dev.c 150.7 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 77
 *				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 <asm/system.h>
#include <linux/bitops.h>
78
#include <linux/capability.h>
L
Linus Torvalds 已提交
79 80 81
#include <linux/cpu.h>
#include <linux/types.h>
#include <linux/kernel.h>
82
#include <linux/hash.h>
83
#include <linux/slab.h>
L
Linus Torvalds 已提交
84
#include <linux/sched.h>
A
Arjan van de Ven 已提交
85
#include <linux/mutex.h>
L
Linus Torvalds 已提交
86 87 88 89 90 91 92 93 94
#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>
95
#include <linux/ethtool.h>
L
Linus Torvalds 已提交
96 97
#include <linux/notifier.h>
#include <linux/skbuff.h>
98
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
99 100 101 102 103 104 105 106
#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>
107
#include <net/xfrm.h>
L
Linus Torvalds 已提交
108 109 110 111 112 113 114
#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>
115
#include <net/wext.h>
L
Linus Torvalds 已提交
116 117
#include <net/iw_handler.h>
#include <asm/current.h>
S
Steve Grubb 已提交
118
#include <linux/audit.h>
119
#include <linux/dmaengine.h>
120
#include <linux/err.h>
121
#include <linux/ctype.h>
122
#include <linux/if_arp.h>
123
#include <linux/if_vlan.h>
124
#include <linux/ip.h>
125
#include <net/ip.h>
126 127
#include <linux/ipv6.h>
#include <linux/in.h>
D
David S. Miller 已提交
128 129
#include <linux/jhash.h>
#include <linux/random.h>
130
#include <trace/events/napi.h>
131
#include <linux/pci.h>
132
#include <linux/inetdevice.h>
L
Linus Torvalds 已提交
133

134 135
#include "net-sysfs.h"

136 137 138
/* Instead of increasing this, you should create a hash table. */
#define MAX_GRO_SKBS 8

H
Herbert Xu 已提交
139 140 141
/* 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
142 143 144 145 146 147 148 149 150 151 152
/*
 *	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
153
 *             the average user (w/out VLANs) will not be adversely affected.
L
Linus Torvalds 已提交
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169
 *             --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
 */

170 171 172
#define PTYPE_HASH_SIZE	(16)
#define PTYPE_HASH_MASK	(PTYPE_HASH_SIZE - 1)

L
Linus Torvalds 已提交
173
static DEFINE_SPINLOCK(ptype_lock);
174
static struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
175
static struct list_head ptype_all __read_mostly;	/* Taps */
L
Linus Torvalds 已提交
176 177

/*
178
 * The @dev_base_head list is protected by @dev_base_lock and the rtnl
L
Linus Torvalds 已提交
179 180
 * semaphore.
 *
181
 * Pure readers hold dev_base_lock for reading, or rcu_read_lock()
L
Linus Torvalds 已提交
182 183
 *
 * Writers must hold the rtnl semaphore while they loop through the
184
 * dev_base_head list, and hold dev_base_lock for writing when they do the
L
Linus Torvalds 已提交
185 186 187 188 189 190 191 192 193 194 195 196 197 198
 * 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);

199
static inline struct hlist_head *dev_name_hash(struct net *net, const char *name)
L
Linus Torvalds 已提交
200 201
{
	unsigned hash = full_name_hash(name, strnlen(name, IFNAMSIZ));
202
	return &net->dev_name_head[hash_32(hash, NETDEV_HASHBITS)];
L
Linus Torvalds 已提交
203 204
}

205
static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex)
L
Linus Torvalds 已提交
206
{
207
	return &net->dev_index_head[ifindex & (NETDEV_HASHENTRIES - 1)];
L
Linus Torvalds 已提交
208 209
}

E
Eric Dumazet 已提交
210
static inline void rps_lock(struct softnet_data *sd)
211 212
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
213
	spin_lock(&sd->input_pkt_queue.lock);
214 215 216
#endif
}

E
Eric Dumazet 已提交
217
static inline void rps_unlock(struct softnet_data *sd)
218 219
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
220
	spin_unlock(&sd->input_pkt_queue.lock);
221 222 223
#endif
}

224 225 226
/* Device list insertion */
static int list_netdevice(struct net_device *dev)
{
227
	struct net *net = dev_net(dev);
228 229 230 231

	ASSERT_RTNL();

	write_lock_bh(&dev_base_lock);
232
	list_add_tail_rcu(&dev->dev_list, &net->dev_base_head);
233
	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net, dev->name));
234 235
	hlist_add_head_rcu(&dev->index_hlist,
			   dev_index_hash(net, dev->ifindex));
236 237 238 239
	write_unlock_bh(&dev_base_lock);
	return 0;
}

240 241 242
/* Device list removal
 * caller must respect a RCU grace period before freeing/reusing dev
 */
243 244 245 246 247 248
static void unlist_netdevice(struct net_device *dev)
{
	ASSERT_RTNL();

	/* Unlink dev from the device chain */
	write_lock_bh(&dev_base_lock);
249
	list_del_rcu(&dev->dev_list);
250
	hlist_del_rcu(&dev->name_hlist);
251
	hlist_del_rcu(&dev->index_hlist);
252 253 254
	write_unlock_bh(&dev_base_lock);
}

L
Linus Torvalds 已提交
255 256 257 258
/*
 *	Our notifier list
 */

259
static RAW_NOTIFIER_HEAD(netdev_chain);
L
Linus Torvalds 已提交
260 261 262 263 264

/*
 *	Device drivers call our routines to queue packets here. We empty the
 *	queue in the local softnet handler.
 */
265

266
DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
E
Eric Dumazet 已提交
267
EXPORT_PER_CPU_SYMBOL(softnet_data);
L
Linus Torvalds 已提交
268

269
#ifdef CONFIG_LOCKDEP
270
/*
271
 * register_netdevice() inits txq->_xmit_lock and sets lockdep class
272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
 * 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,
	 ARPHRD_FCFABRIC, ARPHRD_IEEE802_TR, ARPHRD_IEEE80211,
288
	 ARPHRD_IEEE80211_PRISM, ARPHRD_IEEE80211_RADIOTAP, ARPHRD_PHONET,
289
	 ARPHRD_PHONET_PIPE, ARPHRD_IEEE802154,
290
	 ARPHRD_VOID, ARPHRD_NONE};
291

292
static const char *const netdev_lock_name[] =
293 294 295 296 297 298 299 300 301 302 303 304 305
	{"_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",
	 "_xmit_FCFABRIC", "_xmit_IEEE802_TR", "_xmit_IEEE80211",
306
	 "_xmit_IEEE80211_PRISM", "_xmit_IEEE80211_RADIOTAP", "_xmit_PHONET",
307
	 "_xmit_PHONET_PIPE", "_xmit_IEEE802154",
308
	 "_xmit_VOID", "_xmit_NONE"};
309 310

static struct lock_class_key netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)];
311
static struct lock_class_key netdev_addr_lock_key[ARRAY_SIZE(netdev_lock_type)];
312 313 314 315 316 317 318 319 320 321 322 323

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;
}

324 325
static inline void netdev_set_xmit_lockdep_class(spinlock_t *lock,
						 unsigned short dev_type)
326 327 328 329 330 331 332
{
	int i;

	i = netdev_lock_pos(dev_type);
	lockdep_set_class_and_name(lock, &netdev_xmit_lock_key[i],
				   netdev_lock_name[i]);
}
333 334 335 336 337 338 339 340 341 342

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]);
}
343
#else
344 345 346 347 348
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)
349 350 351
{
}
#endif
L
Linus Torvalds 已提交
352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374

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

		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)
 */

375 376 377 378 379 380 381 382
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
383 384 385 386 387 388 389 390
/**
 *	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.
 *
391
 *	This call does not sleep therefore it can not
L
Linus Torvalds 已提交
392 393 394 395 396 397
 *	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)
{
398
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
399

400 401 402
	spin_lock(&ptype_lock);
	list_add_rcu(&pt->list, head);
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
403
}
E
Eric Dumazet 已提交
404
EXPORT_SYMBOL(dev_add_pack);
L
Linus Torvalds 已提交
405 406 407 408 409 410 411 412

/**
 *	__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
413
 *	returns.
L
Linus Torvalds 已提交
414 415 416 417 418 419 420
 *
 *      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)
{
421
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
422 423
	struct packet_type *pt1;

424
	spin_lock(&ptype_lock);
L
Linus Torvalds 已提交
425 426 427 428 429 430 431 432 433 434

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

	printk(KERN_WARNING "dev_remove_pack: %p not found.\n", pt);
out:
435
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
436
}
E
Eric Dumazet 已提交
437 438
EXPORT_SYMBOL(__dev_remove_pack);

L
Linus Torvalds 已提交
439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
/**
 *	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);
454

L
Linus Torvalds 已提交
455 456
	synchronize_net();
}
E
Eric Dumazet 已提交
457
EXPORT_SYMBOL(dev_remove_pack);
L
Linus Torvalds 已提交
458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485

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

		      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));
486
			strlcpy(s[i].name, name, IFNAMSIZ);
L
Linus Torvalds 已提交
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
			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] != ' ' &&
511
		    !strcmp(dev->name, s[i].name)) {
L
Linus Torvalds 已提交
512 513 514 515 516 517 518 519 520
			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
521
EXPORT_SYMBOL(netdev_boot_setup_check);
L
Linus Torvalds 已提交
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545


/**
 *	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
	 */
546
	if (__dev_get_by_name(&init_net, name))
L
Linus Torvalds 已提交
547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
		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
592
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
593 594 595 596 597 598 599 600 601
 *	@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.
 */

602
struct net_device *__dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
603 604
{
	struct hlist_node *p;
E
Eric Dumazet 已提交
605 606
	struct net_device *dev;
	struct hlist_head *head = dev_name_hash(net, name);
L
Linus Torvalds 已提交
607

E
Eric Dumazet 已提交
608
	hlist_for_each_entry(dev, p, head, name_hlist)
L
Linus Torvalds 已提交
609 610
		if (!strncmp(dev->name, name, IFNAMSIZ))
			return dev;
E
Eric Dumazet 已提交
611

L
Linus Torvalds 已提交
612 613
	return NULL;
}
E
Eric Dumazet 已提交
614
EXPORT_SYMBOL(__dev_get_by_name);
L
Linus Torvalds 已提交
615

616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641
/**
 *	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
642 643
/**
 *	dev_get_by_name		- find a device by its name
644
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
645 646 647 648 649 650 651 652 653
 *	@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.
 */

654
struct net_device *dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
655 656 657
{
	struct net_device *dev;

658 659
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
L
Linus Torvalds 已提交
660 661
	if (dev)
		dev_hold(dev);
662
	rcu_read_unlock();
L
Linus Torvalds 已提交
663 664
	return dev;
}
E
Eric Dumazet 已提交
665
EXPORT_SYMBOL(dev_get_by_name);
L
Linus Torvalds 已提交
666 667 668

/**
 *	__dev_get_by_index - find a device by its ifindex
669
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
670 671 672 673 674 675 676 677 678
 *	@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.
 */

679
struct net_device *__dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
680 681
{
	struct hlist_node *p;
E
Eric Dumazet 已提交
682 683
	struct net_device *dev;
	struct hlist_head *head = dev_index_hash(net, ifindex);
L
Linus Torvalds 已提交
684

E
Eric Dumazet 已提交
685
	hlist_for_each_entry(dev, p, head, index_hlist)
L
Linus Torvalds 已提交
686 687
		if (dev->ifindex == ifindex)
			return dev;
E
Eric Dumazet 已提交
688

L
Linus Torvalds 已提交
689 690
	return NULL;
}
E
Eric Dumazet 已提交
691
EXPORT_SYMBOL(__dev_get_by_index);
L
Linus Torvalds 已提交
692

693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
/**
 *	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
718 719 720

/**
 *	dev_get_by_index - find a device by its ifindex
721
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
722 723 724 725 726 727 728 729
 *	@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.
 */

730
struct net_device *dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
731 732 733
{
	struct net_device *dev;

734 735
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifindex);
L
Linus Torvalds 已提交
736 737
	if (dev)
		dev_hold(dev);
738
	rcu_read_unlock();
L
Linus Torvalds 已提交
739 740
	return dev;
}
E
Eric Dumazet 已提交
741
EXPORT_SYMBOL(dev_get_by_index);
L
Linus Torvalds 已提交
742 743 744

/**
 *	dev_getbyhwaddr - find a device by its hardware address
745
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
746 747 748 749 750 751 752 753 754 755 756 757
 *	@type: media type of device
 *	@ha: hardware address
 *
 *	Search for an interface by MAC address. Returns NULL if the device
 *	is not found or a pointer to the device. The caller must hold the
 *	rtnl semaphore. The returned device has not had its ref count increased
 *	and the caller must therefore be careful about locking
 *
 *	BUGS:
 *	If the API was consistent this would be __dev_get_by_hwaddr
 */

758
struct net_device *dev_getbyhwaddr(struct net *net, unsigned short type, char *ha)
L
Linus Torvalds 已提交
759 760 761 762 763
{
	struct net_device *dev;

	ASSERT_RTNL();

764
	for_each_netdev(net, dev)
L
Linus Torvalds 已提交
765 766
		if (dev->type == type &&
		    !memcmp(dev->dev_addr, ha, dev->addr_len))
767 768 769
			return dev;

	return NULL;
L
Linus Torvalds 已提交
770
}
771 772
EXPORT_SYMBOL(dev_getbyhwaddr);

773
struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type)
L
Linus Torvalds 已提交
774 775 776
{
	struct net_device *dev;

777
	ASSERT_RTNL();
778
	for_each_netdev(net, dev)
779
		if (dev->type == type)
780 781 782
			return dev;

	return NULL;
783 784 785
}
EXPORT_SYMBOL(__dev_getfirstbyhwtype);

786
struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
787
{
788
	struct net_device *dev, *ret = NULL;
789

790 791 792 793 794 795 796 797 798
	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
799 800 801 802
}
EXPORT_SYMBOL(dev_getfirstbyhwtype);

/**
E
Eric Dumazet 已提交
803
 *	dev_get_by_flags_rcu - find any device with given flags
804
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
805 806 807 808
 *	@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
809 810
 *	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
811 812
 */

E
Eric Dumazet 已提交
813
struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
E
Eric Dumazet 已提交
814
				    unsigned short mask)
L
Linus Torvalds 已提交
815
{
816
	struct net_device *dev, *ret;
L
Linus Torvalds 已提交
817

818
	ret = NULL;
819
	for_each_netdev_rcu(net, dev) {
L
Linus Torvalds 已提交
820
		if (((dev->flags ^ if_flags) & mask) == 0) {
821
			ret = dev;
L
Linus Torvalds 已提交
822 823 824
			break;
		}
	}
825
	return ret;
L
Linus Torvalds 已提交
826
}
E
Eric Dumazet 已提交
827
EXPORT_SYMBOL(dev_get_by_flags_rcu);
L
Linus Torvalds 已提交
828 829 830 831 832 833

/**
 *	dev_valid_name - check if name is okay for network device
 *	@name: name string
 *
 *	Network device names need to be valid file names to
834 835
 *	to allow sysfs to work.  We also disallow any kind of
 *	whitespace.
L
Linus Torvalds 已提交
836
 */
837
int dev_valid_name(const char *name)
L
Linus Torvalds 已提交
838
{
839 840
	if (*name == '\0')
		return 0;
841 842
	if (strlen(name) >= IFNAMSIZ)
		return 0;
843 844 845 846 847 848 849 850 851
	if (!strcmp(name, ".") || !strcmp(name, ".."))
		return 0;

	while (*name) {
		if (*name == '/' || isspace(*name))
			return 0;
		name++;
	}
	return 1;
L
Linus Torvalds 已提交
852
}
E
Eric Dumazet 已提交
853
EXPORT_SYMBOL(dev_valid_name);
L
Linus Torvalds 已提交
854 855

/**
856 857
 *	__dev_alloc_name - allocate a name for a device
 *	@net: network namespace to allocate the device name in
L
Linus Torvalds 已提交
858
 *	@name: name format string
859
 *	@buf:  scratch buffer and result name string
L
Linus Torvalds 已提交
860 861
 *
 *	Passed a format string - eg "lt%d" it will try and find a suitable
S
Stephen Hemminger 已提交
862 863 864 865 866 867
 *	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
868 869
 */

870
static int __dev_alloc_name(struct net *net, const char *name, char *buf)
L
Linus Torvalds 已提交
871 872 873 874
{
	int i = 0;
	const char *p;
	const int max_netdevices = 8*PAGE_SIZE;
S
Stephen Hemminger 已提交
875
	unsigned long *inuse;
L
Linus Torvalds 已提交
876 877 878 879 880 881 882 883 884 885 886 887 888
	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
889
		inuse = (unsigned long *) get_zeroed_page(GFP_ATOMIC);
L
Linus Torvalds 已提交
890 891 892
		if (!inuse)
			return -ENOMEM;

893
		for_each_netdev(net, d) {
L
Linus Torvalds 已提交
894 895 896 897 898 899
			if (!sscanf(d->name, name, &i))
				continue;
			if (i < 0 || i >= max_netdevices)
				continue;

			/*  avoid cases where sscanf is not exact inverse of printf */
900
			snprintf(buf, IFNAMSIZ, name, i);
L
Linus Torvalds 已提交
901 902 903 904 905 906 907 908
			if (!strncmp(buf, d->name, IFNAMSIZ))
				set_bit(i, inuse);
		}

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

909 910
	if (buf != name)
		snprintf(buf, IFNAMSIZ, name, i);
911
	if (!__dev_get_by_name(net, buf))
L
Linus Torvalds 已提交
912 913 914 915 916 917 918 919 920
		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;
}

921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
/**
 *	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;

941 942
	BUG_ON(!dev_net(dev));
	net = dev_net(dev);
943 944 945 946 947
	ret = __dev_alloc_name(net, name, buf);
	if (ret >= 0)
		strlcpy(dev->name, buf, IFNAMSIZ);
	return ret;
}
E
Eric Dumazet 已提交
948
EXPORT_SYMBOL(dev_alloc_name);
949

950
static int dev_get_valid_name(struct net_device *dev, const char *name, bool fmt)
951
{
952 953 954 955 956
	struct net *net;

	BUG_ON(!dev_net(dev));
	net = dev_net(dev);

957 958 959 960
	if (!dev_valid_name(name))
		return -EINVAL;

	if (fmt && strchr(name, '%'))
961
		return dev_alloc_name(dev, name);
962 963
	else if (__dev_get_by_name(net, name))
		return -EEXIST;
964 965
	else if (dev->name != name)
		strlcpy(dev->name, name, IFNAMSIZ);
966 967 968

	return 0;
}
L
Linus Torvalds 已提交
969 970 971 972 973 974 975 976 977

/**
 *	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.
 */
978
int dev_change_name(struct net_device *dev, const char *newname)
L
Linus Torvalds 已提交
979
{
980
	char oldname[IFNAMSIZ];
L
Linus Torvalds 已提交
981
	int err = 0;
982
	int ret;
983
	struct net *net;
L
Linus Torvalds 已提交
984 985

	ASSERT_RTNL();
986
	BUG_ON(!dev_net(dev));
L
Linus Torvalds 已提交
987

988
	net = dev_net(dev);
L
Linus Torvalds 已提交
989 990 991
	if (dev->flags & IFF_UP)
		return -EBUSY;

992 993 994
	if (strncmp(newname, dev->name, IFNAMSIZ) == 0)
		return 0;

995 996
	memcpy(oldname, dev->name, IFNAMSIZ);

997
	err = dev_get_valid_name(dev, newname, 1);
998 999
	if (err < 0)
		return err;
L
Linus Torvalds 已提交
1000

1001
rollback:
1002 1003 1004 1005
	ret = device_rename(&dev->dev, dev->name);
	if (ret) {
		memcpy(dev->name, oldname, IFNAMSIZ);
		return ret;
1006
	}
1007 1008

	write_lock_bh(&dev_base_lock);
1009
	hlist_del(&dev->name_hlist);
1010 1011 1012 1013 1014 1015
	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));
1016 1017
	write_unlock_bh(&dev_base_lock);

1018
	ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev);
1019 1020 1021
	ret = notifier_to_errno(ret);

	if (ret) {
1022 1023
		/* err >= 0 after dev_alloc_name() or stores the first errno */
		if (err >= 0) {
1024 1025 1026
			err = ret;
			memcpy(dev->name, oldname, IFNAMSIZ);
			goto rollback;
1027 1028 1029 1030
		} else {
			printk(KERN_ERR
			       "%s: name change rollback failed: %d.\n",
			       dev->name, ret);
1031 1032
		}
	}
L
Linus Torvalds 已提交
1033 1034 1035 1036

	return err;
}

1037 1038 1039 1040
/**
 *	dev_set_alias - change ifalias of a device
 *	@dev: device
 *	@alias: name up to IFALIASZ
1041
 *	@len: limit of bytes to copy from info
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
 *
 *	Set ifalias for a device,
 */
int dev_set_alias(struct net_device *dev, const char *alias, size_t len)
{
	ASSERT_RTNL();

	if (len >= IFALIASZ)
		return -EINVAL;

1052 1053 1054 1055 1056 1057 1058 1059
	if (!len) {
		if (dev->ifalias) {
			kfree(dev->ifalias);
			dev->ifalias = NULL;
		}
		return 0;
	}

E
Eric Dumazet 已提交
1060
	dev->ifalias = krealloc(dev->ifalias, len + 1, GFP_KERNEL);
1061 1062 1063 1064 1065 1066 1067 1068
	if (!dev->ifalias)
		return -ENOMEM;

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


1069
/**
S
Stephen Hemminger 已提交
1070
 *	netdev_features_change - device changes features
1071 1072 1073 1074 1075 1076
 *	@dev: device to cause notification
 *
 *	Called to indicate a device has changed features.
 */
void netdev_features_change(struct net_device *dev)
{
1077
	call_netdevice_notifiers(NETDEV_FEAT_CHANGE, dev);
1078 1079 1080
}
EXPORT_SYMBOL(netdev_features_change);

L
Linus Torvalds 已提交
1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
/**
 *	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) {
1092
		call_netdevice_notifiers(NETDEV_CHANGE, dev);
L
Linus Torvalds 已提交
1093 1094 1095
		rtmsg_ifinfo(RTM_NEWLINK, dev, 0);
	}
}
E
Eric Dumazet 已提交
1096
EXPORT_SYMBOL(netdev_state_change);
L
Linus Torvalds 已提交
1097

1098
int netdev_bonding_change(struct net_device *dev, unsigned long event)
1099
{
1100
	return call_netdevice_notifiers(event, dev);
1101 1102 1103
}
EXPORT_SYMBOL(netdev_bonding_change);

L
Linus Torvalds 已提交
1104 1105
/**
 *	dev_load 	- load a network module
1106
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
1107 1108 1109 1110 1111 1112 1113
 *	@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.
 */

1114
void dev_load(struct net *net, const char *name)
L
Linus Torvalds 已提交
1115
{
1116
	struct net_device *dev;
L
Linus Torvalds 已提交
1117

1118 1119 1120
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
	rcu_read_unlock();
L
Linus Torvalds 已提交
1121

1122
	if (!dev && capable(CAP_NET_ADMIN))
L
Linus Torvalds 已提交
1123 1124
		request_module("%s", name);
}
E
Eric Dumazet 已提交
1125
EXPORT_SYMBOL(dev_load);
L
Linus Torvalds 已提交
1126

1127
static int __dev_open(struct net_device *dev)
L
Linus Torvalds 已提交
1128
{
1129
	const struct net_device_ops *ops = dev->netdev_ops;
1130
	int ret;
L
Linus Torvalds 已提交
1131

1132 1133
	ASSERT_RTNL();

L
Linus Torvalds 已提交
1134 1135 1136 1137 1138 1139
	/*
	 *	Is it even present?
	 */
	if (!netif_device_present(dev))
		return -ENODEV;

1140 1141 1142 1143 1144
	ret = call_netdevice_notifiers(NETDEV_PRE_UP, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		return ret;

L
Linus Torvalds 已提交
1145 1146 1147 1148
	/*
	 *	Call device private open method
	 */
	set_bit(__LINK_STATE_START, &dev->state);
1149

1150 1151
	if (ops->ndo_validate_addr)
		ret = ops->ndo_validate_addr(dev);
1152

1153 1154
	if (!ret && ops->ndo_open)
		ret = ops->ndo_open(dev);
L
Linus Torvalds 已提交
1155

1156
	/*
L
Linus Torvalds 已提交
1157 1158 1159
	 *	If it went open OK then:
	 */

1160 1161 1162
	if (ret)
		clear_bit(__LINK_STATE_START, &dev->state);
	else {
L
Linus Torvalds 已提交
1163 1164 1165 1166 1167
		/*
		 *	Set the flags.
		 */
		dev->flags |= IFF_UP;

1168 1169 1170
		/*
		 *	Enable NET_DMA
		 */
1171
		net_dmaengine_get();
1172

L
Linus Torvalds 已提交
1173 1174 1175
		/*
		 *	Initialize multicasting status
		 */
1176
		dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
1177 1178 1179 1180 1181 1182

		/*
		 *	Wakeup transmit queue engine
		 */
		dev_activate(dev);
	}
1183

L
Linus Torvalds 已提交
1184 1185 1186 1187
	return ret;
}

/**
1188 1189
 *	dev_open	- prepare an interface for use.
 *	@dev:	device to open
L
Linus Torvalds 已提交
1190
 *
1191 1192 1193 1194 1195 1196 1197
 *	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
1198
 */
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226
int dev_open(struct net_device *dev)
{
	int ret;

	/*
	 *	Is it already up?
	 */
	if (dev->flags & IFF_UP)
		return 0;

	/*
	 *	Open device
	 */
	ret = __dev_open(dev);
	if (ret < 0)
		return ret;

	/*
	 *	... and announce new interface.
	 */
	rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING);
	call_netdevice_notifiers(NETDEV_UP, dev);

	return ret;
}
EXPORT_SYMBOL(dev_open);

static int __dev_close(struct net_device *dev)
L
Linus Torvalds 已提交
1227
{
1228
	const struct net_device_ops *ops = dev->netdev_ops;
1229

1230
	ASSERT_RTNL();
1231 1232
	might_sleep();

L
Linus Torvalds 已提交
1233 1234 1235 1236
	/*
	 *	Tell people we are going down, so that they can
	 *	prepare to death, when device is still operating.
	 */
1237
	call_netdevice_notifiers(NETDEV_GOING_DOWN, dev);
L
Linus Torvalds 已提交
1238 1239 1240 1241

	clear_bit(__LINK_STATE_START, &dev->state);

	/* Synchronize to scheduled poll. We cannot touch poll list,
1242 1243 1244 1245 1246
	 * 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.
	 */
L
Linus Torvalds 已提交
1247 1248
	smp_mb__after_clear_bit(); /* Commit netif_running(). */

1249 1250
	dev_deactivate(dev);

L
Linus Torvalds 已提交
1251 1252 1253 1254 1255 1256 1257
	/*
	 *	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.
	 */
1258 1259
	if (ops->ndo_stop)
		ops->ndo_stop(dev);
L
Linus Torvalds 已提交
1260 1261 1262 1263 1264 1265 1266 1267

	/*
	 *	Device is now down.
	 */

	dev->flags &= ~IFF_UP;

	/*
1268
	 *	Shutdown NET_DMA
L
Linus Torvalds 已提交
1269
	 */
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
	net_dmaengine_put();

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

	__dev_close(dev);
L
Linus Torvalds 已提交
1290

1291
	/*
1292
	 * Tell people we are down
1293
	 */
1294 1295
	rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING);
	call_netdevice_notifiers(NETDEV_DOWN, dev);
1296

L
Linus Torvalds 已提交
1297 1298
	return 0;
}
E
Eric Dumazet 已提交
1299
EXPORT_SYMBOL(dev_close);
L
Linus Torvalds 已提交
1300 1301


1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324
/**
 *	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)
{
	if (dev->ethtool_ops && dev->ethtool_ops->get_flags &&
	    dev->ethtool_ops->set_flags) {
		u32 flags = dev->ethtool_ops->get_flags(dev);
		if (flags & ETH_FLAG_LRO) {
			flags &= ~ETH_FLAG_LRO;
			dev->ethtool_ops->set_flags(dev, flags);
		}
	}
	WARN_ON(dev->features & NETIF_F_LRO);
}
EXPORT_SYMBOL(dev_disable_lro);


1325 1326
static int dev_boot_phase = 1;

L
Linus Torvalds 已提交
1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341
/*
 *	Device change register/unregister. These are not inline or static
 *	as we export them to the world.
 */

/**
 *	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
1342
 *	to the new notifier to allow device to have a race free
L
Linus Torvalds 已提交
1343 1344 1345 1346 1347 1348
 *	view of the network device list.
 */

int register_netdevice_notifier(struct notifier_block *nb)
{
	struct net_device *dev;
1349
	struct net_device *last;
1350
	struct net *net;
L
Linus Torvalds 已提交
1351 1352 1353
	int err;

	rtnl_lock();
1354
	err = raw_notifier_chain_register(&netdev_chain, nb);
1355 1356
	if (err)
		goto unlock;
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
	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
1368

1369 1370
			nb->notifier_call(nb, NETDEV_UP, dev);
		}
L
Linus Torvalds 已提交
1371
	}
1372 1373

unlock:
L
Linus Torvalds 已提交
1374 1375
	rtnl_unlock();
	return err;
1376 1377 1378

rollback:
	last = dev;
1379 1380 1381 1382
	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev == last)
				break;
1383

1384 1385 1386 1387 1388
			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);
1389
			nb->notifier_call(nb, NETDEV_UNREGISTER_BATCH, dev);
1390 1391
		}
	}
1392 1393

	raw_notifier_chain_unregister(&netdev_chain, nb);
1394
	goto unlock;
L
Linus Torvalds 已提交
1395
}
E
Eric Dumazet 已提交
1396
EXPORT_SYMBOL(register_netdevice_notifier);
L
Linus Torvalds 已提交
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409

/**
 *	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.
 */

int unregister_netdevice_notifier(struct notifier_block *nb)
{
1410 1411 1412
	int err;

	rtnl_lock();
1413
	err = raw_notifier_chain_unregister(&netdev_chain, nb);
1414 1415
	rtnl_unlock();
	return err;
L
Linus Torvalds 已提交
1416
}
E
Eric Dumazet 已提交
1417
EXPORT_SYMBOL(unregister_netdevice_notifier);
L
Linus Torvalds 已提交
1418 1419 1420 1421

/**
 *	call_netdevice_notifiers - call all network notifier blocks
 *      @val: value passed unmodified to notifier function
1422
 *      @dev: net_device pointer passed unmodified to notifier function
L
Linus Torvalds 已提交
1423 1424
 *
 *	Call all network notifier blocks.  Parameters and return value
1425
 *	are as for raw_notifier_call_chain().
L
Linus Torvalds 已提交
1426 1427
 */

1428
int call_netdevice_notifiers(unsigned long val, struct net_device *dev)
L
Linus Torvalds 已提交
1429
{
1430
	ASSERT_RTNL();
1431
	return raw_notifier_call_chain(&netdev_chain, val, dev);
L
Linus Torvalds 已提交
1432 1433 1434 1435 1436 1437 1438 1439 1440
}

/* When > 0 there are consumers of rx skb time stamps */
static atomic_t netstamp_needed = ATOMIC_INIT(0);

void net_enable_timestamp(void)
{
	atomic_inc(&netstamp_needed);
}
E
Eric Dumazet 已提交
1441
EXPORT_SYMBOL(net_enable_timestamp);
L
Linus Torvalds 已提交
1442 1443 1444 1445 1446

void net_disable_timestamp(void)
{
	atomic_dec(&netstamp_needed);
}
E
Eric Dumazet 已提交
1447
EXPORT_SYMBOL(net_disable_timestamp);
L
Linus Torvalds 已提交
1448

E
Eric Dumazet 已提交
1449
static inline void net_timestamp_set(struct sk_buff *skb)
L
Linus Torvalds 已提交
1450 1451
{
	if (atomic_read(&netstamp_needed))
1452
		__net_timestamp(skb);
1453 1454
	else
		skb->tstamp.tv64 = 0;
L
Linus Torvalds 已提交
1455 1456
}

E
Eric Dumazet 已提交
1457 1458 1459 1460 1461 1462
static inline void net_timestamp_check(struct sk_buff *skb)
{
	if (!skb->tstamp.tv64 && atomic_read(&netstamp_needed))
		__net_timestamp(skb);
}

1463 1464 1465 1466 1467 1468 1469 1470
/**
 * dev_forward_skb - loopback an skb to another netif
 *
 * @dev: destination network device
 * @skb: buffer to forward
 *
 * return values:
 *	NET_RX_SUCCESS	(no congestion)
1471
 *	NET_RX_DROP     (packet was dropped, but freed)
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
 *
 * 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)
{
	skb_orphan(skb);
1484
	nf_reset(skb);
1485

1486 1487 1488
	if (unlikely(!(dev->flags & IFF_UP) ||
		     (skb->len > (dev->mtu + dev->hard_header_len)))) {
		atomic_long_inc(&dev->rx_dropped);
1489
		kfree_skb(skb);
1490
		return NET_RX_DROP;
1491
	}
1492
	skb_set_dev(skb, dev);
1493 1494 1495 1496 1497 1498 1499
	skb->tstamp.tv64 = 0;
	skb->pkt_type = PACKET_HOST;
	skb->protocol = eth_type_trans(skb, dev);
	return netif_rx(skb);
}
EXPORT_SYMBOL_GPL(dev_forward_skb);

L
Linus Torvalds 已提交
1500 1501 1502 1503 1504
/*
 *	Support routine. Sends outgoing frames to any network
 *	taps currently in use.
 */

1505
static void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
L
Linus Torvalds 已提交
1506 1507
{
	struct packet_type *ptype;
1508

1509 1510
#ifdef CONFIG_NET_CLS_ACT
	if (!(skb->tstamp.tv64 && (G_TC_FROM(skb->tc_verd) & AT_INGRESS)))
E
Eric Dumazet 已提交
1511
		net_timestamp_set(skb);
1512
#else
E
Eric Dumazet 已提交
1513
	net_timestamp_set(skb);
1514
#endif
L
Linus Torvalds 已提交
1515 1516 1517 1518 1519 1520 1521 1522 1523

	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) &&
		    (ptype->af_packet_priv == NULL ||
		     (struct sock *)ptype->af_packet_priv != skb->sk)) {
E
Eric Dumazet 已提交
1524
			struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
L
Linus Torvalds 已提交
1525 1526 1527 1528 1529 1530 1531
			if (!skb2)
				break;

			/* skb->nh should be correctly
			   set by sender, so that the second statement is
			   just protection against buggy protocols.
			 */
1532
			skb_reset_mac_header(skb2);
L
Linus Torvalds 已提交
1533

1534
			if (skb_network_header(skb2) < skb2->data ||
1535
			    skb2->network_header > skb2->tail) {
L
Linus Torvalds 已提交
1536 1537 1538
				if (net_ratelimit())
					printk(KERN_CRIT "protocol %04x is "
					       "buggy, dev %s\n",
1539 1540
					       ntohs(skb2->protocol),
					       dev->name);
1541
				skb_reset_network_header(skb2);
L
Linus Torvalds 已提交
1542 1543
			}

1544
			skb2->transport_header = skb2->network_header;
L
Linus Torvalds 已提交
1545
			skb2->pkt_type = PACKET_OUTGOING;
D
David S. Miller 已提交
1546
			ptype->func(skb2, skb->dev, ptype, skb->dev);
L
Linus Torvalds 已提交
1547 1548 1549 1550 1551
		}
	}
	rcu_read_unlock();
}

1552 1553 1554 1555
/*
 * 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.
 */
1556
int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq)
1557
{
1558 1559
	if (txq < 1 || txq > dev->num_tx_queues)
		return -EINVAL;
1560

1561 1562 1563 1564 1565
	if (dev->reg_state == NETREG_REGISTERED) {
		ASSERT_RTNL();

		if (txq < dev->real_num_tx_queues)
			qdisc_reset_all_tx_gt(dev, txq);
1566
	}
1567 1568 1569

	dev->real_num_tx_queues = txq;
	return 0;
1570 1571
}
EXPORT_SYMBOL(netif_set_real_num_tx_queues);
1572

1573 1574 1575 1576 1577 1578 1579 1580
#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
1581 1582
 *	negative error code.  If called before registration, it always
 *	succeeds.
1583 1584 1585 1586 1587
 */
int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
{
	int rc;

T
Tom Herbert 已提交
1588 1589 1590
	if (rxq < 1 || rxq > dev->num_rx_queues)
		return -EINVAL;

1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605
	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

1606
static inline void __netif_reschedule(struct Qdisc *q)
1607
{
1608 1609
	struct softnet_data *sd;
	unsigned long flags;
1610

1611 1612
	local_irq_save(flags);
	sd = &__get_cpu_var(softnet_data);
1613 1614 1615
	q->next_sched = NULL;
	*sd->output_queue_tailp = q;
	sd->output_queue_tailp = &q->next_sched;
1616 1617 1618 1619 1620 1621 1622 1623
	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);
1624 1625 1626
}
EXPORT_SYMBOL(__netif_schedule);

1627
void dev_kfree_skb_irq(struct sk_buff *skb)
1628
{
1629
	if (atomic_dec_and_test(&skb->users)) {
1630 1631
		struct softnet_data *sd;
		unsigned long flags;
1632

1633 1634 1635 1636 1637 1638 1639
		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);
	}
1640
}
1641
EXPORT_SYMBOL(dev_kfree_skb_irq);
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652

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);


1653 1654 1655 1656 1657 1658
/**
 * netif_device_detach - mark device as removed
 * @dev: network device
 *
 * Mark device as removed from system and therefore no longer available.
 */
1659 1660 1661 1662
void netif_device_detach(struct net_device *dev)
{
	if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1663
		netif_tx_stop_all_queues(dev);
1664 1665 1666 1667
	}
}
EXPORT_SYMBOL(netif_device_detach);

1668 1669 1670 1671 1672 1673
/**
 * netif_device_attach - mark device as attached
 * @dev: network device
 *
 * Mark device as attached from system and restart if needed.
 */
1674 1675 1676 1677
void netif_device_attach(struct net_device *dev)
{
	if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1678
		netif_tx_wake_all_queues(dev);
1679
		__netdev_watchdog_up(dev);
1680 1681 1682 1683
	}
}
EXPORT_SYMBOL(netif_device_attach);

1684 1685 1686 1687 1688 1689
static bool can_checksum_protocol(unsigned long features, __be16 protocol)
{
	return ((features & NETIF_F_GEN_CSUM) ||
		((features & NETIF_F_IP_CSUM) &&
		 protocol == htons(ETH_P_IP)) ||
		((features & NETIF_F_IPV6_CSUM) &&
1690 1691 1692
		 protocol == htons(ETH_P_IPV6)) ||
		((features & NETIF_F_FCOE_CRC) &&
		 protocol == htons(ETH_P_FCOE)));
1693 1694 1695 1696
}

static bool dev_can_checksum(struct net_device *dev, struct sk_buff *skb)
{
1697 1698 1699 1700 1701 1702
	int features = dev->features;

	if (vlan_tx_tag_present(skb))
		features &= dev->vlan_features;

	if (can_checksum_protocol(features, skb->protocol))
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713
		return true;

	if (skb->protocol == htons(ETH_P_8021Q)) {
		struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data;
		if (can_checksum_protocol(dev->features & dev->vlan_features,
					  veh->h_vlan_encapsulated_proto))
			return true;
	}

	return false;
}
1714

1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744
/**
 * skb_dev_set -- assign a new device to a buffer
 * @skb: buffer for the new device
 * @dev: network device
 *
 * If an skb is owned by a device already, we have to reset
 * all data private to the namespace a device belongs to
 * before assigning it a new device.
 */
#ifdef CONFIG_NET_NS
void skb_set_dev(struct sk_buff *skb, struct net_device *dev)
{
	skb_dst_drop(skb);
	if (skb->dev && !net_eq(dev_net(skb->dev), dev_net(dev))) {
		secpath_reset(skb);
		nf_reset(skb);
		skb_init_secmark(skb);
		skb->mark = 0;
		skb->priority = 0;
		skb->nf_trace = 0;
		skb->ipvs_property = 0;
#ifdef CONFIG_NET_SCHED
		skb->tc_index = 0;
#endif
	}
	skb->dev = dev;
}
EXPORT_SYMBOL(skb_set_dev);
#endif /* CONFIG_NET_NS */

L
Linus Torvalds 已提交
1745 1746 1747 1748
/*
 * Invalidate hardware checksum when packet is to be mangled, and
 * complete checksum manually on outgoing path.
 */
1749
int skb_checksum_help(struct sk_buff *skb)
L
Linus Torvalds 已提交
1750
{
1751
	__wsum csum;
1752
	int ret = 0, offset;
L
Linus Torvalds 已提交
1753

1754
	if (skb->ip_summed == CHECKSUM_COMPLETE)
1755 1756 1757 1758 1759
		goto out_set_summed;

	if (unlikely(skb_shinfo(skb)->gso_size)) {
		/* Let GSO fix up the checksum. */
		goto out_set_summed;
L
Linus Torvalds 已提交
1760 1761
	}

1762 1763 1764 1765 1766 1767 1768 1769 1770
	offset = skb->csum_start - skb_headroom(skb);
	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
1771 1772 1773 1774 1775
		ret = pskb_expand_head(skb, 0, 0, GFP_ATOMIC);
		if (ret)
			goto out;
	}

1776
	*(__sum16 *)(skb->data + offset) = csum_fold(csum);
1777
out_set_summed:
L
Linus Torvalds 已提交
1778
	skb->ip_summed = CHECKSUM_NONE;
1779
out:
L
Linus Torvalds 已提交
1780 1781
	return ret;
}
E
Eric Dumazet 已提交
1782
EXPORT_SYMBOL(skb_checksum_help);
L
Linus Torvalds 已提交
1783

1784 1785 1786
/**
 *	skb_gso_segment - Perform segmentation on skb.
 *	@skb: buffer to segment
1787
 *	@features: features for the output path (see dev->features)
1788 1789
 *
 *	This function segments the given skb and returns a list of segments.
1790 1791 1792
 *
 *	It may return NULL if the skb requires no segmentation.  This is
 *	only possible when GSO is used for verifying header integrity.
1793
 */
1794
struct sk_buff *skb_gso_segment(struct sk_buff *skb, int features)
1795 1796 1797
{
	struct sk_buff *segs = ERR_PTR(-EPROTONOSUPPORT);
	struct packet_type *ptype;
A
Al Viro 已提交
1798
	__be16 type = skb->protocol;
1799
	int err;
1800

1801 1802 1803 1804 1805 1806 1807 1808 1809 1810
	if (type == htons(ETH_P_8021Q)) {
		struct vlan_ethhdr *veh;

		if (unlikely(!pskb_may_pull(skb, VLAN_ETH_HLEN)))
			return ERR_PTR(-EINVAL);

		veh = (struct vlan_ethhdr *)skb->data;
		type = veh->h_vlan_encapsulated_proto;
	}

1811
	skb_reset_mac_header(skb);
1812
	skb->mac_len = skb->network_header - skb->mac_header;
1813 1814
	__skb_pull(skb, skb->mac_len);

1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827
	if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
		struct net_device *dev = skb->dev;
		struct ethtool_drvinfo info = {};

		if (dev && dev->ethtool_ops && dev->ethtool_ops->get_drvinfo)
			dev->ethtool_ops->get_drvinfo(dev, &info);

		WARN(1, "%s: caps=(0x%lx, 0x%lx) len=%d data_len=%d "
			"ip_summed=%d",
		     info.driver, dev ? dev->features : 0L,
		     skb->sk ? skb->sk->sk_route_caps : 0L,
		     skb->len, skb->data_len, skb->ip_summed);

1828 1829 1830 1831 1832
		if (skb_header_cloned(skb) &&
		    (err = pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
			return ERR_PTR(err);
	}

1833
	rcu_read_lock();
1834 1835
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
1836
		if (ptype->type == type && !ptype->dev && ptype->gso_segment) {
1837
			if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
1838 1839 1840 1841
				err = ptype->gso_send_check(skb);
				segs = ERR_PTR(err);
				if (err || skb_gso_ok(skb, features))
					break;
1842 1843
				__skb_push(skb, (skb->data -
						 skb_network_header(skb)));
1844
			}
1845
			segs = ptype->gso_segment(skb, features);
1846 1847 1848 1849 1850
			break;
		}
	}
	rcu_read_unlock();

1851
	__skb_push(skb, skb->data - skb_mac_header(skb));
1852

1853 1854 1855 1856
	return segs;
}
EXPORT_SYMBOL(skb_gso_segment);

1857 1858 1859 1860 1861
/* Take action when hardware reception checksum errors are detected. */
#ifdef CONFIG_BUG
void netdev_rx_csum_fault(struct net_device *dev)
{
	if (net_ratelimit()) {
1862
		printk(KERN_ERR "%s: hw csum failure.\n",
1863
			dev ? dev->name : "<unknown>");
1864 1865 1866 1867 1868 1869
		dump_stack();
	}
}
EXPORT_SYMBOL(netdev_rx_csum_fault);
#endif

L
Linus Torvalds 已提交
1870 1871 1872 1873 1874
/* 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
1875
static int illegal_highdma(struct net_device *dev, struct sk_buff *skb)
L
Linus Torvalds 已提交
1876
{
1877
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
1878
	int i;
1879 1880 1881 1882 1883
	if (!(dev->features & NETIF_F_HIGHDMA)) {
		for (i = 0; i < skb_shinfo(skb)->nr_frags; i++)
			if (PageHighMem(skb_shinfo(skb)->frags[i].page))
				return 1;
	}
L
Linus Torvalds 已提交
1884

1885 1886
	if (PCI_DMA_BUS_IS_PHYS) {
		struct device *pdev = dev->dev.parent;
L
Linus Torvalds 已提交
1887

E
Eric Dumazet 已提交
1888 1889
		if (!pdev)
			return 0;
1890 1891 1892 1893 1894 1895
		for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
			dma_addr_t addr = page_to_phys(skb_shinfo(skb)->frags[i].page);
			if (!pdev->dma_mask || addr + PAGE_SIZE - 1 > *pdev->dma_mask)
				return 1;
		}
	}
1896
#endif
L
Linus Torvalds 已提交
1897 1898 1899
	return 0;
}

1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
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
 *
 *	This function segments the given skb and stores the list of segments
 *	in skb->next.
 */
static int dev_gso_segment(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
	struct sk_buff *segs;
1934 1935 1936 1937 1938 1939 1940 1941
	int features = dev->features & ~(illegal_highdma(dev, skb) ?
					 NETIF_F_SG : 0);

	segs = skb_gso_segment(skb, features);

	/* Verifying header integrity only. */
	if (!segs)
		return 0;
1942

1943
	if (IS_ERR(segs))
1944 1945 1946 1947 1948 1949 1950 1951 1952
		return PTR_ERR(segs);

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

	return 0;
}

E
Eric Dumazet 已提交
1953 1954
/*
 * Try to orphan skb early, right before transmission by the device.
1955 1956
 * We cannot orphan skb if tx timestamp is requested or the sk-reference
 * is needed on driver level for other reasons, e.g. see net/can/raw.c
E
Eric Dumazet 已提交
1957 1958 1959
 */
static inline void skb_orphan_try(struct sk_buff *skb)
{
1960 1961
	struct sock *sk = skb->sk;

1962
	if (sk && !skb_shinfo(skb)->tx_flags) {
1963 1964 1965 1966 1967
		/* skb_tx_hash() wont be able to get sk.
		 * We copy sk_hash into skb->rxhash
		 */
		if (!skb->rxhash)
			skb->rxhash = sk->sk_hash;
E
Eric Dumazet 已提交
1968
		skb_orphan(skb);
1969
	}
E
Eric Dumazet 已提交
1970 1971
}

1972 1973 1974 1975 1976 1977 1978 1979 1980 1981
/*
 * Returns true if either:
 *	1. skb has frag_list and the device doesn't support FRAGLIST, or
 *	2. skb is fragmented and the device does not support SG, or if
 *	   at least one of fragments is in highmem and device does not
 *	   support DMA from it.
 */
static inline int skb_needs_linearize(struct sk_buff *skb,
				      struct net_device *dev)
{
1982 1983 1984 1985 1986
	int features = dev->features;

	if (skb->protocol == htons(ETH_P_8021Q) || vlan_tx_tag_present(skb))
		features &= dev->vlan_features;

1987
	return skb_is_nonlinear(skb) &&
1988 1989
	       ((skb_has_frag_list(skb) && !(features & NETIF_F_FRAGLIST)) ||
		(skb_shinfo(skb)->nr_frags && (!(features & NETIF_F_SG) ||
1990 1991 1992
					      illegal_highdma(dev, skb))));
}

1993 1994
int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
			struct netdev_queue *txq)
1995
{
1996
	const struct net_device_ops *ops = dev->netdev_ops;
1997
	int rc = NETDEV_TX_OK;
1998

1999
	if (likely(!skb->next)) {
S
Stephen Hemminger 已提交
2000
		if (!list_empty(&ptype_all))
2001 2002
			dev_queue_xmit_nit(skb, dev);

2003 2004 2005 2006
		/*
		 * If device doesnt need skb->dst, release it right now while
		 * its hot in this cpu cache
		 */
E
Eric Dumazet 已提交
2007 2008 2009
		if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
			skb_dst_drop(skb);

E
Eric Dumazet 已提交
2010
		skb_orphan_try(skb);
2011

2012 2013 2014 2015 2016 2017 2018 2019 2020
		if (vlan_tx_tag_present(skb) &&
		    !(dev->features & NETIF_F_HW_VLAN_TX)) {
			skb = __vlan_put_tag(skb, vlan_tx_tag_get(skb));
			if (unlikely(!skb))
				goto out;

			skb->vlan_tci = 0;
		}

2021 2022 2023 2024 2025
		if (netif_needs_gso(dev, skb)) {
			if (unlikely(dev_gso_segment(skb)))
				goto out_kfree_skb;
			if (skb->next)
				goto gso;
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041
		} else {
			if (skb_needs_linearize(skb, dev) &&
			    __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) {
				skb_set_transport_header(skb, skb->csum_start -
					      skb_headroom(skb));
				if (!dev_can_checksum(dev, skb) &&
				     skb_checksum_help(skb))
					goto out_kfree_skb;
			}
2042 2043
		}

2044
		rc = ops->ndo_start_xmit(skb, dev);
2045
		if (rc == NETDEV_TX_OK)
E
Eric Dumazet 已提交
2046
			txq_trans_update(txq);
2047
		return rc;
2048 2049
	}

2050
gso:
2051 2052 2053 2054 2055
	do {
		struct sk_buff *nskb = skb->next;

		skb->next = nskb->next;
		nskb->next = NULL;
2056 2057 2058 2059 2060 2061 2062 2063

		/*
		 * If device doesnt need nskb->dst, release it right now while
		 * its hot in this cpu cache
		 */
		if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
			skb_dst_drop(nskb);

2064
		rc = ops->ndo_start_xmit(nskb, dev);
2065
		if (unlikely(rc != NETDEV_TX_OK)) {
2066 2067
			if (rc & ~NETDEV_TX_MASK)
				goto out_kfree_gso_skb;
2068
			nskb->next = skb->next;
2069 2070 2071
			skb->next = nskb;
			return rc;
		}
E
Eric Dumazet 已提交
2072
		txq_trans_update(txq);
2073
		if (unlikely(netif_tx_queue_stopped(txq) && skb->next))
2074
			return NETDEV_TX_BUSY;
2075
	} while (skb->next);
2076

2077 2078 2079
out_kfree_gso_skb:
	if (likely(skb->next == NULL))
		skb->destructor = DEV_GSO_CB(skb)->destructor;
2080 2081
out_kfree_skb:
	kfree_skb(skb);
2082
out:
2083
	return rc;
2084 2085
}

T
Tom Herbert 已提交
2086
static u32 hashrnd __read_mostly;
D
David S. Miller 已提交
2087

2088
u16 skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb)
2089
{
2090
	u32 hash;
D
David S. Miller 已提交
2091

2092 2093
	if (skb_rx_queue_recorded(skb)) {
		hash = skb_get_rx_queue(skb);
E
Eric Dumazet 已提交
2094
		while (unlikely(hash >= dev->real_num_tx_queues))
2095 2096 2097
			hash -= dev->real_num_tx_queues;
		return hash;
	}
2098 2099

	if (skb->sk && skb->sk->sk_hash)
2100
		hash = skb->sk->sk_hash;
2101
	else
2102
		hash = (__force u16) skb->protocol ^ skb->rxhash;
T
Tom Herbert 已提交
2103
	hash = jhash_1word(hash, hashrnd);
D
David S. Miller 已提交
2104 2105

	return (u16) (((u64) hash * dev->real_num_tx_queues) >> 32);
2106
}
2107
EXPORT_SYMBOL(skb_tx_hash);
2108

2109 2110 2111 2112
static inline u16 dev_cap_txqueue(struct net_device *dev, u16 queue_index)
{
	if (unlikely(queue_index >= dev->real_num_tx_queues)) {
		if (net_ratelimit()) {
E
Eric Dumazet 已提交
2113 2114 2115
			pr_warning("%s selects TX queue %d, but "
				"real number of TX queues is %d\n",
				dev->name, queue_index, dev->real_num_tx_queues);
2116 2117 2118 2119 2120 2121
		}
		return 0;
	}
	return queue_index;
}

2122 2123 2124
static struct netdev_queue *dev_pick_tx(struct net_device *dev,
					struct sk_buff *skb)
{
2125
	int queue_index;
2126
	const struct net_device_ops *ops = dev->netdev_ops;
2127

2128 2129 2130 2131 2132 2133 2134
	if (ops->ndo_select_queue) {
		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);
		if (queue_index < 0) {
2135 2136 2137 2138

			queue_index = 0;
			if (dev->real_num_tx_queues > 1)
				queue_index = skb_tx_hash(dev, skb);
2139

E
Eric Dumazet 已提交
2140
			if (sk) {
2141
				struct dst_entry *dst = rcu_dereference_check(sk->sk_dst_cache, 1);
E
Eric Dumazet 已提交
2142 2143 2144 2145

				if (dst && skb_dst(skb) == dst)
					sk_tx_queue_set(sk, queue_index);
			}
2146 2147
		}
	}
2148

2149 2150
	skb_set_queue_mapping(skb, queue_index);
	return netdev_get_tx_queue(dev, queue_index);
2151 2152
}

2153 2154 2155 2156 2157
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);
2158
	bool contended = qdisc_is_running(q);
2159 2160
	int rc;

2161 2162 2163 2164 2165 2166 2167 2168 2169
	/*
	 * 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.
	 */
	if (unlikely(contended))
		spin_lock(&q->busylock);

2170 2171 2172 2173 2174
	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) &&
2175
		   qdisc_run_begin(q)) {
2176 2177 2178 2179 2180
		/*
		 * 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
2181 2182
		if (!(dev->priv_flags & IFF_XMIT_DST_RELEASE))
			skb_dst_force(skb);
2183
		__qdisc_update_bstats(q, skb->len);
2184 2185 2186 2187 2188
		if (sch_direct_xmit(skb, q, dev, txq, root_lock)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
2189
			__qdisc_run(q);
2190
		} else
2191
			qdisc_run_end(q);
2192 2193 2194

		rc = NET_XMIT_SUCCESS;
	} else {
E
Eric Dumazet 已提交
2195
		skb_dst_force(skb);
2196
		rc = qdisc_enqueue_root(skb, q);
2197 2198 2199 2200 2201 2202 2203
		if (qdisc_run_begin(q)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
			__qdisc_run(q);
		}
2204 2205
	}
	spin_unlock(root_lock);
2206 2207
	if (unlikely(contended))
		spin_unlock(&q->busylock);
2208 2209 2210
	return rc;
}

2211 2212 2213
static DEFINE_PER_CPU(int, xmit_recursion);
#define RECURSION_LIMIT 3

2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238
/**
 *	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
2239 2240 2241
int dev_queue_xmit(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
2242
	struct netdev_queue *txq;
L
Linus Torvalds 已提交
2243 2244 2245
	struct Qdisc *q;
	int rc = -ENOMEM;

2246 2247
	/* Disable soft irqs for various locks below. Also
	 * stops preemption for RCU.
L
Linus Torvalds 已提交
2248
	 */
2249
	rcu_read_lock_bh();
L
Linus Torvalds 已提交
2250

2251
	txq = dev_pick_tx(dev, skb);
2252
	q = rcu_dereference_bh(txq->qdisc);
2253

L
Linus Torvalds 已提交
2254
#ifdef CONFIG_NET_CLS_ACT
E
Eric Dumazet 已提交
2255
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_EGRESS);
L
Linus Torvalds 已提交
2256 2257
#endif
	if (q->enqueue) {
2258
		rc = __dev_xmit_skb(skb, q, dev, txq);
2259
		goto out;
L
Linus Torvalds 已提交
2260 2261 2262 2263 2264
	}

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

H
Herbert Xu 已提交
2265 2266
	   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
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276
	   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 */

2277
		if (txq->xmit_lock_owner != cpu) {
L
Linus Torvalds 已提交
2278

2279 2280 2281
			if (__this_cpu_read(xmit_recursion) > RECURSION_LIMIT)
				goto recursion_alert;

2282
			HARD_TX_LOCK(dev, txq, cpu);
L
Linus Torvalds 已提交
2283

2284
			if (!netif_tx_queue_stopped(txq)) {
2285
				__this_cpu_inc(xmit_recursion);
2286
				rc = dev_hard_start_xmit(skb, dev, txq);
2287
				__this_cpu_dec(xmit_recursion);
2288
				if (dev_xmit_complete(rc)) {
2289
					HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2290 2291 2292
					goto out;
				}
			}
2293
			HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2294 2295 2296 2297 2298
			if (net_ratelimit())
				printk(KERN_CRIT "Virtual device %s asks to "
				       "queue packet!\n", dev->name);
		} else {
			/* Recursion is detected! It is possible,
2299 2300 2301
			 * unfortunately
			 */
recursion_alert:
L
Linus Torvalds 已提交
2302 2303 2304 2305 2306 2307 2308
			if (net_ratelimit())
				printk(KERN_CRIT "Dead loop on virtual device "
				       "%s, fix it urgently!\n", dev->name);
		}
	}

	rc = -ENETDOWN;
2309
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2310 2311 2312 2313

	kfree_skb(skb);
	return rc;
out:
2314
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2315 2316
	return rc;
}
E
Eric Dumazet 已提交
2317
EXPORT_SYMBOL(dev_queue_xmit);
L
Linus Torvalds 已提交
2318 2319 2320 2321 2322 2323


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

2324
int netdev_max_backlog __read_mostly = 1000;
E
Eric Dumazet 已提交
2325
int netdev_tstamp_prequeue __read_mostly = 1;
2326 2327
int netdev_budget __read_mostly = 300;
int weight_p __read_mostly = 64;            /* old backlog weight */
L
Linus Torvalds 已提交
2328

E
Eric Dumazet 已提交
2329 2330 2331 2332 2333 2334 2335 2336
/* 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
2337
/*
2338 2339 2340
 * __skb_get_rxhash: calculate a flow hash based on src/dst addresses
 * and src/dst port numbers. Returns a non-zero hash number on success
 * and 0 on failure.
T
Tom Herbert 已提交
2341
 */
2342
__u32 __skb_get_rxhash(struct sk_buff *skb)
T
Tom Herbert 已提交
2343
{
2344
	int nhoff, hash = 0, poff;
T
Tom Herbert 已提交
2345 2346 2347
	struct ipv6hdr *ip6;
	struct iphdr *ip;
	u8 ip_proto;
C
Changli Gao 已提交
2348 2349 2350 2351 2352
	u32 addr1, addr2, ihl;
	union {
		u32 v32;
		u16 v16[2];
	} ports;
T
Tom Herbert 已提交
2353

2354
	nhoff = skb_network_offset(skb);
T
Tom Herbert 已提交
2355 2356 2357

	switch (skb->protocol) {
	case __constant_htons(ETH_P_IP):
2358
		if (!pskb_may_pull(skb, sizeof(*ip) + nhoff))
T
Tom Herbert 已提交
2359 2360
			goto done;

2361
		ip = (struct iphdr *) (skb->data + nhoff);
2362 2363 2364 2365
		if (ip->frag_off & htons(IP_MF | IP_OFFSET))
			ip_proto = 0;
		else
			ip_proto = ip->protocol;
E
Eric Dumazet 已提交
2366 2367
		addr1 = (__force u32) ip->saddr;
		addr2 = (__force u32) ip->daddr;
T
Tom Herbert 已提交
2368 2369 2370
		ihl = ip->ihl;
		break;
	case __constant_htons(ETH_P_IPV6):
2371
		if (!pskb_may_pull(skb, sizeof(*ip6) + nhoff))
T
Tom Herbert 已提交
2372 2373
			goto done;

2374
		ip6 = (struct ipv6hdr *) (skb->data + nhoff);
T
Tom Herbert 已提交
2375
		ip_proto = ip6->nexthdr;
E
Eric Dumazet 已提交
2376 2377
		addr1 = (__force u32) ip6->saddr.s6_addr32[3];
		addr2 = (__force u32) ip6->daddr.s6_addr32[3];
T
Tom Herbert 已提交
2378 2379 2380 2381 2382
		ihl = (40 >> 2);
		break;
	default:
		goto done;
	}
2383

2384 2385 2386 2387 2388 2389
	ports.v32 = 0;
	poff = proto_ports_offset(ip_proto);
	if (poff >= 0) {
		nhoff += ihl * 4 + poff;
		if (pskb_may_pull(skb, nhoff + 4)) {
			ports.v32 = * (__force u32 *) (skb->data + nhoff);
C
Changli Gao 已提交
2390 2391
			if (ports.v16[1] < ports.v16[0])
				swap(ports.v16[0], ports.v16[1]);
E
Eric Dumazet 已提交
2392
		}
T
Tom Herbert 已提交
2393 2394
	}

E
Eric Dumazet 已提交
2395 2396 2397
	/* get a consistent hash (same value on both flow directions) */
	if (addr2 < addr1)
		swap(addr1, addr2);
T
Tom Herbert 已提交
2398

2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
	hash = jhash_3words(addr1, addr2, ports.v32, hashrnd);
	if (!hash)
		hash = 1;

done:
	return hash;
}
EXPORT_SYMBOL(__skb_get_rxhash);

#ifdef CONFIG_RPS

/* One global table that all flow-based protocols share. */
struct rps_sock_flow_table *rps_sock_flow_table __read_mostly;
EXPORT_SYMBOL(rps_sock_flow_table);

/*
 * 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;
2423
	struct rps_map *map = NULL;
2424 2425 2426 2427 2428 2429 2430
	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);
2431 2432 2433 2434 2435
		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);
2436 2437 2438 2439 2440 2441
			goto done;
		}
		rxqueue = dev->_rx + index;
	} else
		rxqueue = dev->_rx;

2442 2443 2444 2445 2446 2447 2448 2449 2450
	if (rxqueue->rps_map) {
		map = rcu_dereference(rxqueue->rps_map);
		if (map && map->len == 1) {
			tcpu = map->cpus[0];
			if (cpu_online(tcpu))
				cpu = tcpu;
			goto done;
		}
	} else if (!rxqueue->rps_flow_table) {
2451
		goto done;
2452
	}
2453

2454
	skb_reset_network_header(skb);
2455 2456 2457
	if (!skb_get_rxhash(skb))
		goto done;

T
Tom Herbert 已提交
2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496
	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 -
		      rflow->last_qtail)) >= 0)) {
			tcpu = rflow->cpu = next_cpu;
			if (tcpu != RPS_NO_CPU)
				rflow->last_qtail = per_cpu(softnet_data,
				    tcpu).input_queue_head;
		}
		if (tcpu != RPS_NO_CPU && cpu_online(tcpu)) {
			*rflowp = rflow;
			cpu = tcpu;
			goto done;
		}
	}

T
Tom Herbert 已提交
2497
	if (map) {
T
Tom Herbert 已提交
2498
		tcpu = map->cpus[((u64) skb->rxhash * map->len) >> 32];
T
Tom Herbert 已提交
2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510

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

done:
	return cpu;
}

/* Called from hardirq (IPI) context */
E
Eric Dumazet 已提交
2511
static void rps_trigger_softirq(void *data)
T
Tom Herbert 已提交
2512
{
E
Eric Dumazet 已提交
2513 2514
	struct softnet_data *sd = data;

E
Eric Dumazet 已提交
2515
	____napi_schedule(sd, &sd->backlog);
C
Changli Gao 已提交
2516
	sd->received_rps++;
T
Tom Herbert 已提交
2517
}
E
Eric Dumazet 已提交
2518

T
Tom Herbert 已提交
2519
#endif /* CONFIG_RPS */
T
Tom Herbert 已提交
2520

E
Eric Dumazet 已提交
2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
/*
 * 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
2542 2543 2544 2545
/*
 * 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
2546 2547
static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
			      unsigned int *qtail)
T
Tom Herbert 已提交
2548
{
E
Eric Dumazet 已提交
2549
	struct softnet_data *sd;
T
Tom Herbert 已提交
2550 2551
	unsigned long flags;

E
Eric Dumazet 已提交
2552
	sd = &per_cpu(softnet_data, cpu);
T
Tom Herbert 已提交
2553 2554 2555

	local_irq_save(flags);

E
Eric Dumazet 已提交
2556
	rps_lock(sd);
2557 2558
	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
2559
enqueue:
E
Eric Dumazet 已提交
2560
			__skb_queue_tail(&sd->input_pkt_queue, skb);
2561
			input_queue_tail_incr_save(sd, qtail);
E
Eric Dumazet 已提交
2562
			rps_unlock(sd);
2563
			local_irq_restore(flags);
T
Tom Herbert 已提交
2564 2565 2566
			return NET_RX_SUCCESS;
		}

2567 2568 2569 2570
		/* 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
2571
			if (!rps_ipi_queued(sd))
E
Eric Dumazet 已提交
2572
				____napi_schedule(sd, &sd->backlog);
T
Tom Herbert 已提交
2573 2574 2575 2576
		}
		goto enqueue;
	}

C
Changli Gao 已提交
2577
	sd->dropped++;
E
Eric Dumazet 已提交
2578
	rps_unlock(sd);
T
Tom Herbert 已提交
2579 2580 2581

	local_irq_restore(flags);

2582
	atomic_long_inc(&skb->dev->rx_dropped);
T
Tom Herbert 已提交
2583 2584 2585
	kfree_skb(skb);
	return NET_RX_DROP;
}
L
Linus Torvalds 已提交
2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603

/**
 *	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)
{
2604
	int ret;
L
Linus Torvalds 已提交
2605 2606 2607 2608 2609

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

E
Eric Dumazet 已提交
2610 2611
	if (netdev_tstamp_prequeue)
		net_timestamp_check(skb);
L
Linus Torvalds 已提交
2612

E
Eric Dumazet 已提交
2613
#ifdef CONFIG_RPS
2614
	{
T
Tom Herbert 已提交
2615
		struct rps_dev_flow voidflow, *rflow = &voidflow;
2616 2617
		int cpu;

2618
		preempt_disable();
2619
		rcu_read_lock();
T
Tom Herbert 已提交
2620 2621

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
2622 2623
		if (cpu < 0)
			cpu = smp_processor_id();
T
Tom Herbert 已提交
2624 2625 2626

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

2627
		rcu_read_unlock();
2628
		preempt_enable();
2629
	}
2630
#else
T
Tom Herbert 已提交
2631 2632 2633 2634 2635
	{
		unsigned int qtail;
		ret = enqueue_to_backlog(skb, get_cpu(), &qtail);
		put_cpu();
	}
2636
#endif
2637
	return ret;
L
Linus Torvalds 已提交
2638
}
E
Eric Dumazet 已提交
2639
EXPORT_SYMBOL(netif_rx);
L
Linus Torvalds 已提交
2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670

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;

2671
			WARN_ON(atomic_read(&skb->users));
L
Linus Torvalds 已提交
2672 2673 2674 2675 2676
			__kfree_skb(skb);
		}
	}

	if (sd->output_queue) {
2677
		struct Qdisc *head;
L
Linus Torvalds 已提交
2678 2679 2680 2681

		local_irq_disable();
		head = sd->output_queue;
		sd->output_queue = NULL;
2682
		sd->output_queue_tailp = &sd->output_queue;
L
Linus Torvalds 已提交
2683 2684 2685
		local_irq_enable();

		while (head) {
2686 2687 2688
			struct Qdisc *q = head;
			spinlock_t *root_lock;

L
Linus Torvalds 已提交
2689 2690
			head = head->next_sched;

2691
			root_lock = qdisc_lock(q);
2692
			if (spin_trylock(root_lock)) {
2693 2694 2695
				smp_mb__before_clear_bit();
				clear_bit(__QDISC_STATE_SCHED,
					  &q->state);
2696 2697
				qdisc_run(q);
				spin_unlock(root_lock);
L
Linus Torvalds 已提交
2698
			} else {
2699
				if (!test_bit(__QDISC_STATE_DEACTIVATED,
2700
					      &q->state)) {
2701
					__netif_reschedule(q);
2702 2703 2704 2705 2706
				} else {
					smp_mb__before_clear_bit();
					clear_bit(__QDISC_STATE_SCHED,
						  &q->state);
				}
L
Linus Torvalds 已提交
2707 2708 2709 2710 2711
			}
		}
	}
}

2712 2713 2714
static inline int deliver_skb(struct sk_buff *skb,
			      struct packet_type *pt_prev,
			      struct net_device *orig_dev)
L
Linus Torvalds 已提交
2715 2716
{
	atomic_inc(&skb->users);
D
David S. Miller 已提交
2717
	return pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
2718 2719
}

2720 2721
#if (defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE)) && \
    (defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE))
2722 2723 2724
/* This hook is defined here for ATM LANE */
int (*br_fdb_test_addr_hook)(struct net_device *dev,
			     unsigned char *addr) __read_mostly;
2725
EXPORT_SYMBOL_GPL(br_fdb_test_addr_hook);
2726
#endif
L
Linus Torvalds 已提交
2727 2728 2729 2730 2731 2732

#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
2733
 * NOTE: This doesnt stop any functionality; if you dont have
L
Linus Torvalds 已提交
2734 2735 2736
 * the ingress scheduler, you just cant add policies on ingress.
 *
 */
2737
static int ing_filter(struct sk_buff *skb, struct netdev_queue *rxq)
L
Linus Torvalds 已提交
2738 2739
{
	struct net_device *dev = skb->dev;
2740
	u32 ttl = G_TC_RTTL(skb->tc_verd);
2741 2742
	int result = TC_ACT_OK;
	struct Qdisc *q;
2743

2744 2745 2746 2747
	if (unlikely(MAX_RED_LOOP < ttl++)) {
		if (net_ratelimit())
			pr_warning( "Redir loop detected Dropping packet (%d->%d)\n",
			       skb->skb_iif, dev->ifindex);
2748 2749
		return TC_ACT_SHOT;
	}
L
Linus Torvalds 已提交
2750

2751 2752
	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
2753

2754
	q = rxq->qdisc;
2755
	if (q != &noop_qdisc) {
2756
		spin_lock(qdisc_lock(q));
2757 2758
		if (likely(!test_bit(__QDISC_STATE_DEACTIVATED, &q->state)))
			result = qdisc_enqueue_root(skb, q);
2759 2760
		spin_unlock(qdisc_lock(q));
	}
2761 2762 2763

	return result;
}
2764

2765 2766 2767 2768
static inline struct sk_buff *handle_ing(struct sk_buff *skb,
					 struct packet_type **pt_prev,
					 int *ret, struct net_device *orig_dev)
{
2769 2770 2771
	struct netdev_queue *rxq = rcu_dereference(skb->dev->ingress_queue);

	if (!rxq || rxq->qdisc == &noop_qdisc)
2772
		goto out;
L
Linus Torvalds 已提交
2773

2774 2775 2776
	if (*pt_prev) {
		*ret = deliver_skb(skb, *pt_prev, orig_dev);
		*pt_prev = NULL;
L
Linus Torvalds 已提交
2777 2778
	}

2779
	switch (ing_filter(skb, rxq)) {
2780 2781 2782 2783 2784 2785 2786 2787 2788
	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
2789 2790 2791
}
#endif

2792 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
/*
 * 	netif_nit_deliver - deliver received packets to network taps
 * 	@skb: buffer
 *
 * 	This function is used to deliver incoming packets to network
 * 	taps. It should be used when the normal netif_receive_skb path
 * 	is bypassed, for example because of VLAN acceleration.
 */
void netif_nit_deliver(struct sk_buff *skb)
{
	struct packet_type *ptype;

	if (list_empty(&ptype_all))
		return;

	skb_reset_network_header(skb);
	skb_reset_transport_header(skb);
	skb->mac_len = skb->network_header - skb->mac_header;

	rcu_read_lock();
	list_for_each_entry_rcu(ptype, &ptype_all, list) {
		if (!ptype->dev || ptype->dev == skb->dev)
			deliver_skb(skb, ptype, skb->dev);
	}
	rcu_read_unlock();
}

2819 2820 2821 2822
/**
 *	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
2823
 *	@rx_handler_data: data pointer that is used by rx handler
2824 2825 2826 2827 2828 2829 2830 2831
 *
 *	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.
 */
int netdev_rx_handler_register(struct net_device *dev,
J
Jiri Pirko 已提交
2832 2833
			       rx_handler_func_t *rx_handler,
			       void *rx_handler_data)
2834 2835 2836 2837 2838 2839
{
	ASSERT_RTNL();

	if (dev->rx_handler)
		return -EBUSY;

J
Jiri Pirko 已提交
2840
	rcu_assign_pointer(dev->rx_handler_data, rx_handler_data);
2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859
	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();
	rcu_assign_pointer(dev->rx_handler, NULL);
J
Jiri Pirko 已提交
2860
	rcu_assign_pointer(dev->rx_handler_data, NULL);
2861 2862 2863
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);

2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884
static inline void skb_bond_set_mac_by_master(struct sk_buff *skb,
					      struct net_device *master)
{
	if (skb->pkt_type == PACKET_HOST) {
		u16 *dest = (u16 *) eth_hdr(skb)->h_dest;

		memcpy(dest, master->dev_addr, ETH_ALEN);
	}
}

/* On bonding slaves other than the currently active slave, suppress
 * duplicates except for 802.3ad ETH_P_SLOW, alb non-mcast/bcast, and
 * ARP on active-backup slaves with arp_validate enabled.
 */
int __skb_bond_should_drop(struct sk_buff *skb, struct net_device *master)
{
	struct net_device *dev = skb->dev;

	if (master->priv_flags & IFF_MASTER_ARPMON)
		dev->last_rx = jiffies;

2885 2886
	if ((master->priv_flags & IFF_MASTER_ALB) &&
	    (master->priv_flags & IFF_BRIDGE_PORT)) {
2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913
		/* Do address unmangle. The local destination address
		 * will be always the one master has. Provides the right
		 * functionality in a bridge.
		 */
		skb_bond_set_mac_by_master(skb, master);
	}

	if (dev->priv_flags & IFF_SLAVE_INACTIVE) {
		if ((dev->priv_flags & IFF_SLAVE_NEEDARP) &&
		    skb->protocol == __cpu_to_be16(ETH_P_ARP))
			return 0;

		if (master->priv_flags & IFF_MASTER_ALB) {
			if (skb->pkt_type != PACKET_BROADCAST &&
			    skb->pkt_type != PACKET_MULTICAST)
				return 0;
		}
		if (master->priv_flags & IFF_MASTER_8023AD &&
		    skb->protocol == __cpu_to_be16(ETH_P_SLOW))
			return 0;

		return 1;
	}
	return 0;
}
EXPORT_SYMBOL(__skb_bond_should_drop);

2914
static int __netif_receive_skb(struct sk_buff *skb)
L
Linus Torvalds 已提交
2915 2916
{
	struct packet_type *ptype, *pt_prev;
2917
	rx_handler_func_t *rx_handler;
D
David S. Miller 已提交
2918
	struct net_device *orig_dev;
2919
	struct net_device *master;
2920
	struct net_device *null_or_orig;
2921
	struct net_device *orig_or_bond;
L
Linus Torvalds 已提交
2922
	int ret = NET_RX_DROP;
A
Al Viro 已提交
2923
	__be16 type;
L
Linus Torvalds 已提交
2924

E
Eric Dumazet 已提交
2925 2926
	if (!netdev_tstamp_prequeue)
		net_timestamp_check(skb);
2927

2928 2929
	if (vlan_tx_tag_present(skb))
		vlan_hwaccel_do_receive(skb);
2930

L
Linus Torvalds 已提交
2931
	/* if we've gotten here through NAPI, check netpoll */
2932
	if (netpoll_receive_skb(skb))
L
Linus Torvalds 已提交
2933 2934
		return NET_RX_DROP;

2935 2936
	if (!skb->skb_iif)
		skb->skb_iif = skb->dev->ifindex;
2937

2938 2939 2940 2941 2942 2943 2944 2945
	/*
	 * bonding note: skbs received on inactive slaves should only
	 * be delivered to pkt handlers that are exact matches.  Also
	 * the deliver_no_wcard flag will be set.  If packet handlers
	 * are sensitive to duplicate packets these skbs will need to
	 * be dropped at the handler.  The vlan accel path may have
	 * already set the deliver_no_wcard flag.
	 */
2946
	null_or_orig = NULL;
J
Joe Eykholt 已提交
2947
	orig_dev = skb->dev;
2948
	master = ACCESS_ONCE(orig_dev->master);
2949 2950 2951 2952 2953
	if (skb->deliver_no_wcard)
		null_or_orig = orig_dev;
	else if (master) {
		if (skb_bond_should_drop(skb, master)) {
			skb->deliver_no_wcard = 1;
2954
			null_or_orig = orig_dev; /* deliver only exact match */
2955
		} else
2956
			skb->dev = master;
J
Joe Eykholt 已提交
2957
	}
2958

2959
	__this_cpu_inc(softnet_data.processed);
2960
	skb_reset_network_header(skb);
2961
	skb_reset_transport_header(skb);
2962
	skb->mac_len = skb->network_header - skb->mac_header;
L
Linus Torvalds 已提交
2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975

	pt_prev = NULL;

	rcu_read_lock();

#ifdef CONFIG_NET_CLS_ACT
	if (skb->tc_verd & TC_NCLS) {
		skb->tc_verd = CLR_TC_NCLS(skb->tc_verd);
		goto ncls;
	}
#endif

	list_for_each_entry_rcu(ptype, &ptype_all, list) {
2976 2977
		if (ptype->dev == null_or_orig || ptype->dev == skb->dev ||
		    ptype->dev == orig_dev) {
2978
			if (pt_prev)
D
David S. Miller 已提交
2979
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
2980 2981 2982 2983 2984
			pt_prev = ptype;
		}
	}

#ifdef CONFIG_NET_CLS_ACT
2985 2986
	skb = handle_ing(skb, &pt_prev, &ret, orig_dev);
	if (!skb)
L
Linus Torvalds 已提交
2987 2988 2989 2990
		goto out;
ncls:
#endif

2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001
	/* Handle special case of bridge or macvlan */
	rx_handler = rcu_dereference(skb->dev->rx_handler);
	if (rx_handler) {
		if (pt_prev) {
			ret = deliver_skb(skb, pt_prev, orig_dev);
			pt_prev = NULL;
		}
		skb = rx_handler(skb);
		if (!skb)
			goto out;
	}
L
Linus Torvalds 已提交
3002

3003 3004 3005 3006 3007 3008
	/*
	 * Make sure frames received on VLAN interfaces stacked on
	 * bonding interfaces still make their way to any base bonding
	 * device that may have registered for a specific ptype.  The
	 * handler may have to adjust skb->dev and orig_dev.
	 */
3009
	orig_or_bond = orig_dev;
3010 3011
	if ((skb->dev->priv_flags & IFF_802_1Q_VLAN) &&
	    (vlan_dev_real_dev(skb->dev)->priv_flags & IFF_BONDING)) {
3012
		orig_or_bond = vlan_dev_real_dev(skb->dev);
3013 3014
	}

L
Linus Torvalds 已提交
3015
	type = skb->protocol;
3016 3017
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
3018
		if (ptype->type == type && (ptype->dev == null_or_orig ||
3019
		     ptype->dev == skb->dev || ptype->dev == orig_dev ||
3020
		     ptype->dev == orig_or_bond)) {
3021
			if (pt_prev)
D
David S. Miller 已提交
3022
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3023 3024 3025 3026 3027
			pt_prev = ptype;
		}
	}

	if (pt_prev) {
D
David S. Miller 已提交
3028
		ret = pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3029
	} else {
3030
		atomic_long_inc(&skb->dev->rx_dropped);
L
Linus Torvalds 已提交
3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041
		kfree_skb(skb);
		/* Jamal, now you will not able to escape explaining
		 * me how you were going to use this. :-)
		 */
		ret = NET_RX_DROP;
	}

out:
	rcu_read_unlock();
	return ret;
}
T
Tom Herbert 已提交
3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059

/**
 *	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)
{
E
Eric Dumazet 已提交
3060 3061 3062
	if (netdev_tstamp_prequeue)
		net_timestamp_check(skb);

3063 3064 3065
	if (skb_defer_rx_timestamp(skb))
		return NET_RX_SUCCESS;

E
Eric Dumazet 已提交
3066
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
3067 3068 3069
	{
		struct rps_dev_flow voidflow, *rflow = &voidflow;
		int cpu, ret;
T
Tom Herbert 已提交
3070

E
Eric Dumazet 已提交
3071 3072 3073
		rcu_read_lock();

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
T
Tom Herbert 已提交
3074

E
Eric Dumazet 已提交
3075 3076 3077 3078 3079 3080 3081
		if (cpu >= 0) {
			ret = enqueue_to_backlog(skb, cpu, &rflow->last_qtail);
			rcu_read_unlock();
		} else {
			rcu_read_unlock();
			ret = __netif_receive_skb(skb);
		}
T
Tom Herbert 已提交
3082

E
Eric Dumazet 已提交
3083
		return ret;
T
Tom Herbert 已提交
3084
	}
3085 3086 3087
#else
	return __netif_receive_skb(skb);
#endif
T
Tom Herbert 已提交
3088
}
E
Eric Dumazet 已提交
3089
EXPORT_SYMBOL(netif_receive_skb);
L
Linus Torvalds 已提交
3090

E
Eric Dumazet 已提交
3091 3092 3093
/* Network device is going away, flush any packets still pending
 * Called with irqs disabled.
 */
3094
static void flush_backlog(void *arg)
3095
{
3096
	struct net_device *dev = arg;
E
Eric Dumazet 已提交
3097
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3098 3099
	struct sk_buff *skb, *tmp;

E
Eric Dumazet 已提交
3100
	rps_lock(sd);
3101
	skb_queue_walk_safe(&sd->input_pkt_queue, skb, tmp) {
3102
		if (skb->dev == dev) {
E
Eric Dumazet 已提交
3103
			__skb_unlink(skb, &sd->input_pkt_queue);
3104
			kfree_skb(skb);
3105
			input_queue_head_incr(sd);
3106
		}
3107
	}
E
Eric Dumazet 已提交
3108
	rps_unlock(sd);
3109 3110 3111 3112 3113

	skb_queue_walk_safe(&sd->process_queue, skb, tmp) {
		if (skb->dev == dev) {
			__skb_unlink(skb, &sd->process_queue);
			kfree_skb(skb);
3114
			input_queue_head_incr(sd);
3115 3116
		}
	}
3117 3118
}

3119 3120 3121 3122 3123 3124 3125
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;

3126 3127
	if (NAPI_GRO_CB(skb)->count == 1) {
		skb_shinfo(skb)->gso_size = 0;
3128
		goto out;
3129
	}
3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150

	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
3151
inline void napi_gro_flush(struct napi_struct *napi)
3152 3153 3154 3155 3156 3157 3158 3159 3160
{
	struct sk_buff *skb, *next;

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

3161
	napi->gro_count = 0;
3162 3163
	napi->gro_list = NULL;
}
E
Eric Dumazet 已提交
3164
EXPORT_SYMBOL(napi_gro_flush);
3165

3166
enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3167 3168 3169 3170 3171
{
	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
3172
	int same_flow;
3173
	int mac_len;
3174
	enum gro_result ret;
3175

3176
	if (!(skb->dev->features & NETIF_F_GRO) || netpoll_rx_on(skb))
3177 3178
		goto normal;

3179
	if (skb_is_gso(skb) || skb_has_frag_list(skb))
3180 3181
		goto normal;

3182 3183 3184 3185 3186
	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
		if (ptype->type != type || ptype->dev || !ptype->gro_receive)
			continue;

3187
		skb_set_network_header(skb, skb_gro_offset(skb));
3188 3189 3190 3191
		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
3192
		NAPI_GRO_CB(skb)->free = 0;
3193 3194 3195 3196 3197 3198 3199 3200 3201

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

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

H
Herbert Xu 已提交
3202
	same_flow = NAPI_GRO_CB(skb)->same_flow;
3203
	ret = NAPI_GRO_CB(skb)->free ? GRO_MERGED_FREE : GRO_MERGED;
H
Herbert Xu 已提交
3204

3205 3206 3207 3208 3209 3210
	if (pp) {
		struct sk_buff *nskb = *pp;

		*pp = nskb->next;
		nskb->next = NULL;
		napi_gro_complete(nskb);
3211
		napi->gro_count--;
3212 3213
	}

H
Herbert Xu 已提交
3214
	if (same_flow)
3215 3216
		goto ok;

3217
	if (NAPI_GRO_CB(skb)->flush || napi->gro_count >= MAX_GRO_SKBS)
3218 3219
		goto normal;

3220
	napi->gro_count++;
3221
	NAPI_GRO_CB(skb)->count = 1;
3222
	skb_shinfo(skb)->gso_size = skb_gro_len(skb);
3223 3224
	skb->next = napi->gro_list;
	napi->gro_list = skb;
3225
	ret = GRO_HELD;
3226

3227
pull:
H
Herbert Xu 已提交
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244
	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;
		skb_shinfo(skb)->frags[0].size -= grow;

		if (unlikely(!skb_shinfo(skb)->frags[0].size)) {
			put_page(skb_shinfo(skb)->frags[0].page);
			memmove(skb_shinfo(skb)->frags,
				skb_shinfo(skb)->frags + 1,
3245
				--skb_shinfo(skb)->nr_frags * sizeof(skb_frag_t));
H
Herbert Xu 已提交
3246
		}
3247 3248
	}

3249
ok:
3250
	return ret;
3251 3252

normal:
3253 3254
	ret = GRO_NORMAL;
	goto pull;
H
Herbert Xu 已提交
3255
}
3256 3257
EXPORT_SYMBOL(dev_gro_receive);

3258
static inline gro_result_t
3259
__napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3260 3261 3262 3263
{
	struct sk_buff *p;

	for (p = napi->gro_list; p; p = p->next) {
3264 3265 3266 3267
		unsigned long diffs;

		diffs = (unsigned long)p->dev ^ (unsigned long)skb->dev;
		diffs |= compare_ether_header(skb_mac_header(p),
3268
					      skb_gro_mac_header(skb));
3269
		NAPI_GRO_CB(p)->same_flow = !diffs;
3270 3271 3272 3273 3274
		NAPI_GRO_CB(p)->flush = 0;
	}

	return dev_gro_receive(napi, skb);
}
H
Herbert Xu 已提交
3275

3276
gro_result_t napi_skb_finish(gro_result_t ret, struct sk_buff *skb)
H
Herbert Xu 已提交
3277
{
3278 3279
	switch (ret) {
	case GRO_NORMAL:
3280 3281 3282
		if (netif_receive_skb(skb))
			ret = GRO_DROP;
		break;
H
Herbert Xu 已提交
3283

3284 3285
	case GRO_DROP:
	case GRO_MERGED_FREE:
H
Herbert Xu 已提交
3286 3287
		kfree_skb(skb);
		break;
3288 3289 3290 3291

	case GRO_HELD:
	case GRO_MERGED:
		break;
H
Herbert Xu 已提交
3292 3293
	}

3294
	return ret;
3295 3296 3297
}
EXPORT_SYMBOL(napi_skb_finish);

3298 3299 3300 3301
void skb_gro_reset_offset(struct sk_buff *skb)
{
	NAPI_GRO_CB(skb)->data_offset = 0;
	NAPI_GRO_CB(skb)->frag0 = NULL;
3302
	NAPI_GRO_CB(skb)->frag0_len = 0;
3303

3304
	if (skb->mac_header == skb->tail &&
3305
	    !PageHighMem(skb_shinfo(skb)->frags[0].page)) {
3306 3307 3308
		NAPI_GRO_CB(skb)->frag0 =
			page_address(skb_shinfo(skb)->frags[0].page) +
			skb_shinfo(skb)->frags[0].page_offset;
3309 3310
		NAPI_GRO_CB(skb)->frag0_len = skb_shinfo(skb)->frags[0].size;
	}
3311 3312 3313
}
EXPORT_SYMBOL(skb_gro_reset_offset);

3314
gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3315
{
3316 3317
	skb_gro_reset_offset(skb);

3318
	return napi_skb_finish(__napi_gro_receive(napi, skb), skb);
3319 3320 3321
}
EXPORT_SYMBOL(napi_gro_receive);

3322 3323 3324 3325 3326 3327 3328 3329 3330
void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
{
	__skb_pull(skb, skb_headlen(skb));
	skb_reserve(skb, NET_IP_ALIGN - skb_headroom(skb));

	napi->skb = skb;
}
EXPORT_SYMBOL(napi_reuse_skb);

3331
struct sk_buff *napi_get_frags(struct napi_struct *napi)
H
Herbert Xu 已提交
3332 3333 3334 3335
{
	struct sk_buff *skb = napi->skb;

	if (!skb) {
3336 3337 3338
		skb = netdev_alloc_skb_ip_align(napi->dev, GRO_MAX_HEAD);
		if (skb)
			napi->skb = skb;
3339
	}
3340 3341
	return skb;
}
3342
EXPORT_SYMBOL(napi_get_frags);
3343

3344 3345
gro_result_t napi_frags_finish(struct napi_struct *napi, struct sk_buff *skb,
			       gro_result_t ret)
3346
{
3347 3348
	switch (ret) {
	case GRO_NORMAL:
3349
	case GRO_HELD:
3350
		skb->protocol = eth_type_trans(skb, skb->dev);
3351

3352 3353 3354 3355
		if (ret == GRO_HELD)
			skb_gro_pull(skb, -ETH_HLEN);
		else if (netif_receive_skb(skb))
			ret = GRO_DROP;
3356
		break;
H
Herbert Xu 已提交
3357

3358 3359 3360 3361
	case GRO_DROP:
	case GRO_MERGED_FREE:
		napi_reuse_skb(napi, skb);
		break;
3362 3363 3364

	case GRO_MERGED:
		break;
3365
	}
H
Herbert Xu 已提交
3366

3367
	return ret;
H
Herbert Xu 已提交
3368
}
3369 3370
EXPORT_SYMBOL(napi_frags_finish);

3371 3372 3373 3374
struct sk_buff *napi_frags_skb(struct napi_struct *napi)
{
	struct sk_buff *skb = napi->skb;
	struct ethhdr *eth;
3375 3376
	unsigned int hlen;
	unsigned int off;
3377 3378 3379 3380 3381 3382

	napi->skb = NULL;

	skb_reset_mac_header(skb);
	skb_gro_reset_offset(skb);

3383 3384 3385 3386 3387 3388 3389 3390 3391 3392
	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;
		}
3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407
	}

	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;
}
EXPORT_SYMBOL(napi_frags_skb);

3408
gro_result_t napi_gro_frags(struct napi_struct *napi)
3409
{
3410
	struct sk_buff *skb = napi_frags_skb(napi);
3411 3412

	if (!skb)
3413
		return GRO_DROP;
3414 3415 3416

	return napi_frags_finish(napi, skb, __napi_gro_receive(napi, skb));
}
H
Herbert Xu 已提交
3417 3418
EXPORT_SYMBOL(napi_gro_frags);

3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446
/*
 * 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();
}

3447
static int process_backlog(struct napi_struct *napi, int quota)
L
Linus Torvalds 已提交
3448 3449
{
	int work = 0;
E
Eric Dumazet 已提交
3450
	struct softnet_data *sd = container_of(napi, struct softnet_data, backlog);
L
Linus Torvalds 已提交
3451

3452 3453 3454 3455 3456 3457 3458 3459 3460
#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
3461
	napi->weight = weight_p;
3462 3463
	local_irq_disable();
	while (work < quota) {
L
Linus Torvalds 已提交
3464
		struct sk_buff *skb;
3465 3466 3467 3468 3469 3470
		unsigned int qlen;

		while ((skb = __skb_dequeue(&sd->process_queue))) {
			local_irq_enable();
			__netif_receive_skb(skb);
			local_irq_disable();
3471 3472 3473 3474 3475
			input_queue_head_incr(sd);
			if (++work >= quota) {
				local_irq_enable();
				return work;
			}
3476
		}
L
Linus Torvalds 已提交
3477

E
Eric Dumazet 已提交
3478
		rps_lock(sd);
3479
		qlen = skb_queue_len(&sd->input_pkt_queue);
3480
		if (qlen)
3481 3482
			skb_queue_splice_tail_init(&sd->input_pkt_queue,
						   &sd->process_queue);
3483

3484
		if (qlen < quota - work) {
E
Eric Dumazet 已提交
3485 3486 3487 3488 3489 3490 3491 3492 3493 3494
			/*
			 * 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;

3495
			quota = work + qlen;
3496
		}
E
Eric Dumazet 已提交
3497
		rps_unlock(sd);
3498 3499
	}
	local_irq_enable();
L
Linus Torvalds 已提交
3500

3501 3502
	return work;
}
L
Linus Torvalds 已提交
3503

3504 3505
/**
 * __napi_schedule - schedule for receive
3506
 * @n: entry to schedule
3507 3508 3509
 *
 * The entry's receive function will be scheduled to run
 */
H
Harvey Harrison 已提交
3510
void __napi_schedule(struct napi_struct *n)
3511 3512
{
	unsigned long flags;
L
Linus Torvalds 已提交
3513

3514
	local_irq_save(flags);
E
Eric Dumazet 已提交
3515
	____napi_schedule(&__get_cpu_var(softnet_data), n);
3516
	local_irq_restore(flags);
L
Linus Torvalds 已提交
3517
}
3518 3519
EXPORT_SYMBOL(__napi_schedule);

3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552
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);
3553
	napi->gro_count = 0;
3554
	napi->gro_list = NULL;
H
Herbert Xu 已提交
3555
	napi->skb = NULL;
3556 3557 3558 3559
	napi->poll = poll;
	napi->weight = weight;
	list_add(&napi->dev_list, &dev->napi_list);
	napi->dev = dev;
H
Herbert Xu 已提交
3560
#ifdef CONFIG_NETPOLL
3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571
	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;

3572
	list_del_init(&napi->dev_list);
3573
	napi_free_frags(napi);
3574 3575 3576 3577 3578 3579 3580 3581

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

	napi->gro_list = NULL;
3582
	napi->gro_count = 0;
3583 3584 3585
}
EXPORT_SYMBOL(netif_napi_del);

L
Linus Torvalds 已提交
3586 3587
static void net_rx_action(struct softirq_action *h)
{
3588
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3589
	unsigned long time_limit = jiffies + 2;
3590
	int budget = netdev_budget;
3591 3592
	void *have;

L
Linus Torvalds 已提交
3593 3594
	local_irq_disable();

3595
	while (!list_empty(&sd->poll_list)) {
3596 3597
		struct napi_struct *n;
		int work, weight;
L
Linus Torvalds 已提交
3598

3599
		/* If softirq window is exhuasted then punt.
3600 3601
		 * Allow this to run for 2 jiffies since which will allow
		 * an average latency of 1.5/HZ.
3602
		 */
3603
		if (unlikely(budget <= 0 || time_after(jiffies, time_limit)))
L
Linus Torvalds 已提交
3604 3605 3606 3607
			goto softnet_break;

		local_irq_enable();

3608 3609 3610 3611 3612
		/* 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.
		 */
3613
		n = list_first_entry(&sd->poll_list, struct napi_struct, poll_list);
L
Linus Torvalds 已提交
3614

3615 3616 3617 3618
		have = netpoll_poll_lock(n);

		weight = n->weight;

3619 3620 3621 3622 3623 3624 3625
		/* 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
		 * accidently calling ->poll() when NAPI is not scheduled.
		 */
		work = 0;
3626
		if (test_bit(NAPI_STATE_SCHED, &n->state)) {
3627
			work = n->poll(n, weight);
3628 3629
			trace_napi_poll(n);
		}
3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641

		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.
		 */
3642
		if (unlikely(work == weight)) {
3643 3644 3645 3646 3647
			if (unlikely(napi_disable_pending(n))) {
				local_irq_enable();
				napi_complete(n);
				local_irq_disable();
			} else
3648
				list_move_tail(&n->poll_list, &sd->poll_list);
3649
		}
3650 3651

		netpoll_poll_unlock(have);
L
Linus Torvalds 已提交
3652 3653
	}
out:
3654
	net_rps_action_and_irq_enable(sd);
T
Tom Herbert 已提交
3655

3656 3657 3658 3659 3660
#ifdef CONFIG_NET_DMA
	/*
	 * There may not be any more sk_buffs coming right now, so push
	 * any pending DMA copies to hardware
	 */
3661
	dma_issue_pending_all();
3662
#endif
3663

L
Linus Torvalds 已提交
3664 3665 3666
	return;

softnet_break:
C
Changli Gao 已提交
3667
	sd->time_squeeze++;
L
Linus Torvalds 已提交
3668 3669 3670 3671
	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
	goto out;
}

E
Eric Dumazet 已提交
3672
static gifconf_func_t *gifconf_list[NPROTO];
L
Linus Torvalds 已提交
3673 3674 3675 3676 3677 3678 3679 3680 3681 3682

/**
 *	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
3683
int register_gifconf(unsigned int family, gifconf_func_t *gifconf)
L
Linus Torvalds 已提交
3684 3685 3686 3687 3688 3689
{
	if (family >= NPROTO)
		return -EINVAL;
	gifconf_list[family] = gifconf;
	return 0;
}
E
Eric Dumazet 已提交
3690
EXPORT_SYMBOL(register_gifconf);
L
Linus Torvalds 已提交
3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703


/*
 *	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
 */

3704
static int dev_ifname(struct net *net, struct ifreq __user *arg)
L
Linus Torvalds 已提交
3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715
{
	struct net_device *dev;
	struct ifreq ifr;

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

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

3716 3717
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifr.ifr_ifindex);
L
Linus Torvalds 已提交
3718
	if (!dev) {
3719
		rcu_read_unlock();
L
Linus Torvalds 已提交
3720 3721 3722 3723
		return -ENODEV;
	}

	strcpy(ifr.ifr_name, dev->name);
3724
	rcu_read_unlock();
L
Linus Torvalds 已提交
3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736

	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'.
 */

3737
static int dev_ifconf(struct net *net, char __user *arg)
L
Linus Torvalds 已提交
3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760
{
	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;
3761
	for_each_netdev(net, dev) {
L
Linus Torvalds 已提交
3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774
		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;
			}
		}
3775
	}
L
Linus Torvalds 已提交
3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792

	/*
	 *	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
/*
 *	This is invoked by the /proc filesystem handler to display a device
 *	in detail.
 */
3793
void *dev_seq_start(struct seq_file *seq, loff_t *pos)
3794
	__acquires(RCU)
L
Linus Torvalds 已提交
3795
{
3796
	struct net *net = seq_file_net(seq);
3797
	loff_t off;
L
Linus Torvalds 已提交
3798 3799
	struct net_device *dev;

3800
	rcu_read_lock();
3801 3802
	if (!*pos)
		return SEQ_START_TOKEN;
L
Linus Torvalds 已提交
3803

3804
	off = 1;
3805
	for_each_netdev_rcu(net, dev)
3806 3807
		if (off++ == *pos)
			return dev;
L
Linus Torvalds 已提交
3808

3809
	return NULL;
L
Linus Torvalds 已提交
3810 3811 3812 3813
}

void *dev_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
3814 3815 3816 3817
	struct net_device *dev = (v == SEQ_START_TOKEN) ?
				  first_net_device(seq_file_net(seq)) :
				  next_net_device((struct net_device *)v);

L
Linus Torvalds 已提交
3818
	++*pos;
3819
	return rcu_dereference(dev);
L
Linus Torvalds 已提交
3820 3821 3822
}

void dev_seq_stop(struct seq_file *seq, void *v)
3823
	__releases(RCU)
L
Linus Torvalds 已提交
3824
{
3825
	rcu_read_unlock();
L
Linus Torvalds 已提交
3826 3827 3828 3829
}

static void dev_seq_printf_stats(struct seq_file *seq, struct net_device *dev)
{
3830 3831
	struct rtnl_link_stats64 temp;
	const struct rtnl_link_stats64 *stats = dev_get_stats(dev, &temp);
L
Linus Torvalds 已提交
3832

3833 3834
	seq_printf(seq, "%6s: %7llu %7llu %4llu %4llu %4llu %5llu %10llu %9llu "
		   "%8llu %7llu %4llu %4llu %4llu %5llu %7llu %10llu\n",
3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849
		   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
3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868
}

/*
 *	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
3869
static struct softnet_data *softnet_get_online(loff_t *pos)
L
Linus Torvalds 已提交
3870
{
C
Changli Gao 已提交
3871
	struct softnet_data *sd = NULL;
L
Linus Torvalds 已提交
3872

3873
	while (*pos < nr_cpu_ids)
3874
		if (cpu_online(*pos)) {
C
Changli Gao 已提交
3875
			sd = &per_cpu(softnet_data, *pos);
L
Linus Torvalds 已提交
3876 3877 3878
			break;
		} else
			++*pos;
C
Changli Gao 已提交
3879
	return sd;
L
Linus Torvalds 已提交
3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898
}

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
3899
	struct softnet_data *sd = v;
L
Linus Torvalds 已提交
3900

T
Tom Herbert 已提交
3901
	seq_printf(seq, "%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
C
Changli Gao 已提交
3902
		   sd->processed, sd->dropped, sd->time_squeeze, 0,
3903
		   0, 0, 0, 0, /* was fastroute */
C
Changli Gao 已提交
3904
		   sd->cpu_collision, sd->received_rps);
L
Linus Torvalds 已提交
3905 3906 3907
	return 0;
}

3908
static const struct seq_operations dev_seq_ops = {
L
Linus Torvalds 已提交
3909 3910 3911 3912 3913 3914 3915 3916
	.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)
{
3917 3918
	return seq_open_net(inode, file, &dev_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
3919 3920
}

3921
static const struct file_operations dev_seq_fops = {
L
Linus Torvalds 已提交
3922 3923 3924 3925
	.owner	 = THIS_MODULE,
	.open    = dev_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
3926
	.release = seq_release_net,
L
Linus Torvalds 已提交
3927 3928
};

3929
static const struct seq_operations softnet_seq_ops = {
L
Linus Torvalds 已提交
3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940
	.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);
}

3941
static const struct file_operations softnet_seq_fops = {
L
Linus Torvalds 已提交
3942 3943 3944 3945 3946 3947 3948
	.owner	 = THIS_MODULE,
	.open    = softnet_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release,
};

3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960
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;
	}

3961
	for (t = 0; t < PTYPE_HASH_SIZE; t++) {
3962 3963 3964 3965 3966 3967 3968 3969 3970 3971
		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)
3972
	__acquires(RCU)
3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995
{
	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
3996
		hash = ntohs(pt->type) & PTYPE_HASH_MASK;
3997 3998

	while (nxt == &ptype_base[hash]) {
3999
		if (++hash >= PTYPE_HASH_SIZE)
4000 4001 4002 4003 4004 4005 4006 4007
			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)
4008
	__releases(RCU)
4009 4010 4011 4012 4013 4014 4015 4016 4017 4018
{
	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");
4019
	else if (pt->dev == NULL || dev_net(pt->dev) == seq_file_net(seq)) {
4020 4021 4022 4023 4024
		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
4025 4026
		seq_printf(seq, " %-8s %pF\n",
			   pt->dev ? pt->dev->name : "", pt->func);
4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040
	}

	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)
{
4041 4042
	return seq_open_net(inode, file, &ptype_seq_ops,
			sizeof(struct seq_net_private));
4043 4044 4045 4046 4047 4048 4049
}

static const struct file_operations ptype_seq_fops = {
	.owner	 = THIS_MODULE,
	.open    = ptype_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
4050
	.release = seq_release_net,
4051 4052 4053
};


4054
static int __net_init dev_proc_net_init(struct net *net)
L
Linus Torvalds 已提交
4055 4056 4057
{
	int rc = -ENOMEM;

4058
	if (!proc_net_fops_create(net, "dev", S_IRUGO, &dev_seq_fops))
L
Linus Torvalds 已提交
4059
		goto out;
4060
	if (!proc_net_fops_create(net, "softnet_stat", S_IRUGO, &softnet_seq_fops))
L
Linus Torvalds 已提交
4061
		goto out_dev;
4062
	if (!proc_net_fops_create(net, "ptype", S_IRUGO, &ptype_seq_fops))
4063
		goto out_softnet;
4064

4065
	if (wext_proc_init(net))
4066
		goto out_ptype;
L
Linus Torvalds 已提交
4067 4068 4069
	rc = 0;
out:
	return rc;
4070
out_ptype:
4071
	proc_net_remove(net, "ptype");
L
Linus Torvalds 已提交
4072
out_softnet:
4073
	proc_net_remove(net, "softnet_stat");
L
Linus Torvalds 已提交
4074
out_dev:
4075
	proc_net_remove(net, "dev");
L
Linus Torvalds 已提交
4076 4077
	goto out;
}
4078

4079
static void __net_exit dev_proc_net_exit(struct net *net)
4080 4081 4082 4083 4084 4085 4086 4087
{
	wext_proc_exit(net);

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

4088
static struct pernet_operations __net_initdata dev_proc_ops = {
4089 4090 4091 4092 4093 4094 4095 4096
	.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
4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125
#else
#define dev_proc_init() 0
#endif	/* CONFIG_PROC_FS */


/**
 *	netdev_set_master	-	set up 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 the reference counts
 *	are adjusted, %RTM_NEWLINK is sent to the routing socket and the
 *	function returns zero.
 */
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;
4126

E
Eric Dumazet 已提交
4127 4128
	if (old) {
		synchronize_net();
L
Linus Torvalds 已提交
4129
		dev_put(old);
E
Eric Dumazet 已提交
4130
	}
L
Linus Torvalds 已提交
4131 4132 4133 4134 4135 4136 4137 4138
	if (master)
		slave->flags |= IFF_SLAVE;
	else
		slave->flags &= ~IFF_SLAVE;

	rtmsg_ifinfo(RTM_NEWLINK, slave, IFF_SLAVE);
	return 0;
}
E
Eric Dumazet 已提交
4139
EXPORT_SYMBOL(netdev_set_master);
L
Linus Torvalds 已提交
4140

4141 4142
static void dev_change_rx_flags(struct net_device *dev, int flags)
{
4143 4144 4145 4146
	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);
4147 4148
}

4149
static int __dev_set_promiscuity(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4150 4151
{
	unsigned short old_flags = dev->flags;
4152 4153
	uid_t uid;
	gid_t gid;
L
Linus Torvalds 已提交
4154

4155 4156
	ASSERT_RTNL();

4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173
	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;
			printk(KERN_WARNING "%s: promiscuity touches roof, "
				"set promiscuity failed, promiscuity feature "
				"of device might be broken.\n", dev->name);
			return -EOVERFLOW;
		}
	}
4174
	if (dev->flags != old_flags) {
L
Linus Torvalds 已提交
4175 4176
		printk(KERN_INFO "device %s %s promiscuous mode\n",
		       dev->name, (dev->flags & IFF_PROMISC) ? "entered" :
4177
							       "left");
4178 4179
		if (audit_enabled) {
			current_uid_gid(&uid, &gid);
4180 4181 4182 4183 4184 4185
			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),
4186
				uid, gid,
4187
				audit_get_sessionid(current));
4188
		}
4189

4190
		dev_change_rx_flags(dev, IFF_PROMISC);
L
Linus Torvalds 已提交
4191
	}
4192
	return 0;
L
Linus Torvalds 已提交
4193 4194
}

4195 4196 4197 4198 4199 4200 4201 4202 4203
/**
 *	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.
4204
 *	Return 0 if successful or a negative errno code on error.
4205
 */
4206
int dev_set_promiscuity(struct net_device *dev, int inc)
4207 4208
{
	unsigned short old_flags = dev->flags;
4209
	int err;
4210

4211
	err = __dev_set_promiscuity(dev, inc);
4212
	if (err < 0)
4213
		return err;
4214 4215
	if (dev->flags != old_flags)
		dev_set_rx_mode(dev);
4216
	return err;
4217
}
E
Eric Dumazet 已提交
4218
EXPORT_SYMBOL(dev_set_promiscuity);
4219

L
Linus Torvalds 已提交
4220 4221 4222 4223 4224 4225 4226 4227 4228 4229
/**
 *	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.
4230
 *	Return 0 if successful or a negative errno code on error.
L
Linus Torvalds 已提交
4231 4232
 */

4233
int dev_set_allmulti(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4234 4235 4236
{
	unsigned short old_flags = dev->flags;

4237 4238
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4239
	dev->flags |= IFF_ALLMULTI;
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255
	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;
			printk(KERN_WARNING "%s: allmulti touches roof, "
				"set allmulti failed, allmulti feature of "
				"device might be broken.\n", dev->name);
			return -EOVERFLOW;
		}
	}
4256
	if (dev->flags ^ old_flags) {
4257
		dev_change_rx_flags(dev, IFF_ALLMULTI);
4258
		dev_set_rx_mode(dev);
4259
	}
4260
	return 0;
4261
}
E
Eric Dumazet 已提交
4262
EXPORT_SYMBOL(dev_set_allmulti);
4263 4264 4265 4266

/*
 *	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
4267
 *	filtering it is put in promiscuous mode while unicast addresses
4268 4269 4270 4271
 *	are present.
 */
void __dev_set_rx_mode(struct net_device *dev)
{
4272 4273
	const struct net_device_ops *ops = dev->netdev_ops;

4274 4275 4276 4277 4278
	/* dev_open will call this function so the list will stay sane. */
	if (!(dev->flags&IFF_UP))
		return;

	if (!netif_device_present(dev))
4279
		return;
4280

4281 4282
	if (ops->ndo_set_rx_mode)
		ops->ndo_set_rx_mode(dev);
4283 4284 4285 4286
	else {
		/* Unicast addresses changes may only happen under the rtnl,
		 * therefore calling __dev_set_promiscuity here is safe.
		 */
4287
		if (!netdev_uc_empty(dev) && !dev->uc_promisc) {
4288 4289
			__dev_set_promiscuity(dev, 1);
			dev->uc_promisc = 1;
4290
		} else if (netdev_uc_empty(dev) && dev->uc_promisc) {
4291 4292 4293 4294
			__dev_set_promiscuity(dev, -1);
			dev->uc_promisc = 0;
		}

4295 4296
		if (ops->ndo_set_multicast_list)
			ops->ndo_set_multicast_list(dev);
4297 4298 4299 4300 4301
	}
}

void dev_set_rx_mode(struct net_device *dev)
{
4302
	netif_addr_lock_bh(dev);
4303
	__dev_set_rx_mode(dev);
4304
	netif_addr_unlock_bh(dev);
L
Linus Torvalds 已提交
4305 4306
}

4307 4308 4309 4310 4311 4312
/**
 *	dev_get_flags - get flags reported to userspace
 *	@dev: device
 *
 *	Get the combination of flag bits exported through APIs to userspace.
 */
L
Linus Torvalds 已提交
4313 4314 4315 4316 4317 4318
unsigned dev_get_flags(const struct net_device *dev)
{
	unsigned flags;

	flags = (dev->flags & ~(IFF_PROMISC |
				IFF_ALLMULTI |
S
Stefan Rompf 已提交
4319 4320 4321
				IFF_RUNNING |
				IFF_LOWER_UP |
				IFF_DORMANT)) |
L
Linus Torvalds 已提交
4322 4323 4324
		(dev->gflags & (IFF_PROMISC |
				IFF_ALLMULTI));

S
Stefan Rompf 已提交
4325 4326 4327 4328 4329 4330 4331 4332
	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
4333 4334 4335

	return flags;
}
E
Eric Dumazet 已提交
4336
EXPORT_SYMBOL(dev_get_flags);
L
Linus Torvalds 已提交
4337

4338
int __dev_change_flags(struct net_device *dev, unsigned int flags)
L
Linus Torvalds 已提交
4339 4340
{
	int old_flags = dev->flags;
4341
	int ret;
L
Linus Torvalds 已提交
4342

4343 4344
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358
	/*
	 *	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.
	 */

4359 4360
	if ((old_flags ^ flags) & IFF_MULTICAST)
		dev_change_rx_flags(dev, IFF_MULTICAST);
4361

4362
	dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4363 4364 4365 4366 4367 4368 4369 4370 4371

	/*
	 *	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  ? */
4372
		ret = ((old_flags & IFF_UP) ? __dev_close : __dev_open)(dev);
L
Linus Torvalds 已提交
4373 4374

		if (!ret)
4375
			dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4376 4377 4378
	}

	if ((flags ^ dev->gflags) & IFF_PROMISC) {
E
Eric Dumazet 已提交
4379 4380
		int inc = (flags & IFF_PROMISC) ? 1 : -1;

L
Linus Torvalds 已提交
4381 4382 4383 4384 4385 4386 4387 4388 4389
		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
4390 4391
		int inc = (flags & IFF_ALLMULTI) ? 1 : -1;

L
Linus Torvalds 已提交
4392 4393 4394 4395
		dev->gflags ^= IFF_ALLMULTI;
		dev_set_allmulti(dev, inc);
	}

4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432
	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.
 */
int dev_change_flags(struct net_device *dev, unsigned flags)
{
	int ret, changes;
	int old_flags = dev->flags;

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

	changes = old_flags ^ dev->flags;
4433 4434
	if (changes)
		rtmsg_ifinfo(RTM_NEWLINK, dev, changes);
L
Linus Torvalds 已提交
4435

4436
	__dev_notify_flags(dev, old_flags);
L
Linus Torvalds 已提交
4437 4438
	return ret;
}
E
Eric Dumazet 已提交
4439
EXPORT_SYMBOL(dev_change_flags);
L
Linus Torvalds 已提交
4440

4441 4442 4443 4444 4445 4446 4447
/**
 *	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
4448 4449
int dev_set_mtu(struct net_device *dev, int new_mtu)
{
4450
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463
	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;
4464 4465
	if (ops->ndo_change_mtu)
		err = ops->ndo_change_mtu(dev, new_mtu);
L
Linus Torvalds 已提交
4466 4467
	else
		dev->mtu = new_mtu;
4468

L
Linus Torvalds 已提交
4469
	if (!err && dev->flags & IFF_UP)
4470
		call_netdevice_notifiers(NETDEV_CHANGEMTU, dev);
L
Linus Torvalds 已提交
4471 4472
	return err;
}
E
Eric Dumazet 已提交
4473
EXPORT_SYMBOL(dev_set_mtu);
L
Linus Torvalds 已提交
4474

4475 4476 4477 4478 4479 4480 4481
/**
 *	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
4482 4483
int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa)
{
4484
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4485 4486
	int err;

4487
	if (!ops->ndo_set_mac_address)
L
Linus Torvalds 已提交
4488 4489 4490 4491 4492
		return -EOPNOTSUPP;
	if (sa->sa_family != dev->type)
		return -EINVAL;
	if (!netif_device_present(dev))
		return -ENODEV;
4493
	err = ops->ndo_set_mac_address(dev, sa);
L
Linus Torvalds 已提交
4494
	if (!err)
4495
		call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
L
Linus Torvalds 已提交
4496 4497
	return err;
}
E
Eric Dumazet 已提交
4498
EXPORT_SYMBOL(dev_set_mac_address);
L
Linus Torvalds 已提交
4499 4500

/*
E
Eric Dumazet 已提交
4501
 *	Perform the SIOCxIFxxx calls, inside rcu_read_lock()
L
Linus Torvalds 已提交
4502
 */
4503
static int dev_ifsioc_locked(struct net *net, struct ifreq *ifr, unsigned int cmd)
L
Linus Torvalds 已提交
4504 4505
{
	int err;
E
Eric Dumazet 已提交
4506
	struct net_device *dev = dev_get_by_name_rcu(net, ifr->ifr_name);
L
Linus Torvalds 已提交
4507 4508 4509 4510 4511

	if (!dev)
		return -ENODEV;

	switch (cmd) {
E
Eric Dumazet 已提交
4512 4513 4514
	case SIOCGIFFLAGS:	/* Get interface flags */
		ifr->ifr_flags = (short) dev_get_flags(dev);
		return 0;
L
Linus Torvalds 已提交
4515

E
Eric Dumazet 已提交
4516 4517 4518 4519
	case SIOCGIFMETRIC:	/* Get the metric on the interface
				   (currently unused) */
		ifr->ifr_metric = 0;
		return 0;
L
Linus Torvalds 已提交
4520

E
Eric Dumazet 已提交
4521 4522 4523
	case SIOCGIFMTU:	/* Get the MTU of a device */
		ifr->ifr_mtu = dev->mtu;
		return 0;
L
Linus Torvalds 已提交
4524

E
Eric Dumazet 已提交
4525 4526 4527 4528 4529 4530 4531 4532
	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
4533

E
Eric Dumazet 已提交
4534 4535 4536
	case SIOCGIFSLAVE:
		err = -EINVAL;
		break;
4537

E
Eric Dumazet 已提交
4538 4539 4540 4541 4542 4543 4544 4545
	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;
4546

E
Eric Dumazet 已提交
4547 4548 4549
	case SIOCGIFINDEX:
		ifr->ifr_ifindex = dev->ifindex;
		return 0;
4550

E
Eric Dumazet 已提交
4551 4552 4553
	case SIOCGIFTXQLEN:
		ifr->ifr_qlen = dev->tx_queue_len;
		return 0;
4554

E
Eric Dumazet 已提交
4555 4556 4557 4558 4559 4560 4561
	default:
		/* dev_ioctl() should ensure this case
		 * is never reached
		 */
		WARN_ON(1);
		err = -EINVAL;
		break;
4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573

	}
	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
4574
	const struct net_device_ops *ops;
4575 4576 4577 4578

	if (!dev)
		return -ENODEV;

J
Jarek Poplawski 已提交
4579 4580
	ops = dev->netdev_ops;

4581
	switch (cmd) {
E
Eric Dumazet 已提交
4582 4583
	case SIOCSIFFLAGS:	/* Set interface flags */
		return dev_change_flags(dev, ifr->ifr_flags);
4584

E
Eric Dumazet 已提交
4585 4586 4587
	case SIOCSIFMETRIC:	/* Set the metric on the interface
				   (currently unused) */
		return -EOPNOTSUPP;
4588

E
Eric Dumazet 已提交
4589 4590
	case SIOCSIFMTU:	/* Set the MTU of a device */
		return dev_set_mtu(dev, ifr->ifr_mtu);
L
Linus Torvalds 已提交
4591

E
Eric Dumazet 已提交
4592 4593
	case SIOCSIFHWADDR:
		return dev_set_mac_address(dev, &ifr->ifr_hwaddr);
L
Linus Torvalds 已提交
4594

E
Eric Dumazet 已提交
4595 4596 4597 4598 4599 4600 4601
	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
4602

E
Eric Dumazet 已提交
4603 4604
	case SIOCSIFMAP:
		if (ops->ndo_set_config) {
L
Linus Torvalds 已提交
4605 4606
			if (!netif_device_present(dev))
				return -ENODEV;
E
Eric Dumazet 已提交
4607 4608 4609
			return ops->ndo_set_config(dev, &ifr->ifr_map);
		}
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
4610

E
Eric Dumazet 已提交
4611 4612 4613 4614 4615 4616
	case SIOCADDMULTI:
		if ((!ops->ndo_set_multicast_list && !ops->ndo_set_rx_mode) ||
		    ifr->ifr_hwaddr.sa_family != AF_UNSPEC)
			return -EINVAL;
		if (!netif_device_present(dev))
			return -ENODEV;
4617
		return dev_mc_add_global(dev, ifr->ifr_hwaddr.sa_data);
E
Eric Dumazet 已提交
4618 4619 4620 4621 4622 4623 4624

	case SIOCDELMULTI:
		if ((!ops->ndo_set_multicast_list && !ops->ndo_set_rx_mode) ||
		    ifr->ifr_hwaddr.sa_family != AF_UNSPEC)
			return -EINVAL;
		if (!netif_device_present(dev))
			return -ENODEV;
4625
		return dev_mc_del_global(dev, ifr->ifr_hwaddr.sa_data);
L
Linus Torvalds 已提交
4626

E
Eric Dumazet 已提交
4627 4628 4629 4630 4631
	case SIOCSIFTXQLEN:
		if (ifr->ifr_qlen < 0)
			return -EINVAL;
		dev->tx_queue_len = ifr->ifr_qlen;
		return 0;
L
Linus Torvalds 已提交
4632

E
Eric Dumazet 已提交
4633 4634 4635
	case SIOCSIFNAME:
		ifr->ifr_newname[IFNAMSIZ-1] = '\0';
		return dev_change_name(dev, ifr->ifr_newname);
L
Linus Torvalds 已提交
4636

E
Eric Dumazet 已提交
4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664
	/*
	 *	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
4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676

	}
	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
4677
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
4678 4679 4680 4681 4682 4683 4684 4685 4686
 *	@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.
 */

4687
int dev_ioctl(struct net *net, unsigned int cmd, void __user *arg)
L
Linus Torvalds 已提交
4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698
{
	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) {
4699
		rtnl_lock();
4700
		ret = dev_ifconf(net, (char __user *) arg);
4701
		rtnl_unlock();
L
Linus Torvalds 已提交
4702 4703 4704
		return ret;
	}
	if (cmd == SIOCGIFNAME)
4705
		return dev_ifname(net, (struct ifreq __user *)arg);
L
Linus Torvalds 已提交
4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720

	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
4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735
	/*
	 *	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
4736
		rcu_read_lock();
E
Eric Dumazet 已提交
4737
		ret = dev_ifsioc_locked(net, &ifr, cmd);
E
Eric Dumazet 已提交
4738
		rcu_read_unlock();
E
Eric Dumazet 已提交
4739 4740 4741 4742 4743 4744 4745 4746
		if (!ret) {
			if (colon)
				*colon = ':';
			if (copy_to_user(arg, &ifr,
					 sizeof(struct ifreq)))
				ret = -EFAULT;
		}
		return ret;
L
Linus Torvalds 已提交
4747

E
Eric Dumazet 已提交
4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760
	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
4761

E
Eric Dumazet 已提交
4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784
	/*
	 *	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
4785

E
Eric Dumazet 已提交
4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836
	/*
	 *	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:
		return -EINVAL;

	/*
	 *	Unknown or private ioctl.
	 */
	default:
		if (cmd == SIOCWANDEV ||
		    (cmd >= SIOCDEVPRIVATE &&
		     cmd <= SIOCDEVPRIVATE + 15)) {
4837
			dev_load(net, ifr.ifr_name);
L
Linus Torvalds 已提交
4838
			rtnl_lock();
4839
			ret = dev_ifsioc(net, &ifr, cmd);
L
Linus Torvalds 已提交
4840
			rtnl_unlock();
E
Eric Dumazet 已提交
4841 4842 4843
			if (!ret && copy_to_user(arg, &ifr,
						 sizeof(struct ifreq)))
				ret = -EFAULT;
L
Linus Torvalds 已提交
4844
			return ret;
E
Eric Dumazet 已提交
4845 4846 4847 4848 4849
		}
		/* Take care of Wireless Extensions */
		if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)
			return wext_handle_ioctl(net, &ifr, cmd, arg);
		return -EINVAL;
L
Linus Torvalds 已提交
4850 4851 4852 4853 4854 4855
	}
}


/**
 *	dev_new_index	-	allocate an ifindex
4856
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
4857 4858 4859 4860 4861
 *
 *	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.
 */
4862
static int dev_new_index(struct net *net)
L
Linus Torvalds 已提交
4863 4864 4865 4866 4867
{
	static int ifindex;
	for (;;) {
		if (++ifindex <= 0)
			ifindex = 1;
4868
		if (!__dev_get_by_index(net, ifindex))
L
Linus Torvalds 已提交
4869 4870 4871 4872 4873
			return ifindex;
	}
}

/* Delayed registration/unregisteration */
4874
static LIST_HEAD(net_todo_list);
L
Linus Torvalds 已提交
4875

4876
static void net_set_todo(struct net_device *dev)
L
Linus Torvalds 已提交
4877 4878 4879 4880
{
	list_add_tail(&dev->todo_list, &net_todo_list);
}

4881
static void rollback_registered_many(struct list_head *head)
4882
{
4883
	struct net_device *dev, *tmp;
4884

4885 4886 4887
	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

4888
	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
4889
		/* Some devices call without registering
4890 4891
		 * for initialization unwind. Remove those
		 * devices and proceed with the remaining.
4892 4893 4894 4895
		 */
		if (dev->reg_state == NETREG_UNINITIALIZED) {
			pr_debug("unregister_netdevice: device %s/%p never "
				 "was registered\n", dev->name, dev);
4896

4897
			WARN_ON(1);
4898 4899
			list_del(&dev->unreg_list);
			continue;
4900
		}
4901

4902
		BUG_ON(dev->reg_state != NETREG_REGISTERED);
4903

4904 4905
		/* If device is running, close it first. */
		dev_close(dev);
4906

4907 4908
		/* And unlink it from device chain. */
		unlist_netdevice(dev);
4909

4910 4911
		dev->reg_state = NETREG_UNREGISTERING;
	}
4912 4913 4914

	synchronize_net();

4915 4916 4917
	list_for_each_entry(dev, head, unreg_list) {
		/* Shutdown queueing discipline. */
		dev_shutdown(dev);
4918 4919


4920 4921 4922 4923
		/* Notify protocols, that we are about to destroy
		   this device. They should clean all the things.
		*/
		call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
4924

4925 4926 4927 4928
		if (!dev->rtnl_link_ops ||
		    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
			rtmsg_ifinfo(RTM_DELLINK, dev, ~0U);

4929 4930 4931
		/*
		 *	Flush the unicast and multicast chains
		 */
4932
		dev_uc_flush(dev);
4933
		dev_mc_flush(dev);
4934

4935 4936
		if (dev->netdev_ops->ndo_uninit)
			dev->netdev_ops->ndo_uninit(dev);
4937

4938 4939
		/* Notifier chain MUST detach us from master device. */
		WARN_ON(dev->master);
4940

4941 4942 4943
		/* Remove entries from kobject tree */
		netdev_unregister_kobject(dev);
	}
4944

4945
	/* Process any work delayed until the end of the batch */
4946
	dev = list_first_entry(head, struct net_device, unreg_list);
4947
	call_netdevice_notifiers(NETDEV_UNREGISTER_BATCH, dev);
4948

4949
	rcu_barrier();
4950

4951
	list_for_each_entry(dev, head, unreg_list)
4952 4953 4954 4955 4956 4957 4958 4959 4960
		dev_put(dev);
}

static void rollback_registered(struct net_device *dev)
{
	LIST_HEAD(single);

	list_add(&dev->unreg_list, &single);
	rollback_registered_many(&single);
4961 4962
}

4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002
unsigned long netdev_fix_features(unsigned long features, const char *name)
{
	/* Fix illegal SG+CSUM combinations. */
	if ((features & NETIF_F_SG) &&
	    !(features & NETIF_F_ALL_CSUM)) {
		if (name)
			printk(KERN_NOTICE "%s: Dropping NETIF_F_SG since no "
			       "checksum feature.\n", name);
		features &= ~NETIF_F_SG;
	}

	/* TSO requires that SG is present as well. */
	if ((features & NETIF_F_TSO) && !(features & NETIF_F_SG)) {
		if (name)
			printk(KERN_NOTICE "%s: Dropping NETIF_F_TSO since no "
			       "SG feature.\n", name);
		features &= ~NETIF_F_TSO;
	}

	if (features & NETIF_F_UFO) {
		if (!(features & NETIF_F_GEN_CSUM)) {
			if (name)
				printk(KERN_ERR "%s: Dropping NETIF_F_UFO "
				       "since no NETIF_F_HW_CSUM feature.\n",
				       name);
			features &= ~NETIF_F_UFO;
		}

		if (!(features & NETIF_F_SG)) {
			if (name)
				printk(KERN_ERR "%s: Dropping NETIF_F_UFO "
				       "since no NETIF_F_SG feature.\n", name);
			features &= ~NETIF_F_UFO;
		}
	}

	return features;
}
EXPORT_SYMBOL(netdev_fix_features);

5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029
/**
 *	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);

5030 5031 5032 5033
static int netif_alloc_rx_queues(struct net_device *dev)
{
#ifdef CONFIG_RPS
	unsigned int i, count = dev->num_rx_queues;
T
Tom Herbert 已提交
5034
	struct netdev_rx_queue *rx;
5035

T
Tom Herbert 已提交
5036
	BUG_ON(count < 1);
5037

T
Tom Herbert 已提交
5038 5039 5040 5041
	rx = kcalloc(count, sizeof(struct netdev_rx_queue), GFP_KERNEL);
	if (!rx) {
		pr_err("netdev: Unable to allocate %u rx queues.\n", count);
		return -ENOMEM;
5042
	}
T
Tom Herbert 已提交
5043 5044 5045 5046 5047 5048 5049 5050
	dev->_rx = rx;

	/*
	 * Set a pointer to first element in the array which holds the
	 * reference count.
	 */
	for (i = 0; i < count; i++)
		rx[i].first = rx;
5051 5052 5053 5054
#endif
	return 0;
}

5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089
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) {
		pr_err("netdev: Unable to allocate %u tx queues.\n",
		       count);
		return -ENOMEM;
	}
	dev->_tx = tx;
	return 0;
}

static void netdev_init_one_queue(struct net_device *dev,
				  struct netdev_queue *queue,
				  void *_unused)
{
	queue->dev = dev;

	/* Initialize queue lock */
	spin_lock_init(&queue->_xmit_lock);
	netdev_set_xmit_lockdep_class(&queue->_xmit_lock, dev->type);
	queue->xmit_lock_owner = -1;
}

static void netdev_init_queues(struct net_device *dev)
{
	netdev_for_each_tx_queue(dev, netdev_init_one_queue, NULL);
	spin_lock_init(&dev->tx_global_lock);
}

L
Linus Torvalds 已提交
5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109
/**
 *	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;
5110
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
5111 5112 5113 5114

	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

5115 5116
	might_sleep();

L
Linus Torvalds 已提交
5117 5118
	/* When net_device's are persistent, this will be fatal. */
	BUG_ON(dev->reg_state != NETREG_UNINITIALIZED);
5119
	BUG_ON(!net);
L
Linus Torvalds 已提交
5120

5121
	spin_lock_init(&dev->addr_list_lock);
5122
	netdev_set_addr_lockdep_class(dev);
L
Linus Torvalds 已提交
5123 5124 5125

	dev->iflink = -1;

5126 5127 5128
	ret = netif_alloc_rx_queues(dev);
	if (ret)
		goto out;
T
Tom Herbert 已提交
5129

5130 5131 5132 5133 5134 5135
	ret = netif_alloc_netdev_queues(dev);
	if (ret)
		goto out;

	netdev_init_queues(dev);

L
Linus Torvalds 已提交
5136
	/* Init, if this function is available */
5137 5138
	if (dev->netdev_ops->ndo_init) {
		ret = dev->netdev_ops->ndo_init(dev);
L
Linus Torvalds 已提交
5139 5140 5141
		if (ret) {
			if (ret > 0)
				ret = -EIO;
5142
			goto out;
L
Linus Torvalds 已提交
5143 5144
		}
	}
5145

5146
	ret = dev_get_valid_name(dev, dev->name, 0);
5147
	if (ret)
5148
		goto err_uninit;
L
Linus Torvalds 已提交
5149

5150
	dev->ifindex = dev_new_index(net);
L
Linus Torvalds 已提交
5151 5152 5153
	if (dev->iflink == -1)
		dev->iflink = dev->ifindex;

5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168
	/* Fix illegal checksum combinations */
	if ((dev->features & NETIF_F_HW_CSUM) &&
	    (dev->features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
		printk(KERN_NOTICE "%s: mixed HW and IP checksum settings.\n",
		       dev->name);
		dev->features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
	}

	if ((dev->features & NETIF_F_NO_CSUM) &&
	    (dev->features & (NETIF_F_HW_CSUM|NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
		printk(KERN_NOTICE "%s: mixed no checksumming and other settings.\n",
		       dev->name);
		dev->features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM|NETIF_F_HW_CSUM);
	}

5169
	dev->features = netdev_fix_features(dev->features, dev->name);
L
Linus Torvalds 已提交
5170

5171 5172 5173 5174
	/* Enable software GSO if SG is supported. */
	if (dev->features & NETIF_F_SG)
		dev->features |= NETIF_F_GSO;

5175 5176 5177
	/* Enable GRO and NETIF_F_HIGHDMA for vlans by default,
	 * vlan_dev_init() will do the dev->features check, so these features
	 * are enabled only if supported by underlying device.
5178
	 */
5179
	dev->vlan_features |= (NETIF_F_GRO | NETIF_F_HIGHDMA);
5180

5181 5182 5183 5184 5185
	ret = call_netdevice_notifiers(NETDEV_POST_INIT, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		goto err_uninit;

5186
	ret = netdev_register_kobject(dev);
5187
	if (ret)
5188
		goto err_uninit;
5189 5190
	dev->reg_state = NETREG_REGISTERED;

L
Linus Torvalds 已提交
5191 5192 5193 5194 5195 5196 5197 5198 5199
	/*
	 *	Default initial state at registry is that the
	 *	device is present.
	 */

	set_bit(__LINK_STATE_PRESENT, &dev->state);

	dev_init_scheduler(dev);
	dev_hold(dev);
5200
	list_netdevice(dev);
L
Linus Torvalds 已提交
5201 5202

	/* Notify protocols, that a new device appeared. */
5203
	ret = call_netdevice_notifiers(NETDEV_REGISTER, dev);
5204
	ret = notifier_to_errno(ret);
5205 5206 5207 5208
	if (ret) {
		rollback_registered(dev);
		dev->reg_state = NETREG_UNREGISTERED;
	}
5209 5210 5211 5212
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
5213 5214 5215
	if (!dev->rtnl_link_ops ||
	    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
		rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);
L
Linus Torvalds 已提交
5216 5217 5218

out:
	return ret;
5219 5220

err_uninit:
5221 5222
	if (dev->netdev_ops->ndo_uninit)
		dev->netdev_ops->ndo_uninit(dev);
5223
	goto out;
L
Linus Torvalds 已提交
5224
}
E
Eric Dumazet 已提交
5225
EXPORT_SYMBOL(register_netdevice);
L
Linus Torvalds 已提交
5226

5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257
/**
 *	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
5258 5259 5260 5261 5262
	/* Note : We dont allocate pcpu_refcnt for dummy devices,
	 * because users of this 'device' dont need to change
	 * its refcount.
	 */

5263 5264 5265 5266 5267
	return 0;
}
EXPORT_SYMBOL_GPL(init_dummy_netdev);


L
Linus Torvalds 已提交
5268 5269 5270 5271 5272 5273 5274 5275 5276
/**
 *	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.
 *
5277
 *	This is a wrapper around register_netdevice that takes the rtnl semaphore
L
Linus Torvalds 已提交
5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295
 *	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();

	/*
	 * If the name is a format string the caller wants us to do a
	 * name allocation.
	 */
	if (strchr(dev->name, '%')) {
		err = dev_alloc_name(dev, dev->name);
		if (err < 0)
			goto out;
	}
5296

L
Linus Torvalds 已提交
5297 5298 5299 5300 5301 5302 5303
	err = register_netdevice(dev);
out:
	rtnl_unlock();
	return err;
}
EXPORT_SYMBOL(register_netdev);

E
Eric Dumazet 已提交
5304 5305 5306 5307 5308 5309 5310 5311 5312 5313
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);

L
Linus Torvalds 已提交
5314 5315 5316 5317 5318 5319 5320 5321 5322
/*
 * netdev_wait_allrefs - wait until all references are gone.
 *
 * 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
5323
 * call dev_put.
L
Linus Torvalds 已提交
5324 5325 5326 5327
 */
static void netdev_wait_allrefs(struct net_device *dev)
{
	unsigned long rebroadcast_time, warning_time;
E
Eric Dumazet 已提交
5328
	int refcnt;
L
Linus Torvalds 已提交
5329

5330 5331
	linkwatch_forget_dev(dev);

L
Linus Torvalds 已提交
5332
	rebroadcast_time = warning_time = jiffies;
E
Eric Dumazet 已提交
5333 5334 5335
	refcnt = netdev_refcnt_read(dev);

	while (refcnt != 0) {
L
Linus Torvalds 已提交
5336
		if (time_after(jiffies, rebroadcast_time + 1 * HZ)) {
5337
			rtnl_lock();
L
Linus Torvalds 已提交
5338 5339

			/* Rebroadcast unregister notification */
5340
			call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5341
			/* don't resend NETDEV_UNREGISTER_BATCH, _BATCH users
5342
			 * should have already handle it the first time */
L
Linus Torvalds 已提交
5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354

			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();
			}

5355
			__rtnl_unlock();
L
Linus Torvalds 已提交
5356 5357 5358 5359 5360 5361

			rebroadcast_time = jiffies;
		}

		msleep(250);

E
Eric Dumazet 已提交
5362 5363
		refcnt = netdev_refcnt_read(dev);

L
Linus Torvalds 已提交
5364 5365 5366 5367
		if (time_after(jiffies, warning_time + 10 * HZ)) {
			printk(KERN_EMERG "unregister_netdevice: "
			       "waiting for %s to become free. Usage "
			       "count = %d\n",
E
Eric Dumazet 已提交
5368
			       dev->name, refcnt);
L
Linus Torvalds 已提交
5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387
			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
5388
 * We are invoked by rtnl_unlock().
L
Linus Torvalds 已提交
5389
 * This allows us to deal with problems:
5390
 * 1) We can delete sysfs objects which invoke hotplug
L
Linus Torvalds 已提交
5391 5392 5393
 *    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
5394 5395 5396
 *
 * 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
5397 5398 5399
 */
void netdev_run_todo(void)
{
5400
	struct list_head list;
L
Linus Torvalds 已提交
5401 5402

	/* Snapshot list, allow later requests */
5403
	list_replace_init(&net_todo_list, &list);
H
Herbert Xu 已提交
5404 5405

	__rtnl_unlock();
5406

L
Linus Torvalds 已提交
5407 5408
	while (!list_empty(&list)) {
		struct net_device *dev
5409
			= list_first_entry(&list, struct net_device, todo_list);
L
Linus Torvalds 已提交
5410 5411
		list_del(&dev->todo_list);

5412 5413 5414 5415 5416 5417
		if (unlikely(dev->reg_state != NETREG_UNREGISTERING)) {
			printk(KERN_ERR "network todo '%s' but state %d\n",
			       dev->name, dev->reg_state);
			dump_stack();
			continue;
		}
L
Linus Torvalds 已提交
5418

5419
		dev->reg_state = NETREG_UNREGISTERED;
L
Linus Torvalds 已提交
5420

5421
		on_each_cpu(flush_backlog, dev, 1);
5422

5423
		netdev_wait_allrefs(dev);
L
Linus Torvalds 已提交
5424

5425
		/* paranoia */
E
Eric Dumazet 已提交
5426
		BUG_ON(netdev_refcnt_read(dev));
E
Eric Dumazet 已提交
5427
		WARN_ON(rcu_dereference_raw(dev->ip_ptr));
5428 5429
		WARN_ON(dev->ip6_ptr);
		WARN_ON(dev->dn_ptr);
L
Linus Torvalds 已提交
5430

5431 5432
		if (dev->destructor)
			dev->destructor(dev);
5433 5434 5435

		/* Free network device */
		kobject_put(&dev->dev.kobj);
L
Linus Torvalds 已提交
5436 5437 5438
	}
}

5439 5440 5441
/**
 *	dev_txq_stats_fold - fold tx_queues stats
 *	@dev: device to get statistics from
5442
 *	@stats: struct rtnl_link_stats64 to hold results
5443 5444
 */
void dev_txq_stats_fold(const struct net_device *dev,
5445
			struct rtnl_link_stats64 *stats)
5446
{
E
Eric Dumazet 已提交
5447
	u64 tx_bytes = 0, tx_packets = 0, tx_dropped = 0;
5448 5449 5450 5451 5452
	unsigned int i;
	struct netdev_queue *txq;

	for (i = 0; i < dev->num_tx_queues; i++) {
		txq = netdev_get_tx_queue(dev, i);
E
Eric Dumazet 已提交
5453
		spin_lock_bh(&txq->_xmit_lock);
5454 5455 5456
		tx_bytes   += txq->tx_bytes;
		tx_packets += txq->tx_packets;
		tx_dropped += txq->tx_dropped;
E
Eric Dumazet 已提交
5457
		spin_unlock_bh(&txq->_xmit_lock);
5458 5459 5460 5461 5462 5463 5464 5465 5466
	}
	if (tx_bytes || tx_packets || tx_dropped) {
		stats->tx_bytes   = tx_bytes;
		stats->tx_packets = tx_packets;
		stats->tx_dropped = tx_dropped;
	}
}
EXPORT_SYMBOL(dev_txq_stats_fold);

5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487
/* Convert net_device_stats to rtnl_link_stats64.  They have the same
 * fields in the same order, with only the type differing.
 */
static void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
				    const struct net_device_stats *netdev_stats)
{
#if BITS_PER_LONG == 64
        BUILD_BUG_ON(sizeof(*stats64) != sizeof(*netdev_stats));
        memcpy(stats64, netdev_stats, sizeof(*stats64));
#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
}

5488 5489 5490
/**
 *	dev_get_stats	- get network device statistics
 *	@dev: device to get statistics from
5491
 *	@storage: place to store stats
5492
 *
5493 5494 5495 5496
 *	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.
5497
 */
5498 5499
struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
					struct rtnl_link_stats64 *storage)
5500
{
5501 5502
	const struct net_device_ops *ops = dev->netdev_ops;

5503 5504
	if (ops->ndo_get_stats64) {
		memset(storage, 0, sizeof(*storage));
5505 5506
		ops->ndo_get_stats64(dev, storage);
	} else if (ops->ndo_get_stats) {
5507
		netdev_stats_to_stats64(storage, ops->ndo_get_stats(dev));
5508 5509 5510
	} else {
		netdev_stats_to_stats64(storage, &dev->stats);
		dev_txq_stats_fold(dev, storage);
5511
	}
5512
	storage->rx_dropped += atomic_long_read(&dev->rx_dropped);
5513
	return storage;
R
Rusty Russell 已提交
5514
}
5515
EXPORT_SYMBOL(dev_get_stats);
R
Rusty Russell 已提交
5516

5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534
struct netdev_queue *dev_ingress_queue_create(struct net_device *dev)
{
	struct netdev_queue *queue = dev_ingress_queue(dev);

#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;
}

L
Linus Torvalds 已提交
5535
/**
5536
 *	alloc_netdev_mq - allocate network device
L
Linus Torvalds 已提交
5537 5538 5539
 *	@sizeof_priv:	size of private data to allocate space for
 *	@name:		device name format string
 *	@setup:		callback to initialize device
5540
 *	@queue_count:	the number of subqueues to allocate
L
Linus Torvalds 已提交
5541 5542
 *
 *	Allocates a struct net_device with private data area for driver use
5543 5544
 *	and performs basic initialization.  Also allocates subquue structs
 *	for each queue on the device at the end of the netdevice.
L
Linus Torvalds 已提交
5545
 */
5546 5547
struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
		void (*setup)(struct net_device *), unsigned int queue_count)
L
Linus Torvalds 已提交
5548 5549
{
	struct net_device *dev;
5550
	size_t alloc_size;
5551
	struct net_device *p;
L
Linus Torvalds 已提交
5552

5553 5554
	BUG_ON(strlen(name) >= sizeof(dev->name));

5555 5556 5557 5558 5559 5560
	if (queue_count < 1) {
		pr_err("alloc_netdev: Unable to allocate device "
		       "with zero queues.\n");
		return NULL;
	}

5561
	alloc_size = sizeof(struct net_device);
5562 5563
	if (sizeof_priv) {
		/* ensure 32-byte alignment of private area */
5564
		alloc_size = ALIGN(alloc_size, NETDEV_ALIGN);
5565 5566 5567
		alloc_size += sizeof_priv;
	}
	/* ensure 32-byte alignment of whole construct */
5568
	alloc_size += NETDEV_ALIGN - 1;
L
Linus Torvalds 已提交
5569

5570
	p = kzalloc(alloc_size, GFP_KERNEL);
L
Linus Torvalds 已提交
5571
	if (!p) {
5572
		printk(KERN_ERR "alloc_netdev: Unable to allocate device.\n");
L
Linus Torvalds 已提交
5573 5574 5575
		return NULL;
	}

5576
	dev = PTR_ALIGN(p, NETDEV_ALIGN);
L
Linus Torvalds 已提交
5577
	dev->padded = (char *)dev - (char *)p;
5578

E
Eric Dumazet 已提交
5579 5580
	dev->pcpu_refcnt = alloc_percpu(int);
	if (!dev->pcpu_refcnt)
5581
		goto free_p;
5582

E
Eric Dumazet 已提交
5583 5584 5585
	if (dev_addr_init(dev))
		goto free_pcpu;

5586
	dev_mc_init(dev);
5587
	dev_uc_init(dev);
J
Jiri Pirko 已提交
5588

5589
	dev_net_set(dev, &init_net);
L
Linus Torvalds 已提交
5590

5591
	dev->num_tx_queues = queue_count;
5592
	dev->real_num_tx_queues = queue_count;
5593

E
Eric Dumazet 已提交
5594
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
5595
	dev->num_rx_queues = queue_count;
5596
	dev->real_num_rx_queues = queue_count;
E
Eric Dumazet 已提交
5597
#endif
T
Tom Herbert 已提交
5598

5599
	dev->gso_max_size = GSO_MAX_SIZE;
L
Linus Torvalds 已提交
5600

5601 5602
	INIT_LIST_HEAD(&dev->ethtool_ntuple_list.list);
	dev->ethtool_ntuple_list.count = 0;
5603
	INIT_LIST_HEAD(&dev->napi_list);
5604
	INIT_LIST_HEAD(&dev->unreg_list);
5605
	INIT_LIST_HEAD(&dev->link_watch_list);
5606
	dev->priv_flags = IFF_XMIT_DST_RELEASE;
L
Linus Torvalds 已提交
5607 5608 5609
	setup(dev);
	strcpy(dev->name, name);
	return dev;
5610

E
Eric Dumazet 已提交
5611 5612
free_pcpu:
	free_percpu(dev->pcpu_refcnt);
5613 5614 5615
free_p:
	kfree(p);
	return NULL;
L
Linus Torvalds 已提交
5616
}
5617
EXPORT_SYMBOL(alloc_netdev_mq);
L
Linus Torvalds 已提交
5618 5619 5620 5621 5622

/**
 *	free_netdev - free network device
 *	@dev: device
 *
5623 5624
 *	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
5625 5626 5627 5628
 *	If this is the last reference then it will be freed.
 */
void free_netdev(struct net_device *dev)
{
5629 5630
	struct napi_struct *p, *n;

5631 5632
	release_net(dev_net(dev));

5633 5634
	kfree(dev->_tx);

5635 5636
	kfree(rcu_dereference_raw(dev->ingress_queue));

5637 5638 5639
	/* Flush device addresses */
	dev_addr_flush(dev);

5640 5641 5642
	/* Clear ethtool n-tuple list */
	ethtool_ntuple_flush(dev);

5643 5644 5645
	list_for_each_entry_safe(p, n, &dev->napi_list, dev_list)
		netif_napi_del(p);

E
Eric Dumazet 已提交
5646 5647 5648
	free_percpu(dev->pcpu_refcnt);
	dev->pcpu_refcnt = NULL;

S
Stephen Hemminger 已提交
5649
	/*  Compatibility with error handling in drivers */
L
Linus Torvalds 已提交
5650 5651 5652 5653 5654 5655 5656 5657
	if (dev->reg_state == NETREG_UNINITIALIZED) {
		kfree((char *)dev - dev->padded);
		return;
	}

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

5658 5659
	/* will free via device release */
	put_device(&dev->dev);
L
Linus Torvalds 已提交
5660
}
E
Eric Dumazet 已提交
5661
EXPORT_SYMBOL(free_netdev);
5662

5663 5664 5665 5666 5667 5668
/**
 *	synchronize_net -  Synchronize with packet receive processing
 *
 *	Wait for packets currently being received to be done.
 *	Does not block later packets from starting.
 */
5669
void synchronize_net(void)
L
Linus Torvalds 已提交
5670 5671
{
	might_sleep();
5672
	synchronize_rcu();
L
Linus Torvalds 已提交
5673
}
E
Eric Dumazet 已提交
5674
EXPORT_SYMBOL(synchronize_net);
L
Linus Torvalds 已提交
5675 5676

/**
5677
 *	unregister_netdevice_queue - remove device from the kernel
L
Linus Torvalds 已提交
5678
 *	@dev: device
5679
 *	@head: list
5680
 *
L
Linus Torvalds 已提交
5681
 *	This function shuts down a device interface and removes it
5682
 *	from the kernel tables.
5683
 *	If head not NULL, device is queued to be unregistered later.
L
Linus Torvalds 已提交
5684 5685 5686 5687 5688
 *
 *	Callers must hold the rtnl semaphore.  You may want
 *	unregister_netdev() instead of this.
 */

5689
void unregister_netdevice_queue(struct net_device *dev, struct list_head *head)
L
Linus Torvalds 已提交
5690
{
5691 5692
	ASSERT_RTNL();

5693
	if (head) {
5694
		list_move_tail(&dev->unreg_list, head);
5695 5696 5697 5698 5699
	} else {
		rollback_registered(dev);
		/* Finish processing unregister after unlock */
		net_set_todo(dev);
	}
L
Linus Torvalds 已提交
5700
}
5701
EXPORT_SYMBOL(unregister_netdevice_queue);
L
Linus Torvalds 已提交
5702

5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716
/**
 *	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);
	}
}
5717
EXPORT_SYMBOL(unregister_netdevice_many);
5718

L
Linus Torvalds 已提交
5719 5720 5721 5722 5723
/**
 *	unregister_netdev - remove device from the kernel
 *	@dev: device
 *
 *	This function shuts down a device interface and removes it
5724
 *	from the kernel tables.
L
Linus Torvalds 已提交
5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737
 *
 *	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);

5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769
/**
 *	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;
5770
	if (net_eq(dev_net(dev), net))
5771 5772 5773 5774 5775 5776
		goto out;

	/* Pick the destination device name, and ensure
	 * we can use it in the destination network namespace.
	 */
	err = -EEXIST;
5777
	if (__dev_get_by_name(net, dev->name)) {
5778 5779 5780
		/* We get here if we can't use the current device name */
		if (!pat)
			goto out;
5781
		if (dev_get_valid_name(dev, pat, 1))
5782 5783 5784 5785 5786 5787 5788 5789
			goto out;
	}

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

	/* If device is running close it first. */
5790
	dev_close(dev);
5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802

	/* 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.
5803 5804 5805 5806

	   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.
5807 5808
	*/
	call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5809
	call_netdevice_notifiers(NETDEV_UNREGISTER_BATCH, dev);
5810 5811 5812 5813

	/*
	 *	Flush the unicast and multicast chains
	 */
5814
	dev_uc_flush(dev);
5815
	dev_mc_flush(dev);
5816 5817

	/* Actually switch the network namespace */
5818
	dev_net_set(dev, net);
5819 5820 5821 5822 5823 5824 5825 5826 5827

	/* 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;
	}

5828
	/* Fixup kobjects */
5829
	err = device_rename(&dev->dev, dev->name);
5830
	WARN_ON(err);
5831 5832 5833 5834 5835 5836 5837

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

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

5838 5839 5840 5841 5842 5843
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
	rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);

5844 5845 5846 5847 5848
	synchronize_net();
	err = 0;
out:
	return err;
}
5849
EXPORT_SYMBOL_GPL(dev_change_net_namespace);
5850

L
Linus Torvalds 已提交
5851 5852 5853 5854 5855 5856 5857 5858 5859
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;

5860
	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN)
L
Linus Torvalds 已提交
5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876
		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. */
5877 5878 5879 5880 5881 5882
	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;
	}
L
Linus Torvalds 已提交
5883 5884 5885 5886 5887

	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_enable();

	/* Process offline CPU's input_pkt_queue */
5888
	while ((skb = __skb_dequeue(&oldsd->process_queue))) {
L
Linus Torvalds 已提交
5889
		netif_rx(skb);
5890
		input_queue_head_incr(oldsd);
T
Tom Herbert 已提交
5891
	}
5892
	while ((skb = __skb_dequeue(&oldsd->input_pkt_queue))) {
5893
		netif_rx(skb);
5894 5895
		input_queue_head_incr(oldsd);
	}
L
Linus Torvalds 已提交
5896 5897 5898 5899 5900

	return NOTIFY_OK;
}


5901
/**
5902 5903 5904 5905
 *	netdev_increment_features - increment feature set by one
 *	@all: current feature set
 *	@one: new feature set
 *	@mask: mask feature set
5906 5907
 *
 *	Computes a new feature set after adding a device with feature set
5908 5909
 *	@one to the master device with current feature set @all.  Will not
 *	enable anything that is off in @mask. Returns the new feature set.
5910
 */
5911 5912 5913 5914
unsigned long netdev_increment_features(unsigned long all, unsigned long one,
					unsigned long mask)
{
	/* If device needs checksumming, downgrade to it. */
E
Eric Dumazet 已提交
5915
	if (all & NETIF_F_NO_CSUM && !(one & NETIF_F_NO_CSUM))
5916 5917 5918 5919 5920 5921 5922 5923
		all ^= NETIF_F_NO_CSUM | (one & NETIF_F_ALL_CSUM);
	else if (mask & NETIF_F_ALL_CSUM) {
		/* If one device supports v4/v6 checksumming, set for all. */
		if (one & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM) &&
		    !(all & NETIF_F_GEN_CSUM)) {
			all &= ~NETIF_F_ALL_CSUM;
			all |= one & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM);
		}
5924

5925 5926 5927 5928 5929 5930
		/* If one device supports hw checksumming, set for all. */
		if (one & NETIF_F_GEN_CSUM && !(all & NETIF_F_GEN_CSUM)) {
			all &= ~NETIF_F_ALL_CSUM;
			all |= NETIF_F_HW_CSUM;
		}
	}
5931

5932
	one |= NETIF_F_ALL_CSUM;
5933

5934
	one |= all & NETIF_F_ONE_FOR_ALL;
5935
	all &= one | NETIF_F_LLTX | NETIF_F_GSO | NETIF_F_UFO;
5936
	all |= one & mask & NETIF_F_ONE_FOR_ALL;
5937 5938 5939

	return all;
}
5940
EXPORT_SYMBOL(netdev_increment_features);
5941

5942 5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954
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;
}

5955
/* Initialize per network namespace state */
5956
static int __net_init netdev_init(struct net *net)
5957 5958 5959
{
	INIT_LIST_HEAD(&net->dev_base_head);

5960 5961 5962
	net->dev_name_head = netdev_create_hash();
	if (net->dev_name_head == NULL)
		goto err_name;
5963

5964 5965 5966
	net->dev_index_head = netdev_create_hash();
	if (net->dev_index_head == NULL)
		goto err_idx;
5967 5968

	return 0;
5969 5970 5971 5972 5973

err_idx:
	kfree(net->dev_name_head);
err_name:
	return -ENOMEM;
5974 5975
}

5976 5977 5978 5979 5980 5981 5982 5983
/**
 *	netdev_drivername - network driver for the device
 *	@dev: network device
 *	@buffer: buffer for resulting name
 *	@len: size of buffer
 *
 *	Determine network driver for device.
 */
5984
char *netdev_drivername(const struct net_device *dev, char *buffer, int len)
5985
{
5986 5987
	const struct device_driver *driver;
	const struct device *parent;
5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003

	if (len <= 0 || !buffer)
		return buffer;
	buffer[0] = 0;

	parent = dev->dev.parent;

	if (!parent)
		return buffer;

	driver = parent->driver;
	if (driver && driver->name)
		strlcpy(buffer, driver->name, len);
	return buffer;
}

6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065
static int __netdev_printk(const char *level, const struct net_device *dev,
			   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;
}

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);

6066
static void __net_exit netdev_exit(struct net *net)
6067 6068 6069 6070 6071
{
	kfree(net->dev_name_head);
	kfree(net->dev_index_head);
}

6072
static struct pernet_operations __net_initdata netdev_net_ops = {
6073 6074 6075 6076
	.init = netdev_init,
	.exit = netdev_exit,
};

6077
static void __net_exit default_device_exit(struct net *net)
6078
{
6079
	struct net_device *dev, *aux;
6080
	/*
6081
	 * Push all migratable network devices back to the
6082 6083 6084
	 * initial network namespace
	 */
	rtnl_lock();
6085
	for_each_netdev_safe(net, dev, aux) {
6086
		int err;
6087
		char fb_name[IFNAMSIZ];
6088 6089 6090 6091 6092

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

6093 6094 6095
		/* Leave virtual devices for the generic cleanup */
		if (dev->rtnl_link_ops)
			continue;
6096

6097
		/* Push remaing network devices to init_net */
6098 6099
		snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
		err = dev_change_net_namespace(dev, &init_net, fb_name);
6100
		if (err) {
6101
			printk(KERN_EMERG "%s: failed to move %s to init_net: %d\n",
6102
				__func__, dev->name, err);
6103
			BUG();
6104 6105 6106 6107 6108
		}
	}
	rtnl_unlock();
}

6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132
static void __net_exit default_device_exit_batch(struct list_head *net_list)
{
	/* At exit all network devices most be removed from a network
	 * namespace.  Do this in the reverse order of registeration.
	 * 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);
	rtnl_unlock();
}

6133
static struct pernet_operations __net_initdata default_device_ops = {
6134
	.exit = default_device_exit,
6135
	.exit_batch = default_device_exit_batch,
6136 6137
};

L
Linus Torvalds 已提交
6138 6139 6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157
/*
 *	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;

6158
	if (netdev_kobject_init())
L
Linus Torvalds 已提交
6159 6160 6161
		goto out;

	INIT_LIST_HEAD(&ptype_all);
6162
	for (i = 0; i < PTYPE_HASH_SIZE; i++)
L
Linus Torvalds 已提交
6163 6164
		INIT_LIST_HEAD(&ptype_base[i]);

6165 6166
	if (register_pernet_subsys(&netdev_net_ops))
		goto out;
L
Linus Torvalds 已提交
6167 6168 6169 6170 6171

	/*
	 *	Initialise the packet receive queues.
	 */

6172
	for_each_possible_cpu(i) {
E
Eric Dumazet 已提交
6173
		struct softnet_data *sd = &per_cpu(softnet_data, i);
L
Linus Torvalds 已提交
6174

C
Changli Gao 已提交
6175
		memset(sd, 0, sizeof(*sd));
E
Eric Dumazet 已提交
6176
		skb_queue_head_init(&sd->input_pkt_queue);
6177
		skb_queue_head_init(&sd->process_queue);
E
Eric Dumazet 已提交
6178 6179
		sd->completion_queue = NULL;
		INIT_LIST_HEAD(&sd->poll_list);
6180 6181
		sd->output_queue = NULL;
		sd->output_queue_tailp = &sd->output_queue;
E
Eric Dumazet 已提交
6182
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
6183 6184 6185 6186
		sd->csd.func = rps_trigger_softirq;
		sd->csd.info = sd;
		sd->csd.flags = 0;
		sd->cpu = i;
6187
#endif
T
Tom Herbert 已提交
6188

E
Eric Dumazet 已提交
6189 6190 6191 6192
		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
6193 6194 6195 6196
	}

	dev_boot_phase = 0;

6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211
	/* 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;

6212 6213
	open_softirq(NET_TX_SOFTIRQ, net_tx_action);
	open_softirq(NET_RX_SOFTIRQ, net_rx_action);
L
Linus Torvalds 已提交
6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224

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

subsys_initcall(net_dev_init);

6225 6226
static int __init initialize_hashrnd(void)
{
T
Tom Herbert 已提交
6227
	get_random_bytes(&hashrnd, sizeof(hashrnd));
6228 6229 6230 6231 6232
	return 0;
}

late_initcall_sync(initialize_hashrnd);