dev.c 156.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 * 	NET3	Protocol independent device support routines.
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 *	Derived from the non IP parts of dev.c 1.0.19
10
 * 		Authors:	Ross Biro
L
Linus Torvalds 已提交
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
 *				Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *				Mark Evans, <evansmp@uhura.aston.ac.uk>
 *
 *	Additional Authors:
 *		Florian la Roche <rzsfl@rz.uni-sb.de>
 *		Alan Cox <gw4pts@gw4pts.ampr.org>
 *		David Hinds <dahinds@users.sourceforge.net>
 *		Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
 *		Adam Sulmicki <adam@cfar.umd.edu>
 *              Pekka Riikonen <priikone@poesidon.pspt.fi>
 *
 *	Changes:
 *              D.J. Barrow     :       Fixed bug where dev->refcnt gets set
 *              			to 2 if register_netdev gets called
 *              			before net_dev_init & also removed a
 *              			few lines of code in the process.
 *		Alan Cox	:	device private ioctl copies fields back.
 *		Alan Cox	:	Transmit queue code does relevant
 *					stunts to keep the queue safe.
 *		Alan Cox	:	Fixed double lock.
 *		Alan Cox	:	Fixed promisc NULL pointer trap
 *		????????	:	Support the full private ioctl range
 *		Alan Cox	:	Moved ioctl permission check into
 *					drivers
 *		Tim Kordas	:	SIOCADDMULTI/SIOCDELMULTI
 *		Alan Cox	:	100 backlog just doesn't cut it when
 *					you start doing multicast video 8)
 *		Alan Cox	:	Rewrote net_bh and list manager.
 *		Alan Cox	: 	Fix ETH_P_ALL echoback lengths.
 *		Alan Cox	:	Took out transmit every packet pass
 *					Saved a few bytes in the ioctl handler
 *		Alan Cox	:	Network driver sets packet type before
 *					calling netif_rx. Saves a function
 *					call a packet.
 *		Alan Cox	:	Hashed net_bh()
 *		Richard Kooijman:	Timestamp fixes.
 *		Alan Cox	:	Wrong field in SIOCGIFDSTADDR
 *		Alan Cox	:	Device lock protection.
 *		Alan Cox	: 	Fixed nasty side effect of device close
 *					changes.
 *		Rudi Cilibrasi	:	Pass the right thing to
 *					set_mac_address()
 *		Dave Miller	:	32bit quantity for the device lock to
 *					make it work out on a Sparc.
 *		Bjorn Ekwall	:	Added KERNELD hack.
 *		Alan Cox	:	Cleaned up the backlog initialise.
 *		Craig Metz	:	SIOCGIFCONF fix if space for under
 *					1 device.
 *	    Thomas Bogendoerfer :	Return ENODEV for dev_open, if there
 *					is no device open function.
 *		Andi Kleen	:	Fix error reporting for SIOCGIFCONF
 *	    Michael Chastain	:	Fix signed/unsigned for SIOCGIFCONF
 *		Cyrus Durgin	:	Cleaned for KMOD
 *		Adam Sulmicki   :	Bug Fix : Network Device Unload
 *					A network device unload needs to purge
 *					the backlog queue.
 *	Paul Rusty Russell	:	SIOCSIFNAME
 *              Pekka Riikonen  :	Netdev boot-time settings code
 *              Andrew Morton   :       Make unregister_netdevice wait
 *              			indefinitely on dev->refcnt
 * 		J Hadi Salim	:	- Backlog queue sampling
 *				        - netif_rx() feedback
 */

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

137 138
#include "net-sysfs.h"

139 140 141
/* Instead of increasing this, you should create a hash table. */
#define MAX_GRO_SKBS 8

H
Herbert Xu 已提交
142 143 144
/* 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
145 146 147 148 149 150 151 152 153 154 155
/*
 *	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
156
 *             the average user (w/out VLANs) will not be adversely affected.
L
Linus Torvalds 已提交
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
 *             --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
 */

173 174 175
#define PTYPE_HASH_SIZE	(16)
#define PTYPE_HASH_MASK	(PTYPE_HASH_SIZE - 1)

L
Linus Torvalds 已提交
176
static DEFINE_SPINLOCK(ptype_lock);
177
static struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
178
static struct list_head ptype_all __read_mostly;	/* Taps */
L
Linus Torvalds 已提交
179 180

/*
181
 * The @dev_base_head list is protected by @dev_base_lock and the rtnl
L
Linus Torvalds 已提交
182 183
 * semaphore.
 *
184
 * Pure readers hold dev_base_lock for reading, or rcu_read_lock()
L
Linus Torvalds 已提交
185 186
 *
 * Writers must hold the rtnl semaphore while they loop through the
187
 * dev_base_head list, and hold dev_base_lock for writing when they do the
L
Linus Torvalds 已提交
188 189 190 191 192 193 194 195 196 197 198 199 200 201
 * 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);

202
static inline struct hlist_head *dev_name_hash(struct net *net, const char *name)
L
Linus Torvalds 已提交
203 204
{
	unsigned hash = full_name_hash(name, strnlen(name, IFNAMSIZ));
205
	return &net->dev_name_head[hash_32(hash, NETDEV_HASHBITS)];
L
Linus Torvalds 已提交
206 207
}

208
static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex)
L
Linus Torvalds 已提交
209
{
210
	return &net->dev_index_head[ifindex & (NETDEV_HASHENTRIES - 1)];
L
Linus Torvalds 已提交
211 212
}

E
Eric Dumazet 已提交
213
static inline void rps_lock(struct softnet_data *sd)
214 215
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
216
	spin_lock(&sd->input_pkt_queue.lock);
217 218 219
#endif
}

E
Eric Dumazet 已提交
220
static inline void rps_unlock(struct softnet_data *sd)
221 222
{
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
223
	spin_unlock(&sd->input_pkt_queue.lock);
224 225 226
#endif
}

227 228 229
/* Device list insertion */
static int list_netdevice(struct net_device *dev)
{
230
	struct net *net = dev_net(dev);
231 232 233 234

	ASSERT_RTNL();

	write_lock_bh(&dev_base_lock);
235
	list_add_tail_rcu(&dev->dev_list, &net->dev_base_head);
236
	hlist_add_head_rcu(&dev->name_hlist, dev_name_hash(net, dev->name));
237 238
	hlist_add_head_rcu(&dev->index_hlist,
			   dev_index_hash(net, dev->ifindex));
239 240 241 242
	write_unlock_bh(&dev_base_lock);
	return 0;
}

243 244 245
/* Device list removal
 * caller must respect a RCU grace period before freeing/reusing dev
 */
246 247 248 249 250 251
static void unlist_netdevice(struct net_device *dev)
{
	ASSERT_RTNL();

	/* Unlink dev from the device chain */
	write_lock_bh(&dev_base_lock);
252
	list_del_rcu(&dev->dev_list);
253
	hlist_del_rcu(&dev->name_hlist);
254
	hlist_del_rcu(&dev->index_hlist);
255 256 257
	write_unlock_bh(&dev_base_lock);
}

L
Linus Torvalds 已提交
258 259 260 261
/*
 *	Our notifier list
 */

262
static RAW_NOTIFIER_HEAD(netdev_chain);
L
Linus Torvalds 已提交
263 264 265 266 267

/*
 *	Device drivers call our routines to queue packets here. We empty the
 *	queue in the local softnet handler.
 */
268

269
DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
E
Eric Dumazet 已提交
270
EXPORT_PER_CPU_SYMBOL(softnet_data);
L
Linus Torvalds 已提交
271

272
#ifdef CONFIG_LOCKDEP
273
/*
274
 * register_netdevice() inits txq->_xmit_lock and sets lockdep class
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
 * 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,
291
	 ARPHRD_IEEE80211_PRISM, ARPHRD_IEEE80211_RADIOTAP, ARPHRD_PHONET,
292
	 ARPHRD_PHONET_PIPE, ARPHRD_IEEE802154,
293
	 ARPHRD_VOID, ARPHRD_NONE};
294

295
static const char *const netdev_lock_name[] =
296 297 298 299 300 301 302 303 304 305 306 307 308
	{"_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",
309
	 "_xmit_IEEE80211_PRISM", "_xmit_IEEE80211_RADIOTAP", "_xmit_PHONET",
310
	 "_xmit_PHONET_PIPE", "_xmit_IEEE802154",
311
	 "_xmit_VOID", "_xmit_NONE"};
312 313

static struct lock_class_key netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)];
314
static struct lock_class_key netdev_addr_lock_key[ARRAY_SIZE(netdev_lock_type)];
315 316 317 318 319 320 321 322 323 324 325 326

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;
}

327 328
static inline void netdev_set_xmit_lockdep_class(spinlock_t *lock,
						 unsigned short dev_type)
329 330 331 332 333 334 335
{
	int i;

	i = netdev_lock_pos(dev_type);
	lockdep_set_class_and_name(lock, &netdev_xmit_lock_key[i],
				   netdev_lock_name[i]);
}
336 337 338 339 340 341 342 343 344 345

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

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

		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)
 */

378 379 380 381 382 383 384 385
static inline struct list_head *ptype_head(const struct packet_type *pt)
{
	if (pt->type == htons(ETH_P_ALL))
		return &ptype_all;
	else
		return &ptype_base[ntohs(pt->type) & PTYPE_HASH_MASK];
}

L
Linus Torvalds 已提交
386 387 388 389 390 391 392 393
/**
 *	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.
 *
394
 *	This call does not sleep therefore it can not
L
Linus Torvalds 已提交
395 396 397 398 399 400
 *	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)
{
401
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
402

403 404 405
	spin_lock(&ptype_lock);
	list_add_rcu(&pt->list, head);
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
406
}
E
Eric Dumazet 已提交
407
EXPORT_SYMBOL(dev_add_pack);
L
Linus Torvalds 已提交
408 409 410 411 412 413 414 415

/**
 *	__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
416
 *	returns.
L
Linus Torvalds 已提交
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)
{
424
	struct list_head *head = ptype_head(pt);
L
Linus Torvalds 已提交
425 426
	struct packet_type *pt1;

427
	spin_lock(&ptype_lock);
L
Linus Torvalds 已提交
428 429 430 431 432 433 434 435 436 437

	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:
438
	spin_unlock(&ptype_lock);
L
Linus Torvalds 已提交
439
}
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_rcu - find a device by its hardware address
748
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
749 750 751 752
 *	@type: media type of device
 *	@ha: hardware address
 *
 *	Search for an interface by MAC address. Returns NULL if the device
E
Eric Dumazet 已提交
753 754
 *	is not found or a pointer to the device.
 *	The caller must hold RCU or RTNL.
755
 *	The returned device has not had its ref count increased
L
Linus Torvalds 已提交
756 757 758 759
 *	and the caller must therefore be careful about locking
 *
 */

760 761
struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
				       const char *ha)
L
Linus Torvalds 已提交
762 763 764
{
	struct net_device *dev;

765
	for_each_netdev_rcu(net, dev)
L
Linus Torvalds 已提交
766 767
		if (dev->type == type &&
		    !memcmp(dev->dev_addr, ha, dev->addr_len))
768 769 770
			return dev;

	return NULL;
L
Linus Torvalds 已提交
771
}
772
EXPORT_SYMBOL(dev_getbyhwaddr_rcu);
773

774
struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type)
L
Linus Torvalds 已提交
775 776 777
{
	struct net_device *dev;

778
	ASSERT_RTNL();
779
	for_each_netdev(net, dev)
780
		if (dev->type == type)
781 782 783
			return dev;

	return NULL;
784 785 786
}
EXPORT_SYMBOL(__dev_getfirstbyhwtype);

787
struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
788
{
789
	struct net_device *dev, *ret = NULL;
790

791 792 793 794 795 796 797 798 799
	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
800 801 802 803
}
EXPORT_SYMBOL(dev_getfirstbyhwtype);

/**
E
Eric Dumazet 已提交
804
 *	dev_get_by_flags_rcu - find any device with given flags
805
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
806 807 808 809
 *	@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
810 811
 *	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
812 813
 */

E
Eric Dumazet 已提交
814
struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
E
Eric Dumazet 已提交
815
				    unsigned short mask)
L
Linus Torvalds 已提交
816
{
817
	struct net_device *dev, *ret;
L
Linus Torvalds 已提交
818

819
	ret = NULL;
820
	for_each_netdev_rcu(net, dev) {
L
Linus Torvalds 已提交
821
		if (((dev->flags ^ if_flags) & mask) == 0) {
822
			ret = dev;
L
Linus Torvalds 已提交
823 824 825
			break;
		}
	}
826
	return ret;
L
Linus Torvalds 已提交
827
}
E
Eric Dumazet 已提交
828
EXPORT_SYMBOL(dev_get_by_flags_rcu);
L
Linus Torvalds 已提交
829 830 831 832 833 834

/**
 *	dev_valid_name - check if name is okay for network device
 *	@name: name string
 *
 *	Network device names need to be valid file names to
835 836
 *	to allow sysfs to work.  We also disallow any kind of
 *	whitespace.
L
Linus Torvalds 已提交
837
 */
838
int dev_valid_name(const char *name)
L
Linus Torvalds 已提交
839
{
840 841
	if (*name == '\0')
		return 0;
842 843
	if (strlen(name) >= IFNAMSIZ)
		return 0;
844 845 846 847 848 849 850 851 852
	if (!strcmp(name, ".") || !strcmp(name, ".."))
		return 0;

	while (*name) {
		if (*name == '/' || isspace(*name))
			return 0;
		name++;
	}
	return 1;
L
Linus Torvalds 已提交
853
}
E
Eric Dumazet 已提交
854
EXPORT_SYMBOL(dev_valid_name);
L
Linus Torvalds 已提交
855 856

/**
857 858
 *	__dev_alloc_name - allocate a name for a device
 *	@net: network namespace to allocate the device name in
L
Linus Torvalds 已提交
859
 *	@name: name format string
860
 *	@buf:  scratch buffer and result name string
L
Linus Torvalds 已提交
861 862
 *
 *	Passed a format string - eg "lt%d" it will try and find a suitable
S
Stephen Hemminger 已提交
863 864 865 866 867 868
 *	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
869 870
 */

871
static int __dev_alloc_name(struct net *net, const char *name, char *buf)
L
Linus Torvalds 已提交
872 873 874 875
{
	int i = 0;
	const char *p;
	const int max_netdevices = 8*PAGE_SIZE;
S
Stephen Hemminger 已提交
876
	unsigned long *inuse;
L
Linus Torvalds 已提交
877 878 879 880 881 882 883 884 885 886 887 888 889
	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
890
		inuse = (unsigned long *) get_zeroed_page(GFP_ATOMIC);
L
Linus Torvalds 已提交
891 892 893
		if (!inuse)
			return -ENOMEM;

894
		for_each_netdev(net, d) {
L
Linus Torvalds 已提交
895 896 897 898 899 900
			if (!sscanf(d->name, name, &i))
				continue;
			if (i < 0 || i >= max_netdevices)
				continue;

			/*  avoid cases where sscanf is not exact inverse of printf */
901
			snprintf(buf, IFNAMSIZ, name, i);
L
Linus Torvalds 已提交
902 903 904 905 906 907 908 909
			if (!strncmp(buf, d->name, IFNAMSIZ))
				set_bit(i, inuse);
		}

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

910 911
	if (buf != name)
		snprintf(buf, IFNAMSIZ, name, i);
912
	if (!__dev_get_by_name(net, buf))
L
Linus Torvalds 已提交
913 914 915 916 917 918 919 920 921
		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;
}

922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941
/**
 *	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;

942 943
	BUG_ON(!dev_net(dev));
	net = dev_net(dev);
944 945 946 947 948
	ret = __dev_alloc_name(net, name, buf);
	if (ret >= 0)
		strlcpy(dev->name, buf, IFNAMSIZ);
	return ret;
}
E
Eric Dumazet 已提交
949
EXPORT_SYMBOL(dev_alloc_name);
950

951
static int dev_get_valid_name(struct net_device *dev, const char *name, bool fmt)
952
{
953 954 955 956 957
	struct net *net;

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

958 959 960 961
	if (!dev_valid_name(name))
		return -EINVAL;

	if (fmt && strchr(name, '%'))
962
		return dev_alloc_name(dev, name);
963 964
	else if (__dev_get_by_name(net, name))
		return -EEXIST;
965 966
	else if (dev->name != name)
		strlcpy(dev->name, name, IFNAMSIZ);
967 968 969

	return 0;
}
L
Linus Torvalds 已提交
970 971 972 973 974 975 976 977 978

/**
 *	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.
 */
979
int dev_change_name(struct net_device *dev, const char *newname)
L
Linus Torvalds 已提交
980
{
981
	char oldname[IFNAMSIZ];
L
Linus Torvalds 已提交
982
	int err = 0;
983
	int ret;
984
	struct net *net;
L
Linus Torvalds 已提交
985 986

	ASSERT_RTNL();
987
	BUG_ON(!dev_net(dev));
L
Linus Torvalds 已提交
988

989
	net = dev_net(dev);
L
Linus Torvalds 已提交
990 991 992
	if (dev->flags & IFF_UP)
		return -EBUSY;

993 994 995
	if (strncmp(newname, dev->name, IFNAMSIZ) == 0)
		return 0;

996 997
	memcpy(oldname, dev->name, IFNAMSIZ);

998
	err = dev_get_valid_name(dev, newname, 1);
999 1000
	if (err < 0)
		return err;
L
Linus Torvalds 已提交
1001

1002
rollback:
1003 1004 1005 1006
	ret = device_rename(&dev->dev, dev->name);
	if (ret) {
		memcpy(dev->name, oldname, IFNAMSIZ);
		return ret;
1007
	}
1008 1009

	write_lock_bh(&dev_base_lock);
1010
	hlist_del(&dev->name_hlist);
1011 1012 1013 1014 1015 1016
	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));
1017 1018
	write_unlock_bh(&dev_base_lock);

1019
	ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev);
1020 1021 1022
	ret = notifier_to_errno(ret);

	if (ret) {
1023 1024
		/* err >= 0 after dev_alloc_name() or stores the first errno */
		if (err >= 0) {
1025 1026 1027
			err = ret;
			memcpy(dev->name, oldname, IFNAMSIZ);
			goto rollback;
1028 1029 1030 1031
		} else {
			printk(KERN_ERR
			       "%s: name change rollback failed: %d.\n",
			       dev->name, ret);
1032 1033
		}
	}
L
Linus Torvalds 已提交
1034 1035 1036 1037

	return err;
}

1038 1039 1040 1041
/**
 *	dev_set_alias - change ifalias of a device
 *	@dev: device
 *	@alias: name up to IFALIASZ
1042
 *	@len: limit of bytes to copy from info
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
 *
 *	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;

1053 1054 1055 1056 1057 1058 1059 1060
	if (!len) {
		if (dev->ifalias) {
			kfree(dev->ifalias);
			dev->ifalias = NULL;
		}
		return 0;
	}

E
Eric Dumazet 已提交
1061
	dev->ifalias = krealloc(dev->ifalias, len + 1, GFP_KERNEL);
1062 1063 1064 1065 1066 1067 1068 1069
	if (!dev->ifalias)
		return -ENOMEM;

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


1070
/**
S
Stephen Hemminger 已提交
1071
 *	netdev_features_change - device changes features
1072 1073 1074 1075 1076 1077
 *	@dev: device to cause notification
 *
 *	Called to indicate a device has changed features.
 */
void netdev_features_change(struct net_device *dev)
{
1078
	call_netdevice_notifiers(NETDEV_FEAT_CHANGE, dev);
1079 1080 1081
}
EXPORT_SYMBOL(netdev_features_change);

L
Linus Torvalds 已提交
1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
/**
 *	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) {
1093
		call_netdevice_notifiers(NETDEV_CHANGE, dev);
L
Linus Torvalds 已提交
1094 1095 1096
		rtmsg_ifinfo(RTM_NEWLINK, dev, 0);
	}
}
E
Eric Dumazet 已提交
1097
EXPORT_SYMBOL(netdev_state_change);
L
Linus Torvalds 已提交
1098

1099
int netdev_bonding_change(struct net_device *dev, unsigned long event)
1100
{
1101
	return call_netdevice_notifiers(event, dev);
1102 1103 1104
}
EXPORT_SYMBOL(netdev_bonding_change);

L
Linus Torvalds 已提交
1105 1106
/**
 *	dev_load 	- load a network module
1107
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
1108 1109 1110 1111 1112 1113 1114
 *	@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.
 */

1115
void dev_load(struct net *net, const char *name)
L
Linus Torvalds 已提交
1116
{
1117
	struct net_device *dev;
L
Linus Torvalds 已提交
1118

1119 1120 1121
	rcu_read_lock();
	dev = dev_get_by_name_rcu(net, name);
	rcu_read_unlock();
L
Linus Torvalds 已提交
1122

1123
	if (!dev && capable(CAP_NET_ADMIN))
L
Linus Torvalds 已提交
1124 1125
		request_module("%s", name);
}
E
Eric Dumazet 已提交
1126
EXPORT_SYMBOL(dev_load);
L
Linus Torvalds 已提交
1127

1128
static int __dev_open(struct net_device *dev)
L
Linus Torvalds 已提交
1129
{
1130
	const struct net_device_ops *ops = dev->netdev_ops;
1131
	int ret;
L
Linus Torvalds 已提交
1132

1133 1134
	ASSERT_RTNL();

L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140
	/*
	 *	Is it even present?
	 */
	if (!netif_device_present(dev))
		return -ENODEV;

1141 1142 1143 1144 1145
	ret = call_netdevice_notifiers(NETDEV_PRE_UP, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		return ret;

L
Linus Torvalds 已提交
1146 1147 1148 1149
	/*
	 *	Call device private open method
	 */
	set_bit(__LINK_STATE_START, &dev->state);
1150

1151 1152
	if (ops->ndo_validate_addr)
		ret = ops->ndo_validate_addr(dev);
1153

1154 1155
	if (!ret && ops->ndo_open)
		ret = ops->ndo_open(dev);
L
Linus Torvalds 已提交
1156

1157
	/*
L
Linus Torvalds 已提交
1158 1159 1160
	 *	If it went open OK then:
	 */

1161 1162 1163
	if (ret)
		clear_bit(__LINK_STATE_START, &dev->state);
	else {
L
Linus Torvalds 已提交
1164 1165 1166 1167 1168
		/*
		 *	Set the flags.
		 */
		dev->flags |= IFF_UP;

1169 1170 1171
		/*
		 *	Enable NET_DMA
		 */
1172
		net_dmaengine_get();
1173

L
Linus Torvalds 已提交
1174 1175 1176
		/*
		 *	Initialize multicasting status
		 */
1177
		dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
1178 1179 1180 1181 1182 1183

		/*
		 *	Wakeup transmit queue engine
		 */
		dev_activate(dev);
	}
1184

L
Linus Torvalds 已提交
1185 1186 1187 1188
	return ret;
}

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

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

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

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

	return ret;
}
EXPORT_SYMBOL(dev_open);

1227
static int __dev_close_many(struct list_head *head)
L
Linus Torvalds 已提交
1228
{
1229
	struct net_device *dev;
1230

1231
	ASSERT_RTNL();
1232 1233
	might_sleep();

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

1241
		clear_bit(__LINK_STATE_START, &dev->state);
L
Linus Torvalds 已提交
1242

1243 1244 1245 1246 1247 1248 1249 1250
		/* Synchronize to scheduled poll. We cannot touch poll list, it
		 * can be even on different cpu. So just clear netif_running().
		 *
		 * dev->stop() will invoke napi_disable() on all of it's
		 * napi_struct instances on this device.
		 */
		smp_mb__after_clear_bit(); /* Commit netif_running(). */
	}
L
Linus Torvalds 已提交
1251

1252
	dev_deactivate_many(head);
1253

1254 1255
	list_for_each_entry(dev, head, unreg_list) {
		const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
1256

1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
		/*
		 *	Call the device specific close. This cannot fail.
		 *	Only if device is UP
		 *
		 *	We allow it to be called even after a DETACH hot-plug
		 *	event.
		 */
		if (ops->ndo_stop)
			ops->ndo_stop(dev);

		/*
		 *	Device is now down.
		 */

		dev->flags &= ~IFF_UP;

		/*
		 *	Shutdown NET_DMA
		 */
		net_dmaengine_put();
	}

	return 0;
}

static int __dev_close(struct net_device *dev)
{
	LIST_HEAD(single);

	list_add(&dev->unreg_list, &single);
	return __dev_close_many(&single);
}

E
Eric Dumazet 已提交
1290
static int dev_close_many(struct list_head *head)
1291 1292 1293
{
	struct net_device *dev, *tmp;
	LIST_HEAD(tmp_list);
L
Linus Torvalds 已提交
1294

1295 1296 1297 1298 1299
	list_for_each_entry_safe(dev, tmp, head, unreg_list)
		if (!(dev->flags & IFF_UP))
			list_move(&dev->unreg_list, &tmp_list);

	__dev_close_many(head);
L
Linus Torvalds 已提交
1300 1301

	/*
1302
	 * Tell people we are down
L
Linus Torvalds 已提交
1303
	 */
1304 1305 1306 1307
	list_for_each_entry(dev, head, unreg_list) {
		rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING);
		call_netdevice_notifiers(NETDEV_DOWN, dev);
	}
1308

1309 1310
	/* rollback_registered_many needs the complete original list */
	list_splice(&tmp_list, head);
1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324
	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)
{
1325
	LIST_HEAD(single);
L
Linus Torvalds 已提交
1326

1327 1328
	list_add(&dev->unreg_list, &single);
	dev_close_many(&single);
1329

L
Linus Torvalds 已提交
1330 1331
	return 0;
}
E
Eric Dumazet 已提交
1332
EXPORT_SYMBOL(dev_close);
L
Linus Torvalds 已提交
1333 1334


1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357
/**
 *	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);


1358 1359
static int dev_boot_phase = 1;

L
Linus Torvalds 已提交
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374
/*
 *	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
1375
 *	to the new notifier to allow device to have a race free
L
Linus Torvalds 已提交
1376 1377 1378 1379 1380 1381
 *	view of the network device list.
 */

int register_netdevice_notifier(struct notifier_block *nb)
{
	struct net_device *dev;
1382
	struct net_device *last;
1383
	struct net *net;
L
Linus Torvalds 已提交
1384 1385 1386
	int err;

	rtnl_lock();
1387
	err = raw_notifier_chain_register(&netdev_chain, nb);
1388 1389
	if (err)
		goto unlock;
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400
	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
1401

1402 1403
			nb->notifier_call(nb, NETDEV_UP, dev);
		}
L
Linus Torvalds 已提交
1404
	}
1405 1406

unlock:
L
Linus Torvalds 已提交
1407 1408
	rtnl_unlock();
	return err;
1409 1410 1411

rollback:
	last = dev;
1412 1413 1414 1415
	for_each_net(net) {
		for_each_netdev(net, dev) {
			if (dev == last)
				break;
1416

1417 1418 1419 1420 1421
			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);
1422
			nb->notifier_call(nb, NETDEV_UNREGISTER_BATCH, dev);
1423 1424
		}
	}
1425 1426

	raw_notifier_chain_unregister(&netdev_chain, nb);
1427
	goto unlock;
L
Linus Torvalds 已提交
1428
}
E
Eric Dumazet 已提交
1429
EXPORT_SYMBOL(register_netdevice_notifier);
L
Linus Torvalds 已提交
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442

/**
 *	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)
{
1443 1444 1445
	int err;

	rtnl_lock();
1446
	err = raw_notifier_chain_unregister(&netdev_chain, nb);
1447 1448
	rtnl_unlock();
	return err;
L
Linus Torvalds 已提交
1449
}
E
Eric Dumazet 已提交
1450
EXPORT_SYMBOL(unregister_netdevice_notifier);
L
Linus Torvalds 已提交
1451 1452 1453 1454

/**
 *	call_netdevice_notifiers - call all network notifier blocks
 *      @val: value passed unmodified to notifier function
1455
 *      @dev: net_device pointer passed unmodified to notifier function
L
Linus Torvalds 已提交
1456 1457
 *
 *	Call all network notifier blocks.  Parameters and return value
1458
 *	are as for raw_notifier_call_chain().
L
Linus Torvalds 已提交
1459 1460
 */

1461
int call_netdevice_notifiers(unsigned long val, struct net_device *dev)
L
Linus Torvalds 已提交
1462
{
1463
	ASSERT_RTNL();
1464
	return raw_notifier_call_chain(&netdev_chain, val, dev);
L
Linus Torvalds 已提交
1465 1466 1467 1468 1469 1470 1471 1472 1473
}

/* 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
1474
EXPORT_SYMBOL(net_enable_timestamp);
L
Linus Torvalds 已提交
1475 1476 1477 1478 1479

void net_disable_timestamp(void)
{
	atomic_dec(&netstamp_needed);
}
E
Eric Dumazet 已提交
1480
EXPORT_SYMBOL(net_disable_timestamp);
L
Linus Torvalds 已提交
1481

E
Eric Dumazet 已提交
1482
static inline void net_timestamp_set(struct sk_buff *skb)
L
Linus Torvalds 已提交
1483 1484
{
	if (atomic_read(&netstamp_needed))
1485
		__net_timestamp(skb);
1486 1487
	else
		skb->tstamp.tv64 = 0;
L
Linus Torvalds 已提交
1488 1489
}

E
Eric Dumazet 已提交
1490 1491 1492 1493 1494 1495
static inline void net_timestamp_check(struct sk_buff *skb)
{
	if (!skb->tstamp.tv64 && atomic_read(&netstamp_needed))
		__net_timestamp(skb);
}

1496 1497 1498 1499 1500 1501 1502 1503
/**
 * dev_forward_skb - loopback an skb to another netif
 *
 * @dev: destination network device
 * @skb: buffer to forward
 *
 * return values:
 *	NET_RX_SUCCESS	(no congestion)
1504
 *	NET_RX_DROP     (packet was dropped, but freed)
1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
 *
 * 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);
1517
	nf_reset(skb);
1518

1519
	if (unlikely(!(dev->flags & IFF_UP) ||
1520
		     (skb->len > (dev->mtu + dev->hard_header_len + VLAN_HLEN)))) {
1521
		atomic_long_inc(&dev->rx_dropped);
1522
		kfree_skb(skb);
1523
		return NET_RX_DROP;
1524
	}
1525
	skb_set_dev(skb, dev);
1526 1527 1528 1529 1530 1531 1532
	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);

1533 1534 1535 1536 1537 1538 1539 1540
static inline int deliver_skb(struct sk_buff *skb,
			      struct packet_type *pt_prev,
			      struct net_device *orig_dev)
{
	atomic_inc(&skb->users);
	return pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
}

L
Linus Torvalds 已提交
1541 1542 1543 1544 1545
/*
 *	Support routine. Sends outgoing frames to any network
 *	taps currently in use.
 */

1546
static void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
L
Linus Torvalds 已提交
1547 1548
{
	struct packet_type *ptype;
1549 1550
	struct sk_buff *skb2 = NULL;
	struct packet_type *pt_prev = NULL;
1551

L
Linus Torvalds 已提交
1552 1553 1554 1555 1556 1557 1558 1559
	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)) {
1560 1561 1562 1563 1564 1565 1566
			if (pt_prev) {
				deliver_skb(skb2, pt_prev, skb->dev);
				pt_prev = ptype;
				continue;
			}

			skb2 = skb_clone(skb, GFP_ATOMIC);
L
Linus Torvalds 已提交
1567 1568 1569
			if (!skb2)
				break;

1570 1571
			net_timestamp_set(skb2);

L
Linus Torvalds 已提交
1572 1573 1574 1575
			/* skb->nh should be correctly
			   set by sender, so that the second statement is
			   just protection against buggy protocols.
			 */
1576
			skb_reset_mac_header(skb2);
L
Linus Torvalds 已提交
1577

1578
			if (skb_network_header(skb2) < skb2->data ||
1579
			    skb2->network_header > skb2->tail) {
L
Linus Torvalds 已提交
1580 1581 1582
				if (net_ratelimit())
					printk(KERN_CRIT "protocol %04x is "
					       "buggy, dev %s\n",
1583 1584
					       ntohs(skb2->protocol),
					       dev->name);
1585
				skb_reset_network_header(skb2);
L
Linus Torvalds 已提交
1586 1587
			}

1588
			skb2->transport_header = skb2->network_header;
L
Linus Torvalds 已提交
1589
			skb2->pkt_type = PACKET_OUTGOING;
1590
			pt_prev = ptype;
L
Linus Torvalds 已提交
1591 1592
		}
	}
1593 1594
	if (pt_prev)
		pt_prev->func(skb2, skb->dev, pt_prev, skb->dev);
L
Linus Torvalds 已提交
1595 1596 1597
	rcu_read_unlock();
}

1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609
/* netif_setup_tc - Handle tc mappings on real_num_tx_queues change
 * @dev: Network device
 * @txq: number of queues available
 *
 * If real_num_tx_queues is changed the tc mappings may no longer be
 * valid. To resolve this verify the tc mapping remains valid and if
 * not NULL the mapping. With no priorities mapping to this
 * offset/count pair it will no longer be used. In the worst case TC0
 * is invalid nothing can be done so disable priority mappings. If is
 * expected that drivers will fix this mapping if they can before
 * calling netif_set_real_num_tx_queues.
 */
E
Eric Dumazet 已提交
1610
static void netif_setup_tc(struct net_device *dev, unsigned int txq)
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
{
	int i;
	struct netdev_tc_txq *tc = &dev->tc_to_txq[0];

	/* If TC0 is invalidated disable TC mapping */
	if (tc->offset + tc->count > txq) {
		pr_warning("Number of in use tx queues changed "
			   "invalidating tc mappings. Priority "
			   "traffic classification disabled!\n");
		dev->num_tc = 0;
		return;
	}

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

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

1640 1641 1642 1643
/*
 * Routine to help set real_num_tx_queues. To avoid skbs mapped to queues
 * greater then real_num_tx_queues stale skbs on the qdisc must be flushed.
 */
1644
int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq)
1645
{
T
Tom Herbert 已提交
1646 1647
	int rc;

1648 1649
	if (txq < 1 || txq > dev->num_tx_queues)
		return -EINVAL;
1650

1651 1652 1653
	if (dev->reg_state == NETREG_REGISTERED) {
		ASSERT_RTNL();

T
Tom Herbert 已提交
1654 1655
		rc = netdev_queue_update_kobjects(dev, dev->real_num_tx_queues,
						  txq);
T
Tom Herbert 已提交
1656 1657 1658
		if (rc)
			return rc;

1659 1660 1661
		if (dev->num_tc)
			netif_setup_tc(dev, txq);

1662 1663
		if (txq < dev->real_num_tx_queues)
			qdisc_reset_all_tx_gt(dev, txq);
1664
	}
1665 1666 1667

	dev->real_num_tx_queues = txq;
	return 0;
1668 1669
}
EXPORT_SYMBOL(netif_set_real_num_tx_queues);
1670

1671 1672 1673 1674 1675 1676 1677 1678
#ifdef CONFIG_RPS
/**
 *	netif_set_real_num_rx_queues - set actual number of RX queues used
 *	@dev: Network device
 *	@rxq: Actual number of RX queues
 *
 *	This must be called either with the rtnl_lock held or before
 *	registration of the net device.  Returns 0 on success, or a
1679 1680
 *	negative error code.  If called before registration, it always
 *	succeeds.
1681 1682 1683 1684 1685
 */
int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
{
	int rc;

T
Tom Herbert 已提交
1686 1687 1688
	if (rxq < 1 || rxq > dev->num_rx_queues)
		return -EINVAL;

1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703
	if (dev->reg_state == NETREG_REGISTERED) {
		ASSERT_RTNL();

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

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

1704
static inline void __netif_reschedule(struct Qdisc *q)
1705
{
1706 1707
	struct softnet_data *sd;
	unsigned long flags;
1708

1709 1710
	local_irq_save(flags);
	sd = &__get_cpu_var(softnet_data);
1711 1712 1713
	q->next_sched = NULL;
	*sd->output_queue_tailp = q;
	sd->output_queue_tailp = &q->next_sched;
1714 1715 1716 1717 1718 1719 1720 1721
	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);
1722 1723 1724
}
EXPORT_SYMBOL(__netif_schedule);

1725
void dev_kfree_skb_irq(struct sk_buff *skb)
1726
{
1727
	if (atomic_dec_and_test(&skb->users)) {
1728 1729
		struct softnet_data *sd;
		unsigned long flags;
1730

1731 1732 1733 1734 1735 1736 1737
		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);
	}
1738
}
1739
EXPORT_SYMBOL(dev_kfree_skb_irq);
1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750

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);


1751 1752 1753 1754 1755 1756
/**
 * netif_device_detach - mark device as removed
 * @dev: network device
 *
 * Mark device as removed from system and therefore no longer available.
 */
1757 1758 1759 1760
void netif_device_detach(struct net_device *dev)
{
	if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1761
		netif_tx_stop_all_queues(dev);
1762 1763 1764 1765
	}
}
EXPORT_SYMBOL(netif_device_detach);

1766 1767 1768 1769 1770 1771
/**
 * netif_device_attach - mark device as attached
 * @dev: network device
 *
 * Mark device as attached from system and restart if needed.
 */
1772 1773 1774 1775
void netif_device_attach(struct net_device *dev)
{
	if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
	    netif_running(dev)) {
1776
		netif_tx_wake_all_queues(dev);
1777
		__netdev_watchdog_up(dev);
1778 1779 1780 1781
	}
}
EXPORT_SYMBOL(netif_device_attach);

1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811
/**
 * 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
1812 1813 1814 1815
/*
 * Invalidate hardware checksum when packet is to be mangled, and
 * complete checksum manually on outgoing path.
 */
1816
int skb_checksum_help(struct sk_buff *skb)
L
Linus Torvalds 已提交
1817
{
1818
	__wsum csum;
1819
	int ret = 0, offset;
L
Linus Torvalds 已提交
1820

1821
	if (skb->ip_summed == CHECKSUM_COMPLETE)
1822 1823 1824 1825 1826
		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
1827 1828
	}

1829
	offset = skb_checksum_start_offset(skb);
1830 1831 1832 1833 1834 1835 1836 1837
	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
1838 1839 1840 1841 1842
		ret = pskb_expand_head(skb, 0, 0, GFP_ATOMIC);
		if (ret)
			goto out;
	}

1843
	*(__sum16 *)(skb->data + offset) = csum_fold(csum);
1844
out_set_summed:
L
Linus Torvalds 已提交
1845
	skb->ip_summed = CHECKSUM_NONE;
1846
out:
L
Linus Torvalds 已提交
1847 1848
	return ret;
}
E
Eric Dumazet 已提交
1849
EXPORT_SYMBOL(skb_checksum_help);
L
Linus Torvalds 已提交
1850

1851 1852 1853
/**
 *	skb_gso_segment - Perform segmentation on skb.
 *	@skb: buffer to segment
1854
 *	@features: features for the output path (see dev->features)
1855 1856
 *
 *	This function segments the given skb and returns a list of segments.
1857 1858 1859
 *
 *	It may return NULL if the skb requires no segmentation.  This is
 *	only possible when GSO is used for verifying header integrity.
1860
 */
1861
struct sk_buff *skb_gso_segment(struct sk_buff *skb, int features)
1862 1863 1864
{
	struct sk_buff *segs = ERR_PTR(-EPROTONOSUPPORT);
	struct packet_type *ptype;
A
Al Viro 已提交
1865
	__be16 type = skb->protocol;
1866
	int vlan_depth = ETH_HLEN;
1867
	int err;
1868

1869 1870
	while (type == htons(ETH_P_8021Q)) {
		struct vlan_hdr *vh;
1871

1872
		if (unlikely(!pskb_may_pull(skb, vlan_depth + VLAN_HLEN)))
1873 1874
			return ERR_PTR(-EINVAL);

1875 1876 1877
		vh = (struct vlan_hdr *)(skb->data + vlan_depth);
		type = vh->h_vlan_encapsulated_proto;
		vlan_depth += VLAN_HLEN;
1878 1879
	}

1880
	skb_reset_mac_header(skb);
1881
	skb->mac_len = skb->network_header - skb->mac_header;
1882 1883
	__skb_pull(skb, skb->mac_len);

1884 1885 1886 1887 1888 1889 1890
	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);

J
Joe Perches 已提交
1891
		WARN(1, "%s: caps=(0x%lx, 0x%lx) len=%d data_len=%d ip_summed=%d\n",
1892 1893 1894 1895
		     info.driver, dev ? dev->features : 0L,
		     skb->sk ? skb->sk->sk_route_caps : 0L,
		     skb->len, skb->data_len, skb->ip_summed);

1896 1897 1898 1899 1900
		if (skb_header_cloned(skb) &&
		    (err = pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
			return ERR_PTR(err);
	}

1901
	rcu_read_lock();
1902 1903
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
1904
		if (ptype->type == type && !ptype->dev && ptype->gso_segment) {
1905
			if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
1906 1907 1908 1909
				err = ptype->gso_send_check(skb);
				segs = ERR_PTR(err);
				if (err || skb_gso_ok(skb, features))
					break;
1910 1911
				__skb_push(skb, (skb->data -
						 skb_network_header(skb)));
1912
			}
1913
			segs = ptype->gso_segment(skb, features);
1914 1915 1916 1917 1918
			break;
		}
	}
	rcu_read_unlock();

1919
	__skb_push(skb, skb->data - skb_mac_header(skb));
1920

1921 1922 1923 1924
	return segs;
}
EXPORT_SYMBOL(skb_gso_segment);

1925 1926 1927 1928 1929
/* Take action when hardware reception checksum errors are detected. */
#ifdef CONFIG_BUG
void netdev_rx_csum_fault(struct net_device *dev)
{
	if (net_ratelimit()) {
1930
		printk(KERN_ERR "%s: hw csum failure.\n",
1931
			dev ? dev->name : "<unknown>");
1932 1933 1934 1935 1936 1937
		dump_stack();
	}
}
EXPORT_SYMBOL(netdev_rx_csum_fault);
#endif

L
Linus Torvalds 已提交
1938 1939 1940 1941 1942
/* 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
1943
static int illegal_highdma(struct net_device *dev, struct sk_buff *skb)
L
Linus Torvalds 已提交
1944
{
1945
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
1946
	int i;
1947 1948 1949 1950 1951
	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
1952

1953 1954
	if (PCI_DMA_BUS_IS_PHYS) {
		struct device *pdev = dev->dev.parent;
L
Linus Torvalds 已提交
1955

E
Eric Dumazet 已提交
1956 1957
		if (!pdev)
			return 0;
1958 1959 1960 1961 1962 1963
		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;
		}
	}
1964
#endif
L
Linus Torvalds 已提交
1965 1966 1967
	return 0;
}

1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
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
1994
 *	@features: device features as applicable to this skb
1995 1996 1997 1998
 *
 *	This function segments the given skb and stores the list of segments
 *	in skb->next.
 */
1999
static int dev_gso_segment(struct sk_buff *skb, int features)
2000 2001
{
	struct sk_buff *segs;
2002 2003 2004 2005 2006 2007

	segs = skb_gso_segment(skb, features);

	/* Verifying header integrity only. */
	if (!segs)
		return 0;
2008

2009
	if (IS_ERR(segs))
2010 2011 2012 2013 2014 2015 2016 2017 2018
		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
2019 2020
/*
 * Try to orphan skb early, right before transmission by the device.
2021 2022
 * We cannot orphan skb if tx timestamp is requested or the sk-reference
 * is needed on driver level for other reasons, e.g. see net/can/raw.c
E
Eric Dumazet 已提交
2023 2024 2025
 */
static inline void skb_orphan_try(struct sk_buff *skb)
{
2026 2027
	struct sock *sk = skb->sk;

2028
	if (sk && !skb_shinfo(skb)->tx_flags) {
2029 2030 2031 2032 2033
		/* skb_tx_hash() wont be able to get sk.
		 * We copy sk_hash into skb->rxhash
		 */
		if (!skb->rxhash)
			skb->rxhash = sk->sk_hash;
E
Eric Dumazet 已提交
2034
		skb_orphan(skb);
2035
	}
E
Eric Dumazet 已提交
2036 2037
}

2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048
static bool can_checksum_protocol(unsigned long features, __be16 protocol)
{
	return ((features & NETIF_F_GEN_CSUM) ||
		((features & NETIF_F_V4_CSUM) &&
		 protocol == htons(ETH_P_IP)) ||
		((features & NETIF_F_V6_CSUM) &&
		 protocol == htons(ETH_P_IPV6)) ||
		((features & NETIF_F_FCOE_CRC) &&
		 protocol == htons(ETH_P_FCOE)));
}

2049 2050
static int harmonize_features(struct sk_buff *skb, __be16 protocol, int features)
{
2051
	if (!can_checksum_protocol(features, protocol)) {
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061
		features &= ~NETIF_F_ALL_CSUM;
		features &= ~NETIF_F_SG;
	} else if (illegal_highdma(skb->dev, skb)) {
		features &= ~NETIF_F_SG;
	}

	return features;
}

int netif_skb_features(struct sk_buff *skb)
2062 2063
{
	__be16 protocol = skb->protocol;
2064
	int features = skb->dev->features;
2065 2066 2067 2068

	if (protocol == htons(ETH_P_8021Q)) {
		struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data;
		protocol = veh->h_vlan_encapsulated_proto;
2069 2070 2071
	} else if (!vlan_tx_tag_present(skb)) {
		return harmonize_features(skb, protocol, features);
	}
2072

2073
	features &= (skb->dev->vlan_features | NETIF_F_HW_VLAN_TX);
2074 2075 2076 2077 2078

	if (protocol != htons(ETH_P_8021Q)) {
		return harmonize_features(skb, protocol, features);
	} else {
		features &= NETIF_F_SG | NETIF_F_HIGHDMA | NETIF_F_FRAGLIST |
2079
				NETIF_F_GEN_CSUM | NETIF_F_HW_VLAN_TX;
2080 2081
		return harmonize_features(skb, protocol, features);
	}
2082
}
2083
EXPORT_SYMBOL(netif_skb_features);
2084

2085 2086 2087 2088 2089 2090 2091 2092
/*
 * 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,
2093
				      int features)
2094
{
2095 2096 2097
	return skb_is_nonlinear(skb) &&
			((skb_has_frag_list(skb) &&
				!(features & NETIF_F_FRAGLIST)) ||
2098
			(skb_shinfo(skb)->nr_frags &&
2099
				!(features & NETIF_F_SG)));
2100 2101
}

2102 2103
int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
			struct netdev_queue *txq)
2104
{
2105
	const struct net_device_ops *ops = dev->netdev_ops;
2106
	int rc = NETDEV_TX_OK;
2107

2108
	if (likely(!skb->next)) {
2109 2110
		int features;

2111 2112 2113 2114
		/*
		 * If device doesnt need skb->dst, release it right now while
		 * its hot in this cpu cache
		 */
E
Eric Dumazet 已提交
2115 2116 2117
		if (dev->priv_flags & IFF_XMIT_DST_RELEASE)
			skb_dst_drop(skb);

2118 2119 2120
		if (!list_empty(&ptype_all))
			dev_queue_xmit_nit(skb, dev);

E
Eric Dumazet 已提交
2121
		skb_orphan_try(skb);
2122

2123 2124
		features = netif_skb_features(skb);

2125
		if (vlan_tx_tag_present(skb) &&
2126
		    !(features & NETIF_F_HW_VLAN_TX)) {
2127 2128 2129 2130 2131 2132 2133
			skb = __vlan_put_tag(skb, vlan_tx_tag_get(skb));
			if (unlikely(!skb))
				goto out;

			skb->vlan_tci = 0;
		}

2134
		if (netif_needs_gso(skb, features)) {
2135
			if (unlikely(dev_gso_segment(skb, features)))
2136 2137 2138
				goto out_kfree_skb;
			if (skb->next)
				goto gso;
2139
		} else {
2140
			if (skb_needs_linearize(skb, features) &&
2141 2142 2143 2144 2145 2146 2147 2148
			    __skb_linearize(skb))
				goto out_kfree_skb;

			/* If packet is not checksummed and device does not
			 * support checksumming for this protocol, complete
			 * checksumming here.
			 */
			if (skb->ip_summed == CHECKSUM_PARTIAL) {
2149 2150
				skb_set_transport_header(skb,
					skb_checksum_start_offset(skb));
2151
				if (!(features & NETIF_F_ALL_CSUM) &&
2152 2153 2154
				     skb_checksum_help(skb))
					goto out_kfree_skb;
			}
2155 2156
		}

2157
		rc = ops->ndo_start_xmit(skb, dev);
2158
		trace_net_dev_xmit(skb, rc);
2159
		if (rc == NETDEV_TX_OK)
E
Eric Dumazet 已提交
2160
			txq_trans_update(txq);
2161
		return rc;
2162 2163
	}

2164
gso:
2165 2166 2167 2168 2169
	do {
		struct sk_buff *nskb = skb->next;

		skb->next = nskb->next;
		nskb->next = NULL;
2170 2171 2172 2173 2174 2175 2176 2177

		/*
		 * 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);

2178
		rc = ops->ndo_start_xmit(nskb, dev);
2179
		trace_net_dev_xmit(nskb, rc);
2180
		if (unlikely(rc != NETDEV_TX_OK)) {
2181 2182
			if (rc & ~NETDEV_TX_MASK)
				goto out_kfree_gso_skb;
2183
			nskb->next = skb->next;
2184 2185 2186
			skb->next = nskb;
			return rc;
		}
E
Eric Dumazet 已提交
2187
		txq_trans_update(txq);
2188
		if (unlikely(netif_tx_queue_stopped(txq) && skb->next))
2189
			return NETDEV_TX_BUSY;
2190
	} while (skb->next);
2191

2192 2193 2194
out_kfree_gso_skb:
	if (likely(skb->next == NULL))
		skb->destructor = DEV_GSO_CB(skb)->destructor;
2195 2196
out_kfree_skb:
	kfree_skb(skb);
2197
out:
2198
	return rc;
2199 2200
}

T
Tom Herbert 已提交
2201
static u32 hashrnd __read_mostly;
D
David S. Miller 已提交
2202

2203 2204 2205 2206 2207 2208
/*
 * Returns a Tx hash based on the given packet descriptor a Tx queues' number
 * to be used as a distribution range.
 */
u16 __skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb,
		  unsigned int num_tx_queues)
2209
{
2210
	u32 hash;
2211 2212
	u16 qoffset = 0;
	u16 qcount = num_tx_queues;
D
David S. Miller 已提交
2213

2214 2215
	if (skb_rx_queue_recorded(skb)) {
		hash = skb_get_rx_queue(skb);
2216 2217
		while (unlikely(hash >= num_tx_queues))
			hash -= num_tx_queues;
2218 2219
		return hash;
	}
2220

2221 2222 2223 2224 2225 2226
	if (dev->num_tc) {
		u8 tc = netdev_get_prio_tc_map(dev, skb->priority);
		qoffset = dev->tc_to_txq[tc].offset;
		qcount = dev->tc_to_txq[tc].count;
	}

2227
	if (skb->sk && skb->sk->sk_hash)
2228
		hash = skb->sk->sk_hash;
2229
	else
2230
		hash = (__force u16) skb->protocol ^ skb->rxhash;
T
Tom Herbert 已提交
2231
	hash = jhash_1word(hash, hashrnd);
D
David S. Miller 已提交
2232

2233
	return (u16) (((u64) hash * qcount) >> 32) + qoffset;
2234
}
2235
EXPORT_SYMBOL(__skb_tx_hash);
2236

2237 2238 2239 2240
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
2241 2242 2243
			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);
2244 2245 2246 2247 2248 2249
		}
		return 0;
	}
	return queue_index;
}

T
Tom Herbert 已提交
2250 2251
static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
{
T
Tom Herbert 已提交
2252
#ifdef CONFIG_XPS
T
Tom Herbert 已提交
2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287
	struct xps_dev_maps *dev_maps;
	struct xps_map *map;
	int queue_index = -1;

	rcu_read_lock();
	dev_maps = rcu_dereference(dev->xps_maps);
	if (dev_maps) {
		map = rcu_dereference(
		    dev_maps->cpu_map[raw_smp_processor_id()]);
		if (map) {
			if (map->len == 1)
				queue_index = map->queues[0];
			else {
				u32 hash;
				if (skb->sk && skb->sk->sk_hash)
					hash = skb->sk->sk_hash;
				else
					hash = (__force u16) skb->protocol ^
					    skb->rxhash;
				hash = jhash_1word(hash, hashrnd);
				queue_index = map->queues[
				    ((u64)hash * map->len) >> 32];
			}
			if (unlikely(queue_index >= dev->real_num_tx_queues))
				queue_index = -1;
		}
	}
	rcu_read_unlock();

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

2288 2289 2290
static struct netdev_queue *dev_pick_tx(struct net_device *dev,
					struct sk_buff *skb)
{
2291
	int queue_index;
2292
	const struct net_device_ops *ops = dev->netdev_ops;
2293

2294 2295 2296
	if (dev->real_num_tx_queues == 1)
		queue_index = 0;
	else if (ops->ndo_select_queue) {
2297 2298 2299 2300 2301
		queue_index = ops->ndo_select_queue(dev, skb);
		queue_index = dev_cap_txqueue(dev, queue_index);
	} else {
		struct sock *sk = skb->sk;
		queue_index = sk_tx_queue_get(sk);
2302

2303 2304 2305
		if (queue_index < 0 || skb->ooo_okay ||
		    queue_index >= dev->real_num_tx_queues) {
			int old_index = queue_index;
2306

T
Tom Herbert 已提交
2307 2308 2309
			queue_index = get_xps_queue(dev, skb);
			if (queue_index < 0)
				queue_index = skb_tx_hash(dev, skb);
2310 2311 2312 2313

			if (queue_index != old_index && sk) {
				struct dst_entry *dst =
				    rcu_dereference_check(sk->sk_dst_cache, 1);
E
Eric Dumazet 已提交
2314 2315 2316 2317

				if (dst && skb_dst(skb) == dst)
					sk_tx_queue_set(sk, queue_index);
			}
2318 2319
		}
	}
2320

2321 2322
	skb_set_queue_mapping(skb, queue_index);
	return netdev_get_tx_queue(dev, queue_index);
2323 2324
}

2325 2326 2327 2328 2329
static inline int __dev_xmit_skb(struct sk_buff *skb, struct Qdisc *q,
				 struct net_device *dev,
				 struct netdev_queue *txq)
{
	spinlock_t *root_lock = qdisc_lock(q);
E
Eric Dumazet 已提交
2330
	bool contended;
2331 2332
	int rc;

E
Eric Dumazet 已提交
2333 2334
	qdisc_skb_cb(skb)->pkt_len = skb->len;
	qdisc_calculate_pkt_len(skb, q);
2335 2336 2337 2338 2339 2340
	/*
	 * Heuristic to force contended enqueues to serialize on a
	 * separate lock before trying to get qdisc main lock.
	 * This permits __QDISC_STATE_RUNNING owner to get the lock more often
	 * and dequeue packets faster.
	 */
E
Eric Dumazet 已提交
2341
	contended = qdisc_is_running(q);
2342 2343 2344
	if (unlikely(contended))
		spin_lock(&q->busylock);

2345 2346 2347 2348 2349
	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) &&
2350
		   qdisc_run_begin(q)) {
2351 2352 2353 2354 2355
		/*
		 * 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
2356 2357
		if (!(dev->priv_flags & IFF_XMIT_DST_RELEASE))
			skb_dst_force(skb);
2358 2359 2360

		qdisc_bstats_update(q, skb);

2361 2362 2363 2364 2365
		if (sch_direct_xmit(skb, q, dev, txq, root_lock)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
2366
			__qdisc_run(q);
2367
		} else
2368
			qdisc_run_end(q);
2369 2370 2371

		rc = NET_XMIT_SUCCESS;
	} else {
E
Eric Dumazet 已提交
2372
		skb_dst_force(skb);
E
Eric Dumazet 已提交
2373
		rc = q->enqueue(skb, q) & NET_XMIT_MASK;
2374 2375 2376 2377 2378 2379 2380
		if (qdisc_run_begin(q)) {
			if (unlikely(contended)) {
				spin_unlock(&q->busylock);
				contended = false;
			}
			__qdisc_run(q);
		}
2381 2382
	}
	spin_unlock(root_lock);
2383 2384
	if (unlikely(contended))
		spin_unlock(&q->busylock);
2385 2386 2387
	return rc;
}

2388
static DEFINE_PER_CPU(int, xmit_recursion);
2389
#define RECURSION_LIMIT 10
2390

2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415
/**
 *	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
2416 2417 2418
int dev_queue_xmit(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
2419
	struct netdev_queue *txq;
L
Linus Torvalds 已提交
2420 2421 2422
	struct Qdisc *q;
	int rc = -ENOMEM;

2423 2424
	/* Disable soft irqs for various locks below. Also
	 * stops preemption for RCU.
L
Linus Torvalds 已提交
2425
	 */
2426
	rcu_read_lock_bh();
L
Linus Torvalds 已提交
2427

2428
	txq = dev_pick_tx(dev, skb);
2429
	q = rcu_dereference_bh(txq->qdisc);
2430

L
Linus Torvalds 已提交
2431
#ifdef CONFIG_NET_CLS_ACT
E
Eric Dumazet 已提交
2432
	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_EGRESS);
L
Linus Torvalds 已提交
2433
#endif
2434
	trace_net_dev_queue(skb);
L
Linus Torvalds 已提交
2435
	if (q->enqueue) {
2436
		rc = __dev_xmit_skb(skb, q, dev, txq);
2437
		goto out;
L
Linus Torvalds 已提交
2438 2439 2440 2441 2442
	}

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

H
Herbert Xu 已提交
2443 2444
	   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
2445 2446 2447 2448 2449 2450 2451 2452 2453 2454
	   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 */

2455
		if (txq->xmit_lock_owner != cpu) {
L
Linus Torvalds 已提交
2456

2457 2458 2459
			if (__this_cpu_read(xmit_recursion) > RECURSION_LIMIT)
				goto recursion_alert;

2460
			HARD_TX_LOCK(dev, txq, cpu);
L
Linus Torvalds 已提交
2461

2462
			if (!netif_tx_queue_stopped(txq)) {
2463
				__this_cpu_inc(xmit_recursion);
2464
				rc = dev_hard_start_xmit(skb, dev, txq);
2465
				__this_cpu_dec(xmit_recursion);
2466
				if (dev_xmit_complete(rc)) {
2467
					HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2468 2469 2470
					goto out;
				}
			}
2471
			HARD_TX_UNLOCK(dev, txq);
L
Linus Torvalds 已提交
2472 2473 2474 2475 2476
			if (net_ratelimit())
				printk(KERN_CRIT "Virtual device %s asks to "
				       "queue packet!\n", dev->name);
		} else {
			/* Recursion is detected! It is possible,
2477 2478 2479
			 * unfortunately
			 */
recursion_alert:
L
Linus Torvalds 已提交
2480 2481 2482 2483 2484 2485 2486
			if (net_ratelimit())
				printk(KERN_CRIT "Dead loop on virtual device "
				       "%s, fix it urgently!\n", dev->name);
		}
	}

	rc = -ENETDOWN;
2487
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2488 2489 2490 2491

	kfree_skb(skb);
	return rc;
out:
2492
	rcu_read_unlock_bh();
L
Linus Torvalds 已提交
2493 2494
	return rc;
}
E
Eric Dumazet 已提交
2495
EXPORT_SYMBOL(dev_queue_xmit);
L
Linus Torvalds 已提交
2496 2497 2498 2499 2500 2501


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

2502
int netdev_max_backlog __read_mostly = 1000;
E
Eric Dumazet 已提交
2503
int netdev_tstamp_prequeue __read_mostly = 1;
2504 2505
int netdev_budget __read_mostly = 300;
int weight_p __read_mostly = 64;            /* old backlog weight */
L
Linus Torvalds 已提交
2506

E
Eric Dumazet 已提交
2507 2508 2509 2510 2511 2512 2513 2514
/* Called with irq disabled */
static inline void ____napi_schedule(struct softnet_data *sd,
				     struct napi_struct *napi)
{
	list_add_tail(&napi->poll_list, &sd->poll_list);
	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
}

T
Tom Herbert 已提交
2515
/*
2516 2517 2518
 * __skb_get_rxhash: calculate a flow hash based on src/dst addresses
 * and src/dst port numbers. Returns a non-zero hash number on success
 * and 0 on failure.
T
Tom Herbert 已提交
2519
 */
2520
__u32 __skb_get_rxhash(struct sk_buff *skb)
T
Tom Herbert 已提交
2521
{
2522
	int nhoff, hash = 0, poff;
T
Tom Herbert 已提交
2523 2524 2525
	struct ipv6hdr *ip6;
	struct iphdr *ip;
	u8 ip_proto;
C
Changli Gao 已提交
2526 2527 2528 2529 2530
	u32 addr1, addr2, ihl;
	union {
		u32 v32;
		u16 v16[2];
	} ports;
T
Tom Herbert 已提交
2531

2532
	nhoff = skb_network_offset(skb);
T
Tom Herbert 已提交
2533 2534 2535

	switch (skb->protocol) {
	case __constant_htons(ETH_P_IP):
2536
		if (!pskb_may_pull(skb, sizeof(*ip) + nhoff))
T
Tom Herbert 已提交
2537 2538
			goto done;

2539
		ip = (struct iphdr *) (skb->data + nhoff);
2540 2541 2542 2543
		if (ip->frag_off & htons(IP_MF | IP_OFFSET))
			ip_proto = 0;
		else
			ip_proto = ip->protocol;
E
Eric Dumazet 已提交
2544 2545
		addr1 = (__force u32) ip->saddr;
		addr2 = (__force u32) ip->daddr;
T
Tom Herbert 已提交
2546 2547 2548
		ihl = ip->ihl;
		break;
	case __constant_htons(ETH_P_IPV6):
2549
		if (!pskb_may_pull(skb, sizeof(*ip6) + nhoff))
T
Tom Herbert 已提交
2550 2551
			goto done;

2552
		ip6 = (struct ipv6hdr *) (skb->data + nhoff);
T
Tom Herbert 已提交
2553
		ip_proto = ip6->nexthdr;
E
Eric Dumazet 已提交
2554 2555
		addr1 = (__force u32) ip6->saddr.s6_addr32[3];
		addr2 = (__force u32) ip6->daddr.s6_addr32[3];
T
Tom Herbert 已提交
2556 2557 2558 2559 2560
		ihl = (40 >> 2);
		break;
	default:
		goto done;
	}
2561

2562 2563 2564 2565 2566 2567
	ports.v32 = 0;
	poff = proto_ports_offset(ip_proto);
	if (poff >= 0) {
		nhoff += ihl * 4 + poff;
		if (pskb_may_pull(skb, nhoff + 4)) {
			ports.v32 = * (__force u32 *) (skb->data + nhoff);
C
Changli Gao 已提交
2568 2569
			if (ports.v16[1] < ports.v16[0])
				swap(ports.v16[0], ports.v16[1]);
E
Eric Dumazet 已提交
2570
		}
T
Tom Herbert 已提交
2571 2572
	}

E
Eric Dumazet 已提交
2573 2574 2575
	/* get a consistent hash (same value on both flow directions) */
	if (addr2 < addr1)
		swap(addr1, addr2);
T
Tom Herbert 已提交
2576

2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588
	hash = jhash_3words(addr1, addr2, ports.v32, hashrnd);
	if (!hash)
		hash = 1;

done:
	return hash;
}
EXPORT_SYMBOL(__skb_get_rxhash);

#ifdef CONFIG_RPS

/* One global table that all flow-based protocols share. */
E
Eric Dumazet 已提交
2589
struct rps_sock_flow_table __rcu *rps_sock_flow_table __read_mostly;
2590 2591
EXPORT_SYMBOL(rps_sock_flow_table);

2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638
static struct rps_dev_flow *
set_rps_cpu(struct net_device *dev, struct sk_buff *skb,
	    struct rps_dev_flow *rflow, u16 next_cpu)
{
	u16 tcpu;

	tcpu = rflow->cpu = next_cpu;
	if (tcpu != RPS_NO_CPU) {
#ifdef CONFIG_RFS_ACCEL
		struct netdev_rx_queue *rxqueue;
		struct rps_dev_flow_table *flow_table;
		struct rps_dev_flow *old_rflow;
		u32 flow_id;
		u16 rxq_index;
		int rc;

		/* Should we steer this flow to a different hardware queue? */
		if (!skb_rx_queue_recorded(skb) || !dev->rx_cpu_rmap)
			goto out;
		rxq_index = cpu_rmap_lookup_index(dev->rx_cpu_rmap, next_cpu);
		if (rxq_index == skb_get_rx_queue(skb))
			goto out;

		rxqueue = dev->_rx + rxq_index;
		flow_table = rcu_dereference(rxqueue->rps_flow_table);
		if (!flow_table)
			goto out;
		flow_id = skb->rxhash & flow_table->mask;
		rc = dev->netdev_ops->ndo_rx_flow_steer(dev, skb,
							rxq_index, flow_id);
		if (rc < 0)
			goto out;
		old_rflow = rflow;
		rflow = &flow_table->flows[flow_id];
		rflow->cpu = next_cpu;
		rflow->filter = rc;
		if (old_rflow->filter == rflow->filter)
			old_rflow->filter = RPS_NO_FILTER;
	out:
#endif
		rflow->last_qtail =
			per_cpu(softnet_data, tcpu).input_queue_head;
	}

	return rflow;
}

2639 2640 2641 2642 2643 2644 2645 2646 2647
/*
 * get_rps_cpu is called from netif_receive_skb and returns the target
 * CPU from the RPS map of the receiving queue for a given skb.
 * rcu_read_lock must be held on entry.
 */
static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb,
		       struct rps_dev_flow **rflowp)
{
	struct netdev_rx_queue *rxqueue;
E
Eric Dumazet 已提交
2648
	struct rps_map *map;
2649 2650 2651 2652 2653 2654 2655
	struct rps_dev_flow_table *flow_table;
	struct rps_sock_flow_table *sock_flow_table;
	int cpu = -1;
	u16 tcpu;

	if (skb_rx_queue_recorded(skb)) {
		u16 index = skb_get_rx_queue(skb);
2656 2657 2658 2659 2660
		if (unlikely(index >= dev->real_num_rx_queues)) {
			WARN_ONCE(dev->real_num_rx_queues > 1,
				  "%s received packet on queue %u, but number "
				  "of RX queues is %u\n",
				  dev->name, index, dev->real_num_rx_queues);
2661 2662 2663 2664 2665 2666
			goto done;
		}
		rxqueue = dev->_rx + index;
	} else
		rxqueue = dev->_rx;

E
Eric Dumazet 已提交
2667 2668 2669
	map = rcu_dereference(rxqueue->rps_map);
	if (map) {
		if (map->len == 1) {
2670 2671 2672 2673 2674
			tcpu = map->cpus[0];
			if (cpu_online(tcpu))
				cpu = tcpu;
			goto done;
		}
E
Eric Dumazet 已提交
2675
	} else if (!rcu_dereference_raw(rxqueue->rps_flow_table)) {
2676
		goto done;
2677
	}
2678

2679
	skb_reset_network_header(skb);
2680 2681 2682
	if (!skb_get_rxhash(skb))
		goto done;

T
Tom Herbert 已提交
2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708
	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 -
2709 2710 2711
		      rflow->last_qtail)) >= 0))
			rflow = set_rps_cpu(dev, skb, rflow, next_cpu);

T
Tom Herbert 已提交
2712 2713 2714 2715 2716 2717 2718
		if (tcpu != RPS_NO_CPU && cpu_online(tcpu)) {
			*rflowp = rflow;
			cpu = tcpu;
			goto done;
		}
	}

T
Tom Herbert 已提交
2719
	if (map) {
T
Tom Herbert 已提交
2720
		tcpu = map->cpus[((u64) skb->rxhash * map->len) >> 32];
T
Tom Herbert 已提交
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731

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

done:
	return cpu;
}

2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771
#ifdef CONFIG_RFS_ACCEL

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

	rcu_read_lock();
	flow_table = rcu_dereference(rxqueue->rps_flow_table);
	if (flow_table && flow_id <= flow_table->mask) {
		rflow = &flow_table->flows[flow_id];
		cpu = ACCESS_ONCE(rflow->cpu);
		if (rflow->filter == filter_id && cpu != RPS_NO_CPU &&
		    ((int)(per_cpu(softnet_data, cpu).input_queue_head -
			   rflow->last_qtail) <
		     (int)(10 * flow_table->mask)))
			expire = false;
	}
	rcu_read_unlock();
	return expire;
}
EXPORT_SYMBOL(rps_may_expire_flow);

#endif /* CONFIG_RFS_ACCEL */

T
Tom Herbert 已提交
2772
/* Called from hardirq (IPI) context */
E
Eric Dumazet 已提交
2773
static void rps_trigger_softirq(void *data)
T
Tom Herbert 已提交
2774
{
E
Eric Dumazet 已提交
2775 2776
	struct softnet_data *sd = data;

E
Eric Dumazet 已提交
2777
	____napi_schedule(sd, &sd->backlog);
C
Changli Gao 已提交
2778
	sd->received_rps++;
T
Tom Herbert 已提交
2779
}
E
Eric Dumazet 已提交
2780

T
Tom Herbert 已提交
2781
#endif /* CONFIG_RPS */
T
Tom Herbert 已提交
2782

E
Eric Dumazet 已提交
2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803
/*
 * 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
2804 2805 2806 2807
/*
 * 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
2808 2809
static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
			      unsigned int *qtail)
T
Tom Herbert 已提交
2810
{
E
Eric Dumazet 已提交
2811
	struct softnet_data *sd;
T
Tom Herbert 已提交
2812 2813
	unsigned long flags;

E
Eric Dumazet 已提交
2814
	sd = &per_cpu(softnet_data, cpu);
T
Tom Herbert 已提交
2815 2816 2817

	local_irq_save(flags);

E
Eric Dumazet 已提交
2818
	rps_lock(sd);
2819 2820
	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
2821
enqueue:
E
Eric Dumazet 已提交
2822
			__skb_queue_tail(&sd->input_pkt_queue, skb);
2823
			input_queue_tail_incr_save(sd, qtail);
E
Eric Dumazet 已提交
2824
			rps_unlock(sd);
2825
			local_irq_restore(flags);
T
Tom Herbert 已提交
2826 2827 2828
			return NET_RX_SUCCESS;
		}

2829 2830 2831 2832
		/* 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
2833
			if (!rps_ipi_queued(sd))
E
Eric Dumazet 已提交
2834
				____napi_schedule(sd, &sd->backlog);
T
Tom Herbert 已提交
2835 2836 2837 2838
		}
		goto enqueue;
	}

C
Changli Gao 已提交
2839
	sd->dropped++;
E
Eric Dumazet 已提交
2840
	rps_unlock(sd);
T
Tom Herbert 已提交
2841 2842 2843

	local_irq_restore(flags);

2844
	atomic_long_inc(&skb->dev->rx_dropped);
T
Tom Herbert 已提交
2845 2846 2847
	kfree_skb(skb);
	return NET_RX_DROP;
}
L
Linus Torvalds 已提交
2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865

/**
 *	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)
{
2866
	int ret;
L
Linus Torvalds 已提交
2867 2868 2869 2870 2871

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

E
Eric Dumazet 已提交
2872 2873
	if (netdev_tstamp_prequeue)
		net_timestamp_check(skb);
L
Linus Torvalds 已提交
2874

2875
	trace_netif_rx(skb);
E
Eric Dumazet 已提交
2876
#ifdef CONFIG_RPS
2877
	{
T
Tom Herbert 已提交
2878
		struct rps_dev_flow voidflow, *rflow = &voidflow;
2879 2880
		int cpu;

2881
		preempt_disable();
2882
		rcu_read_lock();
T
Tom Herbert 已提交
2883 2884

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
2885 2886
		if (cpu < 0)
			cpu = smp_processor_id();
T
Tom Herbert 已提交
2887 2888 2889

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

2890
		rcu_read_unlock();
2891
		preempt_enable();
2892
	}
2893
#else
T
Tom Herbert 已提交
2894 2895 2896 2897 2898
	{
		unsigned int qtail;
		ret = enqueue_to_backlog(skb, get_cpu(), &qtail);
		put_cpu();
	}
2899
#endif
2900
	return ret;
L
Linus Torvalds 已提交
2901
}
E
Eric Dumazet 已提交
2902
EXPORT_SYMBOL(netif_rx);
L
Linus Torvalds 已提交
2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933

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;

2934
			WARN_ON(atomic_read(&skb->users));
2935
			trace_kfree_skb(skb, net_tx_action);
L
Linus Torvalds 已提交
2936 2937 2938 2939 2940
			__kfree_skb(skb);
		}
	}

	if (sd->output_queue) {
2941
		struct Qdisc *head;
L
Linus Torvalds 已提交
2942 2943 2944 2945

		local_irq_disable();
		head = sd->output_queue;
		sd->output_queue = NULL;
2946
		sd->output_queue_tailp = &sd->output_queue;
L
Linus Torvalds 已提交
2947 2948 2949
		local_irq_enable();

		while (head) {
2950 2951 2952
			struct Qdisc *q = head;
			spinlock_t *root_lock;

L
Linus Torvalds 已提交
2953 2954
			head = head->next_sched;

2955
			root_lock = qdisc_lock(q);
2956
			if (spin_trylock(root_lock)) {
2957 2958 2959
				smp_mb__before_clear_bit();
				clear_bit(__QDISC_STATE_SCHED,
					  &q->state);
2960 2961
				qdisc_run(q);
				spin_unlock(root_lock);
L
Linus Torvalds 已提交
2962
			} else {
2963
				if (!test_bit(__QDISC_STATE_DEACTIVATED,
2964
					      &q->state)) {
2965
					__netif_reschedule(q);
2966 2967 2968 2969 2970
				} else {
					smp_mb__before_clear_bit();
					clear_bit(__QDISC_STATE_SCHED,
						  &q->state);
				}
L
Linus Torvalds 已提交
2971 2972 2973 2974 2975
			}
		}
	}
}

2976 2977
#if (defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE)) && \
    (defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE))
2978 2979 2980
/* This hook is defined here for ATM LANE */
int (*br_fdb_test_addr_hook)(struct net_device *dev,
			     unsigned char *addr) __read_mostly;
2981
EXPORT_SYMBOL_GPL(br_fdb_test_addr_hook);
2982
#endif
L
Linus Torvalds 已提交
2983 2984 2985 2986 2987 2988

#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
2989
 * NOTE: This doesnt stop any functionality; if you dont have
L
Linus Torvalds 已提交
2990 2991 2992
 * the ingress scheduler, you just cant add policies on ingress.
 *
 */
2993
static int ing_filter(struct sk_buff *skb, struct netdev_queue *rxq)
L
Linus Torvalds 已提交
2994 2995
{
	struct net_device *dev = skb->dev;
2996
	u32 ttl = G_TC_RTTL(skb->tc_verd);
2997 2998
	int result = TC_ACT_OK;
	struct Qdisc *q;
2999

3000 3001 3002 3003
	if (unlikely(MAX_RED_LOOP < ttl++)) {
		if (net_ratelimit())
			pr_warning( "Redir loop detected Dropping packet (%d->%d)\n",
			       skb->skb_iif, dev->ifindex);
3004 3005
		return TC_ACT_SHOT;
	}
L
Linus Torvalds 已提交
3006

3007 3008
	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
3009

3010
	q = rxq->qdisc;
3011
	if (q != &noop_qdisc) {
3012
		spin_lock(qdisc_lock(q));
3013 3014
		if (likely(!test_bit(__QDISC_STATE_DEACTIVATED, &q->state)))
			result = qdisc_enqueue_root(skb, q);
3015 3016
		spin_unlock(qdisc_lock(q));
	}
3017 3018 3019

	return result;
}
3020

3021 3022 3023 3024
static inline struct sk_buff *handle_ing(struct sk_buff *skb,
					 struct packet_type **pt_prev,
					 int *ret, struct net_device *orig_dev)
{
3025 3026 3027
	struct netdev_queue *rxq = rcu_dereference(skb->dev->ingress_queue);

	if (!rxq || rxq->qdisc == &noop_qdisc)
3028
		goto out;
L
Linus Torvalds 已提交
3029

3030 3031 3032
	if (*pt_prev) {
		*ret = deliver_skb(skb, *pt_prev, orig_dev);
		*pt_prev = NULL;
L
Linus Torvalds 已提交
3033 3034
	}

3035
	switch (ing_filter(skb, rxq)) {
3036 3037 3038 3039 3040 3041 3042 3043 3044
	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
3045 3046 3047
}
#endif

3048 3049 3050 3051
/**
 *	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
3052
 *	@rx_handler_data: data pointer that is used by rx handler
3053 3054 3055 3056 3057 3058 3059 3060
 *
 *	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
3061 3062
			       rx_handler_func_t *rx_handler,
			       void *rx_handler_data)
3063 3064 3065 3066 3067 3068
{
	ASSERT_RTNL();

	if (dev->rx_handler)
		return -EBUSY;

J
Jiri Pirko 已提交
3069
	rcu_assign_pointer(dev->rx_handler_data, rx_handler_data);
3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088
	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
3089
	rcu_assign_pointer(dev->rx_handler_data, NULL);
3090 3091 3092
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);

3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113
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;

3114 3115
	if ((master->priv_flags & IFF_MASTER_ALB) &&
	    (master->priv_flags & IFF_BRIDGE_PORT)) {
3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
		/* 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);

3143
static int __netif_receive_skb(struct sk_buff *skb)
L
Linus Torvalds 已提交
3144 3145
{
	struct packet_type *ptype, *pt_prev;
3146
	rx_handler_func_t *rx_handler;
D
David S. Miller 已提交
3147
	struct net_device *orig_dev;
3148
	struct net_device *master;
3149
	struct net_device *null_or_orig;
3150
	struct net_device *orig_or_bond;
L
Linus Torvalds 已提交
3151
	int ret = NET_RX_DROP;
A
Al Viro 已提交
3152
	__be16 type;
L
Linus Torvalds 已提交
3153

E
Eric Dumazet 已提交
3154 3155
	if (!netdev_tstamp_prequeue)
		net_timestamp_check(skb);
3156

3157
	trace_netif_receive_skb(skb);
3158

L
Linus Torvalds 已提交
3159
	/* if we've gotten here through NAPI, check netpoll */
3160
	if (netpoll_receive_skb(skb))
L
Linus Torvalds 已提交
3161 3162
		return NET_RX_DROP;

3163 3164
	if (!skb->skb_iif)
		skb->skb_iif = skb->dev->ifindex;
3165

3166 3167 3168 3169 3170
	/*
	 * 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
3171
	 * be dropped at the handler.
3172
	 */
3173
	null_or_orig = NULL;
J
Joe Eykholt 已提交
3174
	orig_dev = skb->dev;
3175
	master = ACCESS_ONCE(orig_dev->master);
3176 3177 3178 3179 3180
	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;
3181
			null_or_orig = orig_dev; /* deliver only exact match */
3182
		} else
3183
			skb->dev = master;
J
Joe Eykholt 已提交
3184
	}
3185

3186
	__this_cpu_inc(softnet_data.processed);
3187
	skb_reset_network_header(skb);
3188
	skb_reset_transport_header(skb);
3189
	skb->mac_len = skb->network_header - skb->mac_header;
L
Linus Torvalds 已提交
3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202

	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) {
3203 3204
		if (ptype->dev == null_or_orig || ptype->dev == skb->dev ||
		    ptype->dev == orig_dev) {
3205
			if (pt_prev)
D
David S. Miller 已提交
3206
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3207 3208 3209 3210 3211
			pt_prev = ptype;
		}
	}

#ifdef CONFIG_NET_CLS_ACT
3212 3213
	skb = handle_ing(skb, &pt_prev, &ret, orig_dev);
	if (!skb)
L
Linus Torvalds 已提交
3214 3215 3216 3217
		goto out;
ncls:
#endif

3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228
	/* 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
3229

3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241
	if (vlan_tx_tag_present(skb)) {
		if (pt_prev) {
			ret = deliver_skb(skb, pt_prev, orig_dev);
			pt_prev = NULL;
		}
		if (vlan_hwaccel_do_receive(&skb)) {
			ret = __netif_receive_skb(skb);
			goto out;
		} else if (unlikely(!skb))
			goto out;
	}

3242 3243 3244 3245 3246 3247
	/*
	 * 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.
	 */
3248
	orig_or_bond = orig_dev;
3249 3250
	if ((skb->dev->priv_flags & IFF_802_1Q_VLAN) &&
	    (vlan_dev_real_dev(skb->dev)->priv_flags & IFF_BONDING)) {
3251
		orig_or_bond = vlan_dev_real_dev(skb->dev);
3252 3253
	}

L
Linus Torvalds 已提交
3254
	type = skb->protocol;
3255 3256
	list_for_each_entry_rcu(ptype,
			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {
3257
		if (ptype->type == type && (ptype->dev == null_or_orig ||
3258
		     ptype->dev == skb->dev || ptype->dev == orig_dev ||
3259
		     ptype->dev == orig_or_bond)) {
3260
			if (pt_prev)
D
David S. Miller 已提交
3261
				ret = deliver_skb(skb, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3262 3263 3264 3265 3266
			pt_prev = ptype;
		}
	}

	if (pt_prev) {
D
David S. Miller 已提交
3267
		ret = pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
L
Linus Torvalds 已提交
3268
	} else {
3269
		atomic_long_inc(&skb->dev->rx_dropped);
L
Linus Torvalds 已提交
3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280
		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
3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298

/**
 *	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
3299 3300 3301
	if (netdev_tstamp_prequeue)
		net_timestamp_check(skb);

3302 3303 3304
	if (skb_defer_rx_timestamp(skb))
		return NET_RX_SUCCESS;

E
Eric Dumazet 已提交
3305
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
3306 3307 3308
	{
		struct rps_dev_flow voidflow, *rflow = &voidflow;
		int cpu, ret;
T
Tom Herbert 已提交
3309

E
Eric Dumazet 已提交
3310 3311 3312
		rcu_read_lock();

		cpu = get_rps_cpu(skb->dev, skb, &rflow);
T
Tom Herbert 已提交
3313

E
Eric Dumazet 已提交
3314 3315 3316 3317 3318 3319 3320
		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
3321

E
Eric Dumazet 已提交
3322
		return ret;
T
Tom Herbert 已提交
3323
	}
3324 3325 3326
#else
	return __netif_receive_skb(skb);
#endif
T
Tom Herbert 已提交
3327
}
E
Eric Dumazet 已提交
3328
EXPORT_SYMBOL(netif_receive_skb);
L
Linus Torvalds 已提交
3329

E
Eric Dumazet 已提交
3330 3331 3332
/* Network device is going away, flush any packets still pending
 * Called with irqs disabled.
 */
3333
static void flush_backlog(void *arg)
3334
{
3335
	struct net_device *dev = arg;
E
Eric Dumazet 已提交
3336
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3337 3338
	struct sk_buff *skb, *tmp;

E
Eric Dumazet 已提交
3339
	rps_lock(sd);
3340
	skb_queue_walk_safe(&sd->input_pkt_queue, skb, tmp) {
3341
		if (skb->dev == dev) {
E
Eric Dumazet 已提交
3342
			__skb_unlink(skb, &sd->input_pkt_queue);
3343
			kfree_skb(skb);
3344
			input_queue_head_incr(sd);
3345
		}
3346
	}
E
Eric Dumazet 已提交
3347
	rps_unlock(sd);
3348 3349 3350 3351 3352

	skb_queue_walk_safe(&sd->process_queue, skb, tmp) {
		if (skb->dev == dev) {
			__skb_unlink(skb, &sd->process_queue);
			kfree_skb(skb);
3353
			input_queue_head_incr(sd);
3354 3355
		}
	}
3356 3357
}

3358 3359 3360 3361 3362 3363 3364
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;

3365 3366
	if (NAPI_GRO_CB(skb)->count == 1) {
		skb_shinfo(skb)->gso_size = 0;
3367
		goto out;
3368
	}
3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389

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

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

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

out:
	return netif_receive_skb(skb);
}

E
Eric Dumazet 已提交
3390
inline void napi_gro_flush(struct napi_struct *napi)
3391 3392 3393 3394 3395 3396 3397 3398 3399
{
	struct sk_buff *skb, *next;

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

3400
	napi->gro_count = 0;
3401 3402
	napi->gro_list = NULL;
}
E
Eric Dumazet 已提交
3403
EXPORT_SYMBOL(napi_gro_flush);
3404

3405
enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3406 3407 3408 3409 3410
{
	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
3411
	int same_flow;
3412
	int mac_len;
3413
	enum gro_result ret;
3414

3415
	if (!(skb->dev->features & NETIF_F_GRO) || netpoll_rx_on(skb))
3416 3417
		goto normal;

3418
	if (skb_is_gso(skb) || skb_has_frag_list(skb))
3419 3420
		goto normal;

3421 3422 3423 3424 3425
	rcu_read_lock();
	list_for_each_entry_rcu(ptype, head, list) {
		if (ptype->type != type || ptype->dev || !ptype->gro_receive)
			continue;

3426
		skb_set_network_header(skb, skb_gro_offset(skb));
3427 3428 3429 3430
		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
3431
		NAPI_GRO_CB(skb)->free = 0;
3432 3433 3434 3435 3436 3437 3438 3439 3440

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

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

H
Herbert Xu 已提交
3441
	same_flow = NAPI_GRO_CB(skb)->same_flow;
3442
	ret = NAPI_GRO_CB(skb)->free ? GRO_MERGED_FREE : GRO_MERGED;
H
Herbert Xu 已提交
3443

3444 3445 3446 3447 3448 3449
	if (pp) {
		struct sk_buff *nskb = *pp;

		*pp = nskb->next;
		nskb->next = NULL;
		napi_gro_complete(nskb);
3450
		napi->gro_count--;
3451 3452
	}

H
Herbert Xu 已提交
3453
	if (same_flow)
3454 3455
		goto ok;

3456
	if (NAPI_GRO_CB(skb)->flush || napi->gro_count >= MAX_GRO_SKBS)
3457 3458
		goto normal;

3459
	napi->gro_count++;
3460
	NAPI_GRO_CB(skb)->count = 1;
3461
	skb_shinfo(skb)->gso_size = skb_gro_len(skb);
3462 3463
	skb->next = napi->gro_list;
	napi->gro_list = skb;
3464
	ret = GRO_HELD;
3465

3466
pull:
H
Herbert Xu 已提交
3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483
	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,
3484
				--skb_shinfo(skb)->nr_frags * sizeof(skb_frag_t));
H
Herbert Xu 已提交
3485
		}
3486 3487
	}

3488
ok:
3489
	return ret;
3490 3491

normal:
3492 3493
	ret = GRO_NORMAL;
	goto pull;
H
Herbert Xu 已提交
3494
}
3495 3496
EXPORT_SYMBOL(dev_gro_receive);

3497
static inline gro_result_t
3498
__napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3499 3500 3501 3502
{
	struct sk_buff *p;

	for (p = napi->gro_list; p; p = p->next) {
3503 3504 3505
		unsigned long diffs;

		diffs = (unsigned long)p->dev ^ (unsigned long)skb->dev;
3506
		diffs |= p->vlan_tci ^ skb->vlan_tci;
3507
		diffs |= compare_ether_header(skb_mac_header(p),
3508
					      skb_gro_mac_header(skb));
3509
		NAPI_GRO_CB(p)->same_flow = !diffs;
3510 3511 3512 3513 3514
		NAPI_GRO_CB(p)->flush = 0;
	}

	return dev_gro_receive(napi, skb);
}
H
Herbert Xu 已提交
3515

3516
gro_result_t napi_skb_finish(gro_result_t ret, struct sk_buff *skb)
H
Herbert Xu 已提交
3517
{
3518 3519
	switch (ret) {
	case GRO_NORMAL:
3520 3521 3522
		if (netif_receive_skb(skb))
			ret = GRO_DROP;
		break;
H
Herbert Xu 已提交
3523

3524 3525
	case GRO_DROP:
	case GRO_MERGED_FREE:
H
Herbert Xu 已提交
3526 3527
		kfree_skb(skb);
		break;
3528 3529 3530 3531

	case GRO_HELD:
	case GRO_MERGED:
		break;
H
Herbert Xu 已提交
3532 3533
	}

3534
	return ret;
3535 3536 3537
}
EXPORT_SYMBOL(napi_skb_finish);

3538 3539 3540 3541
void skb_gro_reset_offset(struct sk_buff *skb)
{
	NAPI_GRO_CB(skb)->data_offset = 0;
	NAPI_GRO_CB(skb)->frag0 = NULL;
3542
	NAPI_GRO_CB(skb)->frag0_len = 0;
3543

3544
	if (skb->mac_header == skb->tail &&
3545
	    !PageHighMem(skb_shinfo(skb)->frags[0].page)) {
3546 3547 3548
		NAPI_GRO_CB(skb)->frag0 =
			page_address(skb_shinfo(skb)->frags[0].page) +
			skb_shinfo(skb)->frags[0].page_offset;
3549 3550
		NAPI_GRO_CB(skb)->frag0_len = skb_shinfo(skb)->frags[0].size;
	}
3551 3552 3553
}
EXPORT_SYMBOL(skb_gro_reset_offset);

3554
gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
3555
{
3556 3557
	skb_gro_reset_offset(skb);

3558
	return napi_skb_finish(__napi_gro_receive(napi, skb), skb);
3559 3560 3561
}
EXPORT_SYMBOL(napi_gro_receive);

S
stephen hemminger 已提交
3562
static void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
3563 3564 3565
{
	__skb_pull(skb, skb_headlen(skb));
	skb_reserve(skb, NET_IP_ALIGN - skb_headroom(skb));
3566
	skb->vlan_tci = 0;
3567 3568 3569 3570

	napi->skb = skb;
}

3571
struct sk_buff *napi_get_frags(struct napi_struct *napi)
H
Herbert Xu 已提交
3572 3573 3574 3575
{
	struct sk_buff *skb = napi->skb;

	if (!skb) {
3576 3577 3578
		skb = netdev_alloc_skb_ip_align(napi->dev, GRO_MAX_HEAD);
		if (skb)
			napi->skb = skb;
3579
	}
3580 3581
	return skb;
}
3582
EXPORT_SYMBOL(napi_get_frags);
3583

3584 3585
gro_result_t napi_frags_finish(struct napi_struct *napi, struct sk_buff *skb,
			       gro_result_t ret)
3586
{
3587 3588
	switch (ret) {
	case GRO_NORMAL:
3589
	case GRO_HELD:
3590
		skb->protocol = eth_type_trans(skb, skb->dev);
3591

3592 3593 3594 3595
		if (ret == GRO_HELD)
			skb_gro_pull(skb, -ETH_HLEN);
		else if (netif_receive_skb(skb))
			ret = GRO_DROP;
3596
		break;
H
Herbert Xu 已提交
3597

3598 3599 3600 3601
	case GRO_DROP:
	case GRO_MERGED_FREE:
		napi_reuse_skb(napi, skb);
		break;
3602 3603 3604

	case GRO_MERGED:
		break;
3605
	}
H
Herbert Xu 已提交
3606

3607
	return ret;
H
Herbert Xu 已提交
3608
}
3609 3610
EXPORT_SYMBOL(napi_frags_finish);

3611 3612 3613 3614
struct sk_buff *napi_frags_skb(struct napi_struct *napi)
{
	struct sk_buff *skb = napi->skb;
	struct ethhdr *eth;
3615 3616
	unsigned int hlen;
	unsigned int off;
3617 3618 3619 3620 3621 3622

	napi->skb = NULL;

	skb_reset_mac_header(skb);
	skb_gro_reset_offset(skb);

3623 3624 3625 3626 3627 3628 3629 3630 3631 3632
	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;
		}
3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647
	}

	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);

3648
gro_result_t napi_gro_frags(struct napi_struct *napi)
3649
{
3650
	struct sk_buff *skb = napi_frags_skb(napi);
3651 3652

	if (!skb)
3653
		return GRO_DROP;
3654 3655 3656

	return napi_frags_finish(napi, skb, __napi_gro_receive(napi, skb));
}
H
Herbert Xu 已提交
3657 3658
EXPORT_SYMBOL(napi_gro_frags);

3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686
/*
 * 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();
}

3687
static int process_backlog(struct napi_struct *napi, int quota)
L
Linus Torvalds 已提交
3688 3689
{
	int work = 0;
E
Eric Dumazet 已提交
3690
	struct softnet_data *sd = container_of(napi, struct softnet_data, backlog);
L
Linus Torvalds 已提交
3691

3692 3693 3694 3695 3696 3697 3698 3699 3700
#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
3701
	napi->weight = weight_p;
3702 3703
	local_irq_disable();
	while (work < quota) {
L
Linus Torvalds 已提交
3704
		struct sk_buff *skb;
3705 3706 3707 3708 3709 3710
		unsigned int qlen;

		while ((skb = __skb_dequeue(&sd->process_queue))) {
			local_irq_enable();
			__netif_receive_skb(skb);
			local_irq_disable();
3711 3712 3713 3714 3715
			input_queue_head_incr(sd);
			if (++work >= quota) {
				local_irq_enable();
				return work;
			}
3716
		}
L
Linus Torvalds 已提交
3717

E
Eric Dumazet 已提交
3718
		rps_lock(sd);
3719
		qlen = skb_queue_len(&sd->input_pkt_queue);
3720
		if (qlen)
3721 3722
			skb_queue_splice_tail_init(&sd->input_pkt_queue,
						   &sd->process_queue);
3723

3724
		if (qlen < quota - work) {
E
Eric Dumazet 已提交
3725 3726 3727 3728 3729 3730 3731 3732 3733 3734
			/*
			 * 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;

3735
			quota = work + qlen;
3736
		}
E
Eric Dumazet 已提交
3737
		rps_unlock(sd);
3738 3739
	}
	local_irq_enable();
L
Linus Torvalds 已提交
3740

3741 3742
	return work;
}
L
Linus Torvalds 已提交
3743

3744 3745
/**
 * __napi_schedule - schedule for receive
3746
 * @n: entry to schedule
3747 3748 3749
 *
 * The entry's receive function will be scheduled to run
 */
H
Harvey Harrison 已提交
3750
void __napi_schedule(struct napi_struct *n)
3751 3752
{
	unsigned long flags;
L
Linus Torvalds 已提交
3753

3754
	local_irq_save(flags);
E
Eric Dumazet 已提交
3755
	____napi_schedule(&__get_cpu_var(softnet_data), n);
3756
	local_irq_restore(flags);
L
Linus Torvalds 已提交
3757
}
3758 3759
EXPORT_SYMBOL(__napi_schedule);

3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792
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);
3793
	napi->gro_count = 0;
3794
	napi->gro_list = NULL;
H
Herbert Xu 已提交
3795
	napi->skb = NULL;
3796 3797 3798 3799
	napi->poll = poll;
	napi->weight = weight;
	list_add(&napi->dev_list, &dev->napi_list);
	napi->dev = dev;
H
Herbert Xu 已提交
3800
#ifdef CONFIG_NETPOLL
3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811
	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;

3812
	list_del_init(&napi->dev_list);
3813
	napi_free_frags(napi);
3814 3815 3816 3817 3818 3819 3820 3821

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

	napi->gro_list = NULL;
3822
	napi->gro_count = 0;
3823 3824 3825
}
EXPORT_SYMBOL(netif_napi_del);

L
Linus Torvalds 已提交
3826 3827
static void net_rx_action(struct softirq_action *h)
{
3828
	struct softnet_data *sd = &__get_cpu_var(softnet_data);
3829
	unsigned long time_limit = jiffies + 2;
3830
	int budget = netdev_budget;
3831 3832
	void *have;

L
Linus Torvalds 已提交
3833 3834
	local_irq_disable();

3835
	while (!list_empty(&sd->poll_list)) {
3836 3837
		struct napi_struct *n;
		int work, weight;
L
Linus Torvalds 已提交
3838

3839
		/* If softirq window is exhuasted then punt.
3840 3841
		 * Allow this to run for 2 jiffies since which will allow
		 * an average latency of 1.5/HZ.
3842
		 */
3843
		if (unlikely(budget <= 0 || time_after(jiffies, time_limit)))
L
Linus Torvalds 已提交
3844 3845 3846 3847
			goto softnet_break;

		local_irq_enable();

3848 3849 3850 3851 3852
		/* 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.
		 */
3853
		n = list_first_entry(&sd->poll_list, struct napi_struct, poll_list);
L
Linus Torvalds 已提交
3854

3855 3856 3857 3858
		have = netpoll_poll_lock(n);

		weight = n->weight;

3859 3860 3861 3862 3863 3864 3865
		/* 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;
3866
		if (test_bit(NAPI_STATE_SCHED, &n->state)) {
3867
			work = n->poll(n, weight);
3868 3869
			trace_napi_poll(n);
		}
3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881

		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.
		 */
3882
		if (unlikely(work == weight)) {
3883 3884 3885 3886 3887
			if (unlikely(napi_disable_pending(n))) {
				local_irq_enable();
				napi_complete(n);
				local_irq_disable();
			} else
3888
				list_move_tail(&n->poll_list, &sd->poll_list);
3889
		}
3890 3891

		netpoll_poll_unlock(have);
L
Linus Torvalds 已提交
3892 3893
	}
out:
3894
	net_rps_action_and_irq_enable(sd);
T
Tom Herbert 已提交
3895

3896 3897 3898 3899 3900
#ifdef CONFIG_NET_DMA
	/*
	 * There may not be any more sk_buffs coming right now, so push
	 * any pending DMA copies to hardware
	 */
3901
	dma_issue_pending_all();
3902
#endif
3903

L
Linus Torvalds 已提交
3904 3905 3906
	return;

softnet_break:
C
Changli Gao 已提交
3907
	sd->time_squeeze++;
L
Linus Torvalds 已提交
3908 3909 3910 3911
	__raise_softirq_irqoff(NET_RX_SOFTIRQ);
	goto out;
}

E
Eric Dumazet 已提交
3912
static gifconf_func_t *gifconf_list[NPROTO];
L
Linus Torvalds 已提交
3913 3914 3915 3916 3917 3918 3919 3920 3921 3922

/**
 *	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
3923
int register_gifconf(unsigned int family, gifconf_func_t *gifconf)
L
Linus Torvalds 已提交
3924 3925 3926 3927 3928 3929
{
	if (family >= NPROTO)
		return -EINVAL;
	gifconf_list[family] = gifconf;
	return 0;
}
E
Eric Dumazet 已提交
3930
EXPORT_SYMBOL(register_gifconf);
L
Linus Torvalds 已提交
3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943


/*
 *	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
 */

3944
static int dev_ifname(struct net *net, struct ifreq __user *arg)
L
Linus Torvalds 已提交
3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955
{
	struct net_device *dev;
	struct ifreq ifr;

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

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

3956 3957
	rcu_read_lock();
	dev = dev_get_by_index_rcu(net, ifr.ifr_ifindex);
L
Linus Torvalds 已提交
3958
	if (!dev) {
3959
		rcu_read_unlock();
L
Linus Torvalds 已提交
3960 3961 3962 3963
		return -ENODEV;
	}

	strcpy(ifr.ifr_name, dev->name);
3964
	rcu_read_unlock();
L
Linus Torvalds 已提交
3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976

	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'.
 */

3977
static int dev_ifconf(struct net *net, char __user *arg)
L
Linus Torvalds 已提交
3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000
{
	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;
4001
	for_each_netdev(net, dev) {
L
Linus Torvalds 已提交
4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014
		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;
			}
		}
4015
	}
L
Linus Torvalds 已提交
4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032

	/*
	 *	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.
 */
4033
void *dev_seq_start(struct seq_file *seq, loff_t *pos)
4034
	__acquires(RCU)
L
Linus Torvalds 已提交
4035
{
4036
	struct net *net = seq_file_net(seq);
4037
	loff_t off;
L
Linus Torvalds 已提交
4038 4039
	struct net_device *dev;

4040
	rcu_read_lock();
4041 4042
	if (!*pos)
		return SEQ_START_TOKEN;
L
Linus Torvalds 已提交
4043

4044
	off = 1;
4045
	for_each_netdev_rcu(net, dev)
4046 4047
		if (off++ == *pos)
			return dev;
L
Linus Torvalds 已提交
4048

4049
	return NULL;
L
Linus Torvalds 已提交
4050 4051 4052 4053
}

void *dev_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
4054 4055 4056 4057
	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
4058
	++*pos;
4059
	return rcu_dereference(dev);
L
Linus Torvalds 已提交
4060 4061 4062
}

void dev_seq_stop(struct seq_file *seq, void *v)
4063
	__releases(RCU)
L
Linus Torvalds 已提交
4064
{
4065
	rcu_read_unlock();
L
Linus Torvalds 已提交
4066 4067 4068 4069
}

static void dev_seq_printf_stats(struct seq_file *seq, struct net_device *dev)
{
4070 4071
	struct rtnl_link_stats64 temp;
	const struct rtnl_link_stats64 *stats = dev_get_stats(dev, &temp);
L
Linus Torvalds 已提交
4072

4073 4074
	seq_printf(seq, "%6s: %7llu %7llu %4llu %4llu %4llu %5llu %10llu %9llu "
		   "%8llu %7llu %4llu %4llu %4llu %5llu %7llu %10llu\n",
4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089
		   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
4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108
}

/*
 *	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
4109
static struct softnet_data *softnet_get_online(loff_t *pos)
L
Linus Torvalds 已提交
4110
{
C
Changli Gao 已提交
4111
	struct softnet_data *sd = NULL;
L
Linus Torvalds 已提交
4112

4113
	while (*pos < nr_cpu_ids)
4114
		if (cpu_online(*pos)) {
C
Changli Gao 已提交
4115
			sd = &per_cpu(softnet_data, *pos);
L
Linus Torvalds 已提交
4116 4117 4118
			break;
		} else
			++*pos;
C
Changli Gao 已提交
4119
	return sd;
L
Linus Torvalds 已提交
4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138
}

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
4139
	struct softnet_data *sd = v;
L
Linus Torvalds 已提交
4140

T
Tom Herbert 已提交
4141
	seq_printf(seq, "%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
C
Changli Gao 已提交
4142
		   sd->processed, sd->dropped, sd->time_squeeze, 0,
4143
		   0, 0, 0, 0, /* was fastroute */
C
Changli Gao 已提交
4144
		   sd->cpu_collision, sd->received_rps);
L
Linus Torvalds 已提交
4145 4146 4147
	return 0;
}

4148
static const struct seq_operations dev_seq_ops = {
L
Linus Torvalds 已提交
4149 4150 4151 4152 4153 4154 4155 4156
	.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)
{
4157 4158
	return seq_open_net(inode, file, &dev_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
4159 4160
}

4161
static const struct file_operations dev_seq_fops = {
L
Linus Torvalds 已提交
4162 4163 4164 4165
	.owner	 = THIS_MODULE,
	.open    = dev_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
4166
	.release = seq_release_net,
L
Linus Torvalds 已提交
4167 4168
};

4169
static const struct seq_operations softnet_seq_ops = {
L
Linus Torvalds 已提交
4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180
	.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);
}

4181
static const struct file_operations softnet_seq_fops = {
L
Linus Torvalds 已提交
4182 4183 4184 4185 4186 4187 4188
	.owner	 = THIS_MODULE,
	.open    = softnet_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release,
};

4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200
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;
	}

4201
	for (t = 0; t < PTYPE_HASH_SIZE; t++) {
4202 4203 4204 4205 4206 4207 4208 4209 4210 4211
		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)
4212
	__acquires(RCU)
4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235
{
	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
4236
		hash = ntohs(pt->type) & PTYPE_HASH_MASK;
4237 4238

	while (nxt == &ptype_base[hash]) {
4239
		if (++hash >= PTYPE_HASH_SIZE)
4240 4241 4242 4243 4244 4245 4246 4247
			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)
4248
	__releases(RCU)
4249 4250 4251 4252 4253 4254 4255 4256 4257 4258
{
	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");
4259
	else if (pt->dev == NULL || dev_net(pt->dev) == seq_file_net(seq)) {
4260 4261 4262 4263 4264
		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
4265 4266
		seq_printf(seq, " %-8s %pF\n",
			   pt->dev ? pt->dev->name : "", pt->func);
4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280
	}

	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)
{
4281 4282
	return seq_open_net(inode, file, &ptype_seq_ops,
			sizeof(struct seq_net_private));
4283 4284 4285 4286 4287 4288 4289
}

static const struct file_operations ptype_seq_fops = {
	.owner	 = THIS_MODULE,
	.open    = ptype_seq_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
4290
	.release = seq_release_net,
4291 4292 4293
};


4294
static int __net_init dev_proc_net_init(struct net *net)
L
Linus Torvalds 已提交
4295 4296 4297
{
	int rc = -ENOMEM;

4298
	if (!proc_net_fops_create(net, "dev", S_IRUGO, &dev_seq_fops))
L
Linus Torvalds 已提交
4299
		goto out;
4300
	if (!proc_net_fops_create(net, "softnet_stat", S_IRUGO, &softnet_seq_fops))
L
Linus Torvalds 已提交
4301
		goto out_dev;
4302
	if (!proc_net_fops_create(net, "ptype", S_IRUGO, &ptype_seq_fops))
4303
		goto out_softnet;
4304

4305
	if (wext_proc_init(net))
4306
		goto out_ptype;
L
Linus Torvalds 已提交
4307 4308 4309
	rc = 0;
out:
	return rc;
4310
out_ptype:
4311
	proc_net_remove(net, "ptype");
L
Linus Torvalds 已提交
4312
out_softnet:
4313
	proc_net_remove(net, "softnet_stat");
L
Linus Torvalds 已提交
4314
out_dev:
4315
	proc_net_remove(net, "dev");
L
Linus Torvalds 已提交
4316 4317
	goto out;
}
4318

4319
static void __net_exit dev_proc_net_exit(struct net *net)
4320 4321 4322 4323 4324 4325 4326 4327
{
	wext_proc_exit(net);

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

4328
static struct pernet_operations __net_initdata dev_proc_ops = {
4329 4330 4331 4332 4333 4334 4335 4336
	.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
4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365
#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;
4366

E
Eric Dumazet 已提交
4367 4368
	if (old) {
		synchronize_net();
L
Linus Torvalds 已提交
4369
		dev_put(old);
E
Eric Dumazet 已提交
4370
	}
L
Linus Torvalds 已提交
4371 4372 4373 4374 4375 4376 4377 4378
	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
4379
EXPORT_SYMBOL(netdev_set_master);
L
Linus Torvalds 已提交
4380

4381 4382
static void dev_change_rx_flags(struct net_device *dev, int flags)
{
4383 4384 4385 4386
	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);
4387 4388
}

4389
static int __dev_set_promiscuity(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4390 4391
{
	unsigned short old_flags = dev->flags;
4392 4393
	uid_t uid;
	gid_t gid;
L
Linus Torvalds 已提交
4394

4395 4396
	ASSERT_RTNL();

4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413
	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;
		}
	}
4414
	if (dev->flags != old_flags) {
L
Linus Torvalds 已提交
4415 4416
		printk(KERN_INFO "device %s %s promiscuous mode\n",
		       dev->name, (dev->flags & IFF_PROMISC) ? "entered" :
4417
							       "left");
4418 4419
		if (audit_enabled) {
			current_uid_gid(&uid, &gid);
4420 4421 4422 4423 4424 4425
			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),
4426
				uid, gid,
4427
				audit_get_sessionid(current));
4428
		}
4429

4430
		dev_change_rx_flags(dev, IFF_PROMISC);
L
Linus Torvalds 已提交
4431
	}
4432
	return 0;
L
Linus Torvalds 已提交
4433 4434
}

4435 4436 4437 4438 4439 4440 4441 4442 4443
/**
 *	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.
4444
 *	Return 0 if successful or a negative errno code on error.
4445
 */
4446
int dev_set_promiscuity(struct net_device *dev, int inc)
4447 4448
{
	unsigned short old_flags = dev->flags;
4449
	int err;
4450

4451
	err = __dev_set_promiscuity(dev, inc);
4452
	if (err < 0)
4453
		return err;
4454 4455
	if (dev->flags != old_flags)
		dev_set_rx_mode(dev);
4456
	return err;
4457
}
E
Eric Dumazet 已提交
4458
EXPORT_SYMBOL(dev_set_promiscuity);
4459

L
Linus Torvalds 已提交
4460 4461 4462 4463 4464 4465 4466 4467 4468 4469
/**
 *	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.
4470
 *	Return 0 if successful or a negative errno code on error.
L
Linus Torvalds 已提交
4471 4472
 */

4473
int dev_set_allmulti(struct net_device *dev, int inc)
L
Linus Torvalds 已提交
4474 4475 4476
{
	unsigned short old_flags = dev->flags;

4477 4478
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4479
	dev->flags |= IFF_ALLMULTI;
4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495
	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;
		}
	}
4496
	if (dev->flags ^ old_flags) {
4497
		dev_change_rx_flags(dev, IFF_ALLMULTI);
4498
		dev_set_rx_mode(dev);
4499
	}
4500
	return 0;
4501
}
E
Eric Dumazet 已提交
4502
EXPORT_SYMBOL(dev_set_allmulti);
4503 4504 4505 4506

/*
 *	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
4507
 *	filtering it is put in promiscuous mode while unicast addresses
4508 4509 4510 4511
 *	are present.
 */
void __dev_set_rx_mode(struct net_device *dev)
{
4512 4513
	const struct net_device_ops *ops = dev->netdev_ops;

4514 4515 4516 4517 4518
	/* dev_open will call this function so the list will stay sane. */
	if (!(dev->flags&IFF_UP))
		return;

	if (!netif_device_present(dev))
4519
		return;
4520

4521 4522
	if (ops->ndo_set_rx_mode)
		ops->ndo_set_rx_mode(dev);
4523 4524 4525 4526
	else {
		/* Unicast addresses changes may only happen under the rtnl,
		 * therefore calling __dev_set_promiscuity here is safe.
		 */
4527
		if (!netdev_uc_empty(dev) && !dev->uc_promisc) {
4528 4529
			__dev_set_promiscuity(dev, 1);
			dev->uc_promisc = 1;
4530
		} else if (netdev_uc_empty(dev) && dev->uc_promisc) {
4531 4532 4533 4534
			__dev_set_promiscuity(dev, -1);
			dev->uc_promisc = 0;
		}

4535 4536
		if (ops->ndo_set_multicast_list)
			ops->ndo_set_multicast_list(dev);
4537 4538 4539 4540 4541
	}
}

void dev_set_rx_mode(struct net_device *dev)
{
4542
	netif_addr_lock_bh(dev);
4543
	__dev_set_rx_mode(dev);
4544
	netif_addr_unlock_bh(dev);
L
Linus Torvalds 已提交
4545 4546
}

4547 4548 4549 4550 4551 4552
/**
 *	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
4553 4554 4555 4556 4557 4558
unsigned dev_get_flags(const struct net_device *dev)
{
	unsigned flags;

	flags = (dev->flags & ~(IFF_PROMISC |
				IFF_ALLMULTI |
S
Stefan Rompf 已提交
4559 4560 4561
				IFF_RUNNING |
				IFF_LOWER_UP |
				IFF_DORMANT)) |
L
Linus Torvalds 已提交
4562 4563 4564
		(dev->gflags & (IFF_PROMISC |
				IFF_ALLMULTI));

S
Stefan Rompf 已提交
4565 4566 4567 4568 4569 4570 4571 4572
	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
4573 4574 4575

	return flags;
}
E
Eric Dumazet 已提交
4576
EXPORT_SYMBOL(dev_get_flags);
L
Linus Torvalds 已提交
4577

4578
int __dev_change_flags(struct net_device *dev, unsigned int flags)
L
Linus Torvalds 已提交
4579 4580
{
	int old_flags = dev->flags;
4581
	int ret;
L
Linus Torvalds 已提交
4582

4583 4584
	ASSERT_RTNL();

L
Linus Torvalds 已提交
4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598
	/*
	 *	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.
	 */

4599 4600
	if ((old_flags ^ flags) & IFF_MULTICAST)
		dev_change_rx_flags(dev, IFF_MULTICAST);
4601

4602
	dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4603 4604 4605 4606 4607 4608 4609 4610 4611

	/*
	 *	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  ? */
4612
		ret = ((old_flags & IFF_UP) ? __dev_close : __dev_open)(dev);
L
Linus Torvalds 已提交
4613 4614

		if (!ret)
4615
			dev_set_rx_mode(dev);
L
Linus Torvalds 已提交
4616 4617 4618
	}

	if ((flags ^ dev->gflags) & IFF_PROMISC) {
E
Eric Dumazet 已提交
4619 4620
		int inc = (flags & IFF_PROMISC) ? 1 : -1;

L
Linus Torvalds 已提交
4621 4622 4623 4624 4625 4626 4627 4628 4629
		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
4630 4631
		int inc = (flags & IFF_ALLMULTI) ? 1 : -1;

L
Linus Torvalds 已提交
4632 4633 4634 4635
		dev->gflags ^= IFF_ALLMULTI;
		dev_set_allmulti(dev, inc);
	}

4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672
	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;
4673 4674
	if (changes)
		rtmsg_ifinfo(RTM_NEWLINK, dev, changes);
L
Linus Torvalds 已提交
4675

4676
	__dev_notify_flags(dev, old_flags);
L
Linus Torvalds 已提交
4677 4678
	return ret;
}
E
Eric Dumazet 已提交
4679
EXPORT_SYMBOL(dev_change_flags);
L
Linus Torvalds 已提交
4680

4681 4682 4683 4684 4685 4686 4687
/**
 *	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
4688 4689
int dev_set_mtu(struct net_device *dev, int new_mtu)
{
4690
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703
	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;
4704 4705
	if (ops->ndo_change_mtu)
		err = ops->ndo_change_mtu(dev, new_mtu);
L
Linus Torvalds 已提交
4706 4707
	else
		dev->mtu = new_mtu;
4708

L
Linus Torvalds 已提交
4709
	if (!err && dev->flags & IFF_UP)
4710
		call_netdevice_notifiers(NETDEV_CHANGEMTU, dev);
L
Linus Torvalds 已提交
4711 4712
	return err;
}
E
Eric Dumazet 已提交
4713
EXPORT_SYMBOL(dev_set_mtu);
L
Linus Torvalds 已提交
4714

4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725
/**
 *	dev_set_group - Change group this device belongs to
 *	@dev: device
 *	@new_group: group this device should belong to
 */
void dev_set_group(struct net_device *dev, int new_group)
{
	dev->group = new_group;
}
EXPORT_SYMBOL(dev_set_group);

4726 4727 4728 4729 4730 4731 4732
/**
 *	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
4733 4734
int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa)
{
4735
	const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
4736 4737
	int err;

4738
	if (!ops->ndo_set_mac_address)
L
Linus Torvalds 已提交
4739 4740 4741 4742 4743
		return -EOPNOTSUPP;
	if (sa->sa_family != dev->type)
		return -EINVAL;
	if (!netif_device_present(dev))
		return -ENODEV;
4744
	err = ops->ndo_set_mac_address(dev, sa);
L
Linus Torvalds 已提交
4745
	if (!err)
4746
		call_netdevice_notifiers(NETDEV_CHANGEADDR, dev);
L
Linus Torvalds 已提交
4747 4748
	return err;
}
E
Eric Dumazet 已提交
4749
EXPORT_SYMBOL(dev_set_mac_address);
L
Linus Torvalds 已提交
4750 4751

/*
E
Eric Dumazet 已提交
4752
 *	Perform the SIOCxIFxxx calls, inside rcu_read_lock()
L
Linus Torvalds 已提交
4753
 */
4754
static int dev_ifsioc_locked(struct net *net, struct ifreq *ifr, unsigned int cmd)
L
Linus Torvalds 已提交
4755 4756
{
	int err;
E
Eric Dumazet 已提交
4757
	struct net_device *dev = dev_get_by_name_rcu(net, ifr->ifr_name);
L
Linus Torvalds 已提交
4758 4759 4760 4761 4762

	if (!dev)
		return -ENODEV;

	switch (cmd) {
E
Eric Dumazet 已提交
4763 4764 4765
	case SIOCGIFFLAGS:	/* Get interface flags */
		ifr->ifr_flags = (short) dev_get_flags(dev);
		return 0;
L
Linus Torvalds 已提交
4766

E
Eric Dumazet 已提交
4767 4768 4769 4770
	case SIOCGIFMETRIC:	/* Get the metric on the interface
				   (currently unused) */
		ifr->ifr_metric = 0;
		return 0;
L
Linus Torvalds 已提交
4771

E
Eric Dumazet 已提交
4772 4773 4774
	case SIOCGIFMTU:	/* Get the MTU of a device */
		ifr->ifr_mtu = dev->mtu;
		return 0;
L
Linus Torvalds 已提交
4775

E
Eric Dumazet 已提交
4776 4777 4778 4779 4780 4781 4782 4783
	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
4784

E
Eric Dumazet 已提交
4785 4786 4787
	case SIOCGIFSLAVE:
		err = -EINVAL;
		break;
4788

E
Eric Dumazet 已提交
4789 4790 4791 4792 4793 4794 4795 4796
	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;
4797

E
Eric Dumazet 已提交
4798 4799 4800
	case SIOCGIFINDEX:
		ifr->ifr_ifindex = dev->ifindex;
		return 0;
4801

E
Eric Dumazet 已提交
4802 4803 4804
	case SIOCGIFTXQLEN:
		ifr->ifr_qlen = dev->tx_queue_len;
		return 0;
4805

E
Eric Dumazet 已提交
4806 4807 4808 4809 4810 4811 4812
	default:
		/* dev_ioctl() should ensure this case
		 * is never reached
		 */
		WARN_ON(1);
		err = -EINVAL;
		break;
4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824

	}
	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
4825
	const struct net_device_ops *ops;
4826 4827 4828 4829

	if (!dev)
		return -ENODEV;

J
Jarek Poplawski 已提交
4830 4831
	ops = dev->netdev_ops;

4832
	switch (cmd) {
E
Eric Dumazet 已提交
4833 4834
	case SIOCSIFFLAGS:	/* Set interface flags */
		return dev_change_flags(dev, ifr->ifr_flags);
4835

E
Eric Dumazet 已提交
4836 4837 4838
	case SIOCSIFMETRIC:	/* Set the metric on the interface
				   (currently unused) */
		return -EOPNOTSUPP;
4839

E
Eric Dumazet 已提交
4840 4841
	case SIOCSIFMTU:	/* Set the MTU of a device */
		return dev_set_mtu(dev, ifr->ifr_mtu);
L
Linus Torvalds 已提交
4842

E
Eric Dumazet 已提交
4843 4844
	case SIOCSIFHWADDR:
		return dev_set_mac_address(dev, &ifr->ifr_hwaddr);
L
Linus Torvalds 已提交
4845

E
Eric Dumazet 已提交
4846 4847 4848 4849 4850 4851 4852
	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
4853

E
Eric Dumazet 已提交
4854 4855
	case SIOCSIFMAP:
		if (ops->ndo_set_config) {
L
Linus Torvalds 已提交
4856 4857
			if (!netif_device_present(dev))
				return -ENODEV;
E
Eric Dumazet 已提交
4858 4859 4860
			return ops->ndo_set_config(dev, &ifr->ifr_map);
		}
		return -EOPNOTSUPP;
L
Linus Torvalds 已提交
4861

E
Eric Dumazet 已提交
4862 4863 4864 4865 4866 4867
	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;
4868
		return dev_mc_add_global(dev, ifr->ifr_hwaddr.sa_data);
E
Eric Dumazet 已提交
4869 4870 4871 4872 4873 4874 4875

	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;
4876
		return dev_mc_del_global(dev, ifr->ifr_hwaddr.sa_data);
L
Linus Torvalds 已提交
4877

E
Eric Dumazet 已提交
4878 4879 4880 4881 4882
	case SIOCSIFTXQLEN:
		if (ifr->ifr_qlen < 0)
			return -EINVAL;
		dev->tx_queue_len = ifr->ifr_qlen;
		return 0;
L
Linus Torvalds 已提交
4883

E
Eric Dumazet 已提交
4884 4885 4886
	case SIOCSIFNAME:
		ifr->ifr_newname[IFNAMSIZ-1] = '\0';
		return dev_change_name(dev, ifr->ifr_newname);
L
Linus Torvalds 已提交
4887

E
Eric Dumazet 已提交
4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915
	/*
	 *	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
4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927

	}
	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
4928
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
4929 4930 4931 4932 4933 4934 4935 4936 4937
 *	@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.
 */

4938
int dev_ioctl(struct net *net, unsigned int cmd, void __user *arg)
L
Linus Torvalds 已提交
4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949
{
	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) {
4950
		rtnl_lock();
4951
		ret = dev_ifconf(net, (char __user *) arg);
4952
		rtnl_unlock();
L
Linus Torvalds 已提交
4953 4954 4955
		return ret;
	}
	if (cmd == SIOCGIFNAME)
4956
		return dev_ifname(net, (struct ifreq __user *)arg);
L
Linus Torvalds 已提交
4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971

	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
4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986
	/*
	 *	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
4987
		rcu_read_lock();
E
Eric Dumazet 已提交
4988
		ret = dev_ifsioc_locked(net, &ifr, cmd);
E
Eric Dumazet 已提交
4989
		rcu_read_unlock();
E
Eric Dumazet 已提交
4990 4991 4992 4993 4994 4995 4996 4997
		if (!ret) {
			if (colon)
				*colon = ':';
			if (copy_to_user(arg, &ifr,
					 sizeof(struct ifreq)))
				ret = -EFAULT;
		}
		return ret;
L
Linus Torvalds 已提交
4998

E
Eric Dumazet 已提交
4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011
	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
5012

E
Eric Dumazet 已提交
5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035
	/*
	 *	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
5036

E
Eric Dumazet 已提交
5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087
	/*
	 *	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)) {
5088
			dev_load(net, ifr.ifr_name);
L
Linus Torvalds 已提交
5089
			rtnl_lock();
5090
			ret = dev_ifsioc(net, &ifr, cmd);
L
Linus Torvalds 已提交
5091
			rtnl_unlock();
E
Eric Dumazet 已提交
5092 5093 5094
			if (!ret && copy_to_user(arg, &ifr,
						 sizeof(struct ifreq)))
				ret = -EFAULT;
L
Linus Torvalds 已提交
5095
			return ret;
E
Eric Dumazet 已提交
5096 5097 5098 5099 5100
		}
		/* 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
5101 5102 5103 5104 5105 5106
	}
}


/**
 *	dev_new_index	-	allocate an ifindex
5107
 *	@net: the applicable net namespace
L
Linus Torvalds 已提交
5108 5109 5110 5111 5112
 *
 *	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.
 */
5113
static int dev_new_index(struct net *net)
L
Linus Torvalds 已提交
5114 5115 5116 5117 5118
{
	static int ifindex;
	for (;;) {
		if (++ifindex <= 0)
			ifindex = 1;
5119
		if (!__dev_get_by_index(net, ifindex))
L
Linus Torvalds 已提交
5120 5121 5122 5123 5124
			return ifindex;
	}
}

/* Delayed registration/unregisteration */
5125
static LIST_HEAD(net_todo_list);
L
Linus Torvalds 已提交
5126

5127
static void net_set_todo(struct net_device *dev)
L
Linus Torvalds 已提交
5128 5129 5130 5131
{
	list_add_tail(&dev->todo_list, &net_todo_list);
}

5132
static void rollback_registered_many(struct list_head *head)
5133
{
5134
	struct net_device *dev, *tmp;
5135

5136 5137 5138
	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

5139
	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
5140
		/* Some devices call without registering
5141 5142
		 * for initialization unwind. Remove those
		 * devices and proceed with the remaining.
5143 5144 5145 5146
		 */
		if (dev->reg_state == NETREG_UNINITIALIZED) {
			pr_debug("unregister_netdevice: device %s/%p never "
				 "was registered\n", dev->name, dev);
5147

5148
			WARN_ON(1);
5149 5150
			list_del(&dev->unreg_list);
			continue;
5151
		}
5152

5153
		BUG_ON(dev->reg_state != NETREG_REGISTERED);
5154
	}
5155

5156 5157
	/* If device is running, close it first. */
	dev_close_many(head);
5158

5159
	list_for_each_entry(dev, head, unreg_list) {
5160 5161
		/* And unlink it from device chain. */
		unlist_netdevice(dev);
5162

5163 5164
		dev->reg_state = NETREG_UNREGISTERING;
	}
5165 5166 5167

	synchronize_net();

5168 5169 5170
	list_for_each_entry(dev, head, unreg_list) {
		/* Shutdown queueing discipline. */
		dev_shutdown(dev);
5171 5172


5173 5174 5175 5176
		/* Notify protocols, that we are about to destroy
		   this device. They should clean all the things.
		*/
		call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5177

5178 5179 5180 5181
		if (!dev->rtnl_link_ops ||
		    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
			rtmsg_ifinfo(RTM_DELLINK, dev, ~0U);

5182 5183 5184
		/*
		 *	Flush the unicast and multicast chains
		 */
5185
		dev_uc_flush(dev);
5186
		dev_mc_flush(dev);
5187

5188 5189
		if (dev->netdev_ops->ndo_uninit)
			dev->netdev_ops->ndo_uninit(dev);
5190

5191 5192
		/* Notifier chain MUST detach us from master device. */
		WARN_ON(dev->master);
5193

5194 5195 5196
		/* Remove entries from kobject tree */
		netdev_unregister_kobject(dev);
	}
5197

5198
	/* Process any work delayed until the end of the batch */
5199
	dev = list_first_entry(head, struct net_device, unreg_list);
5200
	call_netdevice_notifiers(NETDEV_UNREGISTER_BATCH, dev);
5201

5202
	rcu_barrier();
5203

5204
	list_for_each_entry(dev, head, unreg_list)
5205 5206 5207 5208 5209 5210 5211 5212 5213
		dev_put(dev);
}

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

	list_add(&dev->unreg_list, &single);
	rollback_registered_many(&single);
5214 5215
}

5216 5217
unsigned long netdev_fix_features(unsigned long features, const char *name)
{
5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234
	/* Fix illegal checksum combinations */
	if ((features & NETIF_F_HW_CSUM) &&
	    (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
		if (name)
			printk(KERN_NOTICE "%s: mixed HW and IP checksum settings.\n",
				name);
		features &= ~(NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
	}

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

5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252
	/* 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) {
5253 5254 5255 5256
		/* maybe split UFO into V4 and V6? */
		if (!((features & NETIF_F_GEN_CSUM) ||
		    (features & (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))
			    == (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM))) {
5257 5258
			if (name)
				printk(KERN_ERR "%s: Dropping NETIF_F_UFO "
5259
				       "since no checksum offload features.\n",
5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275
				       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);

5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302
/**
 *	netif_stacked_transfer_operstate -	transfer operstate
 *	@rootdev: the root or lower level device to transfer state from
 *	@dev: the device to transfer operstate to
 *
 *	Transfer operational state from root to device. This is normally
 *	called when a stacking relationship exists between the root
 *	device and the device(a leaf device).
 */
void netif_stacked_transfer_operstate(const struct net_device *rootdev,
					struct net_device *dev)
{
	if (rootdev->operstate == IF_OPER_DORMANT)
		netif_dormant_on(dev);
	else
		netif_dormant_off(dev);

	if (netif_carrier_ok(rootdev)) {
		if (!netif_carrier_ok(dev))
			netif_carrier_on(dev);
	} else {
		if (netif_carrier_ok(dev))
			netif_carrier_off(dev);
	}
}
EXPORT_SYMBOL(netif_stacked_transfer_operstate);

T
Tom Herbert 已提交
5303
#ifdef CONFIG_RPS
5304 5305 5306
static int netif_alloc_rx_queues(struct net_device *dev)
{
	unsigned int i, count = dev->num_rx_queues;
T
Tom Herbert 已提交
5307
	struct netdev_rx_queue *rx;
5308

T
Tom Herbert 已提交
5309
	BUG_ON(count < 1);
5310

T
Tom Herbert 已提交
5311 5312 5313 5314
	rx = kcalloc(count, sizeof(struct netdev_rx_queue), GFP_KERNEL);
	if (!rx) {
		pr_err("netdev: Unable to allocate %u rx queues.\n", count);
		return -ENOMEM;
5315
	}
T
Tom Herbert 已提交
5316 5317 5318
	dev->_rx = rx;

	for (i = 0; i < count; i++)
T
Tom Herbert 已提交
5319
		rx[i].dev = dev;
5320 5321
	return 0;
}
T
Tom Herbert 已提交
5322
#endif
5323

C
Changli Gao 已提交
5324 5325 5326 5327 5328 5329 5330
static void netdev_init_one_queue(struct net_device *dev,
				  struct netdev_queue *queue, void *_unused)
{
	/* Initialize queue lock */
	spin_lock_init(&queue->_xmit_lock);
	netdev_set_xmit_lockdep_class(&queue->_xmit_lock, dev->type);
	queue->xmit_lock_owner = -1;
5331
	netdev_queue_numa_node_write(queue, NUMA_NO_NODE);
C
Changli Gao 已提交
5332 5333 5334
	queue->dev = dev;
}

5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348
static int netif_alloc_netdev_queues(struct net_device *dev)
{
	unsigned int count = dev->num_tx_queues;
	struct netdev_queue *tx;

	BUG_ON(count < 1);

	tx = kcalloc(count, sizeof(struct netdev_queue), GFP_KERNEL);
	if (!tx) {
		pr_err("netdev: Unable to allocate %u tx queues.\n",
		       count);
		return -ENOMEM;
	}
	dev->_tx = tx;
T
Tom Herbert 已提交
5349

5350 5351
	netdev_for_each_tx_queue(dev, netdev_init_one_queue, NULL);
	spin_lock_init(&dev->tx_global_lock);
C
Changli Gao 已提交
5352 5353

	return 0;
5354 5355
}

L
Linus Torvalds 已提交
5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375
/**
 *	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;
5376
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
5377 5378 5379 5380

	BUG_ON(dev_boot_phase);
	ASSERT_RTNL();

5381 5382
	might_sleep();

L
Linus Torvalds 已提交
5383 5384
	/* When net_device's are persistent, this will be fatal. */
	BUG_ON(dev->reg_state != NETREG_UNINITIALIZED);
5385
	BUG_ON(!net);
L
Linus Torvalds 已提交
5386

5387
	spin_lock_init(&dev->addr_list_lock);
5388
	netdev_set_addr_lockdep_class(dev);
L
Linus Torvalds 已提交
5389 5390 5391 5392

	dev->iflink = -1;

	/* Init, if this function is available */
5393 5394
	if (dev->netdev_ops->ndo_init) {
		ret = dev->netdev_ops->ndo_init(dev);
L
Linus Torvalds 已提交
5395 5396 5397
		if (ret) {
			if (ret > 0)
				ret = -EIO;
5398
			goto out;
L
Linus Torvalds 已提交
5399 5400
		}
	}
5401

5402
	ret = dev_get_valid_name(dev, dev->name, 0);
5403
	if (ret)
5404
		goto err_uninit;
L
Linus Torvalds 已提交
5405

5406
	dev->ifindex = dev_new_index(net);
L
Linus Torvalds 已提交
5407 5408 5409
	if (dev->iflink == -1)
		dev->iflink = dev->ifindex;

5410
	dev->features = netdev_fix_features(dev->features, dev->name);
L
Linus Torvalds 已提交
5411

5412 5413 5414 5415
	/* Enable software GSO if SG is supported. */
	if (dev->features & NETIF_F_SG)
		dev->features |= NETIF_F_GSO;

5416 5417 5418
	/* Enable GRO and NETIF_F_HIGHDMA for vlans by default,
	 * vlan_dev_init() will do the dev->features check, so these features
	 * are enabled only if supported by underlying device.
5419
	 */
5420
	dev->vlan_features |= (NETIF_F_GRO | NETIF_F_HIGHDMA);
5421

5422 5423 5424 5425 5426
	ret = call_netdevice_notifiers(NETDEV_POST_INIT, dev);
	ret = notifier_to_errno(ret);
	if (ret)
		goto err_uninit;

5427
	ret = netdev_register_kobject(dev);
5428
	if (ret)
5429
		goto err_uninit;
5430 5431
	dev->reg_state = NETREG_REGISTERED;

L
Linus Torvalds 已提交
5432 5433 5434 5435 5436 5437 5438 5439 5440
	/*
	 *	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);
5441
	list_netdevice(dev);
L
Linus Torvalds 已提交
5442 5443

	/* Notify protocols, that a new device appeared. */
5444
	ret = call_netdevice_notifiers(NETDEV_REGISTER, dev);
5445
	ret = notifier_to_errno(ret);
5446 5447 5448 5449
	if (ret) {
		rollback_registered(dev);
		dev->reg_state = NETREG_UNREGISTERED;
	}
5450 5451 5452 5453
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
5454 5455 5456
	if (!dev->rtnl_link_ops ||
	    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
		rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);
L
Linus Torvalds 已提交
5457 5458 5459

out:
	return ret;
5460 5461

err_uninit:
5462 5463
	if (dev->netdev_ops->ndo_uninit)
		dev->netdev_ops->ndo_uninit(dev);
5464
	goto out;
L
Linus Torvalds 已提交
5465
}
E
Eric Dumazet 已提交
5466
EXPORT_SYMBOL(register_netdevice);
L
Linus Torvalds 已提交
5467

5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498
/**
 *	init_dummy_netdev	- init a dummy network device for NAPI
 *	@dev: device to init
 *
 *	This takes a network device structure and initialize the minimum
 *	amount of fields so it can be used to schedule NAPI polls without
 *	registering a full blown interface. This is to be used by drivers
 *	that need to tie several hardware interfaces to a single NAPI
 *	poll scheduler due to HW limitations.
 */
int init_dummy_netdev(struct net_device *dev)
{
	/* Clear everything. Note we don't initialize spinlocks
	 * are they aren't supposed to be taken by any of the
	 * NAPI code and this dummy netdev is supposed to be
	 * only ever used for NAPI polls
	 */
	memset(dev, 0, sizeof(struct net_device));

	/* make sure we BUG if trying to hit standard
	 * register/unregister code path
	 */
	dev->reg_state = NETREG_DUMMY;

	/* NAPI wants this */
	INIT_LIST_HEAD(&dev->napi_list);

	/* a dummy interface is started by default */
	set_bit(__LINK_STATE_PRESENT, &dev->state);
	set_bit(__LINK_STATE_START, &dev->state);

E
Eric Dumazet 已提交
5499 5500 5501 5502 5503
	/* Note : We dont allocate pcpu_refcnt for dummy devices,
	 * because users of this 'device' dont need to change
	 * its refcount.
	 */

5504 5505 5506 5507 5508
	return 0;
}
EXPORT_SYMBOL_GPL(init_dummy_netdev);


L
Linus Torvalds 已提交
5509 5510 5511 5512 5513 5514 5515 5516 5517
/**
 *	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.
 *
5518
 *	This is a wrapper around register_netdevice that takes the rtnl semaphore
L
Linus Torvalds 已提交
5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536
 *	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;
	}
5537

L
Linus Torvalds 已提交
5538 5539 5540 5541 5542 5543 5544
	err = register_netdevice(dev);
out:
	rtnl_unlock();
	return err;
}
EXPORT_SYMBOL(register_netdev);

E
Eric Dumazet 已提交
5545 5546 5547 5548 5549 5550 5551 5552 5553 5554
int netdev_refcnt_read(const struct net_device *dev)
{
	int i, refcnt = 0;

	for_each_possible_cpu(i)
		refcnt += *per_cpu_ptr(dev->pcpu_refcnt, i);
	return refcnt;
}
EXPORT_SYMBOL(netdev_refcnt_read);

L
Linus Torvalds 已提交
5555 5556 5557 5558 5559 5560 5561 5562 5563
/*
 * 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
5564
 * call dev_put.
L
Linus Torvalds 已提交
5565 5566 5567 5568
 */
static void netdev_wait_allrefs(struct net_device *dev)
{
	unsigned long rebroadcast_time, warning_time;
E
Eric Dumazet 已提交
5569
	int refcnt;
L
Linus Torvalds 已提交
5570

5571 5572
	linkwatch_forget_dev(dev);

L
Linus Torvalds 已提交
5573
	rebroadcast_time = warning_time = jiffies;
E
Eric Dumazet 已提交
5574 5575 5576
	refcnt = netdev_refcnt_read(dev);

	while (refcnt != 0) {
L
Linus Torvalds 已提交
5577
		if (time_after(jiffies, rebroadcast_time + 1 * HZ)) {
5578
			rtnl_lock();
L
Linus Torvalds 已提交
5579 5580

			/* Rebroadcast unregister notification */
5581
			call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
5582
			/* don't resend NETDEV_UNREGISTER_BATCH, _BATCH users
5583
			 * should have already handle it the first time */
L
Linus Torvalds 已提交
5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595

			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();
			}

5596
			__rtnl_unlock();
L
Linus Torvalds 已提交
5597 5598 5599 5600 5601 5602

			rebroadcast_time = jiffies;
		}

		msleep(250);

E
Eric Dumazet 已提交
5603 5604
		refcnt = netdev_refcnt_read(dev);

L
Linus Torvalds 已提交
5605 5606 5607 5608
		if (time_after(jiffies, warning_time + 10 * HZ)) {
			printk(KERN_EMERG "unregister_netdevice: "
			       "waiting for %s to become free. Usage "
			       "count = %d\n",
E
Eric Dumazet 已提交
5609
			       dev->name, refcnt);
L
Linus Torvalds 已提交
5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628
			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
5629
 * We are invoked by rtnl_unlock().
L
Linus Torvalds 已提交
5630
 * This allows us to deal with problems:
5631
 * 1) We can delete sysfs objects which invoke hotplug
L
Linus Torvalds 已提交
5632 5633 5634
 *    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
5635 5636 5637
 *
 * 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
5638 5639 5640
 */
void netdev_run_todo(void)
{
5641
	struct list_head list;
L
Linus Torvalds 已提交
5642 5643

	/* Snapshot list, allow later requests */
5644
	list_replace_init(&net_todo_list, &list);
H
Herbert Xu 已提交
5645 5646

	__rtnl_unlock();
5647

L
Linus Torvalds 已提交
5648 5649
	while (!list_empty(&list)) {
		struct net_device *dev
5650
			= list_first_entry(&list, struct net_device, todo_list);
L
Linus Torvalds 已提交
5651 5652
		list_del(&dev->todo_list);

5653 5654 5655 5656 5657 5658
		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
5659

5660
		dev->reg_state = NETREG_UNREGISTERED;
L
Linus Torvalds 已提交
5661

5662
		on_each_cpu(flush_backlog, dev, 1);
5663

5664
		netdev_wait_allrefs(dev);
L
Linus Torvalds 已提交
5665

5666
		/* paranoia */
E
Eric Dumazet 已提交
5667
		BUG_ON(netdev_refcnt_read(dev));
E
Eric Dumazet 已提交
5668
		WARN_ON(rcu_dereference_raw(dev->ip_ptr));
E
Eric Dumazet 已提交
5669
		WARN_ON(rcu_dereference_raw(dev->ip6_ptr));
5670
		WARN_ON(dev->dn_ptr);
L
Linus Torvalds 已提交
5671

5672 5673
		if (dev->destructor)
			dev->destructor(dev);
5674 5675 5676

		/* Free network device */
		kobject_put(&dev->dev.kobj);
L
Linus Torvalds 已提交
5677 5678 5679
	}
}

5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700
/* Convert net_device_stats to rtnl_link_stats64.  They have the same
 * fields in the same order, with only the type differing.
 */
static void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
				    const struct net_device_stats *netdev_stats)
{
#if BITS_PER_LONG == 64
        BUILD_BUG_ON(sizeof(*stats64) != sizeof(*netdev_stats));
        memcpy(stats64, netdev_stats, sizeof(*stats64));
#else
	size_t i, n = sizeof(*stats64) / sizeof(u64);
	const unsigned long *src = (const unsigned long *)netdev_stats;
	u64 *dst = (u64 *)stats64;

	BUILD_BUG_ON(sizeof(*netdev_stats) / sizeof(unsigned long) !=
		     sizeof(*stats64) / sizeof(u64));
	for (i = 0; i < n; i++)
		dst[i] = src[i];
#endif
}

5701 5702 5703
/**
 *	dev_get_stats	- get network device statistics
 *	@dev: device to get statistics from
5704
 *	@storage: place to store stats
5705
 *
5706 5707 5708 5709
 *	Get network statistics from device. Return @storage.
 *	The device driver may provide its own method by setting
 *	dev->netdev_ops->get_stats64 or dev->netdev_ops->get_stats;
 *	otherwise the internal statistics structure is used.
5710
 */
5711 5712
struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
					struct rtnl_link_stats64 *storage)
5713
{
5714 5715
	const struct net_device_ops *ops = dev->netdev_ops;

5716 5717
	if (ops->ndo_get_stats64) {
		memset(storage, 0, sizeof(*storage));
5718 5719
		ops->ndo_get_stats64(dev, storage);
	} else if (ops->ndo_get_stats) {
5720
		netdev_stats_to_stats64(storage, ops->ndo_get_stats(dev));
5721 5722
	} else {
		netdev_stats_to_stats64(storage, &dev->stats);
5723
	}
5724
	storage->rx_dropped += atomic_long_read(&dev->rx_dropped);
5725
	return storage;
R
Rusty Russell 已提交
5726
}
5727
EXPORT_SYMBOL(dev_get_stats);
R
Rusty Russell 已提交
5728

5729
struct netdev_queue *dev_ingress_queue_create(struct net_device *dev)
5730
{
5731
	struct netdev_queue *queue = dev_ingress_queue(dev);
5732

5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744
#ifdef CONFIG_NET_CLS_ACT
	if (queue)
		return queue;
	queue = kzalloc(sizeof(*queue), GFP_KERNEL);
	if (!queue)
		return NULL;
	netdev_init_one_queue(dev, queue, NULL);
	queue->qdisc = &noop_qdisc;
	queue->qdisc_sleeping = &noop_qdisc;
	rcu_assign_pointer(dev->ingress_queue, queue);
#endif
	return queue;
5745 5746
}

L
Linus Torvalds 已提交
5747
/**
T
Tom Herbert 已提交
5748
 *	alloc_netdev_mqs - allocate network device
L
Linus Torvalds 已提交
5749 5750 5751
 *	@sizeof_priv:	size of private data to allocate space for
 *	@name:		device name format string
 *	@setup:		callback to initialize device
T
Tom Herbert 已提交
5752 5753
 *	@txqs:		the number of TX subqueues to allocate
 *	@rxqs:		the number of RX subqueues to allocate
L
Linus Torvalds 已提交
5754 5755
 *
 *	Allocates a struct net_device with private data area for driver use
5756
 *	and performs basic initialization.  Also allocates subquue structs
T
Tom Herbert 已提交
5757
 *	for each queue on the device.
L
Linus Torvalds 已提交
5758
 */
T
Tom Herbert 已提交
5759 5760 5761
struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
		void (*setup)(struct net_device *),
		unsigned int txqs, unsigned int rxqs)
L
Linus Torvalds 已提交
5762 5763
{
	struct net_device *dev;
5764
	size_t alloc_size;
5765
	struct net_device *p;
L
Linus Torvalds 已提交
5766

5767 5768
	BUG_ON(strlen(name) >= sizeof(dev->name));

T
Tom Herbert 已提交
5769
	if (txqs < 1) {
5770 5771 5772 5773 5774
		pr_err("alloc_netdev: Unable to allocate device "
		       "with zero queues.\n");
		return NULL;
	}

T
Tom Herbert 已提交
5775 5776 5777 5778 5779 5780 5781 5782
#ifdef CONFIG_RPS
	if (rxqs < 1) {
		pr_err("alloc_netdev: Unable to allocate device "
		       "with zero RX queues.\n");
		return NULL;
	}
#endif

5783
	alloc_size = sizeof(struct net_device);
5784 5785
	if (sizeof_priv) {
		/* ensure 32-byte alignment of private area */
5786
		alloc_size = ALIGN(alloc_size, NETDEV_ALIGN);
5787 5788 5789
		alloc_size += sizeof_priv;
	}
	/* ensure 32-byte alignment of whole construct */
5790
	alloc_size += NETDEV_ALIGN - 1;
L
Linus Torvalds 已提交
5791

5792
	p = kzalloc(alloc_size, GFP_KERNEL);
L
Linus Torvalds 已提交
5793
	if (!p) {
5794
		printk(KERN_ERR "alloc_netdev: Unable to allocate device.\n");
L
Linus Torvalds 已提交
5795 5796 5797
		return NULL;
	}

5798
	dev = PTR_ALIGN(p, NETDEV_ALIGN);
L
Linus Torvalds 已提交
5799
	dev->padded = (char *)dev - (char *)p;
5800

E
Eric Dumazet 已提交
5801 5802
	dev->pcpu_refcnt = alloc_percpu(int);
	if (!dev->pcpu_refcnt)
5803
		goto free_p;
5804 5805

	if (dev_addr_init(dev))
E
Eric Dumazet 已提交
5806
		goto free_pcpu;
5807

5808
	dev_mc_init(dev);
5809
	dev_uc_init(dev);
J
Jiri Pirko 已提交
5810

5811
	dev_net_set(dev, &init_net);
L
Linus Torvalds 已提交
5812

T
Tom Herbert 已提交
5813 5814
	dev->num_tx_queues = txqs;
	dev->real_num_tx_queues = txqs;
5815 5816
	if (netif_alloc_netdev_queues(dev))
		goto free_pcpu;
5817

E
Eric Dumazet 已提交
5818
#ifdef CONFIG_RPS
T
Tom Herbert 已提交
5819 5820
	dev->num_rx_queues = rxqs;
	dev->real_num_rx_queues = rxqs;
T
Tom Herbert 已提交
5821 5822
	if (netif_alloc_rx_queues(dev))
		goto free_pcpu;
E
Eric Dumazet 已提交
5823
#endif
T
Tom Herbert 已提交
5824

5825
	dev->gso_max_size = GSO_MAX_SIZE;
L
Linus Torvalds 已提交
5826

5827 5828
	INIT_LIST_HEAD(&dev->ethtool_ntuple_list.list);
	dev->ethtool_ntuple_list.count = 0;
5829
	INIT_LIST_HEAD(&dev->napi_list);
5830
	INIT_LIST_HEAD(&dev->unreg_list);
5831
	INIT_LIST_HEAD(&dev->link_watch_list);
5832
	dev->priv_flags = IFF_XMIT_DST_RELEASE;
L
Linus Torvalds 已提交
5833 5834
	setup(dev);
	strcpy(dev->name, name);
5835
	dev->group = INIT_NETDEV_GROUP;
L
Linus Torvalds 已提交
5836
	return dev;
5837

E
Eric Dumazet 已提交
5838 5839
free_pcpu:
	free_percpu(dev->pcpu_refcnt);
5840
	kfree(dev->_tx);
T
Tom Herbert 已提交
5841 5842 5843 5844
#ifdef CONFIG_RPS
	kfree(dev->_rx);
#endif

5845 5846 5847
free_p:
	kfree(p);
	return NULL;
L
Linus Torvalds 已提交
5848
}
T
Tom Herbert 已提交
5849
EXPORT_SYMBOL(alloc_netdev_mqs);
L
Linus Torvalds 已提交
5850 5851 5852 5853 5854

/**
 *	free_netdev - free network device
 *	@dev: device
 *
5855 5856
 *	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
5857 5858 5859 5860
 *	If this is the last reference then it will be freed.
 */
void free_netdev(struct net_device *dev)
{
5861 5862
	struct napi_struct *p, *n;

5863 5864
	release_net(dev_net(dev));

5865
	kfree(dev->_tx);
T
Tom Herbert 已提交
5866 5867 5868
#ifdef CONFIG_RPS
	kfree(dev->_rx);
#endif
5869

5870 5871
	kfree(rcu_dereference_raw(dev->ingress_queue));

5872 5873 5874
	/* Flush device addresses */
	dev_addr_flush(dev);

5875 5876 5877
	/* Clear ethtool n-tuple list */
	ethtool_ntuple_flush(dev);

5878 5879 5880
	list_for_each_entry_safe(p, n, &dev->napi_list, dev_list)
		netif_napi_del(p);

E
Eric Dumazet 已提交
5881 5882 5883
	free_percpu(dev->pcpu_refcnt);
	dev->pcpu_refcnt = NULL;

S
Stephen Hemminger 已提交
5884
	/*  Compatibility with error handling in drivers */
L
Linus Torvalds 已提交
5885 5886 5887 5888 5889 5890 5891 5892
	if (dev->reg_state == NETREG_UNINITIALIZED) {
		kfree((char *)dev - dev->padded);
		return;
	}

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

5893 5894
	/* will free via device release */
	put_device(&dev->dev);
L
Linus Torvalds 已提交
5895
}
E
Eric Dumazet 已提交
5896
EXPORT_SYMBOL(free_netdev);
5897

5898 5899 5900 5901 5902 5903
/**
 *	synchronize_net -  Synchronize with packet receive processing
 *
 *	Wait for packets currently being received to be done.
 *	Does not block later packets from starting.
 */
5904
void synchronize_net(void)
L
Linus Torvalds 已提交
5905 5906
{
	might_sleep();
5907
	synchronize_rcu();
L
Linus Torvalds 已提交
5908
}
E
Eric Dumazet 已提交
5909
EXPORT_SYMBOL(synchronize_net);
L
Linus Torvalds 已提交
5910 5911

/**
5912
 *	unregister_netdevice_queue - remove device from the kernel
L
Linus Torvalds 已提交
5913
 *	@dev: device
5914
 *	@head: list
5915
 *
L
Linus Torvalds 已提交
5916
 *	This function shuts down a device interface and removes it
5917
 *	from the kernel tables.
5918
 *	If head not NULL, device is queued to be unregistered later.
L
Linus Torvalds 已提交
5919 5920 5921 5922 5923
 *
 *	Callers must hold the rtnl semaphore.  You may want
 *	unregister_netdev() instead of this.
 */

5924
void unregister_netdevice_queue(struct net_device *dev, struct list_head *head)
L
Linus Torvalds 已提交
5925
{
5926 5927
	ASSERT_RTNL();

5928
	if (head) {
5929
		list_move_tail(&dev->unreg_list, head);
5930 5931 5932 5933 5934
	} else {
		rollback_registered(dev);
		/* Finish processing unregister after unlock */
		net_set_todo(dev);
	}
L
Linus Torvalds 已提交
5935
}
5936
EXPORT_SYMBOL(unregister_netdevice_queue);
L
Linus Torvalds 已提交
5937

5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951
/**
 *	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);
	}
}
5952
EXPORT_SYMBOL(unregister_netdevice_many);
5953

L
Linus Torvalds 已提交
5954 5955 5956 5957 5958
/**
 *	unregister_netdev - remove device from the kernel
 *	@dev: device
 *
 *	This function shuts down a device interface and removes it
5959
 *	from the kernel tables.
L
Linus Torvalds 已提交
5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972
 *
 *	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);

5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004
/**
 *	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;
6005
	if (net_eq(dev_net(dev), net))
6006 6007 6008 6009 6010 6011
		goto out;

	/* Pick the destination device name, and ensure
	 * we can use it in the destination network namespace.
	 */
	err = -EEXIST;
6012
	if (__dev_get_by_name(net, dev->name)) {
6013 6014 6015
		/* We get here if we can't use the current device name */
		if (!pat)
			goto out;
6016
		if (dev_get_valid_name(dev, pat, 1))
6017 6018 6019 6020 6021 6022 6023 6024
			goto out;
	}

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

	/* If device is running close it first. */
6025
	dev_close(dev);
6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037

	/* 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.
6038 6039 6040 6041

	   Note that dev->reg_state stays at NETREG_REGISTERED.
	   This is wanted because this way 8021q and macvlan know
	   the device is just moving and can keep their slaves up.
6042 6043
	*/
	call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
6044
	call_netdevice_notifiers(NETDEV_UNREGISTER_BATCH, dev);
6045 6046 6047 6048

	/*
	 *	Flush the unicast and multicast chains
	 */
6049
	dev_uc_flush(dev);
6050
	dev_mc_flush(dev);
6051 6052

	/* Actually switch the network namespace */
6053
	dev_net_set(dev, net);
6054 6055 6056 6057 6058 6059 6060 6061 6062

	/* 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;
	}

6063
	/* Fixup kobjects */
6064
	err = device_rename(&dev->dev, dev->name);
6065
	WARN_ON(err);
6066 6067 6068 6069 6070 6071 6072

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

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

6073 6074 6075 6076 6077 6078
	/*
	 *	Prevent userspace races by waiting until the network
	 *	device is fully setup before sending notifications.
	 */
	rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U);

6079 6080 6081 6082 6083
	synchronize_net();
	err = 0;
out:
	return err;
}
6084
EXPORT_SYMBOL_GPL(dev_change_net_namespace);
6085

L
Linus Torvalds 已提交
6086 6087 6088 6089 6090 6091 6092 6093 6094
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;

6095
	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN)
L
Linus Torvalds 已提交
6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111
		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. */
6112 6113 6114 6115 6116 6117
	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
6118 6119 6120 6121 6122

	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	local_irq_enable();

	/* Process offline CPU's input_pkt_queue */
6123
	while ((skb = __skb_dequeue(&oldsd->process_queue))) {
L
Linus Torvalds 已提交
6124
		netif_rx(skb);
6125
		input_queue_head_incr(oldsd);
T
Tom Herbert 已提交
6126
	}
6127
	while ((skb = __skb_dequeue(&oldsd->input_pkt_queue))) {
6128
		netif_rx(skb);
6129 6130
		input_queue_head_incr(oldsd);
	}
L
Linus Torvalds 已提交
6131 6132 6133 6134 6135

	return NOTIFY_OK;
}


6136
/**
6137 6138 6139 6140
 *	netdev_increment_features - increment feature set by one
 *	@all: current feature set
 *	@one: new feature set
 *	@mask: mask feature set
6141 6142
 *
 *	Computes a new feature set after adding a device with feature set
6143 6144
 *	@one to the master device with current feature set @all.  Will not
 *	enable anything that is off in @mask. Returns the new feature set.
6145
 */
6146 6147 6148 6149
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
6150
	if (all & NETIF_F_NO_CSUM && !(one & NETIF_F_NO_CSUM))
6151 6152 6153 6154 6155 6156 6157 6158
		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);
		}
6159

6160 6161 6162 6163 6164 6165
		/* 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;
		}
	}
6166

6167
	one |= NETIF_F_ALL_CSUM;
6168

6169
	one |= all & NETIF_F_ONE_FOR_ALL;
6170
	all &= one | NETIF_F_LLTX | NETIF_F_GSO | NETIF_F_UFO;
6171
	all |= one & mask & NETIF_F_ONE_FOR_ALL;
6172 6173 6174

	return all;
}
6175
EXPORT_SYMBOL(netdev_increment_features);
6176

6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189
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;
}

6190
/* Initialize per network namespace state */
6191
static int __net_init netdev_init(struct net *net)
6192 6193 6194
{
	INIT_LIST_HEAD(&net->dev_base_head);

6195 6196 6197
	net->dev_name_head = netdev_create_hash();
	if (net->dev_name_head == NULL)
		goto err_name;
6198

6199 6200 6201
	net->dev_index_head = netdev_create_hash();
	if (net->dev_index_head == NULL)
		goto err_idx;
6202 6203

	return 0;
6204 6205 6206 6207 6208

err_idx:
	kfree(net->dev_name_head);
err_name:
	return -ENOMEM;
6209 6210
}

6211 6212 6213 6214 6215 6216 6217 6218
/**
 *	netdev_drivername - network driver for the device
 *	@dev: network device
 *	@buffer: buffer for resulting name
 *	@len: size of buffer
 *
 *	Determine network driver for device.
 */
6219
char *netdev_drivername(const struct net_device *dev, char *buffer, int len)
6220
{
6221 6222
	const struct device_driver *driver;
	const struct device *parent;
6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238

	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;
}

6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300
static int __netdev_printk(const char *level, const struct net_device *dev,
			   struct va_format *vaf)
{
	int r;

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

	return r;
}

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

	va_start(args, format);

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

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

	return r;
}
EXPORT_SYMBOL(netdev_printk);

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

define_netdev_printk_level(netdev_emerg, KERN_EMERG);
define_netdev_printk_level(netdev_alert, KERN_ALERT);
define_netdev_printk_level(netdev_crit, KERN_CRIT);
define_netdev_printk_level(netdev_err, KERN_ERR);
define_netdev_printk_level(netdev_warn, KERN_WARNING);
define_netdev_printk_level(netdev_notice, KERN_NOTICE);
define_netdev_printk_level(netdev_info, KERN_INFO);

6301
static void __net_exit netdev_exit(struct net *net)
6302 6303 6304 6305 6306
{
	kfree(net->dev_name_head);
	kfree(net->dev_index_head);
}

6307
static struct pernet_operations __net_initdata netdev_net_ops = {
6308 6309 6310 6311
	.init = netdev_init,
	.exit = netdev_exit,
};

6312
static void __net_exit default_device_exit(struct net *net)
6313
{
6314
	struct net_device *dev, *aux;
6315
	/*
6316
	 * Push all migratable network devices back to the
6317 6318 6319
	 * initial network namespace
	 */
	rtnl_lock();
6320
	for_each_netdev_safe(net, dev, aux) {
6321
		int err;
6322
		char fb_name[IFNAMSIZ];
6323 6324 6325 6326 6327

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

6328 6329 6330
		/* Leave virtual devices for the generic cleanup */
		if (dev->rtnl_link_ops)
			continue;
6331

6332
		/* Push remaing network devices to init_net */
6333 6334
		snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
		err = dev_change_net_namespace(dev, &init_net, fb_name);
6335
		if (err) {
6336
			printk(KERN_EMERG "%s: failed to move %s to init_net: %d\n",
6337
				__func__, dev->name, err);
6338
			BUG();
6339 6340 6341 6342 6343
		}
	}
	rtnl_unlock();
}

6344 6345 6346
static void __net_exit default_device_exit_batch(struct list_head *net_list)
{
	/* At exit all network devices most be removed from a network
6347
	 * namespace.  Do this in the reverse order of registration.
6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367
	 * 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();
}

6368
static struct pernet_operations __net_initdata default_device_ops = {
6369
	.exit = default_device_exit,
6370
	.exit_batch = default_device_exit_batch,
6371 6372
};

L
Linus Torvalds 已提交
6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392
/*
 *	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;

6393
	if (netdev_kobject_init())
L
Linus Torvalds 已提交
6394 6395 6396
		goto out;

	INIT_LIST_HEAD(&ptype_all);
6397
	for (i = 0; i < PTYPE_HASH_SIZE; i++)
L
Linus Torvalds 已提交
6398 6399
		INIT_LIST_HEAD(&ptype_base[i]);

6400 6401
	if (register_pernet_subsys(&netdev_net_ops))
		goto out;
L
Linus Torvalds 已提交
6402 6403 6404 6405 6406

	/*
	 *	Initialise the packet receive queues.
	 */

6407
	for_each_possible_cpu(i) {
E
Eric Dumazet 已提交
6408
		struct softnet_data *sd = &per_cpu(softnet_data, i);
L
Linus Torvalds 已提交
6409

C
Changli Gao 已提交
6410
		memset(sd, 0, sizeof(*sd));
E
Eric Dumazet 已提交
6411
		skb_queue_head_init(&sd->input_pkt_queue);
6412
		skb_queue_head_init(&sd->process_queue);
E
Eric Dumazet 已提交
6413 6414
		sd->completion_queue = NULL;
		INIT_LIST_HEAD(&sd->poll_list);
6415 6416
		sd->output_queue = NULL;
		sd->output_queue_tailp = &sd->output_queue;
E
Eric Dumazet 已提交
6417
#ifdef CONFIG_RPS
E
Eric Dumazet 已提交
6418 6419 6420 6421
		sd->csd.func = rps_trigger_softirq;
		sd->csd.info = sd;
		sd->csd.flags = 0;
		sd->cpu = i;
6422
#endif
T
Tom Herbert 已提交
6423

E
Eric Dumazet 已提交
6424 6425 6426 6427
		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
6428 6429 6430 6431
	}

	dev_boot_phase = 0;

6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446
	/* 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;

6447 6448
	open_softirq(NET_TX_SOFTIRQ, net_tx_action);
	open_softirq(NET_RX_SOFTIRQ, net_rx_action);
L
Linus Torvalds 已提交
6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459

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

subsys_initcall(net_dev_init);

6460 6461
static int __init initialize_hashrnd(void)
{
T
Tom Herbert 已提交
6462
	get_random_bytes(&hashrnd, sizeof(hashrnd));
6463 6464 6465 6466 6467
	return 0;
}

late_initcall_sync(initialize_hashrnd);