dev.c 144.2 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
#include <net/sock.h>
#include <linux/rtnetlink.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/stat.h>
#include <linux/if_bridge.h>
P
Patrick McHardy 已提交
105
#include <linux/if_macvlan.h>
L
Linus Torvalds 已提交
106 107 108
#include <net/dst.h>
#include <net/pkt_sched.h>
#include <net/checksum.h>
109
#include <net/xfrm.h>
L
Linus Torvalds 已提交
110 111 112 113 114 115 116
#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>
117
#include <net/wext.h>
L
Linus Torvalds 已提交
118 119
#include <net/iw_handler.h>
#include <asm/current.h>
S
Steve Grubb 已提交
120
#include <linux/audit.h>
121
#include <linux/dmaengine.h>
122
#include <linux/err.h>
123
#include <linux/ctype.h>
124
#include <linux/if_arp.h>
125
#include <linux/if_vlan.h>
126
#include <linux/ip.h>
127
#include <net/ip.h>
128 129
#include <linux/ipv6.h>
#include <linux/in.h>
D
David S. Miller 已提交
130 131
#include <linux/jhash.h>
#include <linux/random.h>
132
#include <trace/events/napi.h>
133
#include <linux/pci.h>
L
Linus Torvalds 已提交
134

135 136
#include "net-sysfs.h"

137 138 139
/* Instead of increasing this, you should create a hash table. */
#define MAX_GRO_SKBS 8

H
Herbert Xu 已提交
140 141 142
/* 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
143 144 145 146 147 148 149 150 151 152 153
/*
 *	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
154
 *             the average user (w/out VLANs) will not be adversely affected.
L
Linus Torvalds 已提交
155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
 *             --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
 */

171 172 173
#define PTYPE_HASH_SIZE	(16)
#define PTYPE_HASH_MASK	(PTYPE_HASH_SIZE - 1)

L
Linus Torvalds 已提交
174
static DEFINE_SPINLOCK(ptype_lock);
175
static struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
176
static struct list_head ptype_all __read_mostly;	/* Taps */
L
Linus Torvalds 已提交
177 178

/*
179
 * The @dev_base_head list is protected by @dev_base_lock and the rtnl
L
Linus Torvalds 已提交
180 181
 * semaphore.
 *
182
 * Pure readers hold dev_base_lock for reading, or rcu_read_lock()
L
Linus Torvalds 已提交
183 184
 *
 * Writers must hold the rtnl semaphore while they loop through the
185
 * dev_base_head list, and hold dev_base_lock for writing when they do the
L
Linus Torvalds 已提交
186 187 188 189 190 191 192 193 194 195 196 197 198 199
 * 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);

200
static inline struct hlist_head *dev_name_hash(struct net *net, const char *name)
L
Linus Torvalds 已提交
201 202
{
	unsigned hash = full_name_hash(name, strnlen(name, IFNAMSIZ));
203
	return &net->dev_name_head[hash_32(hash, NETDEV_HASHBITS)];
L
Linus Torvalds 已提交
204 205
}

206
static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex)
L
Linus Torvalds 已提交
207
{
208
	return &net->dev_index_head[ifindex & (NETDEV_HASHENTRIES - 1)];
L
Linus Torvalds 已提交
209 210
}

E
Eric Dumazet 已提交
211
static inline void rps_lock(struct softnet_data *sd)
212 213
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
214
	spin_lock(&sd->input_pkt_queue.lock);
215 216 217
#endif
}

E
Eric Dumazet 已提交
218
static inline void rps_unlock(struct softnet_data *sd)
219 220
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
221
	spin_unlock(&sd->input_pkt_queue.lock);
222 223 224
#endif
}

225 226 227
/* Device list insertion */
static int list_netdevice(struct net_device *dev)
{
228
	struct net *net = dev_net(dev);
229 230 231 232

	ASSERT_RTNL();

	write_lock_bh(&dev_base_lock);
233
	list_add_tail_rcu(&dev->dev_list, &net->dev_base_head);
234
	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net, dev->name));
235 236
	hlist_add_head_rcu(&dev->index_hlist,
			   dev_index_hash(net, dev->ifindex));
237 238 239 240
	write_unlock_bh(&dev_base_lock);
	return 0;
}

241 242 243
/* Device list removal
 * caller must respect a RCU grace period before freeing/reusing dev
 */
244 245 246 247 248 249
static void unlist_netdevice(struct net_device *dev)
{
	ASSERT_RTNL();

	/* Unlink dev from the device chain */
	write_lock_bh(&dev_base_lock);
250
	list_del_rcu(&dev->dev_list);
251
	hlist_del_rcu(&dev->name_hlist);
252
	hlist_del_rcu(&dev->index_hlist);
253 254 255
	write_unlock_bh(&dev_base_lock);
}

L
Linus Torvalds 已提交
256 257 258 259
/*
 *	Our notifier list
 */

260
static RAW_NOTIFIER_HEAD(netdev_chain);
L
Linus Torvalds 已提交
261 262 263 264 265

/*
 *	Device drivers call our routines to queue packets here. We empty the
 *	queue in the local softnet handler.
 */
266

267
DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
E
Eric Dumazet 已提交
268
EXPORT_PER_CPU_SYMBOL(softnet_data);
L
Linus Torvalds 已提交
269

270
#ifdef CONFIG_LOCKDEP
271
/*
272
 * register_netdevice() inits txq->_xmit_lock and sets lockdep class
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
 * 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,
289
	 ARPHRD_IEEE80211_PRISM, ARPHRD_IEEE80211_RADIOTAP, ARPHRD_PHONET,
290
	 ARPHRD_PHONET_PIPE, ARPHRD_IEEE802154,
291
	 ARPHRD_VOID, ARPHRD_NONE};
292

293
static const char *const netdev_lock_name[] =
294 295 296 297 298 299 300 301 302 303 304 305 306
	{"_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",
307
	 "_xmit_IEEE80211_PRISM", "_xmit_IEEE80211_RADIOTAP", "_xmit_PHONET",
308
	 "_xmit_PHONET_PIPE", "_xmit_IEEE802154",
309
	 "_xmit_VOID", "_xmit_NONE"};
310 311

static struct lock_class_key netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)];
312
static struct lock_class_key netdev_addr_lock_key[ARRAY_SIZE(netdev_lock_type)];
313 314 315 316 317 318 319 320 321 322 323 324

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;
}

325 326
static inline void netdev_set_xmit_lockdep_class(spinlock_t *lock,
						 unsigned short dev_type)
327 328 329 330 331 332 333
{
	int i;

	i = netdev_lock_pos(dev_type);
	lockdep_set_class_and_name(lock, &netdev_xmit_lock_key[i],
				   netdev_lock_name[i]);
}
334 335 336 337 338 339 340 341 342 343

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]);
}
344
#else
345 346 347 348 349
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)
350 351 352
{
}
#endif
L
Linus Torvalds 已提交
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383

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

		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)
 */

/**
 *	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.
 *
384
 *	This call does not sleep therefore it can not
L
Linus Torvalds 已提交
385 386 387 388 389 390 391 392 393
 *	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)
{
	int hash;

	spin_lock_bh(&ptype_lock);
S
Stephen Hemminger 已提交
394
	if (pt->type == htons(ETH_P_ALL))
L
Linus Torvalds 已提交
395
		list_add_rcu(&pt->list, &ptype_all);
S
Stephen Hemminger 已提交
396
	else {
397
		hash = ntohs(pt->type) & PTYPE_HASH_MASK;
L
Linus Torvalds 已提交
398 399 400 401
		list_add_rcu(&pt->list, &ptype_base[hash]);
	}
	spin_unlock_bh(&ptype_lock);
}
E
Eric Dumazet 已提交
402
EXPORT_SYMBOL(dev_add_pack);
L
Linus Torvalds 已提交
403 404 405 406 407 408 409 410

/**
 *	__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
411
 *	returns.
L
Linus Torvalds 已提交
412 413 414 415 416 417 418 419 420 421 422 423
 *
 *      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)
{
	struct list_head *head;
	struct packet_type *pt1;

	spin_lock_bh(&ptype_lock);

S
Stephen Hemminger 已提交
424
	if (pt->type == htons(ETH_P_ALL))
L
Linus Torvalds 已提交
425
		head = &ptype_all;
S
Stephen Hemminger 已提交
426
	else
427
		head = &ptype_base[ntohs(pt->type) & PTYPE_HASH_MASK];
L
Linus Torvalds 已提交
428 429 430 431 432 433 434 435 436 437 438 439

	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:
	spin_unlock_bh(&ptype_lock);
}
E
Eric Dumazet 已提交
440 441
EXPORT_SYMBOL(__dev_remove_pack);

L
Linus Torvalds 已提交
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
/**
 *	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);
457

L
Linus Torvalds 已提交
458 459
	synchronize_net();
}
E
Eric Dumazet 已提交
460
EXPORT_SYMBOL(dev_remove_pack);
L
Linus Torvalds 已提交
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488

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

		      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));
489
			strlcpy(s[i].name, name, IFNAMSIZ);
L
Linus Torvalds 已提交
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513
			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] != ' ' &&
514
		    !strcmp(dev->name, s[i].name)) {
L
Linus Torvalds 已提交
515 516 517 518 519 520 521 522 523
			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
524
EXPORT_SYMBOL(netdev_boot_setup_check);
L
Linus Torvalds 已提交
525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548


/**
 *	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
	 */
549
	if (__dev_get_by_name(&init_net, name))
L
Linus Torvalds 已提交
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594
		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
595
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
596 597 598 599 600 601 602 603 604
 *	@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.
 */

605
struct net_device *__dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
606 607
{
	struct hlist_node *p;
E
Eric Dumazet 已提交
608 609
	struct net_device *dev;
	struct hlist_head *head = dev_name_hash(net, name);
L
Linus Torvalds 已提交
610

E
Eric Dumazet 已提交
611
	hlist_for_each_entry(dev, p, head, name_hlist)
L
Linus Torvalds 已提交
612 613
		if (!strncmp(dev->name, name, IFNAMSIZ))
			return dev;
E
Eric Dumazet 已提交
614

L
Linus Torvalds 已提交
615 616
	return NULL;
}
E
Eric Dumazet 已提交
617
EXPORT_SYMBOL(__dev_get_by_name);
L
Linus Torvalds 已提交
618

619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644
/**
 *	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
645 646
/**
 *	dev_get_by_name		- find a device by its name
647
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
648 649 650 651 652 653 654 655 656
 *	@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.
 */

657
struct net_device *dev_get_by_name(struct net *net, const char *name)
L
Linus Torvalds 已提交
658 659 660
{
	struct net_device *dev;

661 662
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
L
Linus Torvalds 已提交
663 664
	if (dev)
		dev_hold(dev);
665
	rcu_read_unlock();
L
Linus Torvalds 已提交
666 667
	return dev;
}
E
Eric Dumazet 已提交
668
EXPORT_SYMBOL(dev_get_by_name);
L
Linus Torvalds 已提交
669 670 671

/**
 *	__dev_get_by_index - find a device by its ifindex
672
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
673 674 675 676 677 678 679 680 681
 *	@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.
 */

682
struct net_device *__dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
683 684
{
	struct hlist_node *p;
E
Eric Dumazet 已提交
685 686
	struct net_device *dev;
	struct hlist_head *head = dev_index_hash(net, ifindex);
L
Linus Torvalds 已提交
687

E
Eric Dumazet 已提交
688
	hlist_for_each_entry(dev, p, head, index_hlist)
L
Linus Torvalds 已提交
689 690
		if (dev->ifindex == ifindex)
			return dev;
E
Eric Dumazet 已提交
691

L
Linus Torvalds 已提交
692 693
	return NULL;
}
E
Eric Dumazet 已提交
694
EXPORT_SYMBOL(__dev_get_by_index);
L
Linus Torvalds 已提交
695

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720
/**
 *	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
721 722 723

/**
 *	dev_get_by_index - find a device by its ifindex
724
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
725 726 727 728 729 730 731 732
 *	@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.
 */

733
struct net_device *dev_get_by_index(struct net *net, int ifindex)
L
Linus Torvalds 已提交
734 735 736
{
	struct net_device *dev;

737 738
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifindex);
L
Linus Torvalds 已提交
739 740
	if (dev)
		dev_hold(dev);
741
	rcu_read_unlock();
L
Linus Torvalds 已提交
742 743
	return dev;
}
E
Eric Dumazet 已提交
744
EXPORT_SYMBOL(dev_get_by_index);
L
Linus Torvalds 已提交
745 746 747

/**
 *	dev_getbyhwaddr - find a device by its hardware address
748
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
749 750 751 752 753 754 755 756 757 758 759 760
 *	@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
 */

761
struct net_device *dev_getbyhwaddr(struct net *net, unsigned short type, char *ha)
L
Linus Torvalds 已提交
762 763 764 765 766
{
	struct net_device *dev;

	ASSERT_RTNL();

767
	for_each_netdev(net, dev)
L
Linus Torvalds 已提交
768 769
		if (dev->type == type &&
		    !memcmp(dev->dev_addr, ha, dev->addr_len))
770 771 772
			return dev;

	return NULL;
L
Linus Torvalds 已提交
773
}
774 775
EXPORT_SYMBOL(dev_getbyhwaddr);

776
struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type)
L
Linus Torvalds 已提交
777 778 779
{
	struct net_device *dev;

780
	ASSERT_RTNL();
781
	for_each_netdev(net, dev)
782
		if (dev->type == type)
783 784 785
			return dev;

	return NULL;
786 787 788
}
EXPORT_SYMBOL(__dev_getfirstbyhwtype);

789
struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
790
{
791
	struct net_device *dev, *ret = NULL;
792

793 794 795 796 797 798 799 800 801
	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
802 803 804 805
}
EXPORT_SYMBOL(dev_getfirstbyhwtype);

/**
E
Eric Dumazet 已提交
806
 *	dev_get_by_flags_rcu - find any device with given flags
807
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
808 809 810 811
 *	@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
812 813
 *	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
814 815
 */

E
Eric Dumazet 已提交
816
struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
E
Eric Dumazet 已提交
817
				    unsigned short mask)
L
Linus Torvalds 已提交
818
{
819
	struct net_device *dev, *ret;
L
Linus Torvalds 已提交
820

821
	ret = NULL;
822
	for_each_netdev_rcu(net, dev) {
L
Linus Torvalds 已提交
823
		if (((dev->flags ^ if_flags) & mask) == 0) {
824
			ret = dev;
L
Linus Torvalds 已提交
825 826 827
			break;
		}
	}
828
	return ret;
L
Linus Torvalds 已提交
829
}
E
Eric Dumazet 已提交
830
EXPORT_SYMBOL(dev_get_by_flags_rcu);
L
Linus Torvalds 已提交
831 832 833 834 835 836

/**
 *	dev_valid_name - check if name is okay for network device
 *	@name: name string
 *
 *	Network device names need to be valid file names to
837 838
 *	to allow sysfs to work.  We also disallow any kind of
 *	whitespace.
L
Linus Torvalds 已提交
839
 */
840
int dev_valid_name(const char *name)
L
Linus Torvalds 已提交
841
{
842 843
	if (*name == '\0')
		return 0;
844 845
	if (strlen(name) >= IFNAMSIZ)
		return 0;
846 847 848 849 850 851 852 853 854
	if (!strcmp(name, ".") || !strcmp(name, ".."))
		return 0;

	while (*name) {
		if (*name == '/' || isspace(*name))
			return 0;
		name++;
	}
	return 1;
L
Linus Torvalds 已提交
855
}
E
Eric Dumazet 已提交
856
EXPORT_SYMBOL(dev_valid_name);
L
Linus Torvalds 已提交
857 858

/**
859 860
 *	__dev_alloc_name - allocate a name for a device
 *	@net: network namespace to allocate the device name in
L
Linus Torvalds 已提交
861
 *	@name: name format string
862
 *	@buf:  scratch buffer and result name string
L
Linus Torvalds 已提交
863 864
 *
 *	Passed a format string - eg "lt%d" it will try and find a suitable
S
Stephen Hemminger 已提交
865 866 867 868 869 870
 *	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
871 872
 */

873
static int __dev_alloc_name(struct net *net, const char *name, char *buf)
L
Linus Torvalds 已提交
874 875 876 877
{
	int i = 0;
	const char *p;
	const int max_netdevices = 8*PAGE_SIZE;
S
Stephen Hemminger 已提交
878
	unsigned long *inuse;
L
Linus Torvalds 已提交
879 880 881 882 883 884 885 886 887 888 889 890 891
	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
892
		inuse = (unsigned long *) get_zeroed_page(GFP_ATOMIC);
L
Linus Torvalds 已提交
893 894 895
		if (!inuse)
			return -ENOMEM;

896
		for_each_netdev(net, d) {
L
Linus Torvalds 已提交
897 898 899 900 901 902
			if (!sscanf(d->name, name, &i))
				continue;
			if (i < 0 || i >= max_netdevices)
				continue;

			/*  avoid cases where sscanf is not exact inverse of printf */
903
			snprintf(buf, IFNAMSIZ, name, i);
L
Linus Torvalds 已提交
904 905 906 907 908 909 910 911
			if (!strncmp(buf, d->name, IFNAMSIZ))
				set_bit(i, inuse);
		}

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

912 913
	if (buf != name)
		snprintf(buf, IFNAMSIZ, name, i);
914
	if (!__dev_get_by_name(net, buf))
L
Linus Torvalds 已提交
915 916 917 918 919 920 921 922 923
		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;
}

924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943
/**
 *	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;

944 945
	BUG_ON(!dev_net(dev));
	net = dev_net(dev);
946 947 948 949 950
	ret = __dev_alloc_name(net, name, buf);
	if (ret >= 0)
		strlcpy(dev->name, buf, IFNAMSIZ);
	return ret;
}
E
Eric Dumazet 已提交
951
EXPORT_SYMBOL(dev_alloc_name);
952

953
static int dev_get_valid_name(struct net_device *dev, const char *name, bool fmt)
954
{
955 956 957 958 959
	struct net *net;

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

960 961 962 963
	if (!dev_valid_name(name))
		return -EINVAL;

	if (fmt && strchr(name, '%'))
964
		return dev_alloc_name(dev, name);
965 966
	else if (__dev_get_by_name(net, name))
		return -EEXIST;
967 968
	else if (dev->name != name)
		strlcpy(dev->name, name, IFNAMSIZ);
969 970 971

	return 0;
}
L
Linus Torvalds 已提交
972 973 974 975 976 977 978 979 980

/**
 *	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.
 */
981
int dev_change_name(struct net_device *dev, const char *newname)
L
Linus Torvalds 已提交
982
{
983
	char oldname[IFNAMSIZ];
L
Linus Torvalds 已提交
984
	int err = 0;
985
	int ret;
986
	struct net *net;
L
Linus Torvalds 已提交
987 988

	ASSERT_RTNL();
989
	BUG_ON(!dev_net(dev));
L
Linus Torvalds 已提交
990

991
	net = dev_net(dev);
L
Linus Torvalds 已提交
992 993 994
	if (dev->flags & IFF_UP)
		return -EBUSY;

995 996 997
	if (strncmp(newname, dev->name, IFNAMSIZ) == 0)
		return 0;

998 999
	memcpy(oldname, dev->name, IFNAMSIZ);

1000
	err = dev_get_valid_name(dev, newname, 1);
1001 1002
	if (err < 0)
		return err;
L
Linus Torvalds 已提交
1003

1004
rollback:
1005 1006 1007 1008
	ret = device_rename(&dev->dev, dev->name);
	if (ret) {
		memcpy(dev->name, oldname, IFNAMSIZ);
		return ret;
1009
	}
1010 1011

	write_lock_bh(&dev_base_lock);
1012
	hlist_del(&dev->name_hlist);
1013 1014 1015 1016 1017 1018
	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));
1019 1020
	write_unlock_bh(&dev_base_lock);

1021
	ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev);
1022 1023 1024
	ret = notifier_to_errno(ret);

	if (ret) {
1025 1026
		/* err >= 0 after dev_alloc_name() or stores the first errno */
		if (err >= 0) {
1027 1028 1029
			err = ret;
			memcpy(dev->name, oldname, IFNAMSIZ);
			goto rollback;
1030 1031 1032 1033
		} else {
			printk(KERN_ERR
			       "%s: name change rollback failed: %d.\n",
			       dev->name, ret);
1034 1035
		}
	}
L
Linus Torvalds 已提交
1036 1037 1038 1039

	return err;
}

1040 1041 1042 1043
/**
 *	dev_set_alias - change ifalias of a device
 *	@dev: device
 *	@alias: name up to IFALIASZ
1044
 *	@len: limit of bytes to copy from info
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
 *
 *	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;

1055 1056 1057 1058 1059 1060 1061 1062
	if (!len) {
		if (dev->ifalias) {
			kfree(dev->ifalias);
			dev->ifalias = NULL;
		}
		return 0;
	}

E
Eric Dumazet 已提交
1063
	dev->ifalias = krealloc(dev->ifalias, len + 1, GFP_KERNEL);
1064 1065 1066 1067 1068 1069 1070 1071
	if (!dev->ifalias)
		return -ENOMEM;

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


1072
/**
S
Stephen Hemminger 已提交
1073
 *	netdev_features_change - device changes features
1074 1075 1076 1077 1078 1079
 *	@dev: device to cause notification
 *
 *	Called to indicate a device has changed features.
 */
void netdev_features_change(struct net_device *dev)
{
1080
	call_netdevice_notifiers(NETDEV_FEAT_CHANGE, dev);
1081 1082 1083
}
EXPORT_SYMBOL(netdev_features_change);

L
Linus Torvalds 已提交
1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094
/**
 *	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) {
1095
		call_netdevice_notifiers(NETDEV_CHANGE, dev);
L
Linus Torvalds 已提交
1096 1097 1098
		rtmsg_ifinfo(RTM_NEWLINK, dev, 0);
	}
}
E
Eric Dumazet 已提交
1099
EXPORT_SYMBOL(netdev_state_change);
L
Linus Torvalds 已提交
1100

1101
int netdev_bonding_change(struct net_device *dev, unsigned long event)
1102
{
1103
	return call_netdevice_notifiers(event, dev);
1104 1105 1106
}
EXPORT_SYMBOL(netdev_bonding_change);

L
Linus Torvalds 已提交
1107 1108
/**
 *	dev_load 	- load a network module
1109
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
1110 1111 1112 1113 1114 1115 1116
 *	@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.
 */

1117
void dev_load(struct net *net, const char *name)
L
Linus Torvalds 已提交
1118
{
1119
	struct net_device *dev;
L
Linus Torvalds 已提交
1120

1121 1122 1123
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
	rcu_read_unlock();
L
Linus Torvalds 已提交
1124

1125
	if (!dev && capable(CAP_NET_ADMIN))
L
Linus Torvalds 已提交
1126 1127
		request_module("%s", name);
}
E
Eric Dumazet 已提交
1128
EXPORT_SYMBOL(dev_load);
L
Linus Torvalds 已提交
1129

1130
static int __dev_open(struct net_device *dev)
L
Linus Torvalds 已提交
1131
{
1132
	const struct net_device_ops *ops = dev->netdev_ops;
1133
	int ret;
L
Linus Torvalds 已提交
1134

1135 1136
	ASSERT_RTNL();

L
Linus Torvalds 已提交
1137 1138 1139 1140 1141 1142
	/*
	 *	Is it even present?
	 */
	if (!netif_device_present(dev))
		return -ENODEV;

1143 1144 1145 1146 1147
	ret = call_netdevice_notifiers(NETDEV_PRE_UP, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		return ret;

L
Linus Torvalds 已提交
1148 1149 1150 1151
	/*
	 *	Call device private open method
	 */
	set_bit(__LINK_STATE_START, &dev->state);
1152

1153 1154
	if (ops->ndo_validate_addr)
		ret = ops->ndo_validate_addr(dev);
1155

1156 1157
	if (!ret && ops->ndo_open)
		ret = ops->ndo_open(dev);
L
Linus Torvalds 已提交
1158

1159
	/*
L
Linus Torvalds 已提交
1160 1161 1162
	 *	If it went open OK then:
	 */

1163 1164 1165
	if (ret)
		clear_bit(__LINK_STATE_START, &dev->state);
	else {
L
Linus Torvalds 已提交
1166 1167 1168 1169 1170
		/*
		 *	Set the flags.
		 */
		dev->flags |= IFF_UP;

1171 1172 1173
		/*
		 *	Enable NET_DMA
		 */
1174
		net_dmaengine_get();
1175

L
Linus Torvalds 已提交
1176 1177 1178
		/*
		 *	Initialize multicasting status
		 */
1179
		dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
1180 1181 1182 1183 1184 1185

		/*
		 *	Wakeup transmit queue engine
		 */
		dev_activate(dev);
	}
1186

L
Linus Torvalds 已提交
1187 1188 1189 1190
	return ret;
}

/**
1191 1192
 *	dev_open	- prepare an interface for use.
 *	@dev:	device to open
L
Linus Torvalds 已提交
1193
 *
1194 1195 1196 1197 1198 1199 1200
 *	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
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 1227 1228 1229
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
1230
{
1231
	const struct net_device_ops *ops = dev->netdev_ops;
1232

1233
	ASSERT_RTNL();
1234 1235
	might_sleep();

L
Linus Torvalds 已提交
1236 1237 1238 1239
	/*
	 *	Tell people we are going down, so that they can
	 *	prepare to death, when device is still operating.
	 */
1240
	call_netdevice_notifiers(NETDEV_GOING_DOWN, dev);
L
Linus Torvalds 已提交
1241 1242 1243 1244

	clear_bit(__LINK_STATE_START, &dev->state);

	/* Synchronize to scheduled poll. We cannot touch poll list,
1245 1246 1247 1248 1249
	 * 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
1250 1251
	smp_mb__after_clear_bit(); /* Commit netif_running(). */

1252 1253
	dev_deactivate(dev);

L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260
	/*
	 *	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.
	 */
1261 1262
	if (ops->ndo_stop)
		ops->ndo_stop(dev);
L
Linus Torvalds 已提交
1263 1264 1265 1266 1267 1268 1269 1270

	/*
	 *	Device is now down.
	 */

	dev->flags &= ~IFF_UP;

	/*
1271
	 *	Shutdown NET_DMA
L
Linus Torvalds 已提交
1272
	 */
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
	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
1293

1294
	/*
1295
	 * Tell people we are down
1296
	 */
1297 1298
	rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING);
	call_netdevice_notifiers(NETDEV_DOWN, dev);
1299

L
Linus Torvalds 已提交
1300 1301
	return 0;
}
E
Eric Dumazet 已提交
1302
EXPORT_SYMBOL(dev_close);
L
Linus Torvalds 已提交
1303 1304


1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
/**
 *	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);


1328 1329
static int dev_boot_phase = 1;

L
Linus Torvalds 已提交
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344
/*
 *	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
1345
 *	to the new notifier to allow device to have a race free
L
Linus Torvalds 已提交
1346 1347 1348 1349 1350 1351
 *	view of the network device list.
 */

int register_netdevice_notifier(struct notifier_block *nb)
{
	struct net_device *dev;
1352
	struct net_device *last;
1353
	struct net *net;
L
Linus Torvalds 已提交
1354 1355 1356
	int err;

	rtnl_lock();
1357
	err = raw_notifier_chain_register(&netdev_chain, nb);
1358 1359
	if (err)
		goto unlock;
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370
	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
1371

1372 1373
			nb->notifier_call(nb, NETDEV_UP, dev);
		}
L
Linus Torvalds 已提交
1374
	}
1375 1376

unlock:
L
Linus Torvalds 已提交
1377 1378
	rtnl_unlock();
	return err;
1379 1380 1381

rollback:
	last = dev;
1382 1383 1384 1385
	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev == last)
				break;
1386

1387 1388 1389 1390 1391
			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);
1392
			nb->notifier_call(nb, NETDEV_UNREGISTER_BATCH, dev);
1393 1394
		}
	}
1395 1396

	raw_notifier_chain_unregister(&netdev_chain, nb);
1397
	goto unlock;
L
Linus Torvalds 已提交
1398
}
E
Eric Dumazet 已提交
1399
EXPORT_SYMBOL(register_netdevice_notifier);
L
Linus Torvalds 已提交
1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412

/**
 *	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)
{
1413 1414 1415
	int err;

	rtnl_lock();
1416
	err = raw_notifier_chain_unregister(&netdev_chain, nb);
1417 1418
	rtnl_unlock();
	return err;
L
Linus Torvalds 已提交
1419
}
E
Eric Dumazet 已提交
1420
EXPORT_SYMBOL(unregister_netdevice_notifier);
L
Linus Torvalds 已提交
1421 1422 1423 1424

/**
 *	call_netdevice_notifiers - call all network notifier blocks
 *      @val: value passed unmodified to notifier function
1425
 *      @dev: net_device pointer passed unmodified to notifier function
L
Linus Torvalds 已提交
1426 1427
 *
 *	Call all network notifier blocks.  Parameters and return value
1428
 *	are as for raw_notifier_call_chain().
L
Linus Torvalds 已提交
1429 1430
 */

1431
int call_netdevice_notifiers(unsigned long val, struct net_device *dev)
L
Linus Torvalds 已提交
1432
{
1433
	ASSERT_RTNL();
1434
	return raw_notifier_call_chain(&netdev_chain, val, dev);
L
Linus Torvalds 已提交
1435 1436 1437 1438 1439 1440 1441 1442 1443
}

/* 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
1444
EXPORT_SYMBOL(net_enable_timestamp);
L
Linus Torvalds 已提交
1445 1446 1447 1448 1449

void net_disable_timestamp(void)
{
	atomic_dec(&netstamp_needed);
}
E
Eric Dumazet 已提交
1450
EXPORT_SYMBOL(net_disable_timestamp);
L
Linus Torvalds 已提交
1451

E
Eric Dumazet 已提交
1452
static inline void net_timestamp_set(struct sk_buff *skb)
L
Linus Torvalds 已提交
1453 1454
{
	if (atomic_read(&netstamp_needed))
1455
		__net_timestamp(skb);
1456 1457
	else
		skb->tstamp.tv64 = 0;
L
Linus Torvalds 已提交
1458 1459
}

E
Eric Dumazet 已提交
1460 1461 1462 1463 1464 1465
static inline void net_timestamp_check(struct sk_buff *skb)
{
	if (!skb->tstamp.tv64 && atomic_read(&netstamp_needed))
		__net_timestamp(skb);
}

1466 1467 1468 1469 1470 1471 1472 1473
/**
 * dev_forward_skb - loopback an skb to another netif
 *
 * @dev: destination network device
 * @skb: buffer to forward
 *
 * return values:
 *	NET_RX_SUCCESS	(no congestion)
1474
 *	NET_RX_DROP     (packet was dropped, but freed)
1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
 *
 * 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);

1488 1489 1490
	if (!(dev->flags & IFF_UP) ||
	    (skb->len > (dev->mtu + dev->hard_header_len))) {
		kfree_skb(skb);
1491
		return NET_RX_DROP;
1492
	}
1493
	skb_set_dev(skb, dev);
1494 1495 1496 1497 1498 1499 1500
	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
1501 1502 1503 1504 1505
/*
 *	Support routine. Sends outgoing frames to any network
 *	taps currently in use.
 */

1506
static void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
L
Linus Torvalds 已提交
1507 1508
{
	struct packet_type *ptype;
1509

1510 1511
#ifdef CONFIG_NET_CLS_ACT
	if (!(skb->tstamp.tv64 && (G_TC_FROM(skb->tc_verd) & AT_INGRESS)))
E
Eric Dumazet 已提交
1512
		net_timestamp_set(skb);
1513
#else
E
Eric Dumazet 已提交
1514
	net_timestamp_set(skb);
1515
#endif
L
Linus Torvalds 已提交
1516 1517 1518 1519 1520 1521 1522 1523 1524

	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
1525
			struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
L
Linus Torvalds 已提交
1526 1527 1528 1529 1530 1531 1532
			if (!skb2)
				break;

			/* skb->nh should be correctly
			   set by sender, so that the second statement is
			   just protection against buggy protocols.
			 */
1533
			skb_reset_mac_header(skb2);
L
Linus Torvalds 已提交
1534

1535
			if (skb_network_header(skb2) < skb2->data ||
1536
			    skb2->network_header > skb2->tail) {
L
Linus Torvalds 已提交
1537 1538 1539 1540
				if (net_ratelimit())
					printk(KERN_CRIT "protocol %04x is "
					       "buggy, dev %s\n",
					       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
static inline void __netif_reschedule(struct Qdisc *q)
1554
{
1555 1556
	struct softnet_data *sd;
	unsigned long flags;
1557

1558 1559
	local_irq_save(flags);
	sd = &__get_cpu_var(softnet_data);
1560 1561 1562
	q->next_sched = NULL;
	*sd->output_queue_tailp = q;
	sd->output_queue_tailp = &q->next_sched;
1563 1564 1565 1566 1567 1568 1569 1570
	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);
1571 1572 1573
}
EXPORT_SYMBOL(__netif_schedule);

1574
void dev_kfree_skb_irq(struct sk_buff *skb)
1575
{
E
Eric Dumazet 已提交
1576 1577 1578
	if (!skb->destructor)
		dev_kfree_skb(skb);
	else if (atomic_dec_and_test(&skb->users)) {
1579 1580
		struct softnet_data *sd;
		unsigned long flags;
1581

1582 1583 1584 1585 1586 1587 1588
		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);
	}
1589
}
1590
EXPORT_SYMBOL(dev_kfree_skb_irq);
1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601

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);


1602 1603 1604 1605 1606 1607
/**
 * netif_device_detach - mark device as removed
 * @dev: network device
 *
 * Mark device as removed from system and therefore no longer available.
 */
1608 1609 1610 1611
void netif_device_detach(struct net_device *dev)
{
	if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1612
		netif_tx_stop_all_queues(dev);
1613 1614 1615 1616
	}
}
EXPORT_SYMBOL(netif_device_detach);

1617 1618 1619 1620 1621 1622
/**
 * netif_device_attach - mark device as attached
 * @dev: network device
 *
 * Mark device as attached from system and restart if needed.
 */
1623 1624 1625 1626
void netif_device_attach(struct net_device *dev)
{
	if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1627
		netif_tx_wake_all_queues(dev);
1628
		__netdev_watchdog_up(dev);
1629 1630 1631 1632
	}
}
EXPORT_SYMBOL(netif_device_attach);

1633 1634 1635 1636 1637 1638
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) &&
1639 1640 1641
		 protocol == htons(ETH_P_IPV6)) ||
		((features & NETIF_F_FCOE_CRC) &&
		 protocol == htons(ETH_P_FCOE)));
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
}

static bool dev_can_checksum(struct net_device *dev, struct sk_buff *skb)
{
	if (can_checksum_protocol(dev->features, skb->protocol))
		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;
}
1658

1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688
/**
 * 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
1689 1690 1691 1692
/*
 * Invalidate hardware checksum when packet is to be mangled, and
 * complete checksum manually on outgoing path.
 */
1693
int skb_checksum_help(struct sk_buff *skb)
L
Linus Torvalds 已提交
1694
{
1695
	__wsum csum;
1696
	int ret = 0, offset;
L
Linus Torvalds 已提交
1697

1698
	if (skb->ip_summed == CHECKSUM_COMPLETE)
1699 1700 1701 1702 1703
		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
1704 1705
	}

1706 1707 1708 1709 1710 1711 1712 1713 1714
	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
1715 1716 1717 1718 1719
		ret = pskb_expand_head(skb, 0, 0, GFP_ATOMIC);
		if (ret)
			goto out;
	}

1720
	*(__sum16 *)(skb->data + offset) = csum_fold(csum);
1721
out_set_summed:
L
Linus Torvalds 已提交
1722
	skb->ip_summed = CHECKSUM_NONE;
1723
out:
L
Linus Torvalds 已提交
1724 1725
	return ret;
}
E
Eric Dumazet 已提交
1726
EXPORT_SYMBOL(skb_checksum_help);
L
Linus Torvalds 已提交
1727

1728 1729 1730
/**
 *	skb_gso_segment - Perform segmentation on skb.
 *	@skb: buffer to segment
1731
 *	@features: features for the output path (see dev->features)
1732 1733
 *
 *	This function segments the given skb and returns a list of segments.
1734 1735 1736
 *
 *	It may return NULL if the skb requires no segmentation.  This is
 *	only possible when GSO is used for verifying header integrity.
1737
 */
1738
struct sk_buff *skb_gso_segment(struct sk_buff *skb, int features)
1739 1740 1741
{
	struct sk_buff *segs = ERR_PTR(-EPROTONOSUPPORT);
	struct packet_type *ptype;
A
Al Viro 已提交
1742
	__be16 type = skb->protocol;
1743
	int err;
1744

1745
	skb_reset_mac_header(skb);
1746
	skb->mac_len = skb->network_header - skb->mac_header;
1747 1748
	__skb_pull(skb, skb->mac_len);

1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761
	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);

1762 1763 1764 1765 1766
		if (skb_header_cloned(skb) &&
		    (err = pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
			return ERR_PTR(err);
	}

1767
	rcu_read_lock();
1768 1769
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
1770
		if (ptype->type == type && !ptype->dev && ptype->gso_segment) {
1771
			if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
1772 1773 1774 1775
				err = ptype->gso_send_check(skb);
				segs = ERR_PTR(err);
				if (err || skb_gso_ok(skb, features))
					break;
1776 1777
				__skb_push(skb, (skb->data -
						 skb_network_header(skb)));
1778
			}
1779
			segs = ptype->gso_segment(skb, features);
1780 1781 1782 1783 1784
			break;
		}
	}
	rcu_read_unlock();

1785
	__skb_push(skb, skb->data - skb_mac_header(skb));
1786

1787 1788 1789 1790
	return segs;
}
EXPORT_SYMBOL(skb_gso_segment);

1791 1792 1793 1794 1795
/* Take action when hardware reception checksum errors are detected. */
#ifdef CONFIG_BUG
void netdev_rx_csum_fault(struct net_device *dev)
{
	if (net_ratelimit()) {
1796
		printk(KERN_ERR "%s: hw csum failure.\n",
1797
			dev ? dev->name : "<unknown>");
1798 1799 1800 1801 1802 1803
		dump_stack();
	}
}
EXPORT_SYMBOL(netdev_rx_csum_fault);
#endif

L
Linus Torvalds 已提交
1804 1805 1806 1807 1808
/* 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
1809
static int illegal_highdma(struct net_device *dev, struct sk_buff *skb)
L
Linus Torvalds 已提交
1810
{
1811
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
1812
	int i;
1813 1814 1815 1816 1817
	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
1818

1819 1820
	if (PCI_DMA_BUS_IS_PHYS) {
		struct device *pdev = dev->dev.parent;
L
Linus Torvalds 已提交
1821

E
Eric Dumazet 已提交
1822 1823
		if (!pdev)
			return 0;
1824 1825 1826 1827 1828 1829
		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;
		}
	}
1830
#endif
L
Linus Torvalds 已提交
1831 1832 1833
	return 0;
}

1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
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;
1868 1869 1870 1871 1872 1873 1874 1875
	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;
1876

1877
	if (IS_ERR(segs))
1878 1879 1880 1881 1882 1883 1884 1885 1886
		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
1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897
/*
 * Try to orphan skb early, right before transmission by the device.
 * We cannot orphan skb if tx timestamp is requested, since
 * drivers need to call skb_tstamp_tx() to send the timestamp.
 */
static inline void skb_orphan_try(struct sk_buff *skb)
{
	if (!skb_tx(skb)->flags)
		skb_orphan(skb);
}

1898 1899
int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
			struct netdev_queue *txq)
1900
{
1901
	const struct net_device_ops *ops = dev->netdev_ops;
1902
	int rc = NETDEV_TX_OK;
1903

1904
	if (likely(!skb->next)) {
S
Stephen Hemminger 已提交
1905
		if (!list_empty(&ptype_all))
1906 1907
			dev_queue_xmit_nit(skb, dev);

1908 1909 1910 1911
		/*
		 * If device doesnt need skb->dst, release it right now while
		 * its hot in this cpu cache
		 */
E
Eric Dumazet 已提交
1912 1913 1914
		if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
			skb_dst_drop(skb);

E
Eric Dumazet 已提交
1915
		skb_orphan_try(skb);
1916 1917 1918 1919 1920 1921 1922 1923

		if (netif_needs_gso(dev, skb)) {
			if (unlikely(dev_gso_segment(skb)))
				goto out_kfree_skb;
			if (skb->next)
				goto gso;
		}

1924
		rc = ops->ndo_start_xmit(skb, dev);
1925
		if (rc == NETDEV_TX_OK)
E
Eric Dumazet 已提交
1926
			txq_trans_update(txq);
1927
		return rc;
1928 1929
	}

1930
gso:
1931 1932 1933 1934 1935
	do {
		struct sk_buff *nskb = skb->next;

		skb->next = nskb->next;
		nskb->next = NULL;
1936 1937 1938 1939 1940 1941 1942 1943

		/*
		 * 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);

1944
		rc = ops->ndo_start_xmit(nskb, dev);
1945
		if (unlikely(rc != NETDEV_TX_OK)) {
1946 1947
			if (rc & ~NETDEV_TX_MASK)
				goto out_kfree_gso_skb;
1948
			nskb->next = skb->next;
1949 1950 1951
			skb->next = nskb;
			return rc;
		}
E
Eric Dumazet 已提交
1952
		txq_trans_update(txq);
1953
		if (unlikely(netif_tx_queue_stopped(txq) && skb->next))
1954
			return NETDEV_TX_BUSY;
1955
	} while (skb->next);
1956

1957 1958 1959
out_kfree_gso_skb:
	if (likely(skb->next == NULL))
		skb->destructor = DEV_GSO_CB(skb)->destructor;
1960 1961
out_kfree_skb:
	kfree_skb(skb);
1962
	return rc;
1963 1964
}

T
Tom Herbert 已提交
1965
static u32 hashrnd __read_mostly;
D
David S. Miller 已提交
1966

1967
u16 skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb)
1968
{
1969
	u32 hash;
D
David S. Miller 已提交
1970

1971 1972
	if (skb_rx_queue_recorded(skb)) {
		hash = skb_get_rx_queue(skb);
E
Eric Dumazet 已提交
1973
		while (unlikely(hash >= dev->real_num_tx_queues))
1974 1975 1976
			hash -= dev->real_num_tx_queues;
		return hash;
	}
1977 1978

	if (skb->sk && skb->sk->sk_hash)
1979
		hash = skb->sk->sk_hash;
1980
	else
E
Eric Dumazet 已提交
1981
		hash = (__force u16) skb->protocol;
1982

T
Tom Herbert 已提交
1983
	hash = jhash_1word(hash, hashrnd);
D
David S. Miller 已提交
1984 1985

	return (u16) (((u64) hash * dev->real_num_tx_queues) >> 32);
1986
}
1987
EXPORT_SYMBOL(skb_tx_hash);
1988

1989 1990 1991 1992
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
1993 1994 1995
			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);
1996 1997 1998 1999 2000 2001
		}
		return 0;
	}
	return queue_index;
}

2002 2003 2004
static struct netdev_queue *dev_pick_tx(struct net_device *dev,
					struct sk_buff *skb)
{
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
	u16 queue_index;
	struct sock *sk = skb->sk;

	if (sk_tx_queue_recorded(sk)) {
		queue_index = sk_tx_queue_get(sk);
	} else {
		const struct net_device_ops *ops = dev->netdev_ops;

		if (ops->ndo_select_queue) {
			queue_index = ops->ndo_select_queue(dev, skb);
2015
			queue_index = dev_cap_txqueue(dev, queue_index);
2016 2017 2018 2019
		} else {
			queue_index = 0;
			if (dev->real_num_tx_queues > 1)
				queue_index = skb_tx_hash(dev, skb);
2020

E
Eric Dumazet 已提交
2021
			if (sk) {
2022
				struct dst_entry *dst = rcu_dereference_check(sk->sk_dst_cache, 1);
E
Eric Dumazet 已提交
2023 2024 2025 2026

				if (dst && skb_dst(skb) == dst)
					sk_tx_queue_set(sk, queue_index);
			}
2027 2028
		}
	}
2029

2030 2031
	skb_set_queue_mapping(skb, queue_index);
	return netdev_get_tx_queue(dev, queue_index);
2032 2033
}

2034 2035 2036 2037 2038
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);
2039
	bool contended = qdisc_is_running(q);
2040 2041
	int rc;

2042 2043 2044 2045 2046 2047 2048 2049 2050
	/*
	 * 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);

2051 2052 2053 2054 2055
	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) &&
2056
		   qdisc_run_begin(q)) {
2057 2058 2059 2060 2061
		/*
		 * 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
2062 2063
		if (!(dev->priv_flags & IFF_XMIT_DST_RELEASE))
			skb_dst_force(skb);
2064
		__qdisc_update_bstats(q, skb->len);
2065 2066 2067 2068 2069
		if (sch_direct_xmit(skb, q, dev, txq, root_lock)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
2070
			__qdisc_run(q);
2071
		} else
2072
			qdisc_run_end(q);
2073 2074 2075

		rc = NET_XMIT_SUCCESS;
	} else {
E
Eric Dumazet 已提交
2076
		skb_dst_force(skb);
2077
		rc = qdisc_enqueue_root(skb, q);
2078 2079 2080 2081 2082 2083 2084
		if (qdisc_run_begin(q)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
			__qdisc_run(q);
		}
2085 2086
	}
	spin_unlock(root_lock);
2087 2088
	if (unlikely(contended))
		spin_unlock(&q->busylock);
2089 2090 2091
	return rc;
}

2092 2093 2094 2095 2096 2097 2098 2099 2100 2101
/*
 * 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)
{
2102 2103 2104 2105
	return skb_is_nonlinear(skb) &&
	       ((skb_has_frags(skb) && !(dev->features & NETIF_F_FRAGLIST)) ||
	        (skb_shinfo(skb)->nr_frags && (!(dev->features & NETIF_F_SG) ||
					      illegal_highdma(dev, skb))));
2106 2107
}

2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
/**
 *	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
2133 2134 2135
int dev_queue_xmit(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
2136
	struct netdev_queue *txq;
L
Linus Torvalds 已提交
2137 2138 2139
	struct Qdisc *q;
	int rc = -ENOMEM;

2140 2141 2142 2143
	/* GSO will handle the following emulations directly. */
	if (netif_needs_gso(dev, skb))
		goto gso;

2144 2145
	/* Convert a paged skb to linear, if required */
	if (skb_needs_linearize(skb, dev) && __skb_linearize(skb))
L
Linus Torvalds 已提交
2146 2147 2148 2149 2150
		goto out_kfree_skb;

	/* If packet is not checksummed and device does not support
	 * checksumming for this protocol, complete checksumming here.
	 */
2151 2152 2153
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
		skb_set_transport_header(skb, skb->csum_start -
					      skb_headroom(skb));
2154 2155
		if (!dev_can_checksum(dev, skb) && skb_checksum_help(skb))
			goto out_kfree_skb;
2156
	}
L
Linus Torvalds 已提交
2157

2158
gso:
2159 2160
	/* Disable soft irqs for various locks below. Also
	 * stops preemption for RCU.
L
Linus Torvalds 已提交
2161
	 */
2162
	rcu_read_lock_bh();
L
Linus Torvalds 已提交
2163

2164
	txq = dev_pick_tx(dev, skb);
2165
	q = rcu_dereference_bh(txq->qdisc);
2166

L
Linus Torvalds 已提交
2167
#ifdef CONFIG_NET_CLS_ACT
E
Eric Dumazet 已提交
2168
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_EGRESS);
L
Linus Torvalds 已提交
2169 2170
#endif
	if (q->enqueue) {
2171
		rc = __dev_xmit_skb(skb, q, dev, txq);
2172
		goto out;
L
Linus Torvalds 已提交
2173 2174 2175 2176 2177
	}

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

H
Herbert Xu 已提交
2178 2179
	   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
2180 2181 2182 2183 2184 2185 2186 2187 2188 2189
	   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 */

2190
		if (txq->xmit_lock_owner != cpu) {
L
Linus Torvalds 已提交
2191

2192
			HARD_TX_LOCK(dev, txq, cpu);
L
Linus Torvalds 已提交
2193

2194
			if (!netif_tx_queue_stopped(txq)) {
2195 2196
				rc = dev_hard_start_xmit(skb, dev, txq);
				if (dev_xmit_complete(rc)) {
2197
					HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2198 2199 2200
					goto out;
				}
			}
2201
			HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214
			if (net_ratelimit())
				printk(KERN_CRIT "Virtual device %s asks to "
				       "queue packet!\n", dev->name);
		} else {
			/* Recursion is detected! It is possible,
			 * unfortunately */
			if (net_ratelimit())
				printk(KERN_CRIT "Dead loop on virtual device "
				       "%s, fix it urgently!\n", dev->name);
		}
	}

	rc = -ENETDOWN;
2215
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2216 2217 2218 2219 2220

out_kfree_skb:
	kfree_skb(skb);
	return rc;
out:
2221
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2222 2223
	return rc;
}
E
Eric Dumazet 已提交
2224
EXPORT_SYMBOL(dev_queue_xmit);
L
Linus Torvalds 已提交
2225 2226 2227 2228 2229 2230


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

2231
int netdev_max_backlog __read_mostly = 1000;
E
Eric Dumazet 已提交
2232
int netdev_tstamp_prequeue __read_mostly = 1;
2233 2234
int netdev_budget __read_mostly = 300;
int weight_p __read_mostly = 64;            /* old backlog weight */
L
Linus Torvalds 已提交
2235

E
Eric Dumazet 已提交
2236 2237 2238 2239 2240 2241 2242 2243
/* 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);
}

E
Eric Dumazet 已提交
2244
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
2245 2246

/* One global table that all flow-based protocols share. */
2247
struct rps_sock_flow_table *rps_sock_flow_table __read_mostly;
T
Tom Herbert 已提交
2248 2249
EXPORT_SYMBOL(rps_sock_flow_table);

T
Tom Herbert 已提交
2250 2251 2252
/*
 * 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.
2253
 * rcu_read_lock must be held on entry.
T
Tom Herbert 已提交
2254
 */
T
Tom Herbert 已提交
2255 2256
static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb,
		       struct rps_dev_flow **rflowp)
T
Tom Herbert 已提交
2257 2258 2259 2260 2261
{
	struct ipv6hdr *ip6;
	struct iphdr *ip;
	struct netdev_rx_queue *rxqueue;
	struct rps_map *map;
T
Tom Herbert 已提交
2262 2263
	struct rps_dev_flow_table *flow_table;
	struct rps_sock_flow_table *sock_flow_table;
T
Tom Herbert 已提交
2264 2265
	int cpu = -1;
	u8 ip_proto;
T
Tom Herbert 已提交
2266
	u16 tcpu;
C
Changli Gao 已提交
2267 2268 2269 2270 2271
	u32 addr1, addr2, ihl;
	union {
		u32 v32;
		u16 v16[2];
	} ports;
T
Tom Herbert 已提交
2272 2273 2274 2275

	if (skb_rx_queue_recorded(skb)) {
		u16 index = skb_get_rx_queue(skb);
		if (unlikely(index >= dev->num_rx_queues)) {
2276 2277 2278
			WARN_ONCE(dev->num_rx_queues > 1, "%s received packet "
				"on queue %u, but number of RX queues is %u\n",
				dev->name, index, dev->num_rx_queues);
T
Tom Herbert 已提交
2279 2280 2281 2282 2283 2284
			goto done;
		}
		rxqueue = dev->_rx + index;
	} else
		rxqueue = dev->_rx;

T
Tom Herbert 已提交
2285
	if (!rxqueue->rps_map && !rxqueue->rps_flow_table)
T
Tom Herbert 已提交
2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
		goto done;

	if (skb->rxhash)
		goto got_hash; /* Skip hash computation on packet header */

	switch (skb->protocol) {
	case __constant_htons(ETH_P_IP):
		if (!pskb_may_pull(skb, sizeof(*ip)))
			goto done;

		ip = (struct iphdr *) skb->data;
		ip_proto = ip->protocol;
E
Eric Dumazet 已提交
2298 2299
		addr1 = (__force u32) ip->saddr;
		addr2 = (__force u32) ip->daddr;
T
Tom Herbert 已提交
2300 2301 2302 2303 2304 2305 2306 2307
		ihl = ip->ihl;
		break;
	case __constant_htons(ETH_P_IPV6):
		if (!pskb_may_pull(skb, sizeof(*ip6)))
			goto done;

		ip6 = (struct ipv6hdr *) skb->data;
		ip_proto = ip6->nexthdr;
E
Eric Dumazet 已提交
2308 2309
		addr1 = (__force u32) ip6->saddr.s6_addr32[3];
		addr2 = (__force u32) ip6->daddr.s6_addr32[3];
T
Tom Herbert 已提交
2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
		ihl = (40 >> 2);
		break;
	default:
		goto done;
	}
	switch (ip_proto) {
	case IPPROTO_TCP:
	case IPPROTO_UDP:
	case IPPROTO_DCCP:
	case IPPROTO_ESP:
	case IPPROTO_AH:
	case IPPROTO_SCTP:
	case IPPROTO_UDPLITE:
E
Eric Dumazet 已提交
2323
		if (pskb_may_pull(skb, (ihl * 4) + 4)) {
C
Changli Gao 已提交
2324 2325 2326 2327
			ports.v32 = * (__force u32 *) (skb->data + (ihl * 4));
			if (ports.v16[1] < ports.v16[0])
				swap(ports.v16[0], ports.v16[1]);
			break;
E
Eric Dumazet 已提交
2328
		}
T
Tom Herbert 已提交
2329
	default:
C
Changli Gao 已提交
2330
		ports.v32 = 0;
T
Tom Herbert 已提交
2331 2332 2333
		break;
	}

E
Eric Dumazet 已提交
2334 2335 2336
	/* get a consistent hash (same value on both flow directions) */
	if (addr2 < addr1)
		swap(addr1, addr2);
C
Changli Gao 已提交
2337
	skb->rxhash = jhash_3words(addr1, addr2, ports.v32, hashrnd);
T
Tom Herbert 已提交
2338 2339 2340 2341
	if (!skb->rxhash)
		skb->rxhash = 1;

got_hash:
T
Tom Herbert 已提交
2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380
	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
2381 2382
	map = rcu_dereference(rxqueue->rps_map);
	if (map) {
T
Tom Herbert 已提交
2383
		tcpu = map->cpus[((u64) skb->rxhash * map->len) >> 32];
T
Tom Herbert 已提交
2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395

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

done:
	return cpu;
}

/* Called from hardirq (IPI) context */
E
Eric Dumazet 已提交
2396
static void rps_trigger_softirq(void *data)
T
Tom Herbert 已提交
2397
{
E
Eric Dumazet 已提交
2398 2399
	struct softnet_data *sd = data;

E
Eric Dumazet 已提交
2400
	____napi_schedule(sd, &sd->backlog);
C
Changli Gao 已提交
2401
	sd->received_rps++;
T
Tom Herbert 已提交
2402
}
E
Eric Dumazet 已提交
2403

T
Tom Herbert 已提交
2404
#endif /* CONFIG_RPS */
T
Tom Herbert 已提交
2405

E
Eric Dumazet 已提交
2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426
/*
 * 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
2427 2428 2429 2430
/*
 * 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
2431 2432
static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
			      unsigned int *qtail)
T
Tom Herbert 已提交
2433
{
E
Eric Dumazet 已提交
2434
	struct softnet_data *sd;
T
Tom Herbert 已提交
2435 2436
	unsigned long flags;

E
Eric Dumazet 已提交
2437
	sd = &per_cpu(softnet_data, cpu);
T
Tom Herbert 已提交
2438 2439 2440

	local_irq_save(flags);

E
Eric Dumazet 已提交
2441
	rps_lock(sd);
2442 2443
	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
2444
enqueue:
E
Eric Dumazet 已提交
2445
			__skb_queue_tail(&sd->input_pkt_queue, skb);
2446
			input_queue_tail_incr_save(sd, qtail);
E
Eric Dumazet 已提交
2447
			rps_unlock(sd);
2448
			local_irq_restore(flags);
T
Tom Herbert 已提交
2449 2450 2451
			return NET_RX_SUCCESS;
		}

2452 2453 2454 2455
		/* 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
2456
			if (!rps_ipi_queued(sd))
E
Eric Dumazet 已提交
2457
				____napi_schedule(sd, &sd->backlog);
T
Tom Herbert 已提交
2458 2459 2460 2461
		}
		goto enqueue;
	}

C
Changli Gao 已提交
2462
	sd->dropped++;
E
Eric Dumazet 已提交
2463
	rps_unlock(sd);
T
Tom Herbert 已提交
2464 2465 2466 2467 2468 2469

	local_irq_restore(flags);

	kfree_skb(skb);
	return NET_RX_DROP;
}
L
Linus Torvalds 已提交
2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487

/**
 *	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)
{
2488
	int ret;
L
Linus Torvalds 已提交
2489 2490 2491 2492 2493

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

E
Eric Dumazet 已提交
2494 2495
	if (netdev_tstamp_prequeue)
		net_timestamp_check(skb);
L
Linus Torvalds 已提交
2496

E
Eric Dumazet 已提交
2497
#ifdef CONFIG_RPS
2498
	{
T
Tom Herbert 已提交
2499
		struct rps_dev_flow voidflow, *rflow = &voidflow;
2500 2501 2502
		int cpu;

		rcu_read_lock();
T
Tom Herbert 已提交
2503 2504

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
2505 2506
		if (cpu < 0)
			cpu = smp_processor_id();
T
Tom Herbert 已提交
2507 2508 2509

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

2510 2511
		rcu_read_unlock();
	}
2512
#else
T
Tom Herbert 已提交
2513 2514 2515 2516 2517
	{
		unsigned int qtail;
		ret = enqueue_to_backlog(skb, get_cpu(), &qtail);
		put_cpu();
	}
2518
#endif
2519
	return ret;
L
Linus Torvalds 已提交
2520
}
E
Eric Dumazet 已提交
2521
EXPORT_SYMBOL(netif_rx);
L
Linus Torvalds 已提交
2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552

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;

2553
			WARN_ON(atomic_read(&skb->users));
L
Linus Torvalds 已提交
2554 2555 2556 2557 2558
			__kfree_skb(skb);
		}
	}

	if (sd->output_queue) {
2559
		struct Qdisc *head;
L
Linus Torvalds 已提交
2560 2561 2562 2563

		local_irq_disable();
		head = sd->output_queue;
		sd->output_queue = NULL;
2564
		sd->output_queue_tailp = &sd->output_queue;
L
Linus Torvalds 已提交
2565 2566 2567
		local_irq_enable();

		while (head) {
2568 2569 2570
			struct Qdisc *q = head;
			spinlock_t *root_lock;

L
Linus Torvalds 已提交
2571 2572
			head = head->next_sched;

2573
			root_lock = qdisc_lock(q);
2574
			if (spin_trylock(root_lock)) {
2575 2576 2577
				smp_mb__before_clear_bit();
				clear_bit(__QDISC_STATE_SCHED,
					  &q->state);
2578 2579
				qdisc_run(q);
				spin_unlock(root_lock);
L
Linus Torvalds 已提交
2580
			} else {
2581
				if (!test_bit(__QDISC_STATE_DEACTIVATED,
2582
					      &q->state)) {
2583
					__netif_reschedule(q);
2584 2585 2586 2587 2588
				} else {
					smp_mb__before_clear_bit();
					clear_bit(__QDISC_STATE_SCHED,
						  &q->state);
				}
L
Linus Torvalds 已提交
2589 2590 2591 2592 2593
			}
		}
	}
}

2594 2595 2596
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
2597 2598
{
	atomic_inc(&skb->users);
D
David S. Miller 已提交
2599
	return pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
2600 2601
}

2602 2603
#if (defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE)) && \
    (defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE))
2604 2605 2606
/* This hook is defined here for ATM LANE */
int (*br_fdb_test_addr_hook)(struct net_device *dev,
			     unsigned char *addr) __read_mostly;
2607
EXPORT_SYMBOL_GPL(br_fdb_test_addr_hook);
2608
#endif
L
Linus Torvalds 已提交
2609 2610 2611 2612 2613 2614

#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
2615
 * NOTE: This doesnt stop any functionality; if you dont have
L
Linus Torvalds 已提交
2616 2617 2618
 * the ingress scheduler, you just cant add policies on ingress.
 *
 */
2619
static int ing_filter(struct sk_buff *skb)
L
Linus Torvalds 已提交
2620 2621
{
	struct net_device *dev = skb->dev;
2622
	u32 ttl = G_TC_RTTL(skb->tc_verd);
2623 2624 2625
	struct netdev_queue *rxq;
	int result = TC_ACT_OK;
	struct Qdisc *q;
2626

2627 2628 2629
	if (MAX_RED_LOOP < ttl++) {
		printk(KERN_WARNING
		       "Redir loop detected Dropping packet (%d->%d)\n",
2630
		       skb->skb_iif, dev->ifindex);
2631 2632
		return TC_ACT_SHOT;
	}
L
Linus Torvalds 已提交
2633

2634 2635
	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
2636

2637 2638
	rxq = &dev->rx_queue;

2639
	q = rxq->qdisc;
2640
	if (q != &noop_qdisc) {
2641
		spin_lock(qdisc_lock(q));
2642 2643
		if (likely(!test_bit(__QDISC_STATE_DEACTIVATED, &q->state)))
			result = qdisc_enqueue_root(skb, q);
2644 2645
		spin_unlock(qdisc_lock(q));
	}
2646 2647 2648

	return result;
}
2649

2650 2651 2652 2653
static inline struct sk_buff *handle_ing(struct sk_buff *skb,
					 struct packet_type **pt_prev,
					 int *ret, struct net_device *orig_dev)
{
2654
	if (skb->dev->rx_queue.qdisc == &noop_qdisc)
2655
		goto out;
L
Linus Torvalds 已提交
2656

2657 2658 2659
	if (*pt_prev) {
		*ret = deliver_skb(skb, *pt_prev, orig_dev);
		*pt_prev = NULL;
L
Linus Torvalds 已提交
2660 2661
	}

2662 2663 2664 2665 2666 2667 2668 2669 2670 2671
	switch (ing_filter(skb)) {
	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
2672 2673 2674
}
#endif

2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701
/*
 * 	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();
}

2702 2703 2704 2705
/**
 *	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
2706
 *	@rx_handler_data: data pointer that is used by rx handler
2707 2708 2709 2710 2711 2712 2713 2714
 *
 *	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
2715 2716
			       rx_handler_func_t *rx_handler,
			       void *rx_handler_data)
2717 2718 2719 2720 2721 2722
{
	ASSERT_RTNL();

	if (dev->rx_handler)
		return -EBUSY;

J
Jiri Pirko 已提交
2723
	rcu_assign_pointer(dev->rx_handler_data, rx_handler_data);
2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742
	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
2743
	rcu_assign_pointer(dev->rx_handler_data, NULL);
2744 2745 2746
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);

2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
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;

2768 2769
	if ((master->priv_flags & IFF_MASTER_ALB) &&
	    (master->priv_flags & IFF_BRIDGE_PORT)) {
2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796
		/* 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);

2797
static int __netif_receive_skb(struct sk_buff *skb)
L
Linus Torvalds 已提交
2798 2799
{
	struct packet_type *ptype, *pt_prev;
2800
	rx_handler_func_t *rx_handler;
D
David S. Miller 已提交
2801
	struct net_device *orig_dev;
2802
	struct net_device *master;
2803
	struct net_device *null_or_orig;
2804
	struct net_device *orig_or_bond;
L
Linus Torvalds 已提交
2805
	int ret = NET_RX_DROP;
A
Al Viro 已提交
2806
	__be16 type;
L
Linus Torvalds 已提交
2807

E
Eric Dumazet 已提交
2808 2809
	if (!netdev_tstamp_prequeue)
		net_timestamp_check(skb);
2810

E
Eric Dumazet 已提交
2811
	if (vlan_tx_tag_present(skb) && vlan_hwaccel_do_receive(skb))
2812 2813
		return NET_RX_SUCCESS;

L
Linus Torvalds 已提交
2814
	/* if we've gotten here through NAPI, check netpoll */
2815
	if (netpoll_receive_skb(skb))
L
Linus Torvalds 已提交
2816 2817
		return NET_RX_DROP;

2818 2819
	if (!skb->skb_iif)
		skb->skb_iif = skb->dev->ifindex;
2820

2821 2822 2823 2824 2825 2826 2827 2828
	/*
	 * 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.
	 */
2829
	null_or_orig = NULL;
J
Joe Eykholt 已提交
2830
	orig_dev = skb->dev;
2831
	master = ACCESS_ONCE(orig_dev->master);
2832 2833 2834 2835 2836
	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;
2837
			null_or_orig = orig_dev; /* deliver only exact match */
2838
		} else
2839
			skb->dev = master;
J
Joe Eykholt 已提交
2840
	}
2841

2842
	__this_cpu_inc(softnet_data.processed);
2843
	skb_reset_network_header(skb);
2844
	skb_reset_transport_header(skb);
2845
	skb->mac_len = skb->network_header - skb->mac_header;
L
Linus Torvalds 已提交
2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858

	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) {
2859 2860
		if (ptype->dev == null_or_orig || ptype->dev == skb->dev ||
		    ptype->dev == orig_dev) {
2861
			if (pt_prev)
D
David S. Miller 已提交
2862
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
2863 2864 2865 2866 2867
			pt_prev = ptype;
		}
	}

#ifdef CONFIG_NET_CLS_ACT
2868 2869
	skb = handle_ing(skb, &pt_prev, &ret, orig_dev);
	if (!skb)
L
Linus Torvalds 已提交
2870 2871 2872 2873
		goto out;
ncls:
#endif

2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884
	/* 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
2885

2886 2887 2888 2889 2890 2891
	/*
	 * 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.
	 */
2892
	orig_or_bond = orig_dev;
2893 2894
	if ((skb->dev->priv_flags & IFF_802_1Q_VLAN) &&
	    (vlan_dev_real_dev(skb->dev)->priv_flags & IFF_BONDING)) {
2895
		orig_or_bond = vlan_dev_real_dev(skb->dev);
2896 2897
	}

L
Linus Torvalds 已提交
2898
	type = skb->protocol;
2899 2900
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
2901
		if (ptype->type == type && (ptype->dev == null_or_orig ||
2902
		     ptype->dev == skb->dev || ptype->dev == orig_dev ||
2903
		     ptype->dev == orig_or_bond)) {
2904
			if (pt_prev)
D
David S. Miller 已提交
2905
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
2906 2907 2908 2909 2910
			pt_prev = ptype;
		}
	}

	if (pt_prev) {
D
David S. Miller 已提交
2911
		ret = pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923
	} else {
		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
2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941

/**
 *	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
2942 2943 2944
	if (netdev_tstamp_prequeue)
		net_timestamp_check(skb);

E
Eric Dumazet 已提交
2945
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
2946 2947 2948
	{
		struct rps_dev_flow voidflow, *rflow = &voidflow;
		int cpu, ret;
T
Tom Herbert 已提交
2949

E
Eric Dumazet 已提交
2950 2951 2952
		rcu_read_lock();

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
T
Tom Herbert 已提交
2953

E
Eric Dumazet 已提交
2954 2955 2956 2957 2958 2959 2960
		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
2961

E
Eric Dumazet 已提交
2962
		return ret;
T
Tom Herbert 已提交
2963
	}
2964 2965 2966
#else
	return __netif_receive_skb(skb);
#endif
T
Tom Herbert 已提交
2967
}
E
Eric Dumazet 已提交
2968
EXPORT_SYMBOL(netif_receive_skb);
L
Linus Torvalds 已提交
2969

E
Eric Dumazet 已提交
2970 2971 2972
/* Network device is going away, flush any packets still pending
 * Called with irqs disabled.
 */
2973
static void flush_backlog(void *arg)
2974
{
2975
	struct net_device *dev = arg;
E
Eric Dumazet 已提交
2976
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
2977 2978
	struct sk_buff *skb, *tmp;

E
Eric Dumazet 已提交
2979
	rps_lock(sd);
2980
	skb_queue_walk_safe(&sd->input_pkt_queue, skb, tmp) {
2981
		if (skb->dev == dev) {
E
Eric Dumazet 已提交
2982
			__skb_unlink(skb, &sd->input_pkt_queue);
2983
			kfree_skb(skb);
2984
			input_queue_head_incr(sd);
2985
		}
2986
	}
E
Eric Dumazet 已提交
2987
	rps_unlock(sd);
2988 2989 2990 2991 2992

	skb_queue_walk_safe(&sd->process_queue, skb, tmp) {
		if (skb->dev == dev) {
			__skb_unlink(skb, &sd->process_queue);
			kfree_skb(skb);
2993
			input_queue_head_incr(sd);
2994 2995
		}
	}
2996 2997
}

2998 2999 3000 3001 3002 3003 3004
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;

3005 3006
	if (NAPI_GRO_CB(skb)->count == 1) {
		skb_shinfo(skb)->gso_size = 0;
3007
		goto out;
3008
	}
3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029

	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);
}

3030
static void napi_gro_flush(struct napi_struct *napi)
3031 3032 3033 3034 3035 3036 3037 3038 3039
{
	struct sk_buff *skb, *next;

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

3040
	napi->gro_count = 0;
3041 3042 3043
	napi->gro_list = NULL;
}

3044
enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3045 3046 3047 3048 3049
{
	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
3050
	int same_flow;
3051
	int mac_len;
3052
	enum gro_result ret;
3053 3054 3055 3056

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

3057
	if (skb_is_gso(skb) || skb_has_frags(skb))
3058 3059
		goto normal;

3060 3061 3062 3063 3064
	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
		if (ptype->type != type || ptype->dev || !ptype->gro_receive)
			continue;

3065
		skb_set_network_header(skb, skb_gro_offset(skb));
3066 3067 3068 3069
		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
3070
		NAPI_GRO_CB(skb)->free = 0;
3071 3072 3073 3074 3075 3076 3077 3078 3079

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

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

H
Herbert Xu 已提交
3080
	same_flow = NAPI_GRO_CB(skb)->same_flow;
3081
	ret = NAPI_GRO_CB(skb)->free ? GRO_MERGED_FREE : GRO_MERGED;
H
Herbert Xu 已提交
3082

3083 3084 3085 3086 3087 3088
	if (pp) {
		struct sk_buff *nskb = *pp;

		*pp = nskb->next;
		nskb->next = NULL;
		napi_gro_complete(nskb);
3089
		napi->gro_count--;
3090 3091
	}

H
Herbert Xu 已提交
3092
	if (same_flow)
3093 3094
		goto ok;

3095
	if (NAPI_GRO_CB(skb)->flush || napi->gro_count >= MAX_GRO_SKBS)
3096 3097
		goto normal;

3098
	napi->gro_count++;
3099
	NAPI_GRO_CB(skb)->count = 1;
3100
	skb_shinfo(skb)->gso_size = skb_gro_len(skb);
3101 3102
	skb->next = napi->gro_list;
	napi->gro_list = skb;
3103
	ret = GRO_HELD;
3104

3105
pull:
H
Herbert Xu 已提交
3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124
	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,
				--skb_shinfo(skb)->nr_frags);
		}
3125 3126
	}

3127
ok:
3128
	return ret;
3129 3130

normal:
3131 3132
	ret = GRO_NORMAL;
	goto pull;
H
Herbert Xu 已提交
3133
}
3134 3135
EXPORT_SYMBOL(dev_gro_receive);

3136 3137
static gro_result_t
__napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3138 3139 3140
{
	struct sk_buff *p;

3141 3142 3143
	if (netpoll_rx_on(skb))
		return GRO_NORMAL;

3144
	for (p = napi->gro_list; p; p = p->next) {
3145 3146 3147 3148
		NAPI_GRO_CB(p)->same_flow =
			(p->dev == skb->dev) &&
			!compare_ether_header(skb_mac_header(p),
					      skb_gro_mac_header(skb));
3149 3150 3151 3152 3153
		NAPI_GRO_CB(p)->flush = 0;
	}

	return dev_gro_receive(napi, skb);
}
H
Herbert Xu 已提交
3154

3155
gro_result_t napi_skb_finish(gro_result_t ret, struct sk_buff *skb)
H
Herbert Xu 已提交
3156
{
3157 3158
	switch (ret) {
	case GRO_NORMAL:
3159 3160 3161
		if (netif_receive_skb(skb))
			ret = GRO_DROP;
		break;
H
Herbert Xu 已提交
3162

3163 3164
	case GRO_DROP:
	case GRO_MERGED_FREE:
H
Herbert Xu 已提交
3165 3166
		kfree_skb(skb);
		break;
3167 3168 3169 3170

	case GRO_HELD:
	case GRO_MERGED:
		break;
H
Herbert Xu 已提交
3171 3172
	}

3173
	return ret;
3174 3175 3176
}
EXPORT_SYMBOL(napi_skb_finish);

3177 3178 3179 3180
void skb_gro_reset_offset(struct sk_buff *skb)
{
	NAPI_GRO_CB(skb)->data_offset = 0;
	NAPI_GRO_CB(skb)->frag0 = NULL;
3181
	NAPI_GRO_CB(skb)->frag0_len = 0;
3182

3183
	if (skb->mac_header == skb->tail &&
3184
	    !PageHighMem(skb_shinfo(skb)->frags[0].page)) {
3185 3186 3187
		NAPI_GRO_CB(skb)->frag0 =
			page_address(skb_shinfo(skb)->frags[0].page) +
			skb_shinfo(skb)->frags[0].page_offset;
3188 3189
		NAPI_GRO_CB(skb)->frag0_len = skb_shinfo(skb)->frags[0].size;
	}
3190 3191 3192
}
EXPORT_SYMBOL(skb_gro_reset_offset);

3193
gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3194
{
3195 3196
	skb_gro_reset_offset(skb);

3197
	return napi_skb_finish(__napi_gro_receive(napi, skb), skb);
3198 3199 3200
}
EXPORT_SYMBOL(napi_gro_receive);

3201 3202 3203 3204 3205 3206 3207 3208 3209
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);

3210
struct sk_buff *napi_get_frags(struct napi_struct *napi)
H
Herbert Xu 已提交
3211 3212 3213 3214
{
	struct sk_buff *skb = napi->skb;

	if (!skb) {
3215 3216 3217
		skb = netdev_alloc_skb_ip_align(napi->dev, GRO_MAX_HEAD);
		if (skb)
			napi->skb = skb;
3218
	}
3219 3220
	return skb;
}
3221
EXPORT_SYMBOL(napi_get_frags);
3222

3223 3224
gro_result_t napi_frags_finish(struct napi_struct *napi, struct sk_buff *skb,
			       gro_result_t ret)
3225
{
3226 3227
	switch (ret) {
	case GRO_NORMAL:
3228
	case GRO_HELD:
3229
		skb->protocol = eth_type_trans(skb, skb->dev);
3230

3231 3232 3233 3234
		if (ret == GRO_HELD)
			skb_gro_pull(skb, -ETH_HLEN);
		else if (netif_receive_skb(skb))
			ret = GRO_DROP;
3235
		break;
H
Herbert Xu 已提交
3236

3237 3238 3239 3240
	case GRO_DROP:
	case GRO_MERGED_FREE:
		napi_reuse_skb(napi, skb);
		break;
3241 3242 3243

	case GRO_MERGED:
		break;
3244
	}
H
Herbert Xu 已提交
3245

3246
	return ret;
H
Herbert Xu 已提交
3247
}
3248 3249
EXPORT_SYMBOL(napi_frags_finish);

3250 3251 3252 3253
struct sk_buff *napi_frags_skb(struct napi_struct *napi)
{
	struct sk_buff *skb = napi->skb;
	struct ethhdr *eth;
3254 3255
	unsigned int hlen;
	unsigned int off;
3256 3257 3258 3259 3260 3261

	napi->skb = NULL;

	skb_reset_mac_header(skb);
	skb_gro_reset_offset(skb);

3262 3263 3264 3265 3266 3267 3268 3269 3270 3271
	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;
		}
3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286
	}

	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);

3287
gro_result_t napi_gro_frags(struct napi_struct *napi)
3288
{
3289
	struct sk_buff *skb = napi_frags_skb(napi);
3290 3291

	if (!skb)
3292
		return GRO_DROP;
3293 3294 3295

	return napi_frags_finish(napi, skb, __napi_gro_receive(napi, skb));
}
H
Herbert Xu 已提交
3296 3297
EXPORT_SYMBOL(napi_gro_frags);

3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325
/*
 * 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();
}

3326
static int process_backlog(struct napi_struct *napi, int quota)
L
Linus Torvalds 已提交
3327 3328
{
	int work = 0;
E
Eric Dumazet 已提交
3329
	struct softnet_data *sd = container_of(napi, struct softnet_data, backlog);
L
Linus Torvalds 已提交
3330

3331 3332 3333 3334 3335 3336 3337 3338 3339
#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
3340
	napi->weight = weight_p;
3341 3342
	local_irq_disable();
	while (work < quota) {
L
Linus Torvalds 已提交
3343
		struct sk_buff *skb;
3344 3345 3346 3347 3348 3349
		unsigned int qlen;

		while ((skb = __skb_dequeue(&sd->process_queue))) {
			local_irq_enable();
			__netif_receive_skb(skb);
			local_irq_disable();
3350 3351 3352 3353 3354
			input_queue_head_incr(sd);
			if (++work >= quota) {
				local_irq_enable();
				return work;
			}
3355
		}
L
Linus Torvalds 已提交
3356

E
Eric Dumazet 已提交
3357
		rps_lock(sd);
3358
		qlen = skb_queue_len(&sd->input_pkt_queue);
3359
		if (qlen)
3360 3361
			skb_queue_splice_tail_init(&sd->input_pkt_queue,
						   &sd->process_queue);
3362

3363
		if (qlen < quota - work) {
E
Eric Dumazet 已提交
3364 3365 3366 3367 3368 3369 3370 3371 3372 3373
			/*
			 * 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;

3374
			quota = work + qlen;
3375
		}
E
Eric Dumazet 已提交
3376
		rps_unlock(sd);
3377 3378
	}
	local_irq_enable();
L
Linus Torvalds 已提交
3379

3380 3381
	return work;
}
L
Linus Torvalds 已提交
3382

3383 3384
/**
 * __napi_schedule - schedule for receive
3385
 * @n: entry to schedule
3386 3387 3388
 *
 * The entry's receive function will be scheduled to run
 */
H
Harvey Harrison 已提交
3389
void __napi_schedule(struct napi_struct *n)
3390 3391
{
	unsigned long flags;
L
Linus Torvalds 已提交
3392

3393
	local_irq_save(flags);
E
Eric Dumazet 已提交
3394
	____napi_schedule(&__get_cpu_var(softnet_data), n);
3395
	local_irq_restore(flags);
L
Linus Torvalds 已提交
3396
}
3397 3398
EXPORT_SYMBOL(__napi_schedule);

3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431
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);
3432
	napi->gro_count = 0;
3433
	napi->gro_list = NULL;
H
Herbert Xu 已提交
3434
	napi->skb = NULL;
3435 3436 3437 3438
	napi->poll = poll;
	napi->weight = weight;
	list_add(&napi->dev_list, &dev->napi_list);
	napi->dev = dev;
H
Herbert Xu 已提交
3439
#ifdef CONFIG_NETPOLL
3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450
	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;

3451
	list_del_init(&napi->dev_list);
3452
	napi_free_frags(napi);
3453 3454 3455 3456 3457 3458 3459 3460

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

	napi->gro_list = NULL;
3461
	napi->gro_count = 0;
3462 3463 3464
}
EXPORT_SYMBOL(netif_napi_del);

L
Linus Torvalds 已提交
3465 3466
static void net_rx_action(struct softirq_action *h)
{
3467
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3468
	unsigned long time_limit = jiffies + 2;
3469
	int budget = netdev_budget;
3470 3471
	void *have;

L
Linus Torvalds 已提交
3472 3473
	local_irq_disable();

3474
	while (!list_empty(&sd->poll_list)) {
3475 3476
		struct napi_struct *n;
		int work, weight;
L
Linus Torvalds 已提交
3477

3478
		/* If softirq window is exhuasted then punt.
3479 3480
		 * Allow this to run for 2 jiffies since which will allow
		 * an average latency of 1.5/HZ.
3481
		 */
3482
		if (unlikely(budget <= 0 || time_after(jiffies, time_limit)))
L
Linus Torvalds 已提交
3483 3484 3485 3486
			goto softnet_break;

		local_irq_enable();

3487 3488 3489 3490 3491
		/* 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.
		 */
3492
		n = list_first_entry(&sd->poll_list, struct napi_struct, poll_list);
L
Linus Torvalds 已提交
3493

3494 3495 3496 3497
		have = netpoll_poll_lock(n);

		weight = n->weight;

3498 3499 3500 3501 3502 3503 3504
		/* 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;
3505
		if (test_bit(NAPI_STATE_SCHED, &n->state)) {
3506
			work = n->poll(n, weight);
3507 3508
			trace_napi_poll(n);
		}
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520

		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.
		 */
3521
		if (unlikely(work == weight)) {
3522 3523 3524 3525 3526
			if (unlikely(napi_disable_pending(n))) {
				local_irq_enable();
				napi_complete(n);
				local_irq_disable();
			} else
3527
				list_move_tail(&n->poll_list, &sd->poll_list);
3528
		}
3529 3530

		netpoll_poll_unlock(have);
L
Linus Torvalds 已提交
3531 3532
	}
out:
3533
	net_rps_action_and_irq_enable(sd);
T
Tom Herbert 已提交
3534

3535 3536 3537 3538 3539
#ifdef CONFIG_NET_DMA
	/*
	 * There may not be any more sk_buffs coming right now, so push
	 * any pending DMA copies to hardware
	 */
3540
	dma_issue_pending_all();
3541
#endif
3542

L
Linus Torvalds 已提交
3543 3544 3545
	return;

softnet_break:
C
Changli Gao 已提交
3546
	sd->time_squeeze++;
L
Linus Torvalds 已提交
3547 3548 3549 3550
	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
	goto out;
}

E
Eric Dumazet 已提交
3551
static gifconf_func_t *gifconf_list[NPROTO];
L
Linus Torvalds 已提交
3552 3553 3554 3555 3556 3557 3558 3559 3560 3561

/**
 *	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
3562
int register_gifconf(unsigned int family, gifconf_func_t *gifconf)
L
Linus Torvalds 已提交
3563 3564 3565 3566 3567 3568
{
	if (family >= NPROTO)
		return -EINVAL;
	gifconf_list[family] = gifconf;
	return 0;
}
E
Eric Dumazet 已提交
3569
EXPORT_SYMBOL(register_gifconf);
L
Linus Torvalds 已提交
3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582


/*
 *	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
 */

3583
static int dev_ifname(struct net *net, struct ifreq __user *arg)
L
Linus Torvalds 已提交
3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594
{
	struct net_device *dev;
	struct ifreq ifr;

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

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

3595 3596
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifr.ifr_ifindex);
L
Linus Torvalds 已提交
3597
	if (!dev) {
3598
		rcu_read_unlock();
L
Linus Torvalds 已提交
3599 3600 3601 3602
		return -ENODEV;
	}

	strcpy(ifr.ifr_name, dev->name);
3603
	rcu_read_unlock();
L
Linus Torvalds 已提交
3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615

	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'.
 */

3616
static int dev_ifconf(struct net *net, char __user *arg)
L
Linus Torvalds 已提交
3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639
{
	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;
3640
	for_each_netdev(net, dev) {
L
Linus Torvalds 已提交
3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653
		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;
			}
		}
3654
	}
L
Linus Torvalds 已提交
3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671

	/*
	 *	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.
 */
3672
void *dev_seq_start(struct seq_file *seq, loff_t *pos)
3673
	__acquires(RCU)
L
Linus Torvalds 已提交
3674
{
3675
	struct net *net = seq_file_net(seq);
3676
	loff_t off;
L
Linus Torvalds 已提交
3677 3678
	struct net_device *dev;

3679
	rcu_read_lock();
3680 3681
	if (!*pos)
		return SEQ_START_TOKEN;
L
Linus Torvalds 已提交
3682

3683
	off = 1;
3684
	for_each_netdev_rcu(net, dev)
3685 3686
		if (off++ == *pos)
			return dev;
L
Linus Torvalds 已提交
3687

3688
	return NULL;
L
Linus Torvalds 已提交
3689 3690 3691 3692
}

void *dev_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
3693 3694 3695 3696
	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
3697
	++*pos;
3698
	return rcu_dereference(dev);
L
Linus Torvalds 已提交
3699 3700 3701
}

void dev_seq_stop(struct seq_file *seq, void *v)
3702
	__releases(RCU)
L
Linus Torvalds 已提交
3703
{
3704
	rcu_read_unlock();
L
Linus Torvalds 已提交
3705 3706 3707 3708
}

static void dev_seq_printf_stats(struct seq_file *seq, struct net_device *dev)
{
3709
	const struct rtnl_link_stats64 *stats = dev_get_stats(dev);
L
Linus Torvalds 已提交
3710

3711 3712
	seq_printf(seq, "%6s: %7llu %7llu %4llu %4llu %4llu %5llu %10llu %9llu "
		   "%8llu %7llu %4llu %4llu %4llu %5llu %7llu %10llu\n",
3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727
		   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
3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746
}

/*
 *	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
3747
static struct softnet_data *softnet_get_online(loff_t *pos)
L
Linus Torvalds 已提交
3748
{
C
Changli Gao 已提交
3749
	struct softnet_data *sd = NULL;
L
Linus Torvalds 已提交
3750

3751
	while (*pos < nr_cpu_ids)
3752
		if (cpu_online(*pos)) {
C
Changli Gao 已提交
3753
			sd = &per_cpu(softnet_data, *pos);
L
Linus Torvalds 已提交
3754 3755 3756
			break;
		} else
			++*pos;
C
Changli Gao 已提交
3757
	return sd;
L
Linus Torvalds 已提交
3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776
}

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
3777
	struct softnet_data *sd = v;
L
Linus Torvalds 已提交
3778

T
Tom Herbert 已提交
3779
	seq_printf(seq, "%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
C
Changli Gao 已提交
3780
		   sd->processed, sd->dropped, sd->time_squeeze, 0,
3781
		   0, 0, 0, 0, /* was fastroute */
C
Changli Gao 已提交
3782
		   sd->cpu_collision, sd->received_rps);
L
Linus Torvalds 已提交
3783 3784 3785
	return 0;
}

3786
static const struct seq_operations dev_seq_ops = {
L
Linus Torvalds 已提交
3787 3788 3789 3790 3791 3792 3793 3794
	.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)
{
3795 3796
	return seq_open_net(inode, file, &dev_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
3797 3798
}

3799
static const struct file_operations dev_seq_fops = {
L
Linus Torvalds 已提交
3800 3801 3802 3803
	.owner	 = THIS_MODULE,
	.open    = dev_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
3804
	.release = seq_release_net,
L
Linus Torvalds 已提交
3805 3806
};

3807
static const struct seq_operations softnet_seq_ops = {
L
Linus Torvalds 已提交
3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818
	.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);
}

3819
static const struct file_operations softnet_seq_fops = {
L
Linus Torvalds 已提交
3820 3821 3822 3823 3824 3825 3826
	.owner	 = THIS_MODULE,
	.open    = softnet_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release,
};

3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838
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;
	}

3839
	for (t = 0; t < PTYPE_HASH_SIZE; t++) {
3840 3841 3842 3843 3844 3845 3846 3847 3848 3849
		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)
3850
	__acquires(RCU)
3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873
{
	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
3874
		hash = ntohs(pt->type) & PTYPE_HASH_MASK;
3875 3876

	while (nxt == &ptype_base[hash]) {
3877
		if (++hash >= PTYPE_HASH_SIZE)
3878 3879 3880 3881 3882 3883 3884 3885
			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)
3886
	__releases(RCU)
3887 3888 3889 3890 3891 3892 3893 3894 3895 3896
{
	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");
3897
	else if (pt->dev == NULL || dev_net(pt->dev) == seq_file_net(seq)) {
3898 3899 3900 3901 3902
		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
3903 3904
		seq_printf(seq, " %-8s %pF\n",
			   pt->dev ? pt->dev->name : "", pt->func);
3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918
	}

	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)
{
3919 3920
	return seq_open_net(inode, file, &ptype_seq_ops,
			sizeof(struct seq_net_private));
3921 3922 3923 3924 3925 3926 3927
}

static const struct file_operations ptype_seq_fops = {
	.owner	 = THIS_MODULE,
	.open    = ptype_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
3928
	.release = seq_release_net,
3929 3930 3931
};


3932
static int __net_init dev_proc_net_init(struct net *net)
L
Linus Torvalds 已提交
3933 3934 3935
{
	int rc = -ENOMEM;

3936
	if (!proc_net_fops_create(net, "dev", S_IRUGO, &dev_seq_fops))
L
Linus Torvalds 已提交
3937
		goto out;
3938
	if (!proc_net_fops_create(net, "softnet_stat", S_IRUGO, &softnet_seq_fops))
L
Linus Torvalds 已提交
3939
		goto out_dev;
3940
	if (!proc_net_fops_create(net, "ptype", S_IRUGO, &ptype_seq_fops))
3941
		goto out_softnet;
3942

3943
	if (wext_proc_init(net))
3944
		goto out_ptype;
L
Linus Torvalds 已提交
3945 3946 3947
	rc = 0;
out:
	return rc;
3948
out_ptype:
3949
	proc_net_remove(net, "ptype");
L
Linus Torvalds 已提交
3950
out_softnet:
3951
	proc_net_remove(net, "softnet_stat");
L
Linus Torvalds 已提交
3952
out_dev:
3953
	proc_net_remove(net, "dev");
L
Linus Torvalds 已提交
3954 3955
	goto out;
}
3956

3957
static void __net_exit dev_proc_net_exit(struct net *net)
3958 3959 3960 3961 3962 3963 3964 3965
{
	wext_proc_exit(net);

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

3966
static struct pernet_operations __net_initdata dev_proc_ops = {
3967 3968 3969 3970 3971 3972 3973 3974
	.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
3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003
#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;
4004

E
Eric Dumazet 已提交
4005 4006
	if (old) {
		synchronize_net();
L
Linus Torvalds 已提交
4007
		dev_put(old);
E
Eric Dumazet 已提交
4008
	}
L
Linus Torvalds 已提交
4009 4010 4011 4012 4013 4014 4015 4016
	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
4017
EXPORT_SYMBOL(netdev_set_master);
L
Linus Torvalds 已提交
4018

4019 4020
static void dev_change_rx_flags(struct net_device *dev, int flags)
{
4021 4022 4023 4024
	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);
4025 4026
}

4027
static int __dev_set_promiscuity(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4028 4029
{
	unsigned short old_flags = dev->flags;
4030 4031
	uid_t uid;
	gid_t gid;
L
Linus Torvalds 已提交
4032

4033 4034
	ASSERT_RTNL();

4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051
	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;
		}
	}
4052
	if (dev->flags != old_flags) {
L
Linus Torvalds 已提交
4053 4054
		printk(KERN_INFO "device %s %s promiscuous mode\n",
		       dev->name, (dev->flags & IFF_PROMISC) ? "entered" :
4055
							       "left");
4056 4057
		if (audit_enabled) {
			current_uid_gid(&uid, &gid);
4058 4059 4060 4061 4062 4063
			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),
4064
				uid, gid,
4065
				audit_get_sessionid(current));
4066
		}
4067

4068
		dev_change_rx_flags(dev, IFF_PROMISC);
L
Linus Torvalds 已提交
4069
	}
4070
	return 0;
L
Linus Torvalds 已提交
4071 4072
}

4073 4074 4075 4076 4077 4078 4079 4080 4081
/**
 *	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.
4082
 *	Return 0 if successful or a negative errno code on error.
4083
 */
4084
int dev_set_promiscuity(struct net_device *dev, int inc)
4085 4086
{
	unsigned short old_flags = dev->flags;
4087
	int err;
4088

4089
	err = __dev_set_promiscuity(dev, inc);
4090
	if (err < 0)
4091
		return err;
4092 4093
	if (dev->flags != old_flags)
		dev_set_rx_mode(dev);
4094
	return err;
4095
}
E
Eric Dumazet 已提交
4096
EXPORT_SYMBOL(dev_set_promiscuity);
4097

L
Linus Torvalds 已提交
4098 4099 4100 4101 4102 4103 4104 4105 4106 4107
/**
 *	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.
4108
 *	Return 0 if successful or a negative errno code on error.
L
Linus Torvalds 已提交
4109 4110
 */

4111
int dev_set_allmulti(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4112 4113 4114
{
	unsigned short old_flags = dev->flags;

4115 4116
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4117
	dev->flags |= IFF_ALLMULTI;
4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133
	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;
		}
	}
4134
	if (dev->flags ^ old_flags) {
4135
		dev_change_rx_flags(dev, IFF_ALLMULTI);
4136
		dev_set_rx_mode(dev);
4137
	}
4138
	return 0;
4139
}
E
Eric Dumazet 已提交
4140
EXPORT_SYMBOL(dev_set_allmulti);
4141 4142 4143 4144

/*
 *	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
4145
 *	filtering it is put in promiscuous mode while unicast addresses
4146 4147 4148 4149
 *	are present.
 */
void __dev_set_rx_mode(struct net_device *dev)
{
4150 4151
	const struct net_device_ops *ops = dev->netdev_ops;

4152 4153 4154 4155 4156
	/* dev_open will call this function so the list will stay sane. */
	if (!(dev->flags&IFF_UP))
		return;

	if (!netif_device_present(dev))
4157
		return;
4158

4159 4160
	if (ops->ndo_set_rx_mode)
		ops->ndo_set_rx_mode(dev);
4161 4162 4163 4164
	else {
		/* Unicast addresses changes may only happen under the rtnl,
		 * therefore calling __dev_set_promiscuity here is safe.
		 */
4165
		if (!netdev_uc_empty(dev) && !dev->uc_promisc) {
4166 4167
			__dev_set_promiscuity(dev, 1);
			dev->uc_promisc = 1;
4168
		} else if (netdev_uc_empty(dev) && dev->uc_promisc) {
4169 4170 4171 4172
			__dev_set_promiscuity(dev, -1);
			dev->uc_promisc = 0;
		}

4173 4174
		if (ops->ndo_set_multicast_list)
			ops->ndo_set_multicast_list(dev);
4175 4176 4177 4178 4179
	}
}

void dev_set_rx_mode(struct net_device *dev)
{
4180
	netif_addr_lock_bh(dev);
4181
	__dev_set_rx_mode(dev);
4182
	netif_addr_unlock_bh(dev);
L
Linus Torvalds 已提交
4183 4184
}

4185 4186 4187 4188 4189 4190
/**
 *	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
4191 4192 4193 4194 4195 4196
unsigned dev_get_flags(const struct net_device *dev)
{
	unsigned flags;

	flags = (dev->flags & ~(IFF_PROMISC |
				IFF_ALLMULTI |
S
Stefan Rompf 已提交
4197 4198 4199
				IFF_RUNNING |
				IFF_LOWER_UP |
				IFF_DORMANT)) |
L
Linus Torvalds 已提交
4200 4201 4202
		(dev->gflags & (IFF_PROMISC |
				IFF_ALLMULTI));

S
Stefan Rompf 已提交
4203 4204 4205 4206 4207 4208 4209 4210
	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
4211 4212 4213

	return flags;
}
E
Eric Dumazet 已提交
4214
EXPORT_SYMBOL(dev_get_flags);
L
Linus Torvalds 已提交
4215

4216
int __dev_change_flags(struct net_device *dev, unsigned int flags)
L
Linus Torvalds 已提交
4217 4218
{
	int old_flags = dev->flags;
4219
	int ret;
L
Linus Torvalds 已提交
4220

4221 4222
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236
	/*
	 *	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.
	 */

4237 4238
	if ((old_flags ^ flags) & IFF_MULTICAST)
		dev_change_rx_flags(dev, IFF_MULTICAST);
4239

4240
	dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4241 4242 4243 4244 4245 4246 4247 4248 4249

	/*
	 *	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  ? */
4250
		ret = ((old_flags & IFF_UP) ? __dev_close : __dev_open)(dev);
L
Linus Torvalds 已提交
4251 4252

		if (!ret)
4253
			dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4254 4255 4256
	}

	if ((flags ^ dev->gflags) & IFF_PROMISC) {
E
Eric Dumazet 已提交
4257 4258
		int inc = (flags & IFF_PROMISC) ? 1 : -1;

L
Linus Torvalds 已提交
4259 4260 4261 4262 4263 4264 4265 4266 4267
		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
4268 4269
		int inc = (flags & IFF_ALLMULTI) ? 1 : -1;

L
Linus Torvalds 已提交
4270 4271 4272 4273
		dev->gflags ^= IFF_ALLMULTI;
		dev_set_allmulti(dev, inc);
	}

4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310
	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;
4311 4312
	if (changes)
		rtmsg_ifinfo(RTM_NEWLINK, dev, changes);
L
Linus Torvalds 已提交
4313

4314
	__dev_notify_flags(dev, old_flags);
L
Linus Torvalds 已提交
4315 4316
	return ret;
}
E
Eric Dumazet 已提交
4317
EXPORT_SYMBOL(dev_change_flags);
L
Linus Torvalds 已提交
4318

4319 4320 4321 4322 4323 4324 4325
/**
 *	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
4326 4327
int dev_set_mtu(struct net_device *dev, int new_mtu)
{
4328
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341
	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;
4342 4343
	if (ops->ndo_change_mtu)
		err = ops->ndo_change_mtu(dev, new_mtu);
L
Linus Torvalds 已提交
4344 4345
	else
		dev->mtu = new_mtu;
4346

L
Linus Torvalds 已提交
4347
	if (!err && dev->flags & IFF_UP)
4348
		call_netdevice_notifiers(NETDEV_CHANGEMTU, dev);
L
Linus Torvalds 已提交
4349 4350
	return err;
}
E
Eric Dumazet 已提交
4351
EXPORT_SYMBOL(dev_set_mtu);
L
Linus Torvalds 已提交
4352

4353 4354 4355 4356 4357 4358 4359
/**
 *	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
4360 4361
int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa)
{
4362
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4363 4364
	int err;

4365
	if (!ops->ndo_set_mac_address)
L
Linus Torvalds 已提交
4366 4367 4368 4369 4370
		return -EOPNOTSUPP;
	if (sa->sa_family != dev->type)
		return -EINVAL;
	if (!netif_device_present(dev))
		return -ENODEV;
4371
	err = ops->ndo_set_mac_address(dev, sa);
L
Linus Torvalds 已提交
4372
	if (!err)
4373
		call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
L
Linus Torvalds 已提交
4374 4375
	return err;
}
E
Eric Dumazet 已提交
4376
EXPORT_SYMBOL(dev_set_mac_address);
L
Linus Torvalds 已提交
4377 4378

/*
E
Eric Dumazet 已提交
4379
 *	Perform the SIOCxIFxxx calls, inside rcu_read_lock()
L
Linus Torvalds 已提交
4380
 */
4381
static int dev_ifsioc_locked(struct net *net, struct ifreq *ifr, unsigned int cmd)
L
Linus Torvalds 已提交
4382 4383
{
	int err;
E
Eric Dumazet 已提交
4384
	struct net_device *dev = dev_get_by_name_rcu(net, ifr->ifr_name);
L
Linus Torvalds 已提交
4385 4386 4387 4388 4389

	if (!dev)
		return -ENODEV;

	switch (cmd) {
E
Eric Dumazet 已提交
4390 4391 4392
	case SIOCGIFFLAGS:	/* Get interface flags */
		ifr->ifr_flags = (short) dev_get_flags(dev);
		return 0;
L
Linus Torvalds 已提交
4393

E
Eric Dumazet 已提交
4394 4395 4396 4397
	case SIOCGIFMETRIC:	/* Get the metric on the interface
				   (currently unused) */
		ifr->ifr_metric = 0;
		return 0;
L
Linus Torvalds 已提交
4398

E
Eric Dumazet 已提交
4399 4400 4401
	case SIOCGIFMTU:	/* Get the MTU of a device */
		ifr->ifr_mtu = dev->mtu;
		return 0;
L
Linus Torvalds 已提交
4402

E
Eric Dumazet 已提交
4403 4404 4405 4406 4407 4408 4409 4410
	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
4411

E
Eric Dumazet 已提交
4412 4413 4414
	case SIOCGIFSLAVE:
		err = -EINVAL;
		break;
4415

E
Eric Dumazet 已提交
4416 4417 4418 4419 4420 4421 4422 4423
	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;
4424

E
Eric Dumazet 已提交
4425 4426 4427
	case SIOCGIFINDEX:
		ifr->ifr_ifindex = dev->ifindex;
		return 0;
4428

E
Eric Dumazet 已提交
4429 4430 4431
	case SIOCGIFTXQLEN:
		ifr->ifr_qlen = dev->tx_queue_len;
		return 0;
4432

E
Eric Dumazet 已提交
4433 4434 4435 4436 4437 4438 4439
	default:
		/* dev_ioctl() should ensure this case
		 * is never reached
		 */
		WARN_ON(1);
		err = -EINVAL;
		break;
4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451

	}
	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
4452
	const struct net_device_ops *ops;
4453 4454 4455 4456

	if (!dev)
		return -ENODEV;

J
Jarek Poplawski 已提交
4457 4458
	ops = dev->netdev_ops;

4459
	switch (cmd) {
E
Eric Dumazet 已提交
4460 4461
	case SIOCSIFFLAGS:	/* Set interface flags */
		return dev_change_flags(dev, ifr->ifr_flags);
4462

E
Eric Dumazet 已提交
4463 4464 4465
	case SIOCSIFMETRIC:	/* Set the metric on the interface
				   (currently unused) */
		return -EOPNOTSUPP;
4466

E
Eric Dumazet 已提交
4467 4468
	case SIOCSIFMTU:	/* Set the MTU of a device */
		return dev_set_mtu(dev, ifr->ifr_mtu);
L
Linus Torvalds 已提交
4469

E
Eric Dumazet 已提交
4470 4471
	case SIOCSIFHWADDR:
		return dev_set_mac_address(dev, &ifr->ifr_hwaddr);
L
Linus Torvalds 已提交
4472

E
Eric Dumazet 已提交
4473 4474 4475 4476 4477 4478 4479
	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
4480

E
Eric Dumazet 已提交
4481 4482
	case SIOCSIFMAP:
		if (ops->ndo_set_config) {
L
Linus Torvalds 已提交
4483 4484
			if (!netif_device_present(dev))
				return -ENODEV;
E
Eric Dumazet 已提交
4485 4486 4487
			return ops->ndo_set_config(dev, &ifr->ifr_map);
		}
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
4488

E
Eric Dumazet 已提交
4489 4490 4491 4492 4493 4494
	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;
4495
		return dev_mc_add_global(dev, ifr->ifr_hwaddr.sa_data);
E
Eric Dumazet 已提交
4496 4497 4498 4499 4500 4501 4502

	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;
4503
		return dev_mc_del_global(dev, ifr->ifr_hwaddr.sa_data);
L
Linus Torvalds 已提交
4504

E
Eric Dumazet 已提交
4505 4506 4507 4508 4509
	case SIOCSIFTXQLEN:
		if (ifr->ifr_qlen < 0)
			return -EINVAL;
		dev->tx_queue_len = ifr->ifr_qlen;
		return 0;
L
Linus Torvalds 已提交
4510

E
Eric Dumazet 已提交
4511 4512 4513
	case SIOCSIFNAME:
		ifr->ifr_newname[IFNAMSIZ-1] = '\0';
		return dev_change_name(dev, ifr->ifr_newname);
L
Linus Torvalds 已提交
4514

E
Eric Dumazet 已提交
4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542
	/*
	 *	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
4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554

	}
	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
4555
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
4556 4557 4558 4559 4560 4561 4562 4563 4564
 *	@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.
 */

4565
int dev_ioctl(struct net *net, unsigned int cmd, void __user *arg)
L
Linus Torvalds 已提交
4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576
{
	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) {
4577
		rtnl_lock();
4578
		ret = dev_ifconf(net, (char __user *) arg);
4579
		rtnl_unlock();
L
Linus Torvalds 已提交
4580 4581 4582
		return ret;
	}
	if (cmd == SIOCGIFNAME)
4583
		return dev_ifname(net, (struct ifreq __user *)arg);
L
Linus Torvalds 已提交
4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598

	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
4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613
	/*
	 *	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
4614
		rcu_read_lock();
E
Eric Dumazet 已提交
4615
		ret = dev_ifsioc_locked(net, &ifr, cmd);
E
Eric Dumazet 已提交
4616
		rcu_read_unlock();
E
Eric Dumazet 已提交
4617 4618 4619 4620 4621 4622 4623 4624
		if (!ret) {
			if (colon)
				*colon = ':';
			if (copy_to_user(arg, &ifr,
					 sizeof(struct ifreq)))
				ret = -EFAULT;
		}
		return ret;
L
Linus Torvalds 已提交
4625

E
Eric Dumazet 已提交
4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638
	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
4639

E
Eric Dumazet 已提交
4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662
	/*
	 *	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
4663

E
Eric Dumazet 已提交
4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714
	/*
	 *	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)) {
4715
			dev_load(net, ifr.ifr_name);
L
Linus Torvalds 已提交
4716
			rtnl_lock();
4717
			ret = dev_ifsioc(net, &ifr, cmd);
L
Linus Torvalds 已提交
4718
			rtnl_unlock();
E
Eric Dumazet 已提交
4719 4720 4721
			if (!ret && copy_to_user(arg, &ifr,
						 sizeof(struct ifreq)))
				ret = -EFAULT;
L
Linus Torvalds 已提交
4722
			return ret;
E
Eric Dumazet 已提交
4723 4724 4725 4726 4727
		}
		/* 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
4728 4729 4730 4731 4732 4733
	}
}


/**
 *	dev_new_index	-	allocate an ifindex
4734
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
4735 4736 4737 4738 4739
 *
 *	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.
 */
4740
static int dev_new_index(struct net *net)
L
Linus Torvalds 已提交
4741 4742 4743 4744 4745
{
	static int ifindex;
	for (;;) {
		if (++ifindex <= 0)
			ifindex = 1;
4746
		if (!__dev_get_by_index(net, ifindex))
L
Linus Torvalds 已提交
4747 4748 4749 4750 4751
			return ifindex;
	}
}

/* Delayed registration/unregisteration */
4752
static LIST_HEAD(net_todo_list);
L
Linus Torvalds 已提交
4753

4754
static void net_set_todo(struct net_device *dev)
L
Linus Torvalds 已提交
4755 4756 4757 4758
{
	list_add_tail(&dev->todo_list, &net_todo_list);
}

4759
static void rollback_registered_many(struct list_head *head)
4760
{
4761
	struct net_device *dev, *tmp;
4762

4763 4764 4765
	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

4766
	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
4767
		/* Some devices call without registering
4768 4769
		 * for initialization unwind. Remove those
		 * devices and proceed with the remaining.
4770 4771 4772 4773
		 */
		if (dev->reg_state == NETREG_UNINITIALIZED) {
			pr_debug("unregister_netdevice: device %s/%p never "
				 "was registered\n", dev->name, dev);
4774

4775
			WARN_ON(1);
4776 4777
			list_del(&dev->unreg_list);
			continue;
4778
		}
4779

4780
		BUG_ON(dev->reg_state != NETREG_REGISTERED);
4781

4782 4783
		/* If device is running, close it first. */
		dev_close(dev);
4784

4785 4786
		/* And unlink it from device chain. */
		unlist_netdevice(dev);
4787

4788 4789
		dev->reg_state = NETREG_UNREGISTERING;
	}
4790 4791 4792

	synchronize_net();

4793 4794 4795
	list_for_each_entry(dev, head, unreg_list) {
		/* Shutdown queueing discipline. */
		dev_shutdown(dev);
4796 4797


4798 4799 4800 4801
		/* Notify protocols, that we are about to destroy
		   this device. They should clean all the things.
		*/
		call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
4802

4803 4804 4805 4806
		if (!dev->rtnl_link_ops ||
		    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
			rtmsg_ifinfo(RTM_DELLINK, dev, ~0U);

4807 4808 4809
		/*
		 *	Flush the unicast and multicast chains
		 */
4810
		dev_uc_flush(dev);
4811
		dev_mc_flush(dev);
4812

4813 4814
		if (dev->netdev_ops->ndo_uninit)
			dev->netdev_ops->ndo_uninit(dev);
4815

4816 4817
		/* Notifier chain MUST detach us from master device. */
		WARN_ON(dev->master);
4818

4819 4820 4821
		/* Remove entries from kobject tree */
		netdev_unregister_kobject(dev);
	}
4822

4823
	/* Process any work delayed until the end of the batch */
4824
	dev = list_first_entry(head, struct net_device, unreg_list);
4825
	call_netdevice_notifiers(NETDEV_UNREGISTER_BATCH, dev);
4826

4827
	synchronize_net();
4828

4829
	list_for_each_entry(dev, head, unreg_list)
4830 4831 4832 4833 4834 4835 4836 4837 4838
		dev_put(dev);
}

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

	list_add(&dev->unreg_list, &single);
	rollback_registered_many(&single);
4839 4840
}

4841 4842 4843
static void __netdev_init_queue_locks_one(struct net_device *dev,
					  struct netdev_queue *dev_queue,
					  void *_unused)
4844 4845
{
	spin_lock_init(&dev_queue->_xmit_lock);
4846
	netdev_set_xmit_lockdep_class(&dev_queue->_xmit_lock, dev->type);
4847 4848 4849 4850 4851
	dev_queue->xmit_lock_owner = -1;
}

static void netdev_init_queue_locks(struct net_device *dev)
{
4852 4853
	netdev_for_each_tx_queue(dev, __netdev_init_queue_locks_one, NULL);
	__netdev_init_queue_locks_one(dev, &dev->rx_queue, NULL);
4854 4855
}

4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895
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);

4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922
/**
 *	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);

L
Linus Torvalds 已提交
4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942
/**
 *	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;
4943
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
4944 4945 4946 4947

	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

4948 4949
	might_sleep();

L
Linus Torvalds 已提交
4950 4951
	/* When net_device's are persistent, this will be fatal. */
	BUG_ON(dev->reg_state != NETREG_UNINITIALIZED);
4952
	BUG_ON(!net);
L
Linus Torvalds 已提交
4953

4954
	spin_lock_init(&dev->addr_list_lock);
4955
	netdev_set_addr_lockdep_class(dev);
4956
	netdev_init_queue_locks(dev);
L
Linus Torvalds 已提交
4957 4958 4959

	dev->iflink = -1;

E
Eric Dumazet 已提交
4960
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976
	if (!dev->num_rx_queues) {
		/*
		 * Allocate a single RX queue if driver never called
		 * alloc_netdev_mq
		 */

		dev->_rx = kzalloc(sizeof(struct netdev_rx_queue), GFP_KERNEL);
		if (!dev->_rx) {
			ret = -ENOMEM;
			goto out;
		}

		dev->_rx->first = dev->_rx;
		atomic_set(&dev->_rx->count, 1);
		dev->num_rx_queues = 1;
	}
E
Eric Dumazet 已提交
4977
#endif
L
Linus Torvalds 已提交
4978
	/* Init, if this function is available */
4979 4980
	if (dev->netdev_ops->ndo_init) {
		ret = dev->netdev_ops->ndo_init(dev);
L
Linus Torvalds 已提交
4981 4982 4983
		if (ret) {
			if (ret > 0)
				ret = -EIO;
4984
			goto out;
L
Linus Torvalds 已提交
4985 4986
		}
	}
4987

4988
	ret = dev_get_valid_name(dev, dev->name, 0);
4989
	if (ret)
4990
		goto err_uninit;
L
Linus Torvalds 已提交
4991

4992
	dev->ifindex = dev_new_index(net);
L
Linus Torvalds 已提交
4993 4994 4995
	if (dev->iflink == -1)
		dev->iflink = dev->ifindex;

4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010
	/* 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);
	}

5011
	dev->features = netdev_fix_features(dev->features, dev->name);
L
Linus Torvalds 已提交
5012

5013 5014 5015 5016
	/* Enable software GSO if SG is supported. */
	if (dev->features & NETIF_F_SG)
		dev->features |= NETIF_F_GSO;

5017 5018 5019 5020 5021
	ret = call_netdevice_notifiers(NETDEV_POST_INIT, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		goto err_uninit;

5022
	ret = netdev_register_kobject(dev);
5023
	if (ret)
5024
		goto err_uninit;
5025 5026
	dev->reg_state = NETREG_REGISTERED;

L
Linus Torvalds 已提交
5027 5028 5029 5030 5031 5032 5033 5034 5035
	/*
	 *	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);
5036
	list_netdevice(dev);
L
Linus Torvalds 已提交
5037 5038

	/* Notify protocols, that a new device appeared. */
5039
	ret = call_netdevice_notifiers(NETDEV_REGISTER, dev);
5040
	ret = notifier_to_errno(ret);
5041 5042 5043 5044
	if (ret) {
		rollback_registered(dev);
		dev->reg_state = NETREG_UNREGISTERED;
	}
5045 5046 5047 5048
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
5049 5050 5051
	if (!dev->rtnl_link_ops ||
	    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
		rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);
L
Linus Torvalds 已提交
5052 5053 5054

out:
	return ret;
5055 5056

err_uninit:
5057 5058
	if (dev->netdev_ops->ndo_uninit)
		dev->netdev_ops->ndo_uninit(dev);
5059
	goto out;
L
Linus Torvalds 已提交
5060
}
E
Eric Dumazet 已提交
5061
EXPORT_SYMBOL(register_netdevice);
L
Linus Torvalds 已提交
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 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101
/**
 *	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;

	/* initialize the ref count */
	atomic_set(&dev->refcnt, 1);

	/* 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);

	return 0;
}
EXPORT_SYMBOL_GPL(init_dummy_netdev);


L
Linus Torvalds 已提交
5102 5103 5104 5105 5106 5107 5108 5109 5110
/**
 *	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.
 *
5111
 *	This is a wrapper around register_netdevice that takes the rtnl semaphore
L
Linus Torvalds 已提交
5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129
 *	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;
	}
5130

L
Linus Torvalds 已提交
5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146
	err = register_netdevice(dev);
out:
	rtnl_unlock();
	return err;
}
EXPORT_SYMBOL(register_netdev);

/*
 * 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
5147
 * call dev_put.
L
Linus Torvalds 已提交
5148 5149 5150 5151 5152
 */
static void netdev_wait_allrefs(struct net_device *dev)
{
	unsigned long rebroadcast_time, warning_time;

5153 5154
	linkwatch_forget_dev(dev);

L
Linus Torvalds 已提交
5155 5156 5157
	rebroadcast_time = warning_time = jiffies;
	while (atomic_read(&dev->refcnt) != 0) {
		if (time_after(jiffies, rebroadcast_time + 1 * HZ)) {
5158
			rtnl_lock();
L
Linus Torvalds 已提交
5159 5160

			/* Rebroadcast unregister notification */
5161
			call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5162
			/* don't resend NETDEV_UNREGISTER_BATCH, _BATCH users
5163
			 * should have already handle it the first time */
L
Linus Torvalds 已提交
5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175

			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();
			}

5176
			__rtnl_unlock();
L
Linus Torvalds 已提交
5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206

			rebroadcast_time = jiffies;
		}

		msleep(250);

		if (time_after(jiffies, warning_time + 10 * HZ)) {
			printk(KERN_EMERG "unregister_netdevice: "
			       "waiting for %s to become free. Usage "
			       "count = %d\n",
			       dev->name, atomic_read(&dev->refcnt));
			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
5207
 * We are invoked by rtnl_unlock().
L
Linus Torvalds 已提交
5208
 * This allows us to deal with problems:
5209
 * 1) We can delete sysfs objects which invoke hotplug
L
Linus Torvalds 已提交
5210 5211 5212
 *    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
5213 5214 5215
 *
 * 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
5216 5217 5218
 */
void netdev_run_todo(void)
{
5219
	struct list_head list;
L
Linus Torvalds 已提交
5220 5221

	/* Snapshot list, allow later requests */
5222
	list_replace_init(&net_todo_list, &list);
H
Herbert Xu 已提交
5223 5224

	__rtnl_unlock();
5225

L
Linus Torvalds 已提交
5226 5227
	while (!list_empty(&list)) {
		struct net_device *dev
5228
			= list_first_entry(&list, struct net_device, todo_list);
L
Linus Torvalds 已提交
5229 5230
		list_del(&dev->todo_list);

5231 5232 5233 5234 5235 5236
		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
5237

5238
		dev->reg_state = NETREG_UNREGISTERED;
L
Linus Torvalds 已提交
5239

5240
		on_each_cpu(flush_backlog, dev, 1);
5241

5242
		netdev_wait_allrefs(dev);
L
Linus Torvalds 已提交
5243

5244 5245
		/* paranoia */
		BUG_ON(atomic_read(&dev->refcnt));
5246 5247 5248
		WARN_ON(dev->ip_ptr);
		WARN_ON(dev->ip6_ptr);
		WARN_ON(dev->dn_ptr);
L
Linus Torvalds 已提交
5249

5250 5251
		if (dev->destructor)
			dev->destructor(dev);
5252 5253 5254

		/* Free network device */
		kobject_put(&dev->dev.kobj);
L
Linus Torvalds 已提交
5255 5256 5257
	}
}

5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283
/**
 *	dev_txq_stats_fold - fold tx_queues stats
 *	@dev: device to get statistics from
 *	@stats: struct net_device_stats to hold results
 */
void dev_txq_stats_fold(const struct net_device *dev,
			struct net_device_stats *stats)
{
	unsigned long tx_bytes = 0, tx_packets = 0, tx_dropped = 0;
	unsigned int i;
	struct netdev_queue *txq;

	for (i = 0; i < dev->num_tx_queues; i++) {
		txq = netdev_get_tx_queue(dev, i);
		tx_bytes   += txq->tx_bytes;
		tx_packets += txq->tx_packets;
		tx_dropped += txq->tx_dropped;
	}
	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);

5284 5285 5286 5287 5288
/**
 *	dev_get_stats	- get network device statistics
 *	@dev: device to get statistics from
 *
 *	Get network statistics from device. The device driver may provide
5289 5290 5291
 *	its own method by setting dev->netdev_ops->get_stats64 or
 *	dev->netdev_ops->get_stats; otherwise the internal statistics
 *	structure is used.
5292
 */
5293
const struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev)
5294
{
5295 5296
	const struct net_device_ops *ops = dev->netdev_ops;

5297 5298
	if (ops->ndo_get_stats64)
		return ops->ndo_get_stats64(dev);
5299
	if (ops->ndo_get_stats)
5300
		return (struct rtnl_link_stats64 *)ops->ndo_get_stats(dev);
5301 5302

	dev_txq_stats_fold(dev, &dev->stats);
5303
	return &dev->stats64;
R
Rusty Russell 已提交
5304
}
5305
EXPORT_SYMBOL(dev_get_stats);
R
Rusty Russell 已提交
5306

5307
static void netdev_init_one_queue(struct net_device *dev,
5308 5309
				  struct netdev_queue *queue,
				  void *_unused)
5310 5311 5312 5313
{
	queue->dev = dev;
}

5314 5315
static void netdev_init_queues(struct net_device *dev)
{
5316 5317
	netdev_init_one_queue(dev, &dev->rx_queue, NULL);
	netdev_for_each_tx_queue(dev, netdev_init_one_queue, NULL);
5318
	spin_lock_init(&dev->tx_global_lock);
5319 5320
}

L
Linus Torvalds 已提交
5321
/**
5322
 *	alloc_netdev_mq - allocate network device
L
Linus Torvalds 已提交
5323 5324 5325
 *	@sizeof_priv:	size of private data to allocate space for
 *	@name:		device name format string
 *	@setup:		callback to initialize device
5326
 *	@queue_count:	the number of subqueues to allocate
L
Linus Torvalds 已提交
5327 5328
 *
 *	Allocates a struct net_device with private data area for driver use
5329 5330
 *	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
5331
 */
5332 5333
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
5334
{
5335
	struct netdev_queue *tx;
L
Linus Torvalds 已提交
5336
	struct net_device *dev;
5337
	size_t alloc_size;
5338
	struct net_device *p;
E
Eric Dumazet 已提交
5339 5340
#ifdef CONFIG_RPS
	struct netdev_rx_queue *rx;
T
Tom Herbert 已提交
5341
	int i;
E
Eric Dumazet 已提交
5342
#endif
L
Linus Torvalds 已提交
5343

5344 5345
	BUG_ON(strlen(name) >= sizeof(dev->name));

5346
	alloc_size = sizeof(struct net_device);
5347 5348
	if (sizeof_priv) {
		/* ensure 32-byte alignment of private area */
5349
		alloc_size = ALIGN(alloc_size, NETDEV_ALIGN);
5350 5351 5352
		alloc_size += sizeof_priv;
	}
	/* ensure 32-byte alignment of whole construct */
5353
	alloc_size += NETDEV_ALIGN - 1;
L
Linus Torvalds 已提交
5354

5355
	p = kzalloc(alloc_size, GFP_KERNEL);
L
Linus Torvalds 已提交
5356
	if (!p) {
5357
		printk(KERN_ERR "alloc_netdev: Unable to allocate device.\n");
L
Linus Torvalds 已提交
5358 5359 5360
		return NULL;
	}

5361
	tx = kcalloc(queue_count, sizeof(struct netdev_queue), GFP_KERNEL);
5362 5363 5364
	if (!tx) {
		printk(KERN_ERR "alloc_netdev: Unable to allocate "
		       "tx qdiscs.\n");
5365
		goto free_p;
5366 5367
	}

E
Eric Dumazet 已提交
5368
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383
	rx = kcalloc(queue_count, sizeof(struct netdev_rx_queue), GFP_KERNEL);
	if (!rx) {
		printk(KERN_ERR "alloc_netdev: Unable to allocate "
		       "rx queues.\n");
		goto free_tx;
	}

	atomic_set(&rx->count, queue_count);

	/*
	 * Set a pointer to first element in the array which holds the
	 * reference count.
	 */
	for (i = 0; i < queue_count; i++)
		rx[i].first = rx;
E
Eric Dumazet 已提交
5384
#endif
T
Tom Herbert 已提交
5385

5386
	dev = PTR_ALIGN(p, NETDEV_ALIGN);
L
Linus Torvalds 已提交
5387
	dev->padded = (char *)dev - (char *)p;
5388 5389

	if (dev_addr_init(dev))
T
Tom Herbert 已提交
5390
		goto free_rx;
5391

5392
	dev_mc_init(dev);
5393
	dev_uc_init(dev);
J
Jiri Pirko 已提交
5394

5395
	dev_net_set(dev, &init_net);
L
Linus Torvalds 已提交
5396

5397 5398
	dev->_tx = tx;
	dev->num_tx_queues = queue_count;
5399
	dev->real_num_tx_queues = queue_count;
5400

E
Eric Dumazet 已提交
5401
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
5402 5403
	dev->_rx = rx;
	dev->num_rx_queues = queue_count;
E
Eric Dumazet 已提交
5404
#endif
T
Tom Herbert 已提交
5405

5406
	dev->gso_max_size = GSO_MAX_SIZE;
L
Linus Torvalds 已提交
5407

5408 5409
	netdev_init_queues(dev);

5410 5411
	INIT_LIST_HEAD(&dev->ethtool_ntuple_list.list);
	dev->ethtool_ntuple_list.count = 0;
5412
	INIT_LIST_HEAD(&dev->napi_list);
5413
	INIT_LIST_HEAD(&dev->unreg_list);
5414
	INIT_LIST_HEAD(&dev->link_watch_list);
5415
	dev->priv_flags = IFF_XMIT_DST_RELEASE;
L
Linus Torvalds 已提交
5416 5417 5418
	setup(dev);
	strcpy(dev->name, name);
	return dev;
5419

T
Tom Herbert 已提交
5420
free_rx:
E
Eric Dumazet 已提交
5421
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
5422
	kfree(rx);
5423
free_tx:
E
Eric Dumazet 已提交
5424
#endif
5425 5426 5427 5428
	kfree(tx);
free_p:
	kfree(p);
	return NULL;
L
Linus Torvalds 已提交
5429
}
5430
EXPORT_SYMBOL(alloc_netdev_mq);
L
Linus Torvalds 已提交
5431 5432 5433 5434 5435

/**
 *	free_netdev - free network device
 *	@dev: device
 *
5436 5437
 *	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
5438 5439 5440 5441
 *	If this is the last reference then it will be freed.
 */
void free_netdev(struct net_device *dev)
{
5442 5443
	struct napi_struct *p, *n;

5444 5445
	release_net(dev_net(dev));

5446 5447
	kfree(dev->_tx);

5448 5449 5450
	/* Flush device addresses */
	dev_addr_flush(dev);

5451 5452 5453
	/* Clear ethtool n-tuple list */
	ethtool_ntuple_flush(dev);

5454 5455 5456
	list_for_each_entry_safe(p, n, &dev->napi_list, dev_list)
		netif_napi_del(p);

S
Stephen Hemminger 已提交
5457
	/*  Compatibility with error handling in drivers */
L
Linus Torvalds 已提交
5458 5459 5460 5461 5462 5463 5464 5465
	if (dev->reg_state == NETREG_UNINITIALIZED) {
		kfree((char *)dev - dev->padded);
		return;
	}

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

5466 5467
	/* will free via device release */
	put_device(&dev->dev);
L
Linus Torvalds 已提交
5468
}
E
Eric Dumazet 已提交
5469
EXPORT_SYMBOL(free_netdev);
5470

5471 5472 5473 5474 5475 5476
/**
 *	synchronize_net -  Synchronize with packet receive processing
 *
 *	Wait for packets currently being received to be done.
 *	Does not block later packets from starting.
 */
5477
void synchronize_net(void)
L
Linus Torvalds 已提交
5478 5479
{
	might_sleep();
5480
	synchronize_rcu();
L
Linus Torvalds 已提交
5481
}
E
Eric Dumazet 已提交
5482
EXPORT_SYMBOL(synchronize_net);
L
Linus Torvalds 已提交
5483 5484

/**
5485
 *	unregister_netdevice_queue - remove device from the kernel
L
Linus Torvalds 已提交
5486
 *	@dev: device
5487
 *	@head: list
5488
 *
L
Linus Torvalds 已提交
5489
 *	This function shuts down a device interface and removes it
5490
 *	from the kernel tables.
5491
 *	If head not NULL, device is queued to be unregistered later.
L
Linus Torvalds 已提交
5492 5493 5494 5495 5496
 *
 *	Callers must hold the rtnl semaphore.  You may want
 *	unregister_netdev() instead of this.
 */

5497
void unregister_netdevice_queue(struct net_device *dev, struct list_head *head)
L
Linus Torvalds 已提交
5498
{
5499 5500
	ASSERT_RTNL();

5501
	if (head) {
5502
		list_move_tail(&dev->unreg_list, head);
5503 5504 5505 5506 5507
	} else {
		rollback_registered(dev);
		/* Finish processing unregister after unlock */
		net_set_todo(dev);
	}
L
Linus Torvalds 已提交
5508
}
5509
EXPORT_SYMBOL(unregister_netdevice_queue);
L
Linus Torvalds 已提交
5510

5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524
/**
 *	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);
	}
}
5525
EXPORT_SYMBOL(unregister_netdevice_many);
5526

L
Linus Torvalds 已提交
5527 5528 5529 5530 5531
/**
 *	unregister_netdev - remove device from the kernel
 *	@dev: device
 *
 *	This function shuts down a device interface and removes it
5532
 *	from the kernel tables.
L
Linus Torvalds 已提交
5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545
 *
 *	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);

5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577
/**
 *	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;
5578
	if (net_eq(dev_net(dev), net))
5579 5580 5581 5582 5583 5584
		goto out;

	/* Pick the destination device name, and ensure
	 * we can use it in the destination network namespace.
	 */
	err = -EEXIST;
5585
	if (__dev_get_by_name(net, dev->name)) {
5586 5587 5588
		/* We get here if we can't use the current device name */
		if (!pat)
			goto out;
5589
		if (dev_get_valid_name(dev, pat, 1))
5590 5591 5592 5593 5594 5595 5596 5597
			goto out;
	}

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

	/* If device is running close it first. */
5598
	dev_close(dev);
5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612

	/* 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.
	*/
	call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5613
	call_netdevice_notifiers(NETDEV_UNREGISTER_BATCH, dev);
5614 5615 5616 5617

	/*
	 *	Flush the unicast and multicast chains
	 */
5618
	dev_uc_flush(dev);
5619
	dev_mc_flush(dev);
5620 5621

	/* Actually switch the network namespace */
5622
	dev_net_set(dev, net);
5623 5624 5625 5626 5627 5628 5629 5630 5631

	/* 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;
	}

5632
	/* Fixup kobjects */
5633
	err = device_rename(&dev->dev, dev->name);
5634
	WARN_ON(err);
5635 5636 5637 5638 5639 5640 5641

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

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

5642 5643 5644 5645 5646 5647
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
	rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);

5648 5649 5650 5651 5652
	synchronize_net();
	err = 0;
out:
	return err;
}
5653
EXPORT_SYMBOL_GPL(dev_change_net_namespace);
5654

L
Linus Torvalds 已提交
5655 5656 5657 5658 5659 5660 5661 5662 5663
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;

5664
	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN)
L
Linus Torvalds 已提交
5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680
		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. */
5681 5682 5683 5684 5685 5686
	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
5687 5688 5689 5690 5691

	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_enable();

	/* Process offline CPU's input_pkt_queue */
5692
	while ((skb = __skb_dequeue(&oldsd->process_queue))) {
L
Linus Torvalds 已提交
5693
		netif_rx(skb);
5694
		input_queue_head_incr(oldsd);
T
Tom Herbert 已提交
5695
	}
5696
	while ((skb = __skb_dequeue(&oldsd->input_pkt_queue))) {
5697
		netif_rx(skb);
5698 5699
		input_queue_head_incr(oldsd);
	}
L
Linus Torvalds 已提交
5700 5701 5702 5703 5704

	return NOTIFY_OK;
}


5705
/**
5706 5707 5708 5709
 *	netdev_increment_features - increment feature set by one
 *	@all: current feature set
 *	@one: new feature set
 *	@mask: mask feature set
5710 5711
 *
 *	Computes a new feature set after adding a device with feature set
5712 5713
 *	@one to the master device with current feature set @all.  Will not
 *	enable anything that is off in @mask. Returns the new feature set.
5714
 */
5715 5716 5717 5718
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
5719
	if (all & NETIF_F_NO_CSUM && !(one & NETIF_F_NO_CSUM))
5720 5721 5722 5723 5724 5725 5726 5727
		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);
		}
5728

5729 5730 5731 5732 5733 5734
		/* 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;
		}
	}
5735

5736
	one |= NETIF_F_ALL_CSUM;
5737

5738
	one |= all & NETIF_F_ONE_FOR_ALL;
5739
	all &= one | NETIF_F_LLTX | NETIF_F_GSO | NETIF_F_UFO;
5740
	all |= one & mask & NETIF_F_ONE_FOR_ALL;
5741 5742 5743

	return all;
}
5744
EXPORT_SYMBOL(netdev_increment_features);
5745

5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758
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;
}

5759
/* Initialize per network namespace state */
5760
static int __net_init netdev_init(struct net *net)
5761 5762 5763
{
	INIT_LIST_HEAD(&net->dev_base_head);

5764 5765 5766
	net->dev_name_head = netdev_create_hash();
	if (net->dev_name_head == NULL)
		goto err_name;
5767

5768 5769 5770
	net->dev_index_head = netdev_create_hash();
	if (net->dev_index_head == NULL)
		goto err_idx;
5771 5772

	return 0;
5773 5774 5775 5776 5777

err_idx:
	kfree(net->dev_name_head);
err_name:
	return -ENOMEM;
5778 5779
}

5780 5781 5782 5783 5784 5785 5786 5787
/**
 *	netdev_drivername - network driver for the device
 *	@dev: network device
 *	@buffer: buffer for resulting name
 *	@len: size of buffer
 *
 *	Determine network driver for device.
 */
5788
char *netdev_drivername(const struct net_device *dev, char *buffer, int len)
5789
{
5790 5791
	const struct device_driver *driver;
	const struct device *parent;
5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807

	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;
}

5808
static void __net_exit netdev_exit(struct net *net)
5809 5810 5811 5812 5813
{
	kfree(net->dev_name_head);
	kfree(net->dev_index_head);
}

5814
static struct pernet_operations __net_initdata netdev_net_ops = {
5815 5816 5817 5818
	.init = netdev_init,
	.exit = netdev_exit,
};

5819
static void __net_exit default_device_exit(struct net *net)
5820
{
5821
	struct net_device *dev, *aux;
5822
	/*
5823
	 * Push all migratable network devices back to the
5824 5825 5826
	 * initial network namespace
	 */
	rtnl_lock();
5827
	for_each_netdev_safe(net, dev, aux) {
5828
		int err;
5829
		char fb_name[IFNAMSIZ];
5830 5831 5832 5833 5834

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

5835 5836 5837
		/* Leave virtual devices for the generic cleanup */
		if (dev->rtnl_link_ops)
			continue;
5838

5839
		/* Push remaing network devices to init_net */
5840 5841
		snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
		err = dev_change_net_namespace(dev, &init_net, fb_name);
5842
		if (err) {
5843
			printk(KERN_EMERG "%s: failed to move %s to init_net: %d\n",
5844
				__func__, dev->name, err);
5845
			BUG();
5846 5847 5848 5849 5850
		}
	}
	rtnl_unlock();
}

5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874
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();
}

5875
static struct pernet_operations __net_initdata default_device_ops = {
5876
	.exit = default_device_exit,
5877
	.exit_batch = default_device_exit_batch,
5878 5879
};

L
Linus Torvalds 已提交
5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899
/*
 *	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;

5900
	if (netdev_kobject_init())
L
Linus Torvalds 已提交
5901 5902 5903
		goto out;

	INIT_LIST_HEAD(&ptype_all);
5904
	for (i = 0; i < PTYPE_HASH_SIZE; i++)
L
Linus Torvalds 已提交
5905 5906
		INIT_LIST_HEAD(&ptype_base[i]);

5907 5908
	if (register_pernet_subsys(&netdev_net_ops))
		goto out;
L
Linus Torvalds 已提交
5909 5910 5911 5912 5913

	/*
	 *	Initialise the packet receive queues.
	 */

5914
	for_each_possible_cpu(i) {
E
Eric Dumazet 已提交
5915
		struct softnet_data *sd = &per_cpu(softnet_data, i);
L
Linus Torvalds 已提交
5916

C
Changli Gao 已提交
5917
		memset(sd, 0, sizeof(*sd));
E
Eric Dumazet 已提交
5918
		skb_queue_head_init(&sd->input_pkt_queue);
5919
		skb_queue_head_init(&sd->process_queue);
E
Eric Dumazet 已提交
5920 5921
		sd->completion_queue = NULL;
		INIT_LIST_HEAD(&sd->poll_list);
5922 5923
		sd->output_queue = NULL;
		sd->output_queue_tailp = &sd->output_queue;
E
Eric Dumazet 已提交
5924
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
5925 5926 5927 5928
		sd->csd.func = rps_trigger_softirq;
		sd->csd.info = sd;
		sd->csd.flags = 0;
		sd->cpu = i;
5929
#endif
T
Tom Herbert 已提交
5930

E
Eric Dumazet 已提交
5931 5932 5933 5934
		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
5935 5936 5937 5938
	}

	dev_boot_phase = 0;

5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953
	/* 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;

5954 5955
	open_softirq(NET_TX_SOFTIRQ, net_tx_action);
	open_softirq(NET_RX_SOFTIRQ, net_rx_action);
L
Linus Torvalds 已提交
5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966

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

subsys_initcall(net_dev_init);

5967 5968
static int __init initialize_hashrnd(void)
{
T
Tom Herbert 已提交
5969
	get_random_bytes(&hashrnd, sizeof(hashrnd));
5970 5971 5972 5973 5974
	return 0;
}

late_initcall_sync(initialize_hashrnd);