifreq)))
				ret = -EFAULT;
L
Linus Torvalds 已提交
5245
			return ret;
E
Eric Dumazet 已提交
5246 5247 5248 5249
		}
		/* Take care of Wireless Extensions */
		if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)
			return wext_handle_ioctl(net, &ifr, cmd, arg);
5250
		return -ENOTTY;
L
Linus Torvalds 已提交
5251 5252 5253 5254 5255 5256
	}
}


/**
 *	dev_new_index	-	allocate an ifindex
5257
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
5258 5259 5260 5261 5262
 *
 *	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.
 */
5263
static int dev_new_index(struct net *net)
L
Linus Torvalds 已提交
5264
{
5265
	int ifindex = net->ifindex;
L
Linus Torvalds 已提交
5266 5267 5268
	for (;;) {
		if (++ifindex <= 0)
			ifindex = 1;
5269
		if (!__dev_get_by_index(net, ifindex))
5270
			return net->ifindex = ifindex;
L
Linus Torvalds 已提交
5271 5272 5273 5274
	}
}

/* Delayed registration/unregisteration */
5275
static LIST_HEAD(net_todo_list);
L
Linus Torvalds 已提交
5276

5277
static void net_set_todo(struct net_device *dev)
L
Linus Torvalds 已提交
5278 5279 5280 5281
{
	list_add_tail(&dev->todo_list, &net_todo_list);
}

5282
static void rollback_registered_many(struct list_head *head)
5283
{
5284
	struct net_device *dev, *tmp;
5285

5286 5287 5288
	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

5289
	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
5290
		/* Some devices call without registering
5291 5292
		 * for initialization unwind. Remove those
		 * devices and proceed with the remaining.
5293 5294
		 */
		if (dev->reg_state == NETREG_UNINITIALIZED) {
5295 5296
			pr_debug("unregister_netdevice: device %s/%p never was registered\n",
				 dev->name, dev);
5297

5298
			WARN_ON(1);
5299 5300
			list_del(&dev->unreg_list);
			continue;
5301
		}
5302
		dev->dismantle = true;
5303
		BUG_ON(dev->reg_state != NETREG_REGISTERED);
5304
	}
5305

5306 5307
	/* If device is running, close it first. */
	dev_close_many(head);
5308

5309
	list_for_each_entry(dev, head, unreg_list) {
5310 5311
		/* And unlink it from device chain. */
		unlist_netdevice(dev);
5312

5313 5314
		dev->reg_state = NETREG_UNREGISTERING;
	}
5315 5316 5317

	synchronize_net();

5318 5319 5320
	list_for_each_entry(dev, head, unreg_list) {
		/* Shutdown queueing discipline. */
		dev_shutdown(dev);
5321 5322


5323 5324 5325 5326
		/* Notify protocols, that we are about to destroy
		   this device. They should clean all the things.
		*/
		call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5327

5328 5329 5330 5331
		if (!dev->rtnl_link_ops ||
		    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
			rtmsg_ifinfo(RTM_DELLINK, dev, ~0U);

5332 5333 5334
		/*
		 *	Flush the unicast and multicast chains
		 */
5335
		dev_uc_flush(dev);
5336
		dev_mc_flush(dev);
5337

5338 5339
		if (dev->netdev_ops->ndo_uninit)
			dev->netdev_ops->ndo_uninit(dev);
5340

5341 5342
		/* Notifier chain MUST detach us from master device. */
		WARN_ON(dev->master);
5343

5344 5345 5346
		/* Remove entries from kobject tree */
		netdev_unregister_kobject(dev);
	}
5347

5348
	synchronize_net();
5349

5350
	list_for_each_entry(dev, head, unreg_list)
5351 5352 5353 5354 5355 5356 5357 5358 5359
		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
5360
	list_del(&single);
5361 5362
}

5363 5364
static netdev_features_t netdev_fix_features(struct net_device *dev,
	netdev_features_t features)
5365
{
5366 5367 5368
	/* Fix illegal checksum combinations */
	if ((features & NETIF_F_HW_CSUM) &&
	    (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
5369
		netdev_warn(dev, "mixed HW and IP checksum settings.\n");
5370 5371 5372
		features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
	}

5373 5374 5375
	/* Fix illegal SG+CSUM combinations. */
	if ((features & NETIF_F_SG) &&
	    !(features & NETIF_F_ALL_CSUM)) {
5376 5377
		netdev_dbg(dev,
			"Dropping NETIF_F_SG since no checksum feature.\n");
5378 5379 5380 5381
		features &= ~NETIF_F_SG;
	}

	/* TSO requires that SG is present as well. */
5382
	if ((features & NETIF_F_ALL_TSO) && !(features & NETIF_F_SG)) {
5383
		netdev_dbg(dev, "Dropping TSO features since no SG feature.\n");
5384
		features &= ~NETIF_F_ALL_TSO;
5385 5386
	}

5387 5388 5389 5390
	/* TSO ECN requires that TSO is present as well. */
	if ((features & NETIF_F_ALL_TSO) == NETIF_F_TSO_ECN)
		features &= ~NETIF_F_TSO_ECN;

5391 5392
	/* Software GSO depends on SG. */
	if ((features & NETIF_F_GSO) && !(features & NETIF_F_SG)) {
5393
		netdev_dbg(dev, "Dropping NETIF_F_GSO since no SG feature.\n");
5394 5395 5396
		features &= ~NETIF_F_GSO;
	}

5397
	/* UFO needs SG and checksumming */
5398
	if (features & NETIF_F_UFO) {
5399 5400 5401 5402
		/* 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))) {
5403
			netdev_dbg(dev,
5404
				"Dropping NETIF_F_UFO since no checksum offload features.\n");
5405 5406 5407 5408
			features &= ~NETIF_F_UFO;
		}

		if (!(features & NETIF_F_SG)) {
5409
			netdev_dbg(dev,
5410
				"Dropping NETIF_F_UFO since no NETIF_F_SG feature.\n");
5411 5412 5413 5414 5415 5416 5417
			features &= ~NETIF_F_UFO;
		}
	}

	return features;
}

5418
int __netdev_update_features(struct net_device *dev)
5419
{
5420
	netdev_features_t features;
5421 5422
	int err = 0;

5423 5424
	ASSERT_RTNL();

5425 5426 5427 5428 5429 5430 5431 5432 5433
	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)
5434
		return 0;
5435

5436 5437
	netdev_dbg(dev, "Features changed: %pNF -> %pNF\n",
		&dev->features, &features);
5438 5439 5440 5441

	if (dev->netdev_ops->ndo_set_features)
		err = dev->netdev_ops->ndo_set_features(dev, features);

5442
	if (unlikely(err < 0)) {
5443
		netdev_err(dev,
5444 5445
			"set_features() failed (%d); wanted %pNF, left %pNF\n",
			err, &features, &dev->features);
5446 5447 5448 5449 5450 5451 5452 5453 5454
		return -1;
	}

	if (!err)
		dev->features = features;

	return 1;
}

5455 5456 5457 5458 5459 5460 5461 5462
/**
 *	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.
 */
5463 5464 5465 5466
void netdev_update_features(struct net_device *dev)
{
	if (__netdev_update_features(dev))
		netdev_features_change(dev);
5467 5468 5469
}
EXPORT_SYMBOL(netdev_update_features);

5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486
/**
 *	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);

5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513
/**
 *	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);

T
Tom Herbert 已提交
5514
#ifdef CONFIG_RPS
5515 5516 5517
static int netif_alloc_rx_queues(struct net_device *dev)
{
	unsigned int i, count = dev->num_rx_queues;
T
Tom Herbert 已提交
5518
	struct netdev_rx_queue *rx;
5519

T
Tom Herbert 已提交
5520
	BUG_ON(count < 1);
5521

T
Tom Herbert 已提交
5522 5523
	rx = kcalloc(count, sizeof(struct netdev_rx_queue), GFP_KERNEL);
	if (!rx) {
5524
		pr_err("netdev: Unable to allocate %u rx queues\n", count);
T
Tom Herbert 已提交
5525
		return -ENOMEM;
5526
	}
T
Tom Herbert 已提交
5527 5528 5529
	dev->_rx = rx;

	for (i = 0; i < count; i++)
T
Tom Herbert 已提交
5530
		rx[i].dev = dev;
5531 5532
	return 0;
}
T
Tom Herbert 已提交
5533
#endif
5534

C
Changli Gao 已提交
5535 5536 5537 5538 5539 5540 5541
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;
5542
	netdev_queue_numa_node_write(queue, NUMA_NO_NODE);
C
Changli Gao 已提交
5543
	queue->dev = dev;
T
Tom Herbert 已提交
5544 5545 5546
#ifdef CONFIG_BQL
	dql_init(&queue->dql, HZ);
#endif
C
Changli Gao 已提交
5547 5548
}

5549 5550 5551 5552 5553 5554 5555 5556 5557
static int netif_alloc_netdev_queues(struct net_device *dev)
{
	unsigned int count = dev->num_tx_queues;
	struct netdev_queue *tx;

	BUG_ON(count < 1);

	tx = kcalloc(count, sizeof(struct netdev_queue), GFP_KERNEL);
	if (!tx) {
5558
		pr_err("netdev: Unable to allocate %u tx queues\n", count);
5559 5560 5561
		return -ENOMEM;
	}
	dev->_tx = tx;
T
Tom Herbert 已提交
5562

5563 5564
	netdev_for_each_tx_queue(dev, netdev_init_one_queue, NULL);
	spin_lock_init(&dev->tx_global_lock);
C
Changli Gao 已提交
5565 5566

	return 0;
5567 5568
}

L
Linus Torvalds 已提交
5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588
/**
 *	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;
5589
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
5590 5591 5592 5593

	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

5594 5595
	might_sleep();

L
Linus Torvalds 已提交
5596 5597
	/* When net_device's are persistent, this will be fatal. */
	BUG_ON(dev->reg_state != NETREG_UNINITIALIZED);
5598
	BUG_ON(!net);
L
Linus Torvalds 已提交
5599

5600
	spin_lock_init(&dev->addr_list_lock);
5601
	netdev_set_addr_lockdep_class(dev);
L
Linus Torvalds 已提交
5602 5603 5604

	dev->iflink = -1;

5605
	ret = dev_get_valid_name(net, dev, dev->name);
5606 5607 5608
	if (ret < 0)
		goto out;

L
Linus Torvalds 已提交
5609
	/* Init, if this function is available */
5610 5611
	if (dev->netdev_ops->ndo_init) {
		ret = dev->netdev_ops->ndo_init(dev);
L
Linus Torvalds 已提交
5612 5613 5614
		if (ret) {
			if (ret > 0)
				ret = -EIO;
5615
			goto out;
L
Linus Torvalds 已提交
5616 5617
		}
	}
5618

5619 5620 5621 5622 5623 5624
	ret = -EBUSY;
	if (!dev->ifindex)
		dev->ifindex = dev_new_index(net);
	else if (__dev_get_by_index(net, dev->ifindex))
		goto err_uninit;

L
Linus Torvalds 已提交
5625 5626 5627
	if (dev->iflink == -1)
		dev->iflink = dev->ifindex;

5628 5629 5630 5631
	/* Transfer changeable features to wanted_features and enable
	 * software offloads (GSO and GRO).
	 */
	dev->hw_features |= NETIF_F_SOFT_FEATURES;
5632 5633
	dev->features |= NETIF_F_SOFT_FEATURES;
	dev->wanted_features = dev->features & dev->hw_features;
L
Linus Torvalds 已提交
5634

5635
	/* Turn on no cache copy if HW is doing checksum */
5636 5637 5638 5639 5640 5641
	if (!(dev->flags & IFF_LOOPBACK)) {
		dev->hw_features |= NETIF_F_NOCACHE_COPY;
		if (dev->features & NETIF_F_ALL_CSUM) {
			dev->wanted_features |= NETIF_F_NOCACHE_COPY;
			dev->features |= NETIF_F_NOCACHE_COPY;
		}
5642 5643
	}

5644
	/* Make NETIF_F_HIGHDMA inheritable to VLAN devices.
5645
	 */
5646
	dev->vlan_features |= NETIF_F_HIGHDMA;
5647

5648 5649 5650 5651 5652
	ret = call_netdevice_notifiers(NETDEV_POST_INIT, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		goto err_uninit;

5653
	ret = netdev_register_kobject(dev);
5654
	if (ret)
5655
		goto err_uninit;
5656 5657
	dev->reg_state = NETREG_REGISTERED;

5658
	__netdev_update_features(dev);
5659

L
Linus Torvalds 已提交
5660 5661 5662 5663 5664 5665 5666
	/*
	 *	Default initial state at registry is that the
	 *	device is present.
	 */

	set_bit(__LINK_STATE_PRESENT, &dev->state);

5667 5668
	linkwatch_init_dev(dev);

L
Linus Torvalds 已提交
5669 5670
	dev_init_scheduler(dev);
	dev_hold(dev);
5671
	list_netdevice(dev);
5672
	add_device_randomness(dev->dev_addr, dev->addr_len);
L
Linus Torvalds 已提交
5673 5674

	/* Notify protocols, that a new device appeared. */
5675
	ret = call_netdevice_notifiers(NETDEV_REGISTER, dev);
5676
	ret = notifier_to_errno(ret);
5677 5678 5679 5680
	if (ret) {
		rollback_registered(dev);
		dev->reg_state = NETREG_UNREGISTERED;
	}
5681 5682 5683 5684
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
5685 5686 5687
	if (!dev->rtnl_link_ops ||
	    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
		rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);
L
Linus Torvalds 已提交
5688 5689 5690

out:
	return ret;
5691 5692

err_uninit:
5693 5694
	if (dev->netdev_ops->ndo_uninit)
		dev->netdev_ops->ndo_uninit(dev);
5695
	goto out;
L
Linus Torvalds 已提交
5696
}
E
Eric Dumazet 已提交
5697
EXPORT_SYMBOL(register_netdevice);
L
Linus Torvalds 已提交
5698

5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729
/**
 *	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
5730 5731 5732 5733 5734
	/* Note : We dont allocate pcpu_refcnt for dummy devices,
	 * because users of this 'device' dont need to change
	 * its refcount.
	 */

5735 5736 5737 5738 5739
	return 0;
}
EXPORT_SYMBOL_GPL(init_dummy_netdev);


L
Linus Torvalds 已提交
5740 5741 5742 5743 5744 5745 5746 5747 5748
/**
 *	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.
 *
5749
 *	This is a wrapper around register_netdevice that takes the rtnl semaphore
L
Linus Torvalds 已提交
5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763
 *	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
5764 5765 5766 5767 5768 5769 5770 5771 5772 5773
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);

5774
/**
L
Linus Torvalds 已提交
5775
 * netdev_wait_allrefs - wait until all references are gone.
5776
 * @dev: target net_device
L
Linus Torvalds 已提交
5777 5778 5779 5780 5781 5782 5783
 *
 * 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
5784
 * call dev_put.
L
Linus Torvalds 已提交
5785 5786 5787 5788
 */
static void netdev_wait_allrefs(struct net_device *dev)
{
	unsigned long rebroadcast_time, warning_time;
E
Eric Dumazet 已提交
5789
	int refcnt;
L
Linus Torvalds 已提交
5790

5791 5792
	linkwatch_forget_dev(dev);

L
Linus Torvalds 已提交
5793
	rebroadcast_time = warning_time = jiffies;
E
Eric Dumazet 已提交
5794 5795 5796
	refcnt = netdev_refcnt_read(dev);

	while (refcnt != 0) {
L
Linus Torvalds 已提交
5797
		if (time_after(jiffies, rebroadcast_time + 1 * HZ)) {
5798
			rtnl_lock();
L
Linus Torvalds 已提交
5799 5800

			/* Rebroadcast unregister notification */
5801
			call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5802 5803

			__rtnl_unlock();
5804
			rcu_barrier();
5805 5806
			rtnl_lock();

5807
			call_netdevice_notifiers(NETDEV_UNREGISTER_FINAL, dev);
L
Linus Torvalds 已提交
5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818
			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();
			}

5819
			__rtnl_unlock();
L
Linus Torvalds 已提交
5820 5821 5822 5823 5824 5825

			rebroadcast_time = jiffies;
		}

		msleep(250);

E
Eric Dumazet 已提交
5826 5827
		refcnt = netdev_refcnt_read(dev);

L
Linus Torvalds 已提交
5828
		if (time_after(jiffies, warning_time + 10 * HZ)) {
5829 5830
			pr_emerg("unregister_netdevice: waiting for %s to become free. Usage count = %d\n",
				 dev->name, refcnt);
L
Linus Torvalds 已提交
5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848 5849
			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
5850
 * We are invoked by rtnl_unlock().
L
Linus Torvalds 已提交
5851
 * This allows us to deal with problems:
5852
 * 1) We can delete sysfs objects which invoke hotplug
L
Linus Torvalds 已提交
5853 5854 5855
 *    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
5856 5857 5858
 *
 * 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
5859 5860 5861
 */
void netdev_run_todo(void)
{
5862
	struct list_head list;
L
Linus Torvalds 已提交
5863 5864

	/* Snapshot list, allow later requests */
5865
	list_replace_init(&net_todo_list, &list);
H
Herbert Xu 已提交
5866 5867

	__rtnl_unlock();
5868

5869 5870

	/* Wait for rcu callbacks to finish before next phase */
5871 5872 5873
	if (!list_empty(&list))
		rcu_barrier();

L
Linus Torvalds 已提交
5874 5875
	while (!list_empty(&list)) {
		struct net_device *dev
5876
			= list_first_entry(&list, struct net_device, todo_list);
L
Linus Torvalds 已提交
5877 5878
		list_del(&dev->todo_list);

5879
		rtnl_lock();
5880
		call_netdevice_notifiers(NETDEV_UNREGISTER_FINAL, dev);
5881
		__rtnl_unlock();
5882

5883
		if (unlikely(dev->reg_state != NETREG_UNREGISTERING)) {
5884
			pr_err("network todo '%s' but state %d\n",
5885 5886 5887 5888
			       dev->name, dev->reg_state);
			dump_stack();
			continue;
		}
L
Linus Torvalds 已提交
5889

5890
		dev->reg_state = NETREG_UNREGISTERED;
L
Linus Torvalds 已提交
5891

5892
		on_each_cpu(flush_backlog, dev, 1);
5893

5894
		netdev_wait_allrefs(dev);
L
Linus Torvalds 已提交
5895

5896
		/* paranoia */
E
Eric Dumazet 已提交
5897
		BUG_ON(netdev_refcnt_read(dev));
5898 5899
		WARN_ON(rcu_access_pointer(dev->ip_ptr));
		WARN_ON(rcu_access_pointer(dev->ip6_ptr));
5900
		WARN_ON(dev->dn_ptr);
L
Linus Torvalds 已提交
5901

5902 5903
		if (dev->destructor)
			dev->destructor(dev);
5904 5905 5906

		/* Free network device */
		kobject_put(&dev->dev.kobj);
L
Linus Torvalds 已提交
5907 5908 5909
	}
}

5910 5911 5912
/* Convert net_device_stats to rtnl_link_stats64.  They have the same
 * fields in the same order, with only the type differing.
 */
5913 5914
void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
			     const struct net_device_stats *netdev_stats)
5915 5916
{
#if BITS_PER_LONG == 64
5917 5918
	BUILD_BUG_ON(sizeof(*stats64) != sizeof(*netdev_stats));
	memcpy(stats64, netdev_stats, sizeof(*stats64));
5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929
#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
}
5930
EXPORT_SYMBOL(netdev_stats_to_stats64);
5931

5932 5933 5934
/**
 *	dev_get_stats	- get network device statistics
 *	@dev: device to get statistics from
5935
 *	@storage: place to store stats
5936
 *
5937 5938 5939 5940
 *	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.
5941
 */
5942 5943
struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
					struct rtnl_link_stats64 *storage)
5944
{
5945 5946
	const struct net_device_ops *ops = dev->netdev_ops;

5947 5948
	if (ops->ndo_get_stats64) {
		memset(storage, 0, sizeof(*storage));
5949 5950
		ops->ndo_get_stats64(dev, storage);
	} else if (ops->ndo_get_stats) {
5951
		netdev_stats_to_stats64(storage, ops->ndo_get_stats(dev));
5952 5953
	} else {
		netdev_stats_to_stats64(storage, &dev->stats);
5954
	}
5955
	storage->rx_dropped += atomic_long_read(&dev->rx_dropped);
5956
	return storage;
R
Rusty Russell 已提交
5957
}
5958
EXPORT_SYMBOL(dev_get_stats);
R
Rusty Russell 已提交
5959

5960
struct netdev_queue *dev_ingress_queue_create(struct net_device *dev)
5961
{
5962
	struct netdev_queue *queue = dev_ingress_queue(dev);
5963

5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975
#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);
	queue->qdisc = &noop_qdisc;
	queue->qdisc_sleeping = &noop_qdisc;
	rcu_assign_pointer(dev->ingress_queue, queue);
#endif
	return queue;
5976 5977
}

5978 5979
static const struct ethtool_ops default_ethtool_ops;

L
Linus Torvalds 已提交
5980
/**
T
Tom Herbert 已提交
5981
 *	alloc_netdev_mqs - allocate network device
L
Linus Torvalds 已提交
5982 5983 5984
 *	@sizeof_priv:	size of private data to allocate space for
 *	@name:		device name format string
 *	@setup:		callback to initialize device
T
Tom Herbert 已提交
5985 5986
 *	@txqs:		the number of TX subqueues to allocate
 *	@rxqs:		the number of RX subqueues to allocate
L
Linus Torvalds 已提交
5987 5988
 *
 *	Allocates a struct net_device with private data area for driver use
5989
 *	and performs basic initialization.  Also allocates subquue structs
T
Tom Herbert 已提交
5990
 *	for each queue on the device.
L
Linus Torvalds 已提交
5991
 */
T
Tom Herbert 已提交
5992 5993 5994
struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
		void (*setup)(struct net_device *),
		unsigned int txqs, unsigned int rxqs)
L
Linus Torvalds 已提交
5995 5996
{
	struct net_device *dev;
5997
	size_t alloc_size;
5998
	struct net_device *p;
L
Linus Torvalds 已提交
5999

6000 6001
	BUG_ON(strlen(name) >= sizeof(dev->name));

T
Tom Herbert 已提交
6002
	if (txqs < 1) {
6003
		pr_err("alloc_netdev: Unable to allocate device with zero queues\n");
6004 6005 6006
		return NULL;
	}

T
Tom Herbert 已提交
6007 6008
#ifdef CONFIG_RPS
	if (rxqs < 1) {
6009
		pr_err("alloc_netdev: Unable to allocate device with zero RX queues\n");
T
Tom Herbert 已提交
6010 6011 6012 6013
		return NULL;
	}
#endif

6014
	alloc_size = sizeof(struct net_device);
6015 6016
	if (sizeof_priv) {
		/* ensure 32-byte alignment of private area */
6017
		alloc_size = ALIGN(alloc_size, NETDEV_ALIGN);
6018 6019 6020
		alloc_size += sizeof_priv;
	}
	/* ensure 32-byte alignment of whole construct */
6021
	alloc_size += NETDEV_ALIGN - 1;
L
Linus Torvalds 已提交
6022

6023
	p = kzalloc(alloc_size, GFP_KERNEL);
L
Linus Torvalds 已提交
6024
	if (!p) {
6025
		pr_err("alloc_netdev: Unable to allocate device\n");
L
Linus Torvalds 已提交
6026 6027 6028
		return NULL;
	}

6029
	dev = PTR_ALIGN(p, NETDEV_ALIGN);
L
Linus Torvalds 已提交
6030
	dev->padded = (char *)dev - (char *)p;
6031

E
Eric Dumazet 已提交
6032 6033
	dev->pcpu_refcnt = alloc_percpu(int);
	if (!dev->pcpu_refcnt)
6034
		goto free_p;
6035 6036

	if (dev_addr_init(dev))
E
Eric Dumazet 已提交
6037
		goto free_pcpu;
6038

6039
	dev_mc_init(dev);
6040
	dev_uc_init(dev);
J
Jiri Pirko 已提交
6041

6042
	dev_net_set(dev, &init_net);
L
Linus Torvalds 已提交
6043

6044
	dev->gso_max_size = GSO_MAX_SIZE;
6045
	dev->gso_max_segs = GSO_MAX_SEGS;
6046 6047 6048 6049 6050 6051 6052

	INIT_LIST_HEAD(&dev->napi_list);
	INIT_LIST_HEAD(&dev->unreg_list);
	INIT_LIST_HEAD(&dev->link_watch_list);
	dev->priv_flags = IFF_XMIT_DST_RELEASE;
	setup(dev);

T
Tom Herbert 已提交
6053 6054
	dev->num_tx_queues = txqs;
	dev->real_num_tx_queues = txqs;
6055
	if (netif_alloc_netdev_queues(dev))
6056
		goto free_all;
6057

E
Eric Dumazet 已提交
6058
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
6059 6060
	dev->num_rx_queues = rxqs;
	dev->real_num_rx_queues = rxqs;
T
Tom Herbert 已提交
6061
	if (netif_alloc_rx_queues(dev))
6062
		goto free_all;
E
Eric Dumazet 已提交
6063
#endif
T
Tom Herbert 已提交
6064

L
Linus Torvalds 已提交
6065
	strcpy(dev->name, name);
6066
	dev->group = INIT_NETDEV_GROUP;
6067 6068
	if (!dev->ethtool_ops)
		dev->ethtool_ops = &default_ethtool_ops;
L
Linus Torvalds 已提交
6069
	return dev;
6070

6071 6072 6073 6074
free_all:
	free_netdev(dev);
	return NULL;

E
Eric Dumazet 已提交
6075 6076
free_pcpu:
	free_percpu(dev->pcpu_refcnt);
6077
	kfree(dev->_tx);
T
Tom Herbert 已提交
6078 6079 6080 6081
#ifdef CONFIG_RPS
	kfree(dev->_rx);
#endif

6082 6083 6084
free_p:
	kfree(p);
	return NULL;
L
Linus Torvalds 已提交
6085
}
T
Tom Herbert 已提交
6086
EXPORT_SYMBOL(alloc_netdev_mqs);
L
Linus Torvalds 已提交
6087 6088 6089 6090 6091

/**
 *	free_netdev - free network device
 *	@dev: device
 *
6092 6093
 *	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
6094 6095 6096 6097
 *	If this is the last reference then it will be freed.
 */
void free_netdev(struct net_device *dev)
{
6098 6099
	struct napi_struct *p, *n;

6100 6101
	release_net(dev_net(dev));

6102
	kfree(dev->_tx);
T
Tom Herbert 已提交
6103 6104 6105
#ifdef CONFIG_RPS
	kfree(dev->_rx);
#endif
6106

6107
	kfree(rcu_dereference_protected(dev->ingress_queue, 1));
6108

6109 6110 6111
	/* Flush device addresses */
	dev_addr_flush(dev);

6112 6113 6114
	list_for_each_entry_safe(p, n, &dev->napi_list, dev_list)
		netif_napi_del(p);

E
Eric Dumazet 已提交
6115 6116 6117
	free_percpu(dev->pcpu_refcnt);
	dev->pcpu_refcnt = NULL;

S
Stephen Hemminger 已提交
6118
	/*  Compatibility with error handling in drivers */
L
Linus Torvalds 已提交
6119 6120 6121 6122 6123 6124 6125 6126
	if (dev->reg_state == NETREG_UNINITIALIZED) {
		kfree((char *)dev - dev->padded);
		return;
	}

	BUG_ON(dev->reg_state != NETREG_UNREGISTERED);
	dev->reg_state = NETREG_RELEASED;

6127 6128
	/* will free via device release */
	put_device(&dev->dev);
L
Linus Torvalds 已提交
6129
}
E
Eric Dumazet 已提交
6130
EXPORT_SYMBOL(free_netdev);
6131

6132 6133 6134 6135 6136 6137
/**
 *	synchronize_net -  Synchronize with packet receive processing
 *
 *	Wait for packets currently being received to be done.
 *	Does not block later packets from starting.
 */
6138
void synchronize_net(void)
L
Linus Torvalds 已提交
6139 6140
{
	might_sleep();
6141 6142 6143 6144
	if (rtnl_is_locked())
		synchronize_rcu_expedited();
	else
		synchronize_rcu();
L
Linus Torvalds 已提交
6145
}
E
Eric Dumazet 已提交
6146
EXPORT_SYMBOL(synchronize_net);
L
Linus Torvalds 已提交
6147 6148

/**
6149
 *	unregister_netdevice_queue - remove device from the kernel
L
Linus Torvalds 已提交
6150
 *	@dev: device
6151
 *	@head: list
6152
 *
L
Linus Torvalds 已提交
6153
 *	This function shuts down a device interface and removes it
6154
 *	from the kernel tables.
6155
 *	If head not NULL, device is queued to be unregistered later.
L
Linus Torvalds 已提交
6156 6157 6158 6159 6160
 *
 *	Callers must hold the rtnl semaphore.  You may want
 *	unregister_netdev() instead of this.
 */

6161
void unregister_netdevice_queue(struct net_device *dev, struct list_head *head)
L
Linus Torvalds 已提交
6162
{
6163 6164
	ASSERT_RTNL();

6165
	if (head) {
6166
		list_move_tail(&dev->unreg_list, head);
6167 6168 6169 6170 6171
	} else {
		rollback_registered(dev);
		/* Finish processing unregister after unlock */
		net_set_todo(dev);
	}
L
Linus Torvalds 已提交
6172
}
6173
EXPORT_SYMBOL(unregister_netdevice_queue);
L
Linus Torvalds 已提交
6174

6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188
/**
 *	unregister_netdevice_many - unregister many devices
 *	@head: list of devices
 */
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);
	}
}
6189
EXPORT_SYMBOL(unregister_netdevice_many);
6190

L
Linus Torvalds 已提交
6191 6192 6193 6194 6195
/**
 *	unregister_netdev - remove device from the kernel
 *	@dev: device
 *
 *	This function shuts down a device interface and removes it
6196
 *	from the kernel tables.
L
Linus Torvalds 已提交
6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209
 *
 *	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);

6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240 6241
/**
 *	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 */
	err = -EINVAL;
	if (dev->reg_state != NETREG_REGISTERED)
		goto out;

	/* Get out if there is nothing todo */
	err = 0;
6242
	if (net_eq(dev_net(dev), net))
6243 6244 6245 6246 6247 6248
		goto out;

	/* Pick the destination device name, and ensure
	 * we can use it in the destination network namespace.
	 */
	err = -EEXIST;
6249
	if (__dev_get_by_name(net, dev->name)) {
6250 6251 6252
		/* We get here if we can't use the current device name */
		if (!pat)
			goto out;
6253
		if (dev_get_valid_name(net, dev, pat) < 0)
6254 6255 6256 6257 6258 6259 6260 6261
			goto out;
	}

	/*
	 * And now a mini version of register_netdevice unregister_netdevice.
	 */

	/* If device is running close it first. */
6262
	dev_close(dev);
6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274

	/* 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.
6275 6276 6277 6278

	   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.
6279 6280
	*/
	call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
6281 6282
	rcu_barrier();
	call_netdevice_notifiers(NETDEV_UNREGISTER_FINAL, dev);
6283
	rtmsg_ifinfo(RTM_DELLINK, dev, ~0U);
6284 6285 6286 6287

	/*
	 *	Flush the unicast and multicast chains
	 */
6288
	dev_uc_flush(dev);
6289
	dev_mc_flush(dev);
6290 6291

	/* Actually switch the network namespace */
6292
	dev_net_set(dev, net);
6293 6294 6295 6296 6297 6298 6299 6300 6301

	/* If there is an ifindex conflict assign a new one */
	if (__dev_get_by_index(net, dev->ifindex)) {
		int iflink = (dev->iflink == dev->ifindex);
		dev->ifindex = dev_new_index(net);
		if (iflink)
			dev->iflink = dev->ifindex;
	}

6302
	/* Fixup kobjects */
6303
	err = device_rename(&dev->dev, dev->name);
6304
	WARN_ON(err);
6305 6306 6307 6308 6309 6310 6311

	/* Add the device back in the hashes */
	list_netdevice(dev);

	/* Notify protocols, that a new device appeared. */
	call_netdevice_notifiers(NETDEV_REGISTER, dev);

6312 6313 6314 6315 6316 6317
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
	rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);

6318 6319 6320 6321 6322
	synchronize_net();
	err = 0;
out:
	return err;
}
6323
EXPORT_SYMBOL_GPL(dev_change_net_namespace);
6324

L
Linus Torvalds 已提交
6325 6326 6327 6328 6329 6330 6331 6332 6333
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;

6334
	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN)
L
Linus Torvalds 已提交
6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350
		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. */
6351 6352 6353 6354 6355 6356
	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;
	}
6357 6358 6359 6360 6361
	/* Append NAPI poll list from offline CPU. */
	if (!list_empty(&oldsd->poll_list)) {
		list_splice_init(&oldsd->poll_list, &sd->poll_list);
		raise_softirq_irqoff(NET_RX_SOFTIRQ);
	}
L
Linus Torvalds 已提交
6362 6363 6364 6365 6366

	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_enable();

	/* Process offline CPU's input_pkt_queue */
6367
	while ((skb = __skb_dequeue(&oldsd->process_queue))) {
L
Linus Torvalds 已提交
6368
		netif_rx(skb);
6369
		input_queue_head_incr(oldsd);
T
Tom Herbert 已提交
6370
	}
6371
	while ((skb = __skb_dequeue(&oldsd->input_pkt_queue))) {
6372
		netif_rx(skb);
6373 6374
		input_queue_head_incr(oldsd);
	}
L
Linus Torvalds 已提交
6375 6376 6377 6378 6379

	return NOTIFY_OK;
}


6380
/**
6381 6382 6383 6384
 *	netdev_increment_features - increment feature set by one
 *	@all: current feature set
 *	@one: new feature set
 *	@mask: mask feature set
6385 6386
 *
 *	Computes a new feature set after adding a device with feature set
6387 6388
 *	@one to the master device with current feature set @all.  Will not
 *	enable anything that is off in @mask. Returns the new feature set.
6389
 */
6390 6391
netdev_features_t netdev_increment_features(netdev_features_t all,
	netdev_features_t one, netdev_features_t mask)
6392
{
6393 6394 6395
	if (mask & NETIF_F_GEN_CSUM)
		mask |= NETIF_F_ALL_CSUM;
	mask |= NETIF_F_VLAN_CHALLENGED;
6396

6397 6398
	all |= one & (NETIF_F_ONE_FOR_ALL|NETIF_F_ALL_CSUM) & mask;
	all &= one | ~NETIF_F_ALL_FOR_ALL;
6399

6400 6401 6402
	/* If one device supports hw checksumming, set for all. */
	if (all & NETIF_F_GEN_CSUM)
		all &= ~(NETIF_F_ALL_CSUM & ~NETIF_F_GEN_CSUM);
6403 6404 6405

	return all;
}
6406
EXPORT_SYMBOL(netdev_increment_features);
6407

6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420
static struct hlist_head *netdev_create_hash(void)
{
	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;
}

6421
/* Initialize per network namespace state */
6422
static int __net_init netdev_init(struct net *net)
6423
{
6424 6425
	if (net != &init_net)
		INIT_LIST_HEAD(&net->dev_base_head);
6426

6427 6428 6429
	net->dev_name_head = netdev_create_hash();
	if (net->dev_name_head == NULL)
		goto err_name;
6430

6431 6432 6433
	net->dev_index_head = netdev_create_hash();
	if (net->dev_index_head == NULL)
		goto err_idx;
6434 6435

	return 0;
6436 6437 6438 6439 6440

err_idx:
	kfree(net->dev_name_head);
err_name:
	return -ENOMEM;
6441 6442
}

6443 6444 6445 6446 6447 6448
/**
 *	netdev_drivername - network driver for the device
 *	@dev: network device
 *
 *	Determine network driver for device.
 */
6449
const char *netdev_drivername(const struct net_device *dev)
6450
{
6451 6452
	const struct device_driver *driver;
	const struct device *parent;
6453
	const char *empty = "";
6454 6455 6456

	parent = dev->dev.parent;
	if (!parent)
6457
		return empty;
6458 6459 6460

	driver = parent->driver;
	if (driver && driver->name)
6461 6462
		return driver->name;
	return empty;
6463 6464
}

6465
int __netdev_printk(const char *level, const struct net_device *dev,
6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479
			   struct va_format *vaf)
{
	int r;

	if (dev && dev->dev.parent)
		r = dev_printk(level, dev->dev.parent, "%s: %pV",
			       netdev_name(dev), vaf);
	else if (dev)
		r = printk("%s%s: %pV", level, netdev_name(dev), vaf);
	else
		r = printk("%s(NULL net_device): %pV", level, vaf);

	return r;
}
6480
EXPORT_SYMBOL(__netdev_printk);
6481 6482 6483 6484 6485 6486 6487 6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527

int netdev_printk(const char *level, const struct net_device *dev,
		  const char *format, ...)
{
	struct va_format vaf;
	va_list args;
	int r;

	va_start(args, format);

	vaf.fmt = format;
	vaf.va = &args;

	r = __netdev_printk(level, dev, &vaf);
	va_end(args);

	return r;
}
EXPORT_SYMBOL(netdev_printk);

#define define_netdev_printk_level(func, level)			\
int func(const struct net_device *dev, const char *fmt, ...)	\
{								\
	int r;							\
	struct va_format vaf;					\
	va_list args;						\
								\
	va_start(args, fmt);					\
								\
	vaf.fmt = fmt;						\
	vaf.va = &args;						\
								\
	r = __netdev_printk(level, dev, &vaf);			\
	va_end(args);						\
								\
	return r;						\
}								\
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);

6528
static void __net_exit netdev_exit(struct net *net)
6529 6530 6531 6532 6533
{
	kfree(net->dev_name_head);
	kfree(net->dev_index_head);
}

6534
static struct pernet_operations __net_initdata netdev_net_ops = {
6535 6536 6537 6538
	.init = netdev_init,
	.exit = netdev_exit,
};

6539
static void __net_exit default_device_exit(struct net *net)
6540
{
6541
	struct net_device *dev, *aux;
6542
	/*
6543
	 * Push all migratable network devices back to the
6544 6545 6546
	 * initial network namespace
	 */
	rtnl_lock();
6547
	for_each_netdev_safe(net, dev, aux) {
6548
		int err;
6549
		char fb_name[IFNAMSIZ];
6550 6551 6552 6553 6554

		/* Ignore unmoveable devices (i.e. loopback) */
		if (dev->features & NETIF_F_NETNS_LOCAL)
			continue;

6555 6556 6557
		/* Leave virtual devices for the generic cleanup */
		if (dev->rtnl_link_ops)
			continue;
6558

L
Lucas De Marchi 已提交
6559
		/* Push remaining network devices to init_net */
6560 6561
		snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
		err = dev_change_net_namespace(dev, &init_net, fb_name);
6562
		if (err) {
6563 6564
			pr_emerg("%s: failed to move %s to init_net: %d\n",
				 __func__, dev->name, err);
6565
			BUG();
6566 6567 6568 6569 6570
		}
	}
	rtnl_unlock();
}

6571 6572 6573
static void __net_exit default_device_exit_batch(struct list_head *net_list)
{
	/* At exit all network devices most be removed from a network
6574
	 * namespace.  Do this in the reverse order of registration.
6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591
	 * 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);

	rtnl_lock();
	list_for_each_entry(net, net_list, exit_list) {
		for_each_netdev_reverse(net, dev) {
			if (dev->rtnl_link_ops)
				dev->rtnl_link_ops->dellink(dev, &dev_kill_list);
			else
				unregister_netdevice_queue(dev, &dev_kill_list);
		}
	}
	unregister_netdevice_many(&dev_kill_list);
E
Eric Dumazet 已提交
6592
	list_del(&dev_kill_list);
6593 6594 6595
	rtnl_unlock();
}

6596
static struct pernet_operations __net_initdata default_device_ops = {
6597
	.exit = default_device_exit,
6598
	.exit_batch = default_device_exit_batch,
6599 6600
};

L
Linus Torvalds 已提交
6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614 6615 6616 6617 6618 6619 6620
/*
 *	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;

6621
	if (netdev_kobject_init())
L
Linus Torvalds 已提交
6622 6623 6624
		goto out;

	INIT_LIST_HEAD(&ptype_all);
6625
	for (i = 0; i < PTYPE_HASH_SIZE; i++)
L
Linus Torvalds 已提交
6626 6627
		INIT_LIST_HEAD(&ptype_base[i]);

6628 6629
	if (register_pernet_subsys(&netdev_net_ops))
		goto out;
L
Linus Torvalds 已提交
6630 6631 6632 6633 6634

	/*
	 *	Initialise the packet receive queues.
	 */

6635
	for_each_possible_cpu(i) {
E
Eric Dumazet 已提交
6636
		struct softnet_data *sd = &per_cpu(softnet_data, i);
L
Linus Torvalds 已提交
6637

C
Changli Gao 已提交
6638
		memset(sd, 0, sizeof(*sd));
E
Eric Dumazet 已提交
6639
		skb_queue_head_init(&sd->input_pkt_queue);
6640
		skb_queue_head_init(&sd->process_queue);
E
Eric Dumazet 已提交
6641 6642
		sd->completion_queue = NULL;
		INIT_LIST_HEAD(&sd->poll_list);
6643 6644
		sd->output_queue = NULL;
		sd->output_queue_tailp = &sd->output_queue;
E
Eric Dumazet 已提交
6645
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
6646 6647 6648 6649
		sd->csd.func = rps_trigger_softirq;
		sd->csd.info = sd;
		sd->csd.flags = 0;
		sd->cpu = i;
6650
#endif
T
Tom Herbert 已提交
6651

E
Eric Dumazet 已提交
6652 6653 6654 6655
		sd->backlog.poll = process_backlog;
		sd->backlog.weight = weight_p;
		sd->backlog.gro_list = NULL;
		sd->backlog.gro_count = 0;
L
Linus Torvalds 已提交
6656 6657 6658 6659
	}

	dev_boot_phase = 0;

6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674
	/* 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;

6675 6676
	open_softirq(NET_TX_SOFTIRQ, net_tx_action);
	open_softirq(NET_RX_SOFTIRQ, net_rx_action);
L
Linus Torvalds 已提交
6677 6678 6679 6680 6681 6682 6683 6684 6685 6686 6687

	hotcpu_notifier(dev_cpu_callback, 0);
	dst_init();
	dev_mcast_init();
	rc = 0;
out:
	return rc;
}

subsys_initcall(net_dev_init);

6688 6689
static int __init initialize_hashrnd(void)
{
T
Tom Herbert 已提交
6690
	get_random_bytes(&hashrnd, sizeof(hashrnd));
6691 6692 6693 6694 6695
	return 0;
}

late_initcall_sync(initialize_hashrnd);