]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - include/linux/netdevice.h
net: resolve symbol conflicts with generic hashtable.h
[mirror_ubuntu-artful-kernel.git] / include / linux / netdevice.h
CommitLineData
1da177e4
LT
1/*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * Definitions for the Interfaces handler.
7 *
8 * Version: @(#)dev.h 1.0.10 08/12/93
9 *
02c30a84 10 * Authors: Ross Biro
1da177e4
LT
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Corey Minyard <wf-rch!minyard@relay.EU.net>
13 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 14 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
15 * Bjorn Ekwall. <bj0rn@blox.se>
16 * Pekka Riikonen <priikone@poseidon.pspt.fi>
17 *
18 * This program is free software; you can redistribute it and/or
19 * modify it under the terms of the GNU General Public License
20 * as published by the Free Software Foundation; either version
21 * 2 of the License, or (at your option) any later version.
22 *
23 * Moved to /usr/include/linux for NET3
24 */
25#ifndef _LINUX_NETDEVICE_H
26#define _LINUX_NETDEVICE_H
27
d7fe0f24 28#include <linux/timer.h>
187f1882 29#include <linux/bug.h>
bea3348e 30#include <linux/delay.h>
60063497 31#include <linux/atomic.h>
53511453 32#include <linux/prefetch.h>
1da177e4
LT
33#include <asm/cache.h>
34#include <asm/byteorder.h>
35
1da177e4 36#include <linux/percpu.h>
4d5b78c0 37#include <linux/rculist.h>
db217334 38#include <linux/dmaengine.h>
bea3348e 39#include <linux/workqueue.h>
114cf580 40#include <linux/dynamic_queue_limits.h>
1da177e4 41
b1b67dd4 42#include <linux/ethtool.h>
a050c33f 43#include <net/net_namespace.h>
cf85d08f 44#include <net/dsa.h>
7a6b6f51 45#ifdef CONFIG_DCB
2f90b865
AD
46#include <net/dcbnl.h>
47#endif
5bc1421e 48#include <net/netprio_cgroup.h>
a050c33f 49
a59e2ecb 50#include <linux/netdev_features.h>
77162022 51#include <linux/neighbour.h>
607ca46e 52#include <uapi/linux/netdevice.h>
61bd3857 53#include <uapi/linux/if_bonding.h>
e4c6734e 54#include <uapi/linux/pkt_cls.h>
a59e2ecb 55
115c1d6e 56struct netpoll_info;
313162d0 57struct device;
c1f19b51 58struct phy_device;
704232c2
JB
59/* 802.11 specific */
60struct wireless_dev;
98a18b6f
AA
61/* 802.15.4 specific */
62struct wpan_dev;
03c57747 63struct mpls_dev;
7c46a640
AD
64/* UDP Tunnel offloads */
65struct udp_tunnel_info;
a7862b45 66struct bpf_prog;
1da177e4 67
f629d208
JP
68void netdev_set_default_ethtool_ops(struct net_device *dev,
69 const struct ethtool_ops *ops);
d07d7507 70
9a1654ba
JP
71/* Backlog congestion levels */
72#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
73#define NET_RX_DROP 1 /* packet dropped */
74
572a9d7b
PM
75/*
76 * Transmit return codes: transmit return codes originate from three different
77 * namespaces:
78 *
79 * - qdisc return codes
80 * - driver transmit return codes
81 * - errno values
82 *
83 * Drivers are allowed to return any one of those in their hard_start_xmit()
84 * function. Real network devices commonly used with qdiscs should only return
85 * the driver transmit return codes though - when qdiscs are used, the actual
86 * transmission happens asynchronously, so the value is not propagated to
5e82b4b2
BH
87 * higher layers. Virtual network devices transmit synchronously; in this case
88 * the driver transmit return codes are consumed by dev_queue_xmit(), and all
572a9d7b
PM
89 * others are propagated to higher layers.
90 */
91
92/* qdisc ->enqueue() return codes. */
93#define NET_XMIT_SUCCESS 0x00
9a1654ba
JP
94#define NET_XMIT_DROP 0x01 /* skb dropped */
95#define NET_XMIT_CN 0x02 /* congestion notification */
9a1654ba 96#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
1da177e4 97
b9df3cb8
GR
98/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
99 * indicates that the device will soon be dropping packets, or already drops
100 * some packets of the same priority; prompting us to send less aggressively. */
572a9d7b 101#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
1da177e4
LT
102#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
103
dc1f8bf6 104/* Driver transmit return codes */
9a1654ba 105#define NETDEV_TX_MASK 0xf0
572a9d7b 106
dc1f8bf6 107enum netdev_tx {
572a9d7b 108 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
9a1654ba
JP
109 NETDEV_TX_OK = 0x00, /* driver took care of packet */
110 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
dc1f8bf6
SH
111};
112typedef enum netdev_tx netdev_tx_t;
113
9a1654ba
JP
114/*
115 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
116 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
117 */
118static inline bool dev_xmit_complete(int rc)
119{
120 /*
121 * Positive cases with an skb consumed by a driver:
122 * - successful transmission (rc == NETDEV_TX_OK)
123 * - error while transmitting (rc < 0)
124 * - error while queueing to a different device (rc & NET_XMIT_MASK)
125 */
126 if (likely(rc < NET_XMIT_MASK))
127 return true;
128
129 return false;
130}
131
1da177e4 132/*
5e82b4b2 133 * Compute the worst-case header length according to the protocols
1da177e4
LT
134 * used.
135 */
fe2918b0 136
c0eb4540
KS
137#if defined(CONFIG_HYPERV_NET)
138# define LL_MAX_HEADER 128
139#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
8388e3da
DM
140# if defined(CONFIG_MAC80211_MESH)
141# define LL_MAX_HEADER 128
142# else
143# define LL_MAX_HEADER 96
144# endif
1da177e4 145#else
8388e3da 146# define LL_MAX_HEADER 32
1da177e4
LT
147#endif
148
d11ead75
BH
149#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
150 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
1da177e4
LT
151#define MAX_HEADER LL_MAX_HEADER
152#else
153#define MAX_HEADER (LL_MAX_HEADER + 48)
154#endif
155
156/*
be1f3c2c
BH
157 * Old network device statistics. Fields are native words
158 * (unsigned long) so they can be read and written atomically.
1da177e4 159 */
fe2918b0 160
d94d9fee 161struct net_device_stats {
3cfde79c
BH
162 unsigned long rx_packets;
163 unsigned long tx_packets;
164 unsigned long rx_bytes;
165 unsigned long tx_bytes;
166 unsigned long rx_errors;
167 unsigned long tx_errors;
168 unsigned long rx_dropped;
169 unsigned long tx_dropped;
170 unsigned long multicast;
1da177e4 171 unsigned long collisions;
1da177e4 172 unsigned long rx_length_errors;
3cfde79c
BH
173 unsigned long rx_over_errors;
174 unsigned long rx_crc_errors;
175 unsigned long rx_frame_errors;
176 unsigned long rx_fifo_errors;
177 unsigned long rx_missed_errors;
1da177e4
LT
178 unsigned long tx_aborted_errors;
179 unsigned long tx_carrier_errors;
180 unsigned long tx_fifo_errors;
181 unsigned long tx_heartbeat_errors;
182 unsigned long tx_window_errors;
1da177e4
LT
183 unsigned long rx_compressed;
184 unsigned long tx_compressed;
185};
186
1da177e4
LT
187
188#include <linux/cache.h>
189#include <linux/skbuff.h>
190
adc9300e 191#ifdef CONFIG_RPS
c5905afb
IM
192#include <linux/static_key.h>
193extern struct static_key rps_needed;
adc9300e
ED
194#endif
195
1da177e4
LT
196struct neighbour;
197struct neigh_parms;
198struct sk_buff;
199
f001fde5
JP
200struct netdev_hw_addr {
201 struct list_head list;
202 unsigned char addr[MAX_ADDR_LEN];
203 unsigned char type;
ccffad25
JP
204#define NETDEV_HW_ADDR_T_LAN 1
205#define NETDEV_HW_ADDR_T_SAN 2
206#define NETDEV_HW_ADDR_T_SLAVE 3
207#define NETDEV_HW_ADDR_T_UNICAST 4
22bedad3 208#define NETDEV_HW_ADDR_T_MULTICAST 5
22bedad3 209 bool global_use;
4cd729b0 210 int sync_cnt;
8f8f103d 211 int refcount;
4543fbef 212 int synced;
f001fde5
JP
213 struct rcu_head rcu_head;
214};
215
31278e71
JP
216struct netdev_hw_addr_list {
217 struct list_head list;
218 int count;
219};
220
22bedad3
JP
221#define netdev_hw_addr_list_count(l) ((l)->count)
222#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
223#define netdev_hw_addr_list_for_each(ha, l) \
224 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 225
22bedad3
JP
226#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
227#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
228#define netdev_for_each_uc_addr(ha, dev) \
229 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
6683ece3 230
22bedad3
JP
231#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
232#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 233#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 234 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
6683ece3 235
d94d9fee 236struct hh_cache {
f6b72b62 237 u16 hh_len;
5c25f686 238 u16 __pad;
3644f0ce 239 seqlock_t hh_lock;
1da177e4
LT
240
241 /* cached hardware header; allow for machine alignment needs. */
242#define HH_DATA_MOD 16
243#define HH_DATA_OFF(__len) \
5ba0eac6 244 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
1da177e4
LT
245#define HH_DATA_ALIGN(__len) \
246 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
247 unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
248};
249
5e82b4b2 250/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
1da177e4
LT
251 * Alternative is:
252 * dev->hard_header_len ? (dev->hard_header_len +
253 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
254 *
255 * We could use other alignment values, but we must maintain the
256 * relationship HH alignment <= LL alignment.
257 */
258#define LL_RESERVED_SPACE(dev) \
f5184d26 259 ((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 260#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
f5184d26 261 ((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 262
3b04ddde
SH
263struct header_ops {
264 int (*create) (struct sk_buff *skb, struct net_device *dev,
265 unsigned short type, const void *daddr,
95c96174 266 const void *saddr, unsigned int len);
3b04ddde 267 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
e69dd336 268 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
3b04ddde
SH
269 void (*cache_update)(struct hh_cache *hh,
270 const struct net_device *dev,
271 const unsigned char *haddr);
2793a23a 272 bool (*validate)(const char *ll_header, unsigned int len);
3b04ddde
SH
273};
274
1da177e4 275/* These flag bits are private to the generic network queueing
5e82b4b2 276 * layer; they may not be explicitly referenced by any other
1da177e4
LT
277 * code.
278 */
279
d94d9fee 280enum netdev_state_t {
1da177e4
LT
281 __LINK_STATE_START,
282 __LINK_STATE_PRESENT,
1da177e4 283 __LINK_STATE_NOCARRIER,
b00055aa
SR
284 __LINK_STATE_LINKWATCH_PENDING,
285 __LINK_STATE_DORMANT,
1da177e4
LT
286};
287
288
289/*
5e82b4b2 290 * This structure holds boot-time configured netdevice settings. They
fe2918b0 291 * are then used in the device probing.
1da177e4
LT
292 */
293struct netdev_boot_setup {
294 char name[IFNAMSIZ];
295 struct ifmap map;
296};
297#define NETDEV_BOOT_SETUP_MAX 8
298
f629d208 299int __init netdev_boot_setup(char *str);
1da177e4 300
bea3348e
SH
301/*
302 * Structure for NAPI scheduling similar to tasklet but with weighting
303 */
304struct napi_struct {
305 /* The poll_list must only be managed by the entity which
306 * changes the state of the NAPI_STATE_SCHED bit. This means
307 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 308 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
309 * can remove from the list right before clearing the bit.
310 */
311 struct list_head poll_list;
312
313 unsigned long state;
314 int weight;
404f7c9e 315 unsigned int gro_count;
bea3348e
SH
316 int (*poll)(struct napi_struct *, int);
317#ifdef CONFIG_NETPOLL
318 spinlock_t poll_lock;
319 int poll_owner;
bea3348e 320#endif
5d38a079 321 struct net_device *dev;
d565b0a1 322 struct sk_buff *gro_list;
5d38a079 323 struct sk_buff *skb;
3b47d303 324 struct hrtimer timer;
404f7c9e 325 struct list_head dev_list;
af12fa6e
ET
326 struct hlist_node napi_hash_node;
327 unsigned int napi_id;
bea3348e
SH
328};
329
d94d9fee 330enum {
bea3348e 331 NAPI_STATE_SCHED, /* Poll is scheduled */
a0a46196 332 NAPI_STATE_DISABLE, /* Disable pending */
7b363e44 333 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
d64b5e85
ED
334 NAPI_STATE_HASHED, /* In NAPI hash (busy polling possible) */
335 NAPI_STATE_NO_BUSY_POLL,/* Do not add in napi_hash, no busy polling */
bea3348e
SH
336};
337
5b252f0c 338enum gro_result {
d1c76af9
HX
339 GRO_MERGED,
340 GRO_MERGED_FREE,
341 GRO_HELD,
342 GRO_NORMAL,
343 GRO_DROP,
344};
5b252f0c 345typedef enum gro_result gro_result_t;
d1c76af9 346
8a4eb573
JP
347/*
348 * enum rx_handler_result - Possible return values for rx_handlers.
349 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
350 * further.
351 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
352 * case skb->dev was changed by rx_handler.
353 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 354 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
355 *
356 * rx_handlers are functions called from inside __netif_receive_skb(), to do
357 * special processing of the skb, prior to delivery to protocol handlers.
358 *
359 * Currently, a net_device can only have a single rx_handler registered. Trying
360 * to register a second rx_handler will return -EBUSY.
361 *
362 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
363 * To unregister a rx_handler on a net_device, use
364 * netdev_rx_handler_unregister().
365 *
366 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
367 * do with the skb.
368 *
5e82b4b2 369 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 370 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 371 * the skb to be delivered in some other way.
8a4eb573
JP
372 *
373 * If the rx_handler changed skb->dev, to divert the skb to another
374 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
375 * new device will be called if it exists.
376 *
5e82b4b2 377 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 378 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 379 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 380 *
5e82b4b2 381 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
382 * delivered, it should return RX_HANDLER_PASS.
383 *
384 * A device without a registered rx_handler will behave as if rx_handler
385 * returned RX_HANDLER_PASS.
386 */
387
388enum rx_handler_result {
389 RX_HANDLER_CONSUMED,
390 RX_HANDLER_ANOTHER,
391 RX_HANDLER_EXACT,
392 RX_HANDLER_PASS,
393};
394typedef enum rx_handler_result rx_handler_result_t;
395typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 396
f629d208 397void __napi_schedule(struct napi_struct *n);
bc9ad166 398void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 399
4d29515f 400static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
401{
402 return test_bit(NAPI_STATE_DISABLE, &n->state);
403}
404
bea3348e 405/**
5e82b4b2
BH
406 * napi_schedule_prep - check if NAPI can be scheduled
407 * @n: NAPI context
bea3348e
SH
408 *
409 * Test if NAPI routine is already running, and if not mark
5e82b4b2 410 * it as running. This is used as a condition variable to
a0a46196
DM
411 * insure only one NAPI poll instance runs. We also make
412 * sure there is no pending NAPI disable.
bea3348e 413 */
4d29515f 414static inline bool napi_schedule_prep(struct napi_struct *n)
bea3348e 415{
a0a46196
DM
416 return !napi_disable_pending(n) &&
417 !test_and_set_bit(NAPI_STATE_SCHED, &n->state);
bea3348e
SH
418}
419
420/**
421 * napi_schedule - schedule NAPI poll
5e82b4b2 422 * @n: NAPI context
bea3348e
SH
423 *
424 * Schedule NAPI poll routine to be called if it is not already
425 * running.
426 */
427static inline void napi_schedule(struct napi_struct *n)
428{
429 if (napi_schedule_prep(n))
430 __napi_schedule(n);
431}
432
bc9ad166
ED
433/**
434 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 435 * @n: NAPI context
bc9ad166
ED
436 *
437 * Variant of napi_schedule(), assuming hard irqs are masked.
438 */
439static inline void napi_schedule_irqoff(struct napi_struct *n)
440{
441 if (napi_schedule_prep(n))
442 __napi_schedule_irqoff(n);
443}
444
bfe13f54 445/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 446static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
447{
448 if (napi_schedule_prep(napi)) {
449 __napi_schedule(napi);
4d29515f 450 return true;
bfe13f54 451 }
4d29515f 452 return false;
bfe13f54
RD
453}
454
3b47d303
ED
455void __napi_complete(struct napi_struct *n);
456void napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
457/**
458 * napi_complete - NAPI processing complete
5e82b4b2 459 * @n: NAPI context
bea3348e
SH
460 *
461 * Mark NAPI processing as complete.
3b47d303 462 * Consider using napi_complete_done() instead.
bea3348e 463 */
3b47d303
ED
464static inline void napi_complete(struct napi_struct *n)
465{
466 return napi_complete_done(n, 0);
467}
bea3348e 468
af12fa6e
ET
469/**
470 * napi_hash_add - add a NAPI to global hashtable
5e82b4b2 471 * @napi: NAPI context
af12fa6e 472 *
5e82b4b2
BH
473 * Generate a new napi_id and store a @napi under it in napi_hash.
474 * Used for busy polling (CONFIG_NET_RX_BUSY_POLL).
93d05d4a 475 * Note: This is normally automatically done from netif_napi_add(),
5e82b4b2 476 * so might disappear in a future Linux version.
af12fa6e 477 */
f629d208 478void napi_hash_add(struct napi_struct *napi);
af12fa6e
ET
479
480/**
481 * napi_hash_del - remove a NAPI from global table
5e82b4b2 482 * @napi: NAPI context
af12fa6e 483 *
5e82b4b2 484 * Warning: caller must observe RCU grace period
34cbe27e
ED
485 * before freeing memory containing @napi, if
486 * this function returns true.
93d05d4a 487 * Note: core networking stack automatically calls it
5e82b4b2 488 * from netif_napi_del().
93d05d4a 489 * Drivers might want to call this helper to combine all
5e82b4b2 490 * the needed RCU grace periods into a single one.
af12fa6e 491 */
34cbe27e 492bool napi_hash_del(struct napi_struct *napi);
af12fa6e 493
bea3348e
SH
494/**
495 * napi_disable - prevent NAPI from scheduling
5e82b4b2 496 * @n: NAPI context
bea3348e
SH
497 *
498 * Stop NAPI from being scheduled on this context.
499 * Waits till any outstanding processing completes.
500 */
3b47d303 501void napi_disable(struct napi_struct *n);
bea3348e
SH
502
503/**
504 * napi_enable - enable NAPI scheduling
5e82b4b2 505 * @n: NAPI context
bea3348e
SH
506 *
507 * Resume NAPI from being scheduled on this context.
508 * Must be paired with napi_disable.
509 */
510static inline void napi_enable(struct napi_struct *n)
511{
512 BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
4e857c58 513 smp_mb__before_atomic();
bea3348e 514 clear_bit(NAPI_STATE_SCHED, &n->state);
2d8bff12 515 clear_bit(NAPI_STATE_NPSVC, &n->state);
bea3348e
SH
516}
517
c264c3de
SH
518/**
519 * napi_synchronize - wait until NAPI is not running
5e82b4b2 520 * @n: NAPI context
c264c3de
SH
521 *
522 * Wait until NAPI is done being scheduled on this context.
523 * Waits till any outstanding processing completes but
524 * does not disable future activations.
525 */
526static inline void napi_synchronize(const struct napi_struct *n)
527{
facc432f
AB
528 if (IS_ENABLED(CONFIG_SMP))
529 while (test_bit(NAPI_STATE_SCHED, &n->state))
530 msleep(1);
531 else
532 barrier();
c264c3de 533}
c264c3de 534
d94d9fee 535enum netdev_queue_state_t {
73466498
TH
536 __QUEUE_STATE_DRV_XOFF,
537 __QUEUE_STATE_STACK_XOFF,
c3f26a26 538 __QUEUE_STATE_FROZEN,
79d16385 539};
8e2f1a63
DB
540
541#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
542#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
543#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
544
545#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
546#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
547 QUEUE_STATE_FROZEN)
548#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
549 QUEUE_STATE_FROZEN)
550
73466498
TH
551/*
552 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
553 * netif_tx_* functions below are used to manipulate this flag. The
554 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
555 * queue independently. The netif_xmit_*stopped functions below are called
556 * to check if the queue has been stopped by the driver or stack (either
557 * of the XOFF bits are set in the state). Drivers should not need to call
558 * netif_xmit*stopped functions, they should only be using netif_tx_*.
559 */
79d16385 560
bb949fbd 561struct netdev_queue {
6a321cb3 562/*
5e82b4b2 563 * read-mostly part
6a321cb3 564 */
bb949fbd 565 struct net_device *dev;
46e5da40 566 struct Qdisc __rcu *qdisc;
b0e1e646 567 struct Qdisc *qdisc_sleeping;
ccf5ff69 568#ifdef CONFIG_SYSFS
1d24eb48
TH
569 struct kobject kobj;
570#endif
f2cd2d3e
ED
571#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
572 int numa_node;
573#endif
c0ef079c
FW
574 unsigned long tx_maxrate;
575 /*
576 * Number of TX timeouts for this queue
577 * (/sys/class/net/DEV/Q/trans_timeout)
578 */
579 unsigned long trans_timeout;
6a321cb3 580/*
5e82b4b2 581 * write-mostly part
6a321cb3
ED
582 */
583 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
584 int xmit_lock_owner;
9d21493b 585 /*
9b36627a 586 * Time (in jiffies) of last Tx
9d21493b
ED
587 */
588 unsigned long trans_start;
ccf5ff69 589
114cf580
TH
590 unsigned long state;
591
592#ifdef CONFIG_BQL
593 struct dql dql;
594#endif
e8a0464c 595} ____cacheline_aligned_in_smp;
bb949fbd 596
f2cd2d3e
ED
597static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
598{
599#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
600 return q->numa_node;
601#else
b236da69 602 return NUMA_NO_NODE;
f2cd2d3e
ED
603#endif
604}
605
606static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
607{
608#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
609 q->numa_node = node;
610#endif
611}
612
df334545 613#ifdef CONFIG_RPS
0a9627f2
TH
614/*
615 * This structure holds an RPS map which can be of variable length. The
616 * map is an array of CPUs.
617 */
618struct rps_map {
619 unsigned int len;
620 struct rcu_head rcu;
621 u16 cpus[0];
622};
60b778ce 623#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 624
fec5e652 625/*
c445477d
BH
626 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
627 * tail pointer for that CPU's input queue at the time of last enqueue, and
628 * a hardware filter index.
fec5e652
TH
629 */
630struct rps_dev_flow {
631 u16 cpu;
c445477d 632 u16 filter;
fec5e652
TH
633 unsigned int last_qtail;
634};
c445477d 635#define RPS_NO_FILTER 0xffff
fec5e652
TH
636
637/*
638 * The rps_dev_flow_table structure contains a table of flow mappings.
639 */
640struct rps_dev_flow_table {
641 unsigned int mask;
642 struct rcu_head rcu;
fec5e652
TH
643 struct rps_dev_flow flows[0];
644};
645#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 646 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
647
648/*
649 * The rps_sock_flow_table contains mappings of flows to the last CPU
650 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
651 * Each entry is a 32bit value. Upper part is the high-order bits
652 * of flow hash, lower part is CPU number.
567e4b79 653 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
654 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
655 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 656 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
657 */
658struct rps_sock_flow_table {
567e4b79 659 u32 mask;
93c1af6c
ED
660
661 u32 ents[0] ____cacheline_aligned_in_smp;
fec5e652 662};
567e4b79 663#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
664
665#define RPS_NO_CPU 0xffff
666
567e4b79
ED
667extern u32 rps_cpu_mask;
668extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
669
fec5e652
TH
670static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
671 u32 hash)
672{
673 if (table && hash) {
567e4b79
ED
674 unsigned int index = hash & table->mask;
675 u32 val = hash & ~rps_cpu_mask;
fec5e652 676
5e82b4b2 677 /* We only give a hint, preemption can change CPU under us */
567e4b79 678 val |= raw_smp_processor_id();
fec5e652 679
567e4b79
ED
680 if (table->ents[index] != val)
681 table->ents[index] = val;
fec5e652
TH
682 }
683}
684
c445477d 685#ifdef CONFIG_RFS_ACCEL
f629d208
JP
686bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
687 u16 filter_id);
c445477d 688#endif
a953be53 689#endif /* CONFIG_RPS */
c445477d 690
0a9627f2
TH
691/* This structure contains an instance of an RX queue. */
692struct netdev_rx_queue {
a953be53 693#ifdef CONFIG_RPS
6e3f7faf
ED
694 struct rps_map __rcu *rps_map;
695 struct rps_dev_flow_table __rcu *rps_flow_table;
a953be53 696#endif
6e3f7faf 697 struct kobject kobj;
fe822240 698 struct net_device *dev;
0a9627f2 699} ____cacheline_aligned_in_smp;
a953be53
MD
700
701/*
702 * RX queue sysfs structures and functions.
703 */
704struct rx_queue_attribute {
705 struct attribute attr;
706 ssize_t (*show)(struct netdev_rx_queue *queue,
707 struct rx_queue_attribute *attr, char *buf);
708 ssize_t (*store)(struct netdev_rx_queue *queue,
709 struct rx_queue_attribute *attr, const char *buf, size_t len);
710};
d314774c 711
bf264145
TH
712#ifdef CONFIG_XPS
713/*
714 * This structure holds an XPS map which can be of variable length. The
715 * map is an array of queues.
716 */
717struct xps_map {
718 unsigned int len;
719 unsigned int alloc_len;
720 struct rcu_head rcu;
721 u16 queues[0];
722};
60b778ce 723#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
724#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
725 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
726
727/*
728 * This structure holds all XPS maps for device. Maps are indexed by CPU.
729 */
730struct xps_dev_maps {
731 struct rcu_head rcu;
a4177869 732 struct xps_map __rcu *cpu_map[0];
bf264145
TH
733};
734#define XPS_DEV_MAPS_SIZE (sizeof(struct xps_dev_maps) + \
735 (nr_cpu_ids * sizeof(struct xps_map *)))
736#endif /* CONFIG_XPS */
737
4f57c087
JF
738#define TC_MAX_QUEUE 16
739#define TC_BITMASK 15
740/* HW offloaded queuing disciplines txq count and offset maps */
741struct netdev_tc_txq {
742 u16 count;
743 u16 offset;
744};
745
68bad94e
NP
746#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
747/*
748 * This structure is to hold information about the device
749 * configured to run FCoE protocol stack.
750 */
751struct netdev_fcoe_hbainfo {
752 char manufacturer[64];
753 char serial_number[64];
754 char hardware_version[64];
755 char driver_version[64];
756 char optionrom_version[64];
757 char firmware_version[64];
758 char model[256];
759 char model_description[256];
760};
761#endif
762
02637fce 763#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 764
02637fce
JP
765/* This structure holds a unique identifier to identify some
766 * physical item (port for example) used by a netdevice.
66b52b0d 767 */
02637fce
JP
768struct netdev_phys_item_id {
769 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
770 unsigned char id_len;
771};
772
d754f98b
SF
773static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
774 struct netdev_phys_item_id *b)
775{
776 return a->id_len == b->id_len &&
777 memcmp(a->id, b->id, a->id_len) == 0;
778}
779
99932d4f
DB
780typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
781 struct sk_buff *skb);
782
a1b7c5fd 783/* These structures hold the attributes of qdisc and classifiers
16e5cc64
JF
784 * that are being passed to the netdevice through the setup_tc op.
785 */
786enum {
787 TC_SETUP_MQPRIO,
a1b7c5fd 788 TC_SETUP_CLSU32,
5b33f488 789 TC_SETUP_CLSFLOWER,
b87f7936 790 TC_SETUP_MATCHALL,
16e5cc64
JF
791};
792
a1b7c5fd
JF
793struct tc_cls_u32_offload;
794
16e5cc64
JF
795struct tc_to_netdev {
796 unsigned int type;
797 union {
798 u8 tc;
a1b7c5fd 799 struct tc_cls_u32_offload *cls_u32;
5b33f488 800 struct tc_cls_flower_offload *cls_flower;
b87f7936 801 struct tc_cls_matchall_offload *cls_mall;
16e5cc64
JF
802 };
803};
804
a7862b45
BB
805/* These structures hold the attributes of xdp state that are being passed
806 * to the netdevice through the xdp op.
807 */
808enum xdp_netdev_command {
809 /* Set or clear a bpf program used in the earliest stages of packet
810 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
811 * is responsible for calling bpf_prog_put on any old progs that are
812 * stored. In case of error, the callee need not release the new prog
813 * reference, but on success it takes ownership and must bpf_prog_put
814 * when it is no longer used.
815 */
816 XDP_SETUP_PROG,
817 /* Check if a bpf program is set on the device. The callee should
818 * return true if a program is currently attached and running.
819 */
820 XDP_QUERY_PROG,
821};
822
823struct netdev_xdp {
824 enum xdp_netdev_command command;
825 union {
826 /* XDP_SETUP_PROG */
827 struct bpf_prog *prog;
828 /* XDP_QUERY_PROG */
829 bool prog_attached;
830 };
831};
16e5cc64 832
d314774c
SH
833/*
834 * This structure defines the management hooks for network devices.
00829823
SH
835 * The following hooks can be defined; unless noted otherwise, they are
836 * optional and can be filled with a null pointer.
d314774c
SH
837 *
838 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
839 * This function is called once when a network device is registered.
840 * The network device can use this for any late stage initialization
841 * or semantic validation. It can fail with an error code which will
842 * be propagated back to register_netdev.
d314774c
SH
843 *
844 * void (*ndo_uninit)(struct net_device *dev);
845 * This function is called when device is unregistered or when registration
846 * fails. It is not called if init fails.
847 *
848 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 849 * This function is called when a network device transitions to the up
d314774c
SH
850 * state.
851 *
852 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 853 * This function is called when a network device transitions to the down
d314774c
SH
854 * state.
855 *
dc1f8bf6
SH
856 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
857 * struct net_device *dev);
00829823 858 * Called when a packet needs to be transmitted.
e79d8429
RR
859 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
860 * the queue before that can happen; it's for obsolete devices and weird
861 * corner cases, but the stack really does a non-trivial amount
862 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 863 * Required; cannot be NULL.
00829823 864 *
cdba756f
ED
865 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
866 * netdev_features_t features);
867 * Adjusts the requested feature flags according to device-specific
868 * constraints, and returns the resulting flags. Must not modify
869 * the device state.
870 *
f663dd9a 871 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
99932d4f 872 * void *accel_priv, select_queue_fallback_t fallback);
5e82b4b2 873 * Called to decide which queue to use when device supports multiple
00829823
SH
874 * transmit queues.
875 *
d314774c
SH
876 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
877 * This function is called to allow device receiver to make
5e82b4b2 878 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
879 *
880 * void (*ndo_set_rx_mode)(struct net_device *dev);
881 * This function is called device changes address list filtering.
01789349 882 * If driver handles unicast address filtering, it should set
5e82b4b2 883 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
884 *
885 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
886 * This function is called when the Media Access Control address
37b607c5 887 * needs to be changed. If this interface is not defined, the
5e82b4b2 888 * MAC address can not be changed.
d314774c
SH
889 *
890 * int (*ndo_validate_addr)(struct net_device *dev);
891 * Test if Media Access Control address is valid for the device.
892 *
893 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
5e82b4b2
BH
894 * Called when a user requests an ioctl which can't be handled by
895 * the generic interface code. If not defined ioctls return
d314774c
SH
896 * not supported error code.
897 *
898 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
899 * Used to set network devices bus interface parameters. This interface
5e82b4b2 900 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
901 * interface (PCI) for low level management.
902 *
903 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
904 * Called when a user wants to change the Maximum Transfer Unit
905 * of a device. If not defined, any request to change MTU will
906 * will return an error.
907 *
00829823 908 * void (*ndo_tx_timeout)(struct net_device *dev);
5e82b4b2 909 * Callback used when the transmitter has not made any progress
d314774c
SH
910 * for dev->watchdog ticks.
911 *
3cfde79c 912 * struct rtnl_link_stats64* (*ndo_get_stats64)(struct net_device *dev,
28172739 913 * struct rtnl_link_stats64 *storage);
d308e38f 914 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 915 * Called when a user wants to get the network device usage
be1f3c2c 916 * statistics. Drivers must do one of the following:
3cfde79c
BH
917 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
918 * rtnl_link_stats64 structure passed by the caller.
82695d9b 919 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
920 * (which should normally be dev->stats) and return a pointer to
921 * it. The structure may be changed asynchronously only if each
922 * field is written atomically.
923 * 3. Update dev->stats asynchronously and atomically, and define
924 * neither operation.
d314774c 925 *
5d632cb7 926 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 927 * If device supports VLAN filtering this function is called when a
80d5c368 928 * VLAN id is registered.
d314774c 929 *
5d632cb7 930 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 931 * If device supports VLAN filtering this function is called when a
80d5c368 932 * VLAN id is unregistered.
d314774c
SH
933 *
934 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
935 *
936 * SR-IOV management functions.
937 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
938 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan, u8 qos);
ed616689
SC
939 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
940 * int max_tx_rate);
5f8444a3 941 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 942 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
943 * int (*ndo_get_vf_config)(struct net_device *dev,
944 * int vf, struct ifla_vf_info *ivf);
1d8faf48 945 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
946 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
947 * struct nlattr *port[]);
01a3d796
VZ
948 *
949 * Enable or disable the VF ability to query its RSS Redirection Table and
950 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 951 * with PF and querying it may introduce a theoretical security risk.
01a3d796 952 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 953 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
4f57c087
JF
954 * int (*ndo_setup_tc)(struct net_device *dev, u8 tc)
955 * Called to setup 'tc' number of traffic classes in the net device. This
956 * is always called from the stack with the rtnl lock held and netif tx
957 * queues stopped. This allows the netdevice to perform queue management
958 * safely.
c445477d 959 *
e9bce845
YZ
960 * Fiber Channel over Ethernet (FCoE) offload functions.
961 * int (*ndo_fcoe_enable)(struct net_device *dev);
962 * Called when the FCoE protocol stack wants to start using LLD for FCoE
963 * so the underlying device can perform whatever needed configuration or
964 * initialization to support acceleration of FCoE traffic.
965 *
966 * int (*ndo_fcoe_disable)(struct net_device *dev);
967 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
968 * so the underlying device can perform whatever needed clean-ups to
969 * stop supporting acceleration of FCoE traffic.
970 *
971 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
972 * struct scatterlist *sgl, unsigned int sgc);
973 * Called when the FCoE Initiator wants to initialize an I/O that
974 * is a possible candidate for Direct Data Placement (DDP). The LLD can
975 * perform necessary setup and returns 1 to indicate the device is set up
976 * successfully to perform DDP on this I/O, otherwise this returns 0.
977 *
978 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
979 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
980 * indicated by the FC exchange id 'xid', so the underlying device can
981 * clean up and reuse resources for later DDP requests.
982 *
983 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
984 * struct scatterlist *sgl, unsigned int sgc);
985 * Called when the FCoE Target wants to initialize an I/O that
986 * is a possible candidate for Direct Data Placement (DDP). The LLD can
987 * perform necessary setup and returns 1 to indicate the device is set up
988 * successfully to perform DDP on this I/O, otherwise this returns 0.
989 *
68bad94e
NP
990 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
991 * struct netdev_fcoe_hbainfo *hbainfo);
992 * Called when the FCoE Protocol stack wants information on the underlying
993 * device. This information is utilized by the FCoE protocol stack to
994 * register attributes with Fiber Channel management service as per the
995 * FC-GS Fabric Device Management Information(FDMI) specification.
996 *
e9bce845
YZ
997 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
998 * Called when the underlying device wants to override default World Wide
999 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1000 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1001 * protocol stack to use.
1002 *
c445477d
BH
1003 * RFS acceleration.
1004 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1005 * u16 rxq_index, u32 flow_id);
1006 * Set hardware filter for RFS. rxq_index is the target queue index;
1007 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1008 * Return the filter ID on success, or a negative error code.
fbaec0ea 1009 *
8b98a70c 1010 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1011 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1012 * Called to make another netdev an underling.
1013 *
1014 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1015 * Called to release previously enslaved netdev.
5455c699
MM
1016 *
1017 * Feature/offload setting functions.
c8f44aff 1018 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1019 * Called to update device configuration to new features. Passed
1020 * feature set might be less than what was returned by ndo_fix_features()).
1021 * Must return >0 or -errno if it changed dev->features itself.
1022 *
edc7d573 1023 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1024 * struct net_device *dev,
f6f6424b 1025 * const unsigned char *addr, u16 vid, u16 flags)
77162022 1026 * Adds an FDB entry to dev for addr.
1690be63
VY
1027 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1028 * struct net_device *dev,
f6f6424b 1029 * const unsigned char *addr, u16 vid)
77162022
JF
1030 * Deletes the FDB entry from dev coresponding to addr.
1031 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3
JHS
1032 * struct net_device *dev, struct net_device *filter_dev,
1033 * int idx)
77162022
JF
1034 * Used to add FDB entries to dump requests. Implementers should add
1035 * entries to skb and update idx with the number of entries.
e5a55a89 1036 *
ad41faa8
ND
1037 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
1038 * u16 flags)
e5a55a89 1039 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1040 * struct net_device *dev, u32 filter_mask,
1041 * int nlflags)
ad41faa8
ND
1042 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1043 * u16 flags);
4bf84c35
JP
1044 *
1045 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1046 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1047 * which do not represent real hardware may define this to allow their
1048 * userspace components to manage their virtual carrier state. Devices
1049 * that determine carrier state from physical hardware properties (eg
1050 * network cables) or protocol-dependent mechanisms (eg
1051 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1052 *
1053 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1054 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1055 * Called to get ID of physical port of this device. If driver does
1056 * not implement this, it is assumed that the hw is not able to have
1057 * multiple net devices on single physical port.
53cf5275 1058 *
7c46a640
AD
1059 * void (*ndo_udp_tunnel_add)(struct net_device *dev,
1060 * struct udp_tunnel_info *ti);
1061 * Called by UDP tunnel to notify a driver about the UDP port and socket
1062 * address family that a UDP tunnel is listnening to. It is called only
1063 * when a new port starts listening. The operation is protected by the
1064 * RTNL.
1065 *
1066 * void (*ndo_udp_tunnel_del)(struct net_device *dev,
1067 * struct udp_tunnel_info *ti);
1068 * Called by UDP tunnel to notify the driver about a UDP port and socket
1069 * address family that the UDP tunnel is not listening to anymore. The
1070 * operation is protected by the RTNL.
1071 *
a6cc0cfa
JF
1072 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1073 * struct net_device *dev)
1074 * Called by upper layer devices to accelerate switching or other
1075 * station functionality into hardware. 'pdev is the lowerdev
1076 * to use for the offload and 'dev' is the net device that will
1077 * back the offload. Returns a pointer to the private structure
1078 * the upper layer will maintain.
1079 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1080 * Called by upper layer device to delete the station created
1081 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1082 * the station and priv is the structure returned by the add
1083 * operation.
1084 * netdev_tx_t (*ndo_dfwd_start_xmit)(struct sk_buff *skb,
1085 * struct net_device *dev,
1086 * void *priv);
1087 * Callback to use for xmit over the accelerated station. This
1088 * is used in place of ndo_start_xmit on accelerated net
1089 * devices.
5e82b4b2
BH
1090 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1091 * struct net_device *dev
1092 * netdev_features_t features);
04ffcb25 1093 * Called by core transmit path to determine if device is capable of
5f35227e
JG
1094 * performing offload operations on a given packet. This is to give
1095 * the device an opportunity to implement any restrictions that cannot
1096 * be otherwise expressed by feature flags. The check is called with
1097 * the set of features that the stack has calculated and it returns
1098 * those the driver believes to be appropriate.
822b3b2e
JF
1099 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1100 * int queue_index, u32 maxrate);
1101 * Called when a user wants to set a max-rate limitation of specific
1102 * TX queue.
a54acb3a
ND
1103 * int (*ndo_get_iflink)(const struct net_device *dev);
1104 * Called to get the iflink value of this device.
d746d707 1105 * void (*ndo_change_proto_down)(struct net_device *dev,
5e82b4b2 1106 * bool proto_down);
d746d707
AK
1107 * This function is used to pass protocol port error state information
1108 * to the switch driver. The switch driver can react to the proto_down
1109 * by doing a phys down on the associated switch port.
fc4099f1
PS
1110 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1111 * This function is used to get egress tunnel information for given skb.
1112 * This is useful for retrieving outer tunnel header parameters while
1113 * sampling packet.
871b642a
PA
1114 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1115 * This function is used to specify the headroom that the skb must
1116 * consider when allocation skb during packet reception. Setting
1117 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1118 * forward. Setting a negative value resets the rx headroom to the
871b642a 1119 * default value.
a7862b45
BB
1120 * int (*ndo_xdp)(struct net_device *dev, struct netdev_xdp *xdp);
1121 * This function is used to set or query state related to XDP on the
1122 * netdevice. See definition of enum xdp_netdev_command for details.
d746d707 1123 *
d314774c
SH
1124 */
1125struct net_device_ops {
1126 int (*ndo_init)(struct net_device *dev);
1127 void (*ndo_uninit)(struct net_device *dev);
1128 int (*ndo_open)(struct net_device *dev);
1129 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1130 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1131 struct net_device *dev);
1132 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1133 struct net_device *dev,
1134 netdev_features_t features);
00829823 1135 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1136 struct sk_buff *skb,
99932d4f
DB
1137 void *accel_priv,
1138 select_queue_fallback_t fallback);
d314774c
SH
1139 void (*ndo_change_rx_flags)(struct net_device *dev,
1140 int flags);
d314774c 1141 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1142 int (*ndo_set_mac_address)(struct net_device *dev,
1143 void *addr);
d314774c 1144 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1145 int (*ndo_do_ioctl)(struct net_device *dev,
1146 struct ifreq *ifr, int cmd);
d314774c
SH
1147 int (*ndo_set_config)(struct net_device *dev,
1148 struct ifmap *map);
00829823
SH
1149 int (*ndo_change_mtu)(struct net_device *dev,
1150 int new_mtu);
1151 int (*ndo_neigh_setup)(struct net_device *dev,
1152 struct neigh_parms *);
d314774c
SH
1153 void (*ndo_tx_timeout) (struct net_device *dev);
1154
28172739
ED
1155 struct rtnl_link_stats64* (*ndo_get_stats64)(struct net_device *dev,
1156 struct rtnl_link_stats64 *storage);
d314774c
SH
1157 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1158
8e586137 1159 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1160 __be16 proto, u16 vid);
8e586137 1161 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1162 __be16 proto, u16 vid);
d314774c 1163#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1164 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1165 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1166 struct netpoll_info *info);
0e34e931 1167 void (*ndo_netpoll_cleanup)(struct net_device *dev);
06021292 1168#endif
e0d1095a 1169#ifdef CONFIG_NET_RX_BUSY_POLL
8b80cda5 1170 int (*ndo_busy_poll)(struct napi_struct *dev);
d314774c 1171#endif
95c26df8
WM
1172 int (*ndo_set_vf_mac)(struct net_device *dev,
1173 int queue, u8 *mac);
1174 int (*ndo_set_vf_vlan)(struct net_device *dev,
1175 int queue, u16 vlan, u8 qos);
ed616689
SC
1176 int (*ndo_set_vf_rate)(struct net_device *dev,
1177 int vf, int min_tx_rate,
1178 int max_tx_rate);
5f8444a3
GR
1179 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1180 int vf, bool setting);
dd461d6a
HS
1181 int (*ndo_set_vf_trust)(struct net_device *dev,
1182 int vf, bool setting);
95c26df8
WM
1183 int (*ndo_get_vf_config)(struct net_device *dev,
1184 int vf,
1185 struct ifla_vf_info *ivf);
1d8faf48
RE
1186 int (*ndo_set_vf_link_state)(struct net_device *dev,
1187 int vf, int link_state);
3b766cd8
EBE
1188 int (*ndo_get_vf_stats)(struct net_device *dev,
1189 int vf,
1190 struct ifla_vf_stats
1191 *vf_stats);
57b61080
SF
1192 int (*ndo_set_vf_port)(struct net_device *dev,
1193 int vf,
1194 struct nlattr *port[]);
1195 int (*ndo_get_vf_port)(struct net_device *dev,
1196 int vf, struct sk_buff *skb);
cc8e27cc
EC
1197 int (*ndo_set_vf_guid)(struct net_device *dev,
1198 int vf, u64 guid,
1199 int guid_type);
01a3d796
VZ
1200 int (*ndo_set_vf_rss_query_en)(
1201 struct net_device *dev,
1202 int vf, bool setting);
16e5cc64
JF
1203 int (*ndo_setup_tc)(struct net_device *dev,
1204 u32 handle,
1205 __be16 protocol,
1206 struct tc_to_netdev *tc);
d11ead75 1207#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1208 int (*ndo_fcoe_enable)(struct net_device *dev);
1209 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1210 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1211 u16 xid,
1212 struct scatterlist *sgl,
1213 unsigned int sgc);
1214 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1215 u16 xid);
6247e086
YZ
1216 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1217 u16 xid,
1218 struct scatterlist *sgl,
1219 unsigned int sgc);
68bad94e
NP
1220 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1221 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1222#endif
1223
d11ead75 1224#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1225#define NETDEV_FCOE_WWNN 0
1226#define NETDEV_FCOE_WWPN 1
1227 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1228 u64 *wwn, int type);
4d288d57 1229#endif
3c9c36bc 1230
c445477d
BH
1231#ifdef CONFIG_RFS_ACCEL
1232 int (*ndo_rx_flow_steer)(struct net_device *dev,
1233 const struct sk_buff *skb,
1234 u16 rxq_index,
1235 u32 flow_id);
1236#endif
fbaec0ea
JP
1237 int (*ndo_add_slave)(struct net_device *dev,
1238 struct net_device *slave_dev);
1239 int (*ndo_del_slave)(struct net_device *dev,
1240 struct net_device *slave_dev);
c8f44aff
MM
1241 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1242 netdev_features_t features);
5455c699 1243 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1244 netdev_features_t features);
503eebc2
JP
1245 int (*ndo_neigh_construct)(struct net_device *dev,
1246 struct neighbour *n);
1247 void (*ndo_neigh_destroy)(struct net_device *dev,
1248 struct neighbour *n);
77162022
JF
1249
1250 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1251 struct nlattr *tb[],
77162022 1252 struct net_device *dev,
6b6e2725 1253 const unsigned char *addr,
f6f6424b 1254 u16 vid,
77162022
JF
1255 u16 flags);
1256 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1257 struct nlattr *tb[],
77162022 1258 struct net_device *dev,
f6f6424b
JP
1259 const unsigned char *addr,
1260 u16 vid);
77162022
JF
1261 int (*ndo_fdb_dump)(struct sk_buff *skb,
1262 struct netlink_callback *cb,
1263 struct net_device *dev,
5d5eacb3 1264 struct net_device *filter_dev,
77162022 1265 int idx);
e5a55a89
JF
1266
1267 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3
RP
1268 struct nlmsghdr *nlh,
1269 u16 flags);
e5a55a89
JF
1270 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1271 u32 pid, u32 seq,
6cbdceeb 1272 struct net_device *dev,
46c264da
ND
1273 u32 filter_mask,
1274 int nlflags);
407af329 1275 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1276 struct nlmsghdr *nlh,
1277 u16 flags);
4bf84c35
JP
1278 int (*ndo_change_carrier)(struct net_device *dev,
1279 bool new_carrier);
66b52b0d 1280 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1281 struct netdev_phys_item_id *ppid);
db24a904
DA
1282 int (*ndo_get_phys_port_name)(struct net_device *dev,
1283 char *name, size_t len);
7c46a640
AD
1284 void (*ndo_udp_tunnel_add)(struct net_device *dev,
1285 struct udp_tunnel_info *ti);
1286 void (*ndo_udp_tunnel_del)(struct net_device *dev,
1287 struct udp_tunnel_info *ti);
a6cc0cfa
JF
1288 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1289 struct net_device *dev);
1290 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1291 void *priv);
1292
1293 netdev_tx_t (*ndo_dfwd_start_xmit) (struct sk_buff *skb,
1294 struct net_device *dev,
1295 void *priv);
25175ba5 1296 int (*ndo_get_lock_subclass)(struct net_device *dev);
822b3b2e
JF
1297 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1298 int queue_index,
1299 u32 maxrate);
a54acb3a 1300 int (*ndo_get_iflink)(const struct net_device *dev);
d746d707
AK
1301 int (*ndo_change_proto_down)(struct net_device *dev,
1302 bool proto_down);
fc4099f1
PS
1303 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1304 struct sk_buff *skb);
871b642a
PA
1305 void (*ndo_set_rx_headroom)(struct net_device *dev,
1306 int needed_headroom);
a7862b45
BB
1307 int (*ndo_xdp)(struct net_device *dev,
1308 struct netdev_xdp *xdp);
d314774c
SH
1309};
1310
7aa98047
LR
1311/**
1312 * enum net_device_priv_flags - &struct net_device priv_flags
1313 *
1314 * These are the &struct net_device, they are only set internally
1315 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1316 * userspace; this means that the order of these flags can change
7aa98047
LR
1317 * during any kernel release.
1318 *
1319 * You should have a pretty good reason to be extending these flags.
1320 *
1321 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1322 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1323 * @IFF_BONDING: bonding master or slave
7aa98047 1324 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1325 * @IFF_WAN_HDLC: WAN HDLC device
1326 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1327 * release skb->dst
1328 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1329 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1330 * @IFF_MACVLAN_PORT: device used as macvlan port
1331 * @IFF_BRIDGE_PORT: device used as bridge port
1332 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1333 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1334 * @IFF_UNICAST_FLT: Supports unicast filtering
1335 * @IFF_TEAM_PORT: device used as team port
1336 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1337 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1338 * change when it's running
1339 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1340 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1341 * underlying stacked devices
1342 * @IFF_IPVLAN_MASTER: IPvlan master device
1343 * @IFF_IPVLAN_SLAVE: IPvlan slave device
007979ea 1344 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1345 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1346 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1347 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1348 * @IFF_TEAM: device is a team device
d4ab4286 1349 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1350 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1351 * entity (i.e. the master device for bridged veth)
3c175784 1352 * @IFF_MACSEC: device is a MACsec device
7aa98047
LR
1353 */
1354enum netdev_priv_flags {
1355 IFF_802_1Q_VLAN = 1<<0,
1356 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1357 IFF_BONDING = 1<<2,
1358 IFF_ISATAP = 1<<3,
1359 IFF_WAN_HDLC = 1<<4,
1360 IFF_XMIT_DST_RELEASE = 1<<5,
1361 IFF_DONT_BRIDGE = 1<<6,
1362 IFF_DISABLE_NETPOLL = 1<<7,
1363 IFF_MACVLAN_PORT = 1<<8,
1364 IFF_BRIDGE_PORT = 1<<9,
1365 IFF_OVS_DATAPATH = 1<<10,
1366 IFF_TX_SKB_SHARING = 1<<11,
1367 IFF_UNICAST_FLT = 1<<12,
1368 IFF_TEAM_PORT = 1<<13,
1369 IFF_SUPP_NOFCS = 1<<14,
1370 IFF_LIVE_ADDR_CHANGE = 1<<15,
1371 IFF_MACVLAN = 1<<16,
1372 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1373 IFF_IPVLAN_MASTER = 1<<18,
1374 IFF_IPVLAN_SLAVE = 1<<19,
007979ea 1375 IFF_L3MDEV_MASTER = 1<<20,
0dc1549b
JP
1376 IFF_NO_QUEUE = 1<<21,
1377 IFF_OPENVSWITCH = 1<<22,
fee6d4c7 1378 IFF_L3MDEV_SLAVE = 1<<23,
c981e421 1379 IFF_TEAM = 1<<24,
d4ab4286 1380 IFF_RXFH_CONFIGURED = 1<<25,
871b642a 1381 IFF_PHONY_HEADROOM = 1<<26,
3c175784 1382 IFF_MACSEC = 1<<27,
7aa98047
LR
1383};
1384
1385#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1386#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1387#define IFF_BONDING IFF_BONDING
7aa98047 1388#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1389#define IFF_WAN_HDLC IFF_WAN_HDLC
1390#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1391#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1392#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1393#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1394#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1395#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1396#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1397#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1398#define IFF_TEAM_PORT IFF_TEAM_PORT
1399#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1400#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1401#define IFF_MACVLAN IFF_MACVLAN
02875878 1402#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
2ad7bf36
MB
1403#define IFF_IPVLAN_MASTER IFF_IPVLAN_MASTER
1404#define IFF_IPVLAN_SLAVE IFF_IPVLAN_SLAVE
007979ea 1405#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1406#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1407#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1408#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1409#define IFF_TEAM IFF_TEAM
d4ab4286 1410#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
3c175784 1411#define IFF_MACSEC IFF_MACSEC
7aa98047 1412
536721b1
KK
1413/**
1414 * struct net_device - The DEVICE structure.
1415 * Actually, this whole structure is a big mistake. It mixes I/O
1416 * data with strictly "high-level" data, and it has to know about
1417 * almost every data structure used in the INET module.
1418 *
1419 * @name: This is the first field of the "visible" part of this structure
1420 * (i.e. as seen by users in the "Space.c" file). It is the name
1421 * of the interface.
1422 *
1423 * @name_hlist: Device name hash chain, please keep it close to name[]
1424 * @ifalias: SNMP alias
1425 * @mem_end: Shared memory end
1426 * @mem_start: Shared memory start
1427 * @base_addr: Device I/O address
1428 * @irq: Device IRQ number
1429 *
14ffbbb8
TG
1430 * @carrier_changes: Stats to monitor carrier on<->off transitions
1431 *
536721b1
KK
1432 * @state: Generic network queuing layer state, see netdev_state_t
1433 * @dev_list: The global list of network devices
5e82b4b2
BH
1434 * @napi_list: List entry used for polling NAPI devices
1435 * @unreg_list: List entry when we are unregistering the
1436 * device; see the function unregister_netdev
1437 * @close_list: List entry used when we are closing the device
62d885fe
BP
1438 * @ptype_all: Device-specific packet handlers for all protocols
1439 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1440 *
1441 * @adj_list: Directly linked devices, like slaves for bonding
1442 * @all_adj_list: All linked devices, *including* neighbours
1443 * @features: Currently active device features
1444 * @hw_features: User-changeable features
1445 *
1446 * @wanted_features: User-requested features
1447 * @vlan_features: Mask of features inheritable by VLAN devices
1448 *
1449 * @hw_enc_features: Mask of features inherited by encapsulating devices
1450 * This field indicates what encapsulation
1451 * offloads the hardware is capable of doing,
1452 * and drivers will need to set them appropriately.
1453 *
1454 * @mpls_features: Mask of features inheritable by MPLS
1455 *
1456 * @ifindex: interface index
5e82b4b2 1457 * @group: The group the device belongs to
536721b1
KK
1458 *
1459 * @stats: Statistics struct, which was left as a legacy, use
1460 * rtnl_link_stats64 instead
1461 *
1462 * @rx_dropped: Dropped packets by core network,
1463 * do not use this in drivers
1464 * @tx_dropped: Dropped packets by core network,
1465 * do not use this in drivers
6e7333d3
JW
1466 * @rx_nohandler: nohandler dropped packets by core network on
1467 * inactive devices, do not use this in drivers
536721b1 1468 *
536721b1
KK
1469 * @wireless_handlers: List of functions to handle Wireless Extensions,
1470 * instead of ioctl,
1471 * see <net/iw_handler.h> for details.
1472 * @wireless_data: Instance data managed by the core of wireless extensions
1473 *
1474 * @netdev_ops: Includes several pointers to callbacks,
1475 * if one wants to override the ndo_*() functions
1476 * @ethtool_ops: Management operations
f997c55c
AA
1477 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1478 * discovery handling. Necessary for e.g. 6LoWPAN.
d476059e 1479 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1480 * of Layer 2 headers.
1481 *
1482 * @flags: Interface flags (a la BSD)
1483 * @priv_flags: Like 'flags' but invisible to userspace,
1484 * see if.h for the definitions
1485 * @gflags: Global flags ( kept as legacy )
1486 * @padded: How much padding added by alloc_netdev()
1487 * @operstate: RFC2863 operstate
1488 * @link_mode: Mapping policy to operstate
1489 * @if_port: Selectable AUI, TP, ...
1490 * @dma: DMA channel
1491 * @mtu: Interface MTU value
1492 * @type: Interface hardware type
2793a23a 1493 * @hard_header_len: Maximum hardware header length.
536721b1
KK
1494 *
1495 * @needed_headroom: Extra headroom the hardware may need, but not in all
1496 * cases can this be guaranteed
1497 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1498 * cases can this be guaranteed. Some cases also use
1499 * LL_MAX_HEADER instead to allocate the skb
1500 *
1501 * interface address info:
1502 *
1503 * @perm_addr: Permanent hw address
1504 * @addr_assign_type: Hw address assignment type
1505 * @addr_len: Hardware address length
8626a0c8 1506 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1507 * @dev_id: Used to differentiate devices that share
1508 * the same link layer address
1509 * @dev_port: Used to differentiate devices that share
1510 * the same function
1511 * @addr_list_lock: XXX: need comments on this one
5e82b4b2 1512 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1513 * has been enabled due to the need to listen to
1514 * additional unicast addresses in a device that
1515 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1516 * @uc: unicast mac addresses
1517 * @mc: multicast mac addresses
1518 * @dev_addrs: list of device hw addresses
1519 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1520 * @promiscuity: Number of times the NIC is told to work in
1521 * promiscuous mode; if it becomes 0 the NIC will
1522 * exit promiscuous mode
536721b1
KK
1523 * @allmulti: Counter, enables or disables allmulticast mode
1524 *
1525 * @vlan_info: VLAN info
1526 * @dsa_ptr: dsa specific data
1527 * @tipc_ptr: TIPC specific data
1528 * @atalk_ptr: AppleTalk link
1529 * @ip_ptr: IPv4 specific data
1530 * @dn_ptr: DECnet specific data
1531 * @ip6_ptr: IPv6 specific data
1532 * @ax25_ptr: AX.25 specific data
1533 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
1534 *
1535 * @last_rx: Time of last Rx
1536 * @dev_addr: Hw address (before bcast,
1537 * because most packets are unicast)
1538 *
1539 * @_rx: Array of RX queues
1540 * @num_rx_queues: Number of RX queues
1541 * allocated at register_netdev() time
1542 * @real_num_rx_queues: Number of RX queues currently active in device
1543 *
1544 * @rx_handler: handler for received packets
1545 * @rx_handler_data: XXX: need comments on this one
1546 * @ingress_queue: XXX: need comments on this one
1547 * @broadcast: hw bcast address
1548 *
14ffbbb8
TG
1549 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1550 * indexed by RX queue number. Assigned by driver.
1551 * This must only be set if the ndo_rx_flow_steer
1552 * operation is defined
1553 * @index_hlist: Device index hash chain
1554 *
536721b1
KK
1555 * @_tx: Array of TX queues
1556 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1557 * @real_num_tx_queues: Number of TX queues currently active in device
1558 * @qdisc: Root qdisc from userspace point of view
1559 * @tx_queue_len: Max frames per queue allowed
1560 * @tx_global_lock: XXX: need comments on this one
1561 *
1562 * @xps_maps: XXX: need comments on this one
1563 *
0c4f691f
SF
1564 * @offload_fwd_mark: Offload device fwding mark
1565 *
536721b1 1566 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1567 * the watchdog (see dev_watchdog())
536721b1
KK
1568 * @watchdog_timer: List of timers
1569 *
1570 * @pcpu_refcnt: Number of references to this device
1571 * @todo_list: Delayed register/unregister
536721b1
KK
1572 * @link_watch_list: XXX: need comments on this one
1573 *
1574 * @reg_state: Register/unregister state machine
1575 * @dismantle: Device is going to be freed
1576 * @rtnl_link_state: This enum represents the phases of creating
1577 * a new link
1578 *
1579 * @destructor: Called from unregister,
1580 * can be used to call free_netdev
1581 * @npinfo: XXX: need comments on this one
1582 * @nd_net: Network namespace this network device is inside
1583 *
1584 * @ml_priv: Mid-layer private
1585 * @lstats: Loopback statistics
1586 * @tstats: Tunnel statistics
1587 * @dstats: Dummy statistics
1588 * @vstats: Virtual ethernet statistics
1589 *
1590 * @garp_port: GARP
1591 * @mrp_port: MRP
1592 *
1593 * @dev: Class/net/name entry
1594 * @sysfs_groups: Space for optional device, statistics and wireless
1595 * sysfs groups
1596 *
1597 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1598 * @rtnl_link_ops: Rtnl_link_ops
1599 *
1600 * @gso_max_size: Maximum size of generic segmentation offload
1601 * @gso_max_segs: Maximum number of segments that can be passed to the
1602 * NIC for GSO
1603 *
1604 * @dcbnl_ops: Data Center Bridging netlink ops
1605 * @num_tc: Number of traffic classes in the net device
1606 * @tc_to_txq: XXX: need comments on this one
1607 * @prio_tc_map XXX: need comments on this one
1608 *
1609 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1610 *
1611 * @priomap: XXX: need comments on this one
1612 * @phydev: Physical device may attach itself
1613 * for hardware timestamping
1614 *
123b3652
ED
1615 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
1616 * @qdisc_running_key: lockdep class annotating Qdisc->running seqcount
536721b1 1617 *
d746d707
AK
1618 * @proto_down: protocol port state information can be sent to the
1619 * switch driver and used to set the phys state of the
1620 * switch port.
1621 *
1da177e4
LT
1622 * FIXME: cleanup struct net_device such that network protocol info
1623 * moves out.
1624 */
1625
d94d9fee 1626struct net_device {
1da177e4 1627 char name[IFNAMSIZ];
9356b8fc 1628 struct hlist_node name_hlist;
0b815a1a 1629 char *ifalias;
1da177e4
LT
1630 /*
1631 * I/O specific fields
1632 * FIXME: Merge these and struct ifmap into one
1633 */
536721b1
KK
1634 unsigned long mem_end;
1635 unsigned long mem_start;
1636 unsigned long base_addr;
1637 int irq;
1da177e4 1638
14ffbbb8
TG
1639 atomic_t carrier_changes;
1640
1da177e4 1641 /*
536721b1
KK
1642 * Some hardware also needs these fields (state,dev_list,
1643 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1644 * part of the usual set specified in Space.c.
1645 */
1646
1da177e4
LT
1647 unsigned long state;
1648
7562f876 1649 struct list_head dev_list;
bea3348e 1650 struct list_head napi_list;
44a0873d 1651 struct list_head unreg_list;
5cde2829 1652 struct list_head close_list;
7866a621
SN
1653 struct list_head ptype_all;
1654 struct list_head ptype_specific;
2f268f12 1655
2f268f12
VF
1656 struct {
1657 struct list_head upper;
1658 struct list_head lower;
1659 } adj_list;
1660
2f268f12
VF
1661 struct {
1662 struct list_head upper;
1663 struct list_head lower;
1664 } all_adj_list;
4c3d5e7b 1665
c8f44aff 1666 netdev_features_t features;
c8f44aff 1667 netdev_features_t hw_features;
c8f44aff 1668 netdev_features_t wanted_features;
c8f44aff 1669 netdev_features_t vlan_features;
6a674e9c 1670 netdev_features_t hw_enc_features;
0d89d203 1671 netdev_features_t mpls_features;
802ab55a 1672 netdev_features_t gso_partial_features;
04ed3e74 1673
1da177e4 1674 int ifindex;
7a66bbc9 1675 int group;
1da177e4 1676
c45d286e 1677 struct net_device_stats stats;
015f0688 1678
015f0688
ED
1679 atomic_long_t rx_dropped;
1680 atomic_long_t tx_dropped;
6e7333d3 1681 atomic_long_t rx_nohandler;
1da177e4 1682
b86e0280 1683#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
1684 const struct iw_handler_def *wireless_handlers;
1685 struct iw_public_data *wireless_data;
b86e0280 1686#endif
d314774c 1687 const struct net_device_ops *netdev_ops;
76fd8593 1688 const struct ethtool_ops *ethtool_ops;
4170604f 1689#ifdef CONFIG_NET_SWITCHDEV
9d47c0a2 1690 const struct switchdev_ops *switchdev_ops;
4170604f 1691#endif
1b69c6d0
DA
1692#ifdef CONFIG_NET_L3_MASTER_DEV
1693 const struct l3mdev_ops *l3mdev_ops;
1694#endif
f997c55c
AA
1695#if IS_ENABLED(CONFIG_IPV6)
1696 const struct ndisc_ops *ndisc_ops;
1697#endif
1da177e4 1698
3b04ddde
SH
1699 const struct header_ops *header_ops;
1700
536721b1
KK
1701 unsigned int flags;
1702 unsigned int priv_flags;
1703
1da177e4 1704 unsigned short gflags;
536721b1 1705 unsigned short padded;
1da177e4 1706
536721b1
KK
1707 unsigned char operstate;
1708 unsigned char link_mode;
b00055aa 1709
536721b1
KK
1710 unsigned char if_port;
1711 unsigned char dma;
bdc220da 1712
536721b1
KK
1713 unsigned int mtu;
1714 unsigned short type;
1715 unsigned short hard_header_len;
1da177e4 1716
f5184d26
JB
1717 unsigned short needed_headroom;
1718 unsigned short needed_tailroom;
1719
1da177e4 1720 /* Interface address info. */
536721b1
KK
1721 unsigned char perm_addr[MAX_ADDR_LEN];
1722 unsigned char addr_assign_type;
1723 unsigned char addr_len;
a0a9663d 1724 unsigned short neigh_priv_len;
536721b1
KK
1725 unsigned short dev_id;
1726 unsigned short dev_port;
ccffad25 1727 spinlock_t addr_list_lock;
14ffbbb8
TG
1728 unsigned char name_assign_type;
1729 bool uc_promisc;
536721b1
KK
1730 struct netdev_hw_addr_list uc;
1731 struct netdev_hw_addr_list mc;
1732 struct netdev_hw_addr_list dev_addrs;
1733
4c3d5e7b
ED
1734#ifdef CONFIG_SYSFS
1735 struct kset *queues_kset;
1736#endif
9d45abe1
WC
1737 unsigned int promiscuity;
1738 unsigned int allmulti;
1da177e4 1739
1da177e4 1740
5e82b4b2 1741 /* Protocol-specific pointers */
65ac6a5f 1742
d11ead75 1743#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 1744 struct vlan_info __rcu *vlan_info;
65ac6a5f 1745#endif
34a430d7 1746#if IS_ENABLED(CONFIG_NET_DSA)
536721b1 1747 struct dsa_switch_tree *dsa_ptr;
37cb0620
YX
1748#endif
1749#if IS_ENABLED(CONFIG_TIPC)
536721b1 1750 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 1751#endif
536721b1
KK
1752 void *atalk_ptr;
1753 struct in_device __rcu *ip_ptr;
1754 struct dn_dev __rcu *dn_ptr;
1755 struct inet6_dev __rcu *ip6_ptr;
1756 void *ax25_ptr;
1757 struct wireless_dev *ieee80211_ptr;
98a18b6f 1758 struct wpan_dev *ieee802154_ptr;
03c57747
RS
1759#if IS_ENABLED(CONFIG_MPLS_ROUTING)
1760 struct mpls_dev __rcu *mpls_ptr;
1761#endif
1da177e4 1762
9356b8fc 1763/*
cd13539b 1764 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 1765 */
536721b1 1766 unsigned long last_rx;
4dc89133 1767
9356b8fc 1768 /* Interface address info used in eth_type_trans() */
536721b1 1769 unsigned char *dev_addr;
f001fde5 1770
a953be53 1771#ifdef CONFIG_SYSFS
0a9627f2
TH
1772 struct netdev_rx_queue *_rx;
1773
0a9627f2 1774 unsigned int num_rx_queues;
62fe0b40 1775 unsigned int real_num_rx_queues;
df334545 1776#endif
0a9627f2 1777
3b47d303 1778 unsigned long gro_flush_timeout;
61391cde 1779 rx_handler_func_t __rcu *rx_handler;
1780 void __rcu *rx_handler_data;
e8a0464c 1781
4cda01e8 1782#ifdef CONFIG_NET_CLS_ACT
d2788d34
DB
1783 struct tcf_proto __rcu *ingress_cl_list;
1784#endif
24824a09 1785 struct netdev_queue __rcu *ingress_queue;
e687ad60
PN
1786#ifdef CONFIG_NETFILTER_INGRESS
1787 struct list_head nf_hooks_ingress;
1788#endif
d2788d34 1789
536721b1 1790 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
1791#ifdef CONFIG_RFS_ACCEL
1792 struct cpu_rmap *rx_cpu_rmap;
1793#endif
1794 struct hlist_node index_hlist;
cd13539b
ED
1795
1796/*
1797 * Cache lines mostly used on transmit path
1798 */
e8a0464c
DM
1799 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
1800 unsigned int num_tx_queues;
fd2ea0a7 1801 unsigned int real_num_tx_queues;
af356afa 1802 struct Qdisc *qdisc;
536721b1 1803 unsigned long tx_queue_len;
c3f26a26 1804 spinlock_t tx_global_lock;
14ffbbb8 1805 int watchdog_timeo;
cd13539b 1806
bf264145 1807#ifdef CONFIG_XPS
a4177869 1808 struct xps_dev_maps __rcu *xps_maps;
bf264145 1809#endif
1f211a1b
DB
1810#ifdef CONFIG_NET_CLS_ACT
1811 struct tcf_proto __rcu *egress_cl_list;
1812#endif
0c4f691f
SF
1813#ifdef CONFIG_NET_SWITCHDEV
1814 u32 offload_fwd_mark;
1815#endif
1816
9356b8fc 1817 /* These may be needed for future network-power-down code. */
9356b8fc
ED
1818 struct timer_list watchdog_timer;
1819
29b4433d 1820 int __percpu *pcpu_refcnt;
1da177e4 1821 struct list_head todo_list;
1da177e4 1822
e014debe 1823 struct list_head link_watch_list;
572a103d 1824
1da177e4 1825 enum { NETREG_UNINITIALIZED=0,
b17a7c17 1826 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
1827 NETREG_UNREGISTERING, /* called unregister_netdevice */
1828 NETREG_UNREGISTERED, /* completed unregister todo */
1829 NETREG_RELEASED, /* called free_netdev */
937f1ba5 1830 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
1831 } reg_state:8;
1832
536721b1 1833 bool dismantle;
a2835763
PM
1834
1835 enum {
1836 RTNL_LINK_INITIALIZED,
1837 RTNL_LINK_INITIALIZING,
1838 } rtnl_link_state:16;
1da177e4 1839
d314774c 1840 void (*destructor)(struct net_device *dev);
1da177e4 1841
1da177e4 1842#ifdef CONFIG_NETPOLL
5fbee843 1843 struct netpoll_info __rcu *npinfo;
1da177e4 1844#endif
eae792b7 1845
0c5c9fb5 1846 possible_net_t nd_net;
4a1c5371 1847
4951704b 1848 /* mid-layer private */
a7855c78 1849 union {
536721b1
KK
1850 void *ml_priv;
1851 struct pcpu_lstats __percpu *lstats;
8f84985f 1852 struct pcpu_sw_netstats __percpu *tstats;
536721b1
KK
1853 struct pcpu_dstats __percpu *dstats;
1854 struct pcpu_vstats __percpu *vstats;
a7855c78 1855 };
536721b1 1856
3cc77ec7 1857 struct garp_port __rcu *garp_port;
febf018d 1858 struct mrp_port __rcu *mrp_port;
1da177e4 1859
5e82b4b2 1860 struct device dev;
0c509a6c 1861 const struct attribute_group *sysfs_groups[4];
a953be53 1862 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 1863
38f7b870 1864 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 1865
82cc1a7a
PWJ
1866 /* for setting kernel sock attribute on TCP connection setup */
1867#define GSO_MAX_SIZE 65536
1868 unsigned int gso_max_size;
30b678d8
BH
1869#define GSO_MAX_SEGS 65535
1870 u16 gso_max_segs;
743b03a8 1871
7a6b6f51 1872#ifdef CONFIG_DCB
32953543 1873 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 1874#endif
5e82b4b2
BH
1875 u8 num_tc;
1876 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
1877 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 1878
d11ead75 1879#if IS_ENABLED(CONFIG_FCOE)
4d288d57 1880 unsigned int fcoe_ddp_xid;
5bc1421e 1881#endif
86f8515f 1882#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 1883 struct netprio_map __rcu *priomap;
4d288d57 1884#endif
5e82b4b2
BH
1885 struct phy_device *phydev;
1886 struct lock_class_key *qdisc_tx_busylock;
f9eb8aea 1887 struct lock_class_key *qdisc_running_key;
5e82b4b2 1888 bool proto_down;
1da177e4 1889};
43cb76d9 1890#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4
LT
1891
1892#define NETDEV_ALIGN 32
1da177e4 1893
4f57c087
JF
1894static inline
1895int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
1896{
1897 return dev->prio_tc_map[prio & TC_BITMASK];
1898}
1899
1900static inline
1901int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
1902{
1903 if (tc >= dev->num_tc)
1904 return -EINVAL;
1905
1906 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
1907 return 0;
1908}
1909
1910static inline
1911void netdev_reset_tc(struct net_device *dev)
1912{
1913 dev->num_tc = 0;
1914 memset(dev->tc_to_txq, 0, sizeof(dev->tc_to_txq));
1915 memset(dev->prio_tc_map, 0, sizeof(dev->prio_tc_map));
1916}
1917
1918static inline
1919int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset)
1920{
1921 if (tc >= dev->num_tc)
1922 return -EINVAL;
1923
1924 dev->tc_to_txq[tc].count = count;
1925 dev->tc_to_txq[tc].offset = offset;
1926 return 0;
1927}
1928
1929static inline
1930int netdev_set_num_tc(struct net_device *dev, u8 num_tc)
1931{
1932 if (num_tc > TC_MAX_QUEUE)
1933 return -EINVAL;
1934
1935 dev->num_tc = num_tc;
1936 return 0;
1937}
1938
1939static inline
1940int netdev_get_num_tc(struct net_device *dev)
1941{
1942 return dev->num_tc;
1943}
1944
e8a0464c
DM
1945static inline
1946struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
1947 unsigned int index)
1948{
1949 return &dev->_tx[index];
1950}
1951
10c51b56
DB
1952static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
1953 const struct sk_buff *skb)
1954{
1955 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
1956}
1957
e8a0464c
DM
1958static inline void netdev_for_each_tx_queue(struct net_device *dev,
1959 void (*f)(struct net_device *,
1960 struct netdev_queue *,
1961 void *),
1962 void *arg)
1963{
1964 unsigned int i;
1965
1966 for (i = 0; i < dev->num_tx_queues; i++)
1967 f(dev, &dev->_tx[i], arg);
1968}
1969
d3fff6c4
ED
1970#define netdev_lockdep_set_classes(dev) \
1971{ \
1972 static struct lock_class_key qdisc_tx_busylock_key; \
1973 static struct lock_class_key qdisc_running_key; \
1974 static struct lock_class_key qdisc_xmit_lock_key; \
1975 static struct lock_class_key dev_addr_list_lock_key; \
1976 unsigned int i; \
1977 \
1978 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
1979 (dev)->qdisc_running_key = &qdisc_running_key; \
1980 lockdep_set_class(&(dev)->addr_list_lock, \
1981 &dev_addr_list_lock_key); \
1982 for (i = 0; i < (dev)->num_tx_queues; i++) \
1983 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
1984 &qdisc_xmit_lock_key); \
1985}
1986
f629d208 1987struct netdev_queue *netdev_pick_tx(struct net_device *dev,
f663dd9a
JW
1988 struct sk_buff *skb,
1989 void *accel_priv);
8c4c49df 1990
871b642a
PA
1991/* returns the headroom that the master device needs to take in account
1992 * when forwarding to this dev
1993 */
1994static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
1995{
1996 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
1997}
1998
1999static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2000{
2001 if (dev->netdev_ops->ndo_set_rx_headroom)
2002 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2003}
2004
2005/* set the device rx headroom to the dev's default */
2006static inline void netdev_reset_rx_headroom(struct net_device *dev)
2007{
2008 netdev_set_rx_headroom(dev, -1);
2009}
2010
c346dca1
YH
2011/*
2012 * Net namespace inlines
2013 */
2014static inline
2015struct net *dev_net(const struct net_device *dev)
2016{
c2d9ba9b 2017 return read_pnet(&dev->nd_net);
c346dca1
YH
2018}
2019
2020static inline
f5aa23fd 2021void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2022{
0c5c9fb5 2023 write_pnet(&dev->nd_net, net);
c346dca1
YH
2024}
2025
3e8a72d1 2026static inline bool netdev_uses_dsa(struct net_device *dev)
cf85d08f 2027{
3fc88677 2028#if IS_ENABLED(CONFIG_NET_DSA)
5aed85ce
FF
2029 if (dev->dsa_ptr != NULL)
2030 return dsa_uses_tagged_protocol(dev->dsa_ptr);
396138f0 2031#endif
5aed85ce 2032 return false;
396138f0
LB
2033}
2034
bea3348e
SH
2035/**
2036 * netdev_priv - access network device private data
2037 * @dev: network device
2038 *
2039 * Get network device private data
2040 */
6472ce60 2041static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2042{
1ce8e7b5 2043 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2044}
2045
1da177e4
LT
2046/* Set the sysfs physical device reference for the network logical device
2047 * if set prior to registration will cause a symlink during initialization.
2048 */
43cb76d9 2049#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2050
384912ed 2051/* Set the sysfs device type for the network logical device to allow
3f79410c 2052 * fine-grained identification of different network device types. For
5e82b4b2 2053 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2054 */
2055#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2056
82dc3c63
ED
2057/* Default NAPI poll() weight
2058 * Device drivers are strongly advised to not use bigger value
2059 */
2060#define NAPI_POLL_WEIGHT 64
2061
3b582cc1 2062/**
5e82b4b2 2063 * netif_napi_add - initialize a NAPI context
3b582cc1 2064 * @dev: network device
5e82b4b2 2065 * @napi: NAPI context
3b582cc1
SH
2066 * @poll: polling function
2067 * @weight: default weight
2068 *
5e82b4b2
BH
2069 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2070 * *any* of the other NAPI-related functions.
3b582cc1 2071 */
d565b0a1
HX
2072void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2073 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2074
d64b5e85 2075/**
5e82b4b2 2076 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2077 * @dev: network device
5e82b4b2 2078 * @napi: NAPI context
d64b5e85
ED
2079 * @poll: polling function
2080 * @weight: default weight
2081 *
2082 * This variant of netif_napi_add() should be used from drivers using NAPI
2083 * to exclusively poll a TX queue.
2084 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2085 */
2086static inline void netif_tx_napi_add(struct net_device *dev,
2087 struct napi_struct *napi,
2088 int (*poll)(struct napi_struct *, int),
2089 int weight)
2090{
2091 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2092 netif_napi_add(dev, napi, poll, weight);
2093}
2094
d8156534 2095/**
5e82b4b2
BH
2096 * netif_napi_del - remove a NAPI context
2097 * @napi: NAPI context
d8156534 2098 *
5e82b4b2 2099 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2100 */
d565b0a1
HX
2101void netif_napi_del(struct napi_struct *napi);
2102
2103struct napi_gro_cb {
78a478d0 2104 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
5e82b4b2 2105 void *frag0;
78a478d0 2106
7489594c
HX
2107 /* Length of frag0. */
2108 unsigned int frag0_len;
2109
86911732 2110 /* This indicates where we are processing relative to skb->data. */
5e82b4b2 2111 int data_offset;
86911732 2112
d565b0a1 2113 /* This is non-zero if the packet cannot be merged with the new skb. */
bf5a755f
JC
2114 u16 flush;
2115
2116 /* Save the IP ID here and check when we get to the transport layer */
2117 u16 flush_id;
d565b0a1
HX
2118
2119 /* Number of segments aggregated. */
2e71a6f8
ED
2120 u16 count;
2121
15e2396d
TH
2122 /* Start offset for remote checksum offload */
2123 u16 gro_remcsum_start;
2124
2e71a6f8
ED
2125 /* jiffies when first packet was created/queued */
2126 unsigned long age;
86347245 2127
afe93325 2128 /* Used in ipv6_gro_receive() and foo-over-udp */
b582ef09
OG
2129 u16 proto;
2130
baa32ff4
TH
2131 /* This is non-zero if the packet may be of the same flow. */
2132 u8 same_flow:1;
2133
fac8e0f5
JG
2134 /* Used in tunnel GRO receive */
2135 u8 encap_mark:1;
573e8fca
TH
2136
2137 /* GRO checksum is valid */
2138 u8 csum_valid:1;
2139
662880f4
TH
2140 /* Number of checksums via CHECKSUM_UNNECESSARY */
2141 u8 csum_cnt:3;
c3c7c254 2142
baa32ff4
TH
2143 /* Free the skb? */
2144 u8 free:2;
2145#define NAPI_GRO_FREE 1
2146#define NAPI_GRO_FREE_STOLEN_HEAD 2
2147
efc98d08
TH
2148 /* Used in foo-over-udp, set in udp[46]_gro_receive */
2149 u8 is_ipv6:1;
2150
a0ca153f
AD
2151 /* Used in GRE, set in fou/gue_gro_receive */
2152 u8 is_fou:1;
2153
1530545e
AD
2154 /* Used to determine if flush_id can be ignored */
2155 u8 is_atomic:1;
2156
2157 /* 5 bit hole */
baa32ff4 2158
bf5a755f
JC
2159 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
2160 __wsum csum;
2161
c3c7c254
ED
2162 /* used in skb_gro_receive() slow path */
2163 struct sk_buff *last;
d565b0a1
HX
2164};
2165
2166#define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
d8156534 2167
1da177e4 2168struct packet_type {
f2ccd8fa
DM
2169 __be16 type; /* This is really htons(ether_type). */
2170 struct net_device *dev; /* NULL is wildcarded here */
2171 int (*func) (struct sk_buff *,
2172 struct net_device *,
2173 struct packet_type *,
2174 struct net_device *);
c0de08d0
EL
2175 bool (*id_match)(struct packet_type *ptype,
2176 struct sock *sk);
1da177e4
LT
2177 void *af_packet_priv;
2178 struct list_head list;
2179};
2180
f191a1d1 2181struct offload_callbacks {
576a30eb 2182 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2183 netdev_features_t features);
d565b0a1 2184 struct sk_buff **(*gro_receive)(struct sk_buff **head,
a2b12f3c 2185 struct sk_buff *skb);
299603e8 2186 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2187};
2188
2189struct packet_offload {
2190 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2191 u16 priority;
f191a1d1
VY
2192 struct offload_callbacks callbacks;
2193 struct list_head list;
1da177e4
LT
2194};
2195
5e82b4b2 2196/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2197struct pcpu_sw_netstats {
2198 u64 rx_packets;
2199 u64 rx_bytes;
2200 u64 tx_packets;
2201 u64 tx_bytes;
2202 struct u64_stats_sync syncp;
2203};
2204
aabc92bb
PNA
2205#define __netdev_alloc_pcpu_stats(type, gfp) \
2206({ \
2207 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2208 if (pcpu_stats) { \
2209 int __cpu; \
2210 for_each_possible_cpu(__cpu) { \
2211 typeof(type) *stat; \
2212 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2213 u64_stats_init(&stat->syncp); \
2214 } \
2215 } \
2216 pcpu_stats; \
1c213bd2
WC
2217})
2218
aabc92bb 2219#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2220 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2221
764f5e54
JP
2222enum netdev_lag_tx_type {
2223 NETDEV_LAG_TX_TYPE_UNKNOWN,
2224 NETDEV_LAG_TX_TYPE_RANDOM,
2225 NETDEV_LAG_TX_TYPE_BROADCAST,
2226 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2227 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2228 NETDEV_LAG_TX_TYPE_HASH,
2229};
2230
2231struct netdev_lag_upper_info {
2232 enum netdev_lag_tx_type tx_type;
2233};
2234
fb1b2e3c
JP
2235struct netdev_lag_lower_state_info {
2236 u8 link_up : 1,
2237 tx_enabled : 1;
2238};
2239
1da177e4
LT
2240#include <linux/notifier.h>
2241
dcfe1421
AW
2242/* netdevice notifier chain. Please remember to update the rtnetlink
2243 * notification exclusion list in rtnetlink_event() when adding new
2244 * types.
2245 */
2246#define NETDEV_UP 0x0001 /* For now you can't veto a device up/down */
2247#define NETDEV_DOWN 0x0002
2248#define NETDEV_REBOOT 0x0003 /* Tell a protocol stack a network interface
2249 detected a hardware crash and restarted
2250 - we can use this eg to kick tcp sessions
2251 once done */
2252#define NETDEV_CHANGE 0x0004 /* Notify device state change */
2253#define NETDEV_REGISTER 0x0005
2254#define NETDEV_UNREGISTER 0x0006
1d486bfb 2255#define NETDEV_CHANGEMTU 0x0007 /* notify after mtu change happened */
dcfe1421
AW
2256#define NETDEV_CHANGEADDR 0x0008
2257#define NETDEV_GOING_DOWN 0x0009
2258#define NETDEV_CHANGENAME 0x000A
2259#define NETDEV_FEAT_CHANGE 0x000B
2260#define NETDEV_BONDING_FAILOVER 0x000C
2261#define NETDEV_PRE_UP 0x000D
2262#define NETDEV_PRE_TYPE_CHANGE 0x000E
2263#define NETDEV_POST_TYPE_CHANGE 0x000F
2264#define NETDEV_POST_INIT 0x0010
0115e8e3 2265#define NETDEV_UNREGISTER_FINAL 0x0011
dcfe1421
AW
2266#define NETDEV_RELEASE 0x0012
2267#define NETDEV_NOTIFY_PEERS 0x0013
2268#define NETDEV_JOIN 0x0014
42e52bf9 2269#define NETDEV_CHANGEUPPER 0x0015
4aa5dee4 2270#define NETDEV_RESEND_IGMP 0x0016
1d486bfb 2271#define NETDEV_PRECHANGEMTU 0x0017 /* notify before mtu change happened */
d4261e56 2272#define NETDEV_CHANGEINFODATA 0x0018
61bd3857 2273#define NETDEV_BONDING_INFO 0x0019
573c7ba0 2274#define NETDEV_PRECHANGEUPPER 0x001A
04d48266 2275#define NETDEV_CHANGELOWERSTATE 0x001B
7c46a640 2276#define NETDEV_UDP_TUNNEL_PUSH_INFO 0x001C
08294a26 2277#define NETDEV_CHANGE_TX_QUEUE_LEN 0x001E
dcfe1421 2278
f629d208
JP
2279int register_netdevice_notifier(struct notifier_block *nb);
2280int unregister_netdevice_notifier(struct notifier_block *nb);
351638e7
JP
2281
2282struct netdev_notifier_info {
2283 struct net_device *dev;
2284};
2285
be9efd36
JP
2286struct netdev_notifier_change_info {
2287 struct netdev_notifier_info info; /* must be first */
2288 unsigned int flags_changed;
2289};
2290
0e4ead9d
JP
2291struct netdev_notifier_changeupper_info {
2292 struct netdev_notifier_info info; /* must be first */
2293 struct net_device *upper_dev; /* new upper dev */
2294 bool master; /* is upper dev master */
5e82b4b2 2295 bool linking; /* is the notification for link or unlink */
29bf24af 2296 void *upper_info; /* upper dev info */
0e4ead9d
JP
2297};
2298
04d48266
JP
2299struct netdev_notifier_changelowerstate_info {
2300 struct netdev_notifier_info info; /* must be first */
2301 void *lower_state_info; /* is lower dev state */
2302};
2303
75538c2b
CW
2304static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2305 struct net_device *dev)
2306{
2307 info->dev = dev;
2308}
2309
351638e7
JP
2310static inline struct net_device *
2311netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2312{
2313 return info->dev;
2314}
2315
f629d208 2316int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2317
2318
1da177e4
LT
2319extern rwlock_t dev_base_lock; /* Device list lock */
2320
881d966b
EB
2321#define for_each_netdev(net, d) \
2322 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2323#define for_each_netdev_reverse(net, d) \
2324 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2325#define for_each_netdev_rcu(net, d) \
2326 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2327#define for_each_netdev_safe(net, d, n) \
2328 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2329#define for_each_netdev_continue(net, d) \
2330 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
254245d2 2331#define for_each_netdev_continue_rcu(net, d) \
2332 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2333#define for_each_netdev_in_bond_rcu(bond, slave) \
2334 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2335 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2336#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2337
a050c33f
DL
2338static inline struct net_device *next_net_device(struct net_device *dev)
2339{
2340 struct list_head *lh;
2341 struct net *net;
2342
c346dca1 2343 net = dev_net(dev);
a050c33f
DL
2344 lh = dev->dev_list.next;
2345 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2346}
2347
ce81b76a
ED
2348static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2349{
2350 struct list_head *lh;
2351 struct net *net;
2352
2353 net = dev_net(dev);
ccf43438 2354 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2355 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2356}
2357
a050c33f
DL
2358static inline struct net_device *first_net_device(struct net *net)
2359{
2360 return list_empty(&net->dev_base_head) ? NULL :
2361 net_device_entry(net->dev_base_head.next);
2362}
7562f876 2363
ccf43438
ED
2364static inline struct net_device *first_net_device_rcu(struct net *net)
2365{
2366 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2367
2368 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2369}
2370
f629d208
JP
2371int netdev_boot_setup_check(struct net_device *dev);
2372unsigned long netdev_boot_base(const char *prefix, int unit);
2373struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2374 const char *hwaddr);
2375struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
2376struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type);
2377void dev_add_pack(struct packet_type *pt);
2378void dev_remove_pack(struct packet_type *pt);
2379void __dev_remove_pack(struct packet_type *pt);
2380void dev_add_offload(struct packet_offload *po);
2381void dev_remove_offload(struct packet_offload *po);
f629d208 2382
a54acb3a 2383int dev_get_iflink(const struct net_device *dev);
fc4099f1 2384int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
6c555490
WC
2385struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2386 unsigned short mask);
f629d208
JP
2387struct net_device *dev_get_by_name(struct net *net, const char *name);
2388struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2389struct net_device *__dev_get_by_name(struct net *net, const char *name);
2390int dev_alloc_name(struct net_device *dev, const char *name);
2391int dev_open(struct net_device *dev);
2392int dev_close(struct net_device *dev);
99c4a26a 2393int dev_close_many(struct list_head *head, bool unlink);
f629d208 2394void dev_disable_lro(struct net_device *dev);
0c4b51f0 2395int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
2b4aa3ce 2396int dev_queue_xmit(struct sk_buff *skb);
f663dd9a 2397int dev_queue_xmit_accel(struct sk_buff *skb, void *accel_priv);
f629d208
JP
2398int register_netdevice(struct net_device *dev);
2399void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2400void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2401static inline void unregister_netdevice(struct net_device *dev)
2402{
2403 unregister_netdevice_queue(dev, NULL);
2404}
2405
f629d208
JP
2406int netdev_refcnt_read(const struct net_device *dev);
2407void free_netdev(struct net_device *dev);
74d332c1 2408void netdev_freemem(struct net_device *dev);
f629d208
JP
2409void synchronize_net(void);
2410int init_dummy_netdev(struct net_device *dev);
937f1ba5 2411
f60e5990 2412DECLARE_PER_CPU(int, xmit_recursion);
a70b506e
DB
2413#define XMIT_RECURSION_LIMIT 10
2414
f60e5990 2415static inline int dev_recursion_level(void)
2416{
2417 return this_cpu_read(xmit_recursion);
2418}
2419
f629d208
JP
2420struct net_device *dev_get_by_index(struct net *net, int ifindex);
2421struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2422struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
2423int netdev_get_name(struct net *net, char *name, int ifindex);
2424int dev_restart(struct net_device *dev);
f629d208 2425int skb_gro_receive(struct sk_buff **head, struct sk_buff *skb);
86911732
HX
2426
2427static inline unsigned int skb_gro_offset(const struct sk_buff *skb)
2428{
2429 return NAPI_GRO_CB(skb)->data_offset;
2430}
2431
2432static inline unsigned int skb_gro_len(const struct sk_buff *skb)
2433{
2434 return skb->len - NAPI_GRO_CB(skb)->data_offset;
2435}
2436
2437static inline void skb_gro_pull(struct sk_buff *skb, unsigned int len)
2438{
2439 NAPI_GRO_CB(skb)->data_offset += len;
2440}
2441
a5b1cf28
HX
2442static inline void *skb_gro_header_fast(struct sk_buff *skb,
2443 unsigned int offset)
86911732 2444{
a5b1cf28
HX
2445 return NAPI_GRO_CB(skb)->frag0 + offset;
2446}
78a478d0 2447
a5b1cf28
HX
2448static inline int skb_gro_header_hard(struct sk_buff *skb, unsigned int hlen)
2449{
2450 return NAPI_GRO_CB(skb)->frag0_len < hlen;
2451}
78a478d0 2452
a5b1cf28
HX
2453static inline void *skb_gro_header_slow(struct sk_buff *skb, unsigned int hlen,
2454 unsigned int offset)
2455{
17dd759c
HX
2456 if (!pskb_may_pull(skb, hlen))
2457 return NULL;
2458
a5b1cf28
HX
2459 NAPI_GRO_CB(skb)->frag0 = NULL;
2460 NAPI_GRO_CB(skb)->frag0_len = 0;
17dd759c 2461 return skb->data + offset;
86911732 2462}
1da177e4 2463
36e7b1b8
HX
2464static inline void *skb_gro_network_header(struct sk_buff *skb)
2465{
78d3fd0b
HX
2466 return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
2467 skb_network_offset(skb);
36e7b1b8
HX
2468}
2469
bf5a755f
JC
2470static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
2471 const void *start, unsigned int len)
2472{
573e8fca 2473 if (NAPI_GRO_CB(skb)->csum_valid)
bf5a755f
JC
2474 NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
2475 csum_partial(start, len, 0));
2476}
2477
573e8fca
TH
2478/* GRO checksum functions. These are logical equivalents of the normal
2479 * checksum functions (in skbuff.h) except that they operate on the GRO
2480 * offsets and fields in sk_buff.
2481 */
2482
2483__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
2484
15e2396d
TH
2485static inline bool skb_at_gro_remcsum_start(struct sk_buff *skb)
2486{
b7fe10e5 2487 return (NAPI_GRO_CB(skb)->gro_remcsum_start == skb_gro_offset(skb));
15e2396d
TH
2488}
2489
573e8fca
TH
2490static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
2491 bool zero_okay,
2492 __sum16 check)
2493{
6edec0e6
TH
2494 return ((skb->ip_summed != CHECKSUM_PARTIAL ||
2495 skb_checksum_start_offset(skb) <
2496 skb_gro_offset(skb)) &&
15e2396d 2497 !skb_at_gro_remcsum_start(skb) &&
662880f4 2498 NAPI_GRO_CB(skb)->csum_cnt == 0 &&
573e8fca
TH
2499 (!zero_okay || check));
2500}
2501
2502static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
2503 __wsum psum)
2504{
2505 if (NAPI_GRO_CB(skb)->csum_valid &&
2506 !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
2507 return 0;
2508
2509 NAPI_GRO_CB(skb)->csum = psum;
2510
2511 return __skb_gro_checksum_complete(skb);
2512}
2513
573e8fca
TH
2514static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
2515{
662880f4
TH
2516 if (NAPI_GRO_CB(skb)->csum_cnt > 0) {
2517 /* Consume a checksum from CHECKSUM_UNNECESSARY */
2518 NAPI_GRO_CB(skb)->csum_cnt--;
2519 } else {
2520 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
2521 * verified a new top level checksum or an encapsulated one
2522 * during GRO. This saves work if we fallback to normal path.
2523 */
2524 __skb_incr_checksum_unnecessary(skb);
573e8fca
TH
2525 }
2526}
2527
2528#define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
2529 compute_pseudo) \
2530({ \
2531 __sum16 __ret = 0; \
2532 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
2533 __ret = __skb_gro_checksum_validate_complete(skb, \
2534 compute_pseudo(skb, proto)); \
5a212329
TH
2535 if (__ret) \
2536 __skb_mark_checksum_bad(skb); \
2537 else \
573e8fca
TH
2538 skb_gro_incr_csum_unnecessary(skb); \
2539 __ret; \
2540})
2541
2542#define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
2543 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
2544
2545#define skb_gro_checksum_validate_zero_check(skb, proto, check, \
2546 compute_pseudo) \
2547 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
2548
2549#define skb_gro_checksum_simple_validate(skb) \
2550 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
2551
d96535a1
TH
2552static inline bool __skb_gro_checksum_convert_check(struct sk_buff *skb)
2553{
2554 return (NAPI_GRO_CB(skb)->csum_cnt == 0 &&
2555 !NAPI_GRO_CB(skb)->csum_valid);
2556}
2557
2558static inline void __skb_gro_checksum_convert(struct sk_buff *skb,
2559 __sum16 check, __wsum pseudo)
2560{
2561 NAPI_GRO_CB(skb)->csum = ~pseudo;
2562 NAPI_GRO_CB(skb)->csum_valid = 1;
2563}
2564
2565#define skb_gro_checksum_try_convert(skb, proto, check, compute_pseudo) \
2566do { \
2567 if (__skb_gro_checksum_convert_check(skb)) \
2568 __skb_gro_checksum_convert(skb, check, \
2569 compute_pseudo(skb, proto)); \
2570} while (0)
2571
26c4f7da
TH
2572struct gro_remcsum {
2573 int offset;
2574 __wsum delta;
2575};
2576
2577static inline void skb_gro_remcsum_init(struct gro_remcsum *grc)
2578{
846cd667 2579 grc->offset = 0;
26c4f7da
TH
2580 grc->delta = 0;
2581}
2582
b7fe10e5
TH
2583static inline void *skb_gro_remcsum_process(struct sk_buff *skb, void *ptr,
2584 unsigned int off, size_t hdrlen,
2585 int start, int offset,
2586 struct gro_remcsum *grc,
2587 bool nopartial)
dcdc8994
TH
2588{
2589 __wsum delta;
b7fe10e5 2590 size_t plen = hdrlen + max_t(size_t, offset + sizeof(u16), start);
dcdc8994
TH
2591
2592 BUG_ON(!NAPI_GRO_CB(skb)->csum_valid);
2593
15e2396d 2594 if (!nopartial) {
b7fe10e5
TH
2595 NAPI_GRO_CB(skb)->gro_remcsum_start = off + hdrlen + start;
2596 return ptr;
2597 }
2598
2599 ptr = skb_gro_header_fast(skb, off);
2600 if (skb_gro_header_hard(skb, off + plen)) {
2601 ptr = skb_gro_header_slow(skb, off + plen, off);
2602 if (!ptr)
2603 return NULL;
15e2396d
TH
2604 }
2605
b7fe10e5
TH
2606 delta = remcsum_adjust(ptr + hdrlen, NAPI_GRO_CB(skb)->csum,
2607 start, offset);
dcdc8994
TH
2608
2609 /* Adjust skb->csum since we changed the packet */
dcdc8994 2610 NAPI_GRO_CB(skb)->csum = csum_add(NAPI_GRO_CB(skb)->csum, delta);
26c4f7da 2611
b7fe10e5 2612 grc->offset = off + hdrlen + offset;
26c4f7da 2613 grc->delta = delta;
b7fe10e5
TH
2614
2615 return ptr;
dcdc8994
TH
2616}
2617
26c4f7da
TH
2618static inline void skb_gro_remcsum_cleanup(struct sk_buff *skb,
2619 struct gro_remcsum *grc)
2620{
b7fe10e5
TH
2621 void *ptr;
2622 size_t plen = grc->offset + sizeof(u16);
2623
26c4f7da
TH
2624 if (!grc->delta)
2625 return;
2626
b7fe10e5
TH
2627 ptr = skb_gro_header_fast(skb, grc->offset);
2628 if (skb_gro_header_hard(skb, grc->offset + sizeof(u16))) {
2629 ptr = skb_gro_header_slow(skb, plen, grc->offset);
2630 if (!ptr)
2631 return;
2632 }
2633
2634 remcsum_unadjust((__sum16 *)ptr, grc->delta);
26c4f7da 2635}
dcdc8994 2636
6ae23ad3
TH
2637struct skb_csum_offl_spec {
2638 __u16 ipv4_okay:1,
2639 ipv6_okay:1,
2640 encap_okay:1,
2641 ip_options_okay:1,
2642 ext_hdrs_okay:1,
2643 tcp_okay:1,
2644 udp_okay:1,
2645 sctp_okay:1,
2646 vlan_okay:1,
2647 no_encapped_ipv6:1,
2648 no_not_encapped:1;
2649};
2650
2651bool __skb_csum_offload_chk(struct sk_buff *skb,
2652 const struct skb_csum_offl_spec *spec,
2653 bool *csum_encapped,
2654 bool csum_help);
2655
2656static inline bool skb_csum_offload_chk(struct sk_buff *skb,
2657 const struct skb_csum_offl_spec *spec,
2658 bool *csum_encapped,
2659 bool csum_help)
2660{
2661 if (skb->ip_summed != CHECKSUM_PARTIAL)
2662 return false;
2663
2664 return __skb_csum_offload_chk(skb, spec, csum_encapped, csum_help);
2665}
2666
2667static inline bool skb_csum_offload_chk_help(struct sk_buff *skb,
2668 const struct skb_csum_offl_spec *spec)
2669{
2670 bool csum_encapped;
2671
2672 return skb_csum_offload_chk(skb, spec, &csum_encapped, true);
2673}
2674
2675static inline bool skb_csum_off_chk_help_cmn(struct sk_buff *skb)
2676{
2677 static const struct skb_csum_offl_spec csum_offl_spec = {
2678 .ipv4_okay = 1,
2679 .ip_options_okay = 1,
2680 .ipv6_okay = 1,
2681 .vlan_okay = 1,
2682 .tcp_okay = 1,
2683 .udp_okay = 1,
2684 };
2685
2686 return skb_csum_offload_chk_help(skb, &csum_offl_spec);
2687}
2688
2689static inline bool skb_csum_off_chk_help_cmn_v4_only(struct sk_buff *skb)
2690{
2691 static const struct skb_csum_offl_spec csum_offl_spec = {
2692 .ipv4_okay = 1,
2693 .ip_options_okay = 1,
2694 .tcp_okay = 1,
2695 .udp_okay = 1,
2696 .vlan_okay = 1,
2697 };
2698
2699 return skb_csum_offload_chk_help(skb, &csum_offl_spec);
2700}
2701
0c4e8581
SH
2702static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
2703 unsigned short type,
3b04ddde 2704 const void *daddr, const void *saddr,
95c96174 2705 unsigned int len)
0c4e8581 2706{
f1ecfd5d 2707 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 2708 return 0;
3b04ddde
SH
2709
2710 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
2711}
2712
b95cce35
SH
2713static inline int dev_parse_header(const struct sk_buff *skb,
2714 unsigned char *haddr)
2715{
2716 const struct net_device *dev = skb->dev;
2717
1b83336b 2718 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 2719 return 0;
3b04ddde 2720 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
2721}
2722
2793a23a
WB
2723/* ll_header must have at least hard_header_len allocated */
2724static inline bool dev_validate_header(const struct net_device *dev,
2725 char *ll_header, int len)
2726{
2727 if (likely(len >= dev->hard_header_len))
2728 return true;
2729
2730 if (capable(CAP_SYS_RAWIO)) {
2731 memset(ll_header + len, 0, dev->hard_header_len - len);
2732 return true;
2733 }
2734
2735 if (dev->header_ops && dev->header_ops->validate)
2736 return dev->header_ops->validate(ll_header, len);
2737
2738 return false;
2739}
2740
1da177e4 2741typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr, int len);
f629d208 2742int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
1da177e4
LT
2743static inline int unregister_gifconf(unsigned int family)
2744{
2745 return register_gifconf(family, NULL);
2746}
2747
99bbc707 2748#ifdef CONFIG_NET_FLOW_LIMIT
5f121b9a 2749#define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
99bbc707
WB
2750struct sd_flow_limit {
2751 u64 count;
2752 unsigned int num_buckets;
2753 unsigned int history_head;
2754 u16 history[FLOW_LIMIT_HISTORY];
2755 u8 buckets[];
2756};
2757
2758extern int netdev_flow_limit_table_len;
2759#endif /* CONFIG_NET_FLOW_LIMIT */
2760
1da177e4 2761/*
5e82b4b2 2762 * Incoming packets are placed on per-CPU queues
1da177e4 2763 */
d94d9fee 2764struct softnet_data {
1da177e4 2765 struct list_head poll_list;
6e7676c1 2766 struct sk_buff_head process_queue;
1da177e4 2767
dee42870 2768 /* stats */
cd7b5396
DM
2769 unsigned int processed;
2770 unsigned int time_squeeze;
cd7b5396 2771 unsigned int received_rps;
fd793d89 2772#ifdef CONFIG_RPS
88751275 2773 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
2774#endif
2775#ifdef CONFIG_NET_FLOW_LIMIT
2776 struct sd_flow_limit __rcu *flow_limit;
2777#endif
2778 struct Qdisc *output_queue;
2779 struct Qdisc **output_queue_tailp;
2780 struct sk_buff *completion_queue;
88751275 2781
4cdb1e2e 2782#ifdef CONFIG_RPS
501e7ef5
ED
2783 /* input_queue_head should be written by cpu owning this struct,
2784 * and only read by other cpus. Worth using a cache line.
2785 */
2786 unsigned int input_queue_head ____cacheline_aligned_in_smp;
2787
2788 /* Elements below can be accessed between CPUs for RPS/RFS */
0a9627f2 2789 struct call_single_data csd ____cacheline_aligned_in_smp;
88751275
ED
2790 struct softnet_data *rps_ipi_next;
2791 unsigned int cpu;
76cc8b13 2792 unsigned int input_queue_tail;
1e94d72f 2793#endif
95c96174 2794 unsigned int dropped;
0a9627f2 2795 struct sk_buff_head input_pkt_queue;
bea3348e 2796 struct napi_struct backlog;
99bbc707 2797
1da177e4
LT
2798};
2799
76cc8b13 2800static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
2801{
2802#ifdef CONFIG_RPS
76cc8b13
TH
2803 sd->input_queue_head++;
2804#endif
2805}
2806
2807static inline void input_queue_tail_incr_save(struct softnet_data *sd,
2808 unsigned int *qtail)
2809{
2810#ifdef CONFIG_RPS
2811 *qtail = ++sd->input_queue_tail;
fec5e652
TH
2812#endif
2813}
2814
0a9627f2 2815DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 2816
f629d208 2817void __netif_schedule(struct Qdisc *q);
46e5da40 2818void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 2819
fd2ea0a7
DM
2820static inline void netif_tx_schedule_all(struct net_device *dev)
2821{
2822 unsigned int i;
2823
2824 for (i = 0; i < dev->num_tx_queues; i++)
2825 netif_schedule_queue(netdev_get_tx_queue(dev, i));
2826}
2827
f9a7cbbf 2828static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 2829{
73466498 2830 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
2831}
2832
bea3348e
SH
2833/**
2834 * netif_start_queue - allow transmit
2835 * @dev: network device
2836 *
2837 * Allow upper layers to call the device hard_start_xmit routine.
2838 */
1da177e4
LT
2839static inline void netif_start_queue(struct net_device *dev)
2840{
e8a0464c 2841 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
2842}
2843
fd2ea0a7
DM
2844static inline void netif_tx_start_all_queues(struct net_device *dev)
2845{
2846 unsigned int i;
2847
2848 for (i = 0; i < dev->num_tx_queues; i++) {
2849 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
2850 netif_tx_start_queue(txq);
2851 }
2852}
2853
46e5da40 2854void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 2855
d29f749e
DJ
2856/**
2857 * netif_wake_queue - restart transmit
2858 * @dev: network device
2859 *
2860 * Allow upper layers to call the device hard_start_xmit routine.
2861 * Used for flow control when transmit resources are available.
2862 */
79d16385
DM
2863static inline void netif_wake_queue(struct net_device *dev)
2864{
e8a0464c 2865 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
2866}
2867
fd2ea0a7
DM
2868static inline void netif_tx_wake_all_queues(struct net_device *dev)
2869{
2870 unsigned int i;
2871
2872 for (i = 0; i < dev->num_tx_queues; i++) {
2873 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
2874 netif_tx_wake_queue(txq);
2875 }
2876}
2877
f9a7cbbf 2878static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 2879{
73466498 2880 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
2881}
2882
bea3348e
SH
2883/**
2884 * netif_stop_queue - stop transmitted packets
2885 * @dev: network device
2886 *
2887 * Stop upper layers calling the device hard_start_xmit routine.
2888 * Used for flow control when transmit resources are unavailable.
2889 */
1da177e4
LT
2890static inline void netif_stop_queue(struct net_device *dev)
2891{
e8a0464c 2892 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
2893}
2894
a2029240 2895void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 2896
4d29515f 2897static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 2898{
73466498 2899 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
2900}
2901
bea3348e
SH
2902/**
2903 * netif_queue_stopped - test if transmit queue is flowblocked
2904 * @dev: network device
2905 *
2906 * Test if transmit queue on device is currently unable to send.
2907 */
4d29515f 2908static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 2909{
e8a0464c 2910 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
2911}
2912
4d29515f 2913static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 2914{
73466498
TH
2915 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
2916}
2917
8e2f1a63
DB
2918static inline bool
2919netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
2920{
2921 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
2922}
2923
8e2f1a63
DB
2924static inline bool
2925netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
2926{
2927 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
2928}
2929
53511453
ED
2930/**
2931 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
2932 * @dev_queue: pointer to transmit queue
2933 *
2934 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 2935 * to give appropriate hint to the CPU.
53511453
ED
2936 */
2937static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
2938{
2939#ifdef CONFIG_BQL
2940 prefetchw(&dev_queue->dql.num_queued);
2941#endif
2942}
2943
2944/**
2945 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
2946 * @dev_queue: pointer to transmit queue
2947 *
2948 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 2949 * to give appropriate hint to the CPU.
53511453
ED
2950 */
2951static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
2952{
2953#ifdef CONFIG_BQL
2954 prefetchw(&dev_queue->dql.limit);
2955#endif
2956}
2957
c5d67bd7
TH
2958static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
2959 unsigned int bytes)
2960{
114cf580
TH
2961#ifdef CONFIG_BQL
2962 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
2963
2964 if (likely(dql_avail(&dev_queue->dql) >= 0))
2965 return;
2966
2967 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
2968
2969 /*
2970 * The XOFF flag must be set before checking the dql_avail below,
2971 * because in netdev_tx_completed_queue we update the dql_completed
2972 * before checking the XOFF flag.
2973 */
2974 smp_mb();
2975
2976 /* check again in case another CPU has just made room avail */
2977 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
2978 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 2979#endif
c5d67bd7
TH
2980}
2981
0042d0c8
FF
2982/**
2983 * netdev_sent_queue - report the number of bytes queued to hardware
2984 * @dev: network device
2985 * @bytes: number of bytes queued to the hardware device queue
2986 *
2987 * Report the number of bytes queued for sending/completion to the network
2988 * device hardware queue. @bytes should be a good approximation and should
2989 * exactly match netdev_completed_queue() @bytes
2990 */
c5d67bd7
TH
2991static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
2992{
2993 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
2994}
2995
2996static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 2997 unsigned int pkts, unsigned int bytes)
c5d67bd7 2998{
114cf580 2999#ifdef CONFIG_BQL
b37c0fbe
AD
3000 if (unlikely(!bytes))
3001 return;
3002
3003 dql_completed(&dev_queue->dql, bytes);
3004
3005 /*
3006 * Without the memory barrier there is a small possiblity that
3007 * netdev_tx_sent_queue will miss the update and cause the queue to
3008 * be stopped forever
3009 */
3010 smp_mb();
3011
3012 if (dql_avail(&dev_queue->dql) < 0)
3013 return;
3014
3015 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3016 netif_schedule_queue(dev_queue);
114cf580 3017#endif
c5d67bd7
TH
3018}
3019
0042d0c8
FF
3020/**
3021 * netdev_completed_queue - report bytes and packets completed by device
3022 * @dev: network device
3023 * @pkts: actual number of packets sent over the medium
3024 * @bytes: actual number of bytes sent over the medium
3025 *
3026 * Report the number of bytes and packets transmitted by the network device
3027 * hardware queue over the physical medium, @bytes must exactly match the
3028 * @bytes amount passed to netdev_sent_queue()
3029 */
c5d67bd7 3030static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3031 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3032{
3033 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3034}
3035
3036static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3037{
114cf580 3038#ifdef CONFIG_BQL
5c490354 3039 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3040 dql_reset(&q->dql);
3041#endif
c5d67bd7
TH
3042}
3043
0042d0c8
FF
3044/**
3045 * netdev_reset_queue - reset the packets and bytes count of a network device
3046 * @dev_queue: network device
3047 *
3048 * Reset the bytes and packet count of a network device and clear the
3049 * software flow control OFF bit for this network device
3050 */
c5d67bd7
TH
3051static inline void netdev_reset_queue(struct net_device *dev_queue)
3052{
3053 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3054}
3055
b9507bda
DB
3056/**
3057 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3058 * @dev: network device
3059 * @queue_index: given tx queue index
3060 *
3061 * Returns 0 if given tx queue index >= number of device tx queues,
3062 * otherwise returns the originally passed tx queue index.
3063 */
3064static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3065{
3066 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3067 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3068 dev->name, queue_index,
3069 dev->real_num_tx_queues);
3070 return 0;
3071 }
3072
3073 return queue_index;
3074}
3075
bea3348e
SH
3076/**
3077 * netif_running - test if up
3078 * @dev: network device
3079 *
3080 * Test if the device has been brought up.
3081 */
4d29515f 3082static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3083{
3084 return test_bit(__LINK_STATE_START, &dev->state);
3085}
3086
f25f4e44 3087/*
5e82b4b2 3088 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3089 * stop, and a check if it's stopped. All other device management is
3090 * done at the overall netdevice level.
3091 * Also test the device if we're multiqueue.
3092 */
bea3348e
SH
3093
3094/**
3095 * netif_start_subqueue - allow sending packets on subqueue
3096 * @dev: network device
3097 * @queue_index: sub queue index
3098 *
3099 * Start individual transmit queue of a device with multiple transmit queues.
3100 */
f25f4e44
PWJ
3101static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3102{
fd2ea0a7 3103 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3104
3105 netif_tx_start_queue(txq);
f25f4e44
PWJ
3106}
3107
bea3348e
SH
3108/**
3109 * netif_stop_subqueue - stop sending packets on subqueue
3110 * @dev: network device
3111 * @queue_index: sub queue index
3112 *
3113 * Stop individual transmit queue of a device with multiple transmit queues.
3114 */
f25f4e44
PWJ
3115static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3116{
fd2ea0a7 3117 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3118 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3119}
3120
bea3348e
SH
3121/**
3122 * netif_subqueue_stopped - test status of subqueue
3123 * @dev: network device
3124 * @queue_index: sub queue index
3125 *
3126 * Check individual transmit queue of a device with multiple transmit queues.
3127 */
4d29515f
DM
3128static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3129 u16 queue_index)
f25f4e44 3130{
fd2ea0a7 3131 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3132
3133 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3134}
3135
4d29515f
DM
3136static inline bool netif_subqueue_stopped(const struct net_device *dev,
3137 struct sk_buff *skb)
668f895a
PE
3138{
3139 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3140}
bea3348e 3141
46e5da40 3142void netif_wake_subqueue(struct net_device *dev, u16 queue_index);
f25f4e44 3143
537c00de 3144#ifdef CONFIG_XPS
53af53ae 3145int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3146 u16 index);
537c00de
AD
3147#else
3148static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3149 const struct cpumask *mask,
537c00de
AD
3150 u16 index)
3151{
3152 return 0;
3153}
3154#endif
3155
5605c762
JP
3156u16 __skb_tx_hash(const struct net_device *dev, struct sk_buff *skb,
3157 unsigned int num_tx_queues);
3158
a3d22a68
VZ
3159/*
3160 * Returns a Tx hash for the given packet when dev->real_num_tx_queues is used
3161 * as a distribution range limit for the returned value.
3162 */
3163static inline u16 skb_tx_hash(const struct net_device *dev,
0e001614 3164 struct sk_buff *skb)
a3d22a68
VZ
3165{
3166 return __skb_tx_hash(dev, skb, dev->real_num_tx_queues);
3167}
3168
bea3348e
SH
3169/**
3170 * netif_is_multiqueue - test if device has multiple transmit queues
3171 * @dev: network device
3172 *
3173 * Check if device has multiple transmit queues
bea3348e 3174 */
4d29515f 3175static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3176{
a02cec21 3177 return dev->num_tx_queues > 1;
f25f4e44 3178}
1da177e4 3179
f629d208 3180int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3181
a953be53 3182#ifdef CONFIG_SYSFS
f629d208 3183int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3184#else
3185static inline int netif_set_real_num_rx_queues(struct net_device *dev,
3186 unsigned int rxq)
3187{
3188 return 0;
3189}
3190#endif
3191
a953be53
MD
3192#ifdef CONFIG_SYSFS
3193static inline unsigned int get_netdev_rx_queue_index(
3194 struct netdev_rx_queue *queue)
3195{
3196 struct net_device *dev = queue->dev;
3197 int index = queue - dev->_rx;
3198
3199 BUG_ON(index >= dev->num_rx_queues);
3200 return index;
3201}
3202#endif
3203
16917b87 3204#define DEFAULT_MAX_NUM_RSS_QUEUES (8)
f629d208 3205int netif_get_num_default_rss_queues(void);
16917b87 3206
e6247027
ED
3207enum skb_free_reason {
3208 SKB_REASON_CONSUMED,
3209 SKB_REASON_DROPPED,
3210};
3211
3212void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3213void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3214
e6247027
ED
3215/*
3216 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3217 * interrupt context or with hardware interrupts being disabled.
3218 * (in_irq() || irqs_disabled())
3219 *
3220 * We provide four helpers that can be used in following contexts :
3221 *
3222 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3223 * replacing kfree_skb(skb)
3224 *
3225 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3226 * Typically used in place of consume_skb(skb) in TX completion path
3227 *
3228 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3229 * replacing kfree_skb(skb)
3230 *
3231 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3232 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3233 */
e6247027
ED
3234static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3235{
3236 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3237}
3238
3239static inline void dev_consume_skb_irq(struct sk_buff *skb)
3240{
3241 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3242}
3243
3244static inline void dev_kfree_skb_any(struct sk_buff *skb)
3245{
3246 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3247}
3248
3249static inline void dev_consume_skb_any(struct sk_buff *skb)
3250{
3251 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3252}
1da177e4 3253
f629d208
JP
3254int netif_rx(struct sk_buff *skb);
3255int netif_rx_ni(struct sk_buff *skb);
04eb4489 3256int netif_receive_skb(struct sk_buff *skb);
f629d208
JP
3257gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3258void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3259struct sk_buff *napi_get_frags(struct napi_struct *napi);
3260gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3261struct packet_offload *gro_find_receive_by_type(__be16 type);
3262struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3263
3264static inline void napi_free_frags(struct napi_struct *napi)
3265{
3266 kfree_skb(napi->skb);
3267 napi->skb = NULL;
3268}
3269
f629d208
JP
3270int netdev_rx_handler_register(struct net_device *dev,
3271 rx_handler_func_t *rx_handler,
3272 void *rx_handler_data);
3273void netdev_rx_handler_unregister(struct net_device *dev);
3274
3275bool dev_valid_name(const char *name);
3276int dev_ioctl(struct net *net, unsigned int cmd, void __user *);
3277int dev_ethtool(struct net *net, struct ifreq *);
3278unsigned int dev_get_flags(const struct net_device *);
3279int __dev_change_flags(struct net_device *, unsigned int flags);
3280int dev_change_flags(struct net_device *, unsigned int);
cb178190
DM
3281void __dev_notify_flags(struct net_device *, unsigned int old_flags,
3282 unsigned int gchanges);
f629d208
JP
3283int dev_change_name(struct net_device *, const char *);
3284int dev_set_alias(struct net_device *, const char *, size_t);
3285int dev_change_net_namespace(struct net_device *, struct net *, const char *);
3286int dev_set_mtu(struct net_device *, int);
3287void dev_set_group(struct net_device *, int);
3288int dev_set_mac_address(struct net_device *, struct sockaddr *);
3289int dev_change_carrier(struct net_device *, bool new_carrier);
3290int dev_get_phys_port_id(struct net_device *dev,
02637fce 3291 struct netdev_phys_item_id *ppid);
db24a904
DA
3292int dev_get_phys_port_name(struct net_device *dev,
3293 char *name, size_t len);
d746d707 3294int dev_change_proto_down(struct net_device *dev, bool proto_down);
a7862b45 3295int dev_change_xdp_fd(struct net_device *dev, int fd);
55a93b3e 3296struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev);
ce93718f
DM
3297struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3298 struct netdev_queue *txq, int *ret);
a0265d28 3299int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 3300int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
3301bool is_skb_forwardable(const struct net_device *dev,
3302 const struct sk_buff *skb);
1da177e4 3303
74b20582
DA
3304void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
3305
20380731 3306extern int netdev_budget;
1da177e4
LT
3307
3308/* Called by rtnetlink.c:rtnl_unlock() */
f629d208 3309void netdev_run_todo(void);
1da177e4 3310
bea3348e
SH
3311/**
3312 * dev_put - release reference to device
3313 * @dev: network device
3314 *
9ef4429b 3315 * Release reference to device to allow it to be freed.
bea3348e 3316 */
1da177e4
LT
3317static inline void dev_put(struct net_device *dev)
3318{
933393f5 3319 this_cpu_dec(*dev->pcpu_refcnt);
1da177e4
LT
3320}
3321
bea3348e
SH
3322/**
3323 * dev_hold - get reference to device
3324 * @dev: network device
3325 *
9ef4429b 3326 * Hold reference to device to keep it from being freed.
bea3348e 3327 */
15333061
SH
3328static inline void dev_hold(struct net_device *dev)
3329{
933393f5 3330 this_cpu_inc(*dev->pcpu_refcnt);
15333061 3331}
1da177e4
LT
3332
3333/* Carrier loss detection, dial on demand. The functions netif_carrier_on
3334 * and _off may be called from IRQ context, but it is caller
3335 * who is responsible for serialization of these calls.
b00055aa
SR
3336 *
3337 * The name carrier is inappropriate, these functions should really be
3338 * called netif_lowerlayer_*() because they represent the state of any
3339 * kind of lower layer not just hardware media.
1da177e4
LT
3340 */
3341
f629d208
JP
3342void linkwatch_init_dev(struct net_device *dev);
3343void linkwatch_fire_event(struct net_device *dev);
3344void linkwatch_forget_dev(struct net_device *dev);
1da177e4 3345
bea3348e
SH
3346/**
3347 * netif_carrier_ok - test if carrier present
3348 * @dev: network device
3349 *
3350 * Check if carrier is present on device
3351 */
4d29515f 3352static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
3353{
3354 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
3355}
3356
f629d208 3357unsigned long dev_trans_start(struct net_device *dev);
9d21493b 3358
f629d208 3359void __netdev_watchdog_up(struct net_device *dev);
1da177e4 3360
f629d208 3361void netif_carrier_on(struct net_device *dev);
1da177e4 3362
f629d208 3363void netif_carrier_off(struct net_device *dev);
1da177e4 3364
bea3348e
SH
3365/**
3366 * netif_dormant_on - mark device as dormant.
3367 * @dev: network device
3368 *
3369 * Mark device as dormant (as per RFC2863).
3370 *
3371 * The dormant state indicates that the relevant interface is not
3372 * actually in a condition to pass packets (i.e., it is not 'up') but is
3373 * in a "pending" state, waiting for some external event. For "on-
3374 * demand" interfaces, this new state identifies the situation where the
3375 * interface is waiting for events to place it in the up state.
bea3348e 3376 */
b00055aa
SR
3377static inline void netif_dormant_on(struct net_device *dev)
3378{
3379 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
3380 linkwatch_fire_event(dev);
3381}
3382
bea3348e
SH
3383/**
3384 * netif_dormant_off - set device as not dormant.
3385 * @dev: network device
3386 *
3387 * Device is not in dormant state.
3388 */
b00055aa
SR
3389static inline void netif_dormant_off(struct net_device *dev)
3390{
3391 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
3392 linkwatch_fire_event(dev);
3393}
3394
bea3348e
SH
3395/**
3396 * netif_dormant - test if carrier present
3397 * @dev: network device
3398 *
3399 * Check if carrier is present on device
3400 */
4d29515f 3401static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
3402{
3403 return test_bit(__LINK_STATE_DORMANT, &dev->state);
3404}
3405
3406
bea3348e
SH
3407/**
3408 * netif_oper_up - test if device is operational
3409 * @dev: network device
3410 *
3411 * Check if carrier is operational
3412 */
4d29515f 3413static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 3414{
b00055aa
SR
3415 return (dev->operstate == IF_OPER_UP ||
3416 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
3417}
3418
bea3348e
SH
3419/**
3420 * netif_device_present - is device available or removed
3421 * @dev: network device
3422 *
3423 * Check if device has not been removed from system.
3424 */
4d29515f 3425static inline bool netif_device_present(struct net_device *dev)
1da177e4
LT
3426{
3427 return test_bit(__LINK_STATE_PRESENT, &dev->state);
3428}
3429
f629d208 3430void netif_device_detach(struct net_device *dev);
1da177e4 3431
f629d208 3432void netif_device_attach(struct net_device *dev);
1da177e4
LT
3433
3434/*
3435 * Network interface message level settings
3436 */
1da177e4
LT
3437
3438enum {
3439 NETIF_MSG_DRV = 0x0001,
3440 NETIF_MSG_PROBE = 0x0002,
3441 NETIF_MSG_LINK = 0x0004,
3442 NETIF_MSG_TIMER = 0x0008,
3443 NETIF_MSG_IFDOWN = 0x0010,
3444 NETIF_MSG_IFUP = 0x0020,
3445 NETIF_MSG_RX_ERR = 0x0040,
3446 NETIF_MSG_TX_ERR = 0x0080,
3447 NETIF_MSG_TX_QUEUED = 0x0100,
3448 NETIF_MSG_INTR = 0x0200,
3449 NETIF_MSG_TX_DONE = 0x0400,
3450 NETIF_MSG_RX_STATUS = 0x0800,
3451 NETIF_MSG_PKTDATA = 0x1000,
3452 NETIF_MSG_HW = 0x2000,
3453 NETIF_MSG_WOL = 0x4000,
3454};
3455
3456#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
3457#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
3458#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
3459#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
3460#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
3461#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
3462#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
3463#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
3464#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
3465#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
3466#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
3467#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
3468#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
3469#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
3470#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
3471
3472static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
3473{
3474 /* use default */
3475 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
3476 return default_msg_enable_bits;
3477 if (debug_value == 0) /* no output */
3478 return 0;
3479 /* set low N bits */
3480 return (1 << debug_value) - 1;
3481}
3482
c773e847 3483static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 3484{
c773e847
DM
3485 spin_lock(&txq->_xmit_lock);
3486 txq->xmit_lock_owner = cpu;
22dd7495
JHS
3487}
3488
fd2ea0a7
DM
3489static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
3490{
3491 spin_lock_bh(&txq->_xmit_lock);
3492 txq->xmit_lock_owner = smp_processor_id();
3493}
3494
4d29515f 3495static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 3496{
4d29515f 3497 bool ok = spin_trylock(&txq->_xmit_lock);
c3f26a26
DM
3498 if (likely(ok))
3499 txq->xmit_lock_owner = smp_processor_id();
3500 return ok;
3501}
3502
3503static inline void __netif_tx_unlock(struct netdev_queue *txq)
3504{
3505 txq->xmit_lock_owner = -1;
3506 spin_unlock(&txq->_xmit_lock);
3507}
3508
3509static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
3510{
3511 txq->xmit_lock_owner = -1;
3512 spin_unlock_bh(&txq->_xmit_lock);
3513}
3514
08baf561
ED
3515static inline void txq_trans_update(struct netdev_queue *txq)
3516{
3517 if (txq->xmit_lock_owner != -1)
3518 txq->trans_start = jiffies;
3519}
3520
ba162f8e
FW
3521/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
3522static inline void netif_trans_update(struct net_device *dev)
3523{
9b36627a
FW
3524 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
3525
3526 if (txq->trans_start != jiffies)
3527 txq->trans_start = jiffies;
ba162f8e
FW
3528}
3529
d29f749e
DJ
3530/**
3531 * netif_tx_lock - grab network device transmit lock
3532 * @dev: network device
d29f749e
DJ
3533 *
3534 * Get network device transmit lock
3535 */
22dd7495
JHS
3536static inline void netif_tx_lock(struct net_device *dev)
3537{
e8a0464c 3538 unsigned int i;
c3f26a26 3539 int cpu;
c773e847 3540
c3f26a26
DM
3541 spin_lock(&dev->tx_global_lock);
3542 cpu = smp_processor_id();
e8a0464c
DM
3543 for (i = 0; i < dev->num_tx_queues; i++) {
3544 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
3545
3546 /* We are the only thread of execution doing a
3547 * freeze, but we have to grab the _xmit_lock in
3548 * order to synchronize with threads which are in
3549 * the ->hard_start_xmit() handler and already
3550 * checked the frozen bit.
3551 */
e8a0464c 3552 __netif_tx_lock(txq, cpu);
c3f26a26
DM
3553 set_bit(__QUEUE_STATE_FROZEN, &txq->state);
3554 __netif_tx_unlock(txq);
e8a0464c 3555 }
932ff279
HX
3556}
3557
3558static inline void netif_tx_lock_bh(struct net_device *dev)
3559{
e8a0464c
DM
3560 local_bh_disable();
3561 netif_tx_lock(dev);
932ff279
HX
3562}
3563
932ff279
HX
3564static inline void netif_tx_unlock(struct net_device *dev)
3565{
e8a0464c
DM
3566 unsigned int i;
3567
3568 for (i = 0; i < dev->num_tx_queues; i++) {
3569 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c773e847 3570
c3f26a26
DM
3571 /* No need to grab the _xmit_lock here. If the
3572 * queue is not stopped for another reason, we
3573 * force a schedule.
3574 */
3575 clear_bit(__QUEUE_STATE_FROZEN, &txq->state);
7b3d3e4f 3576 netif_schedule_queue(txq);
c3f26a26
DM
3577 }
3578 spin_unlock(&dev->tx_global_lock);
932ff279
HX
3579}
3580
3581static inline void netif_tx_unlock_bh(struct net_device *dev)
3582{
e8a0464c
DM
3583 netif_tx_unlock(dev);
3584 local_bh_enable();
932ff279
HX
3585}
3586
c773e847 3587#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 3588 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 3589 __netif_tx_lock(txq, cpu); \
22dd7495
JHS
3590 } \
3591}
3592
5efeac44
EB
3593#define HARD_TX_TRYLOCK(dev, txq) \
3594 (((dev->features & NETIF_F_LLTX) == 0) ? \
3595 __netif_tx_trylock(txq) : \
3596 true )
3597
c773e847 3598#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 3599 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 3600 __netif_tx_unlock(txq); \
22dd7495
JHS
3601 } \
3602}
3603
1da177e4
LT
3604static inline void netif_tx_disable(struct net_device *dev)
3605{
fd2ea0a7 3606 unsigned int i;
c3f26a26 3607 int cpu;
fd2ea0a7 3608
c3f26a26
DM
3609 local_bh_disable();
3610 cpu = smp_processor_id();
fd2ea0a7
DM
3611 for (i = 0; i < dev->num_tx_queues; i++) {
3612 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
3613
3614 __netif_tx_lock(txq, cpu);
fd2ea0a7 3615 netif_tx_stop_queue(txq);
c3f26a26 3616 __netif_tx_unlock(txq);
fd2ea0a7 3617 }
c3f26a26 3618 local_bh_enable();
1da177e4
LT
3619}
3620
e308a5d8
DM
3621static inline void netif_addr_lock(struct net_device *dev)
3622{
3623 spin_lock(&dev->addr_list_lock);
3624}
3625
2429f7ac
JP
3626static inline void netif_addr_lock_nested(struct net_device *dev)
3627{
25175ba5
VY
3628 int subclass = SINGLE_DEPTH_NESTING;
3629
3630 if (dev->netdev_ops->ndo_get_lock_subclass)
3631 subclass = dev->netdev_ops->ndo_get_lock_subclass(dev);
3632
3633 spin_lock_nested(&dev->addr_list_lock, subclass);
2429f7ac
JP
3634}
3635
e308a5d8
DM
3636static inline void netif_addr_lock_bh(struct net_device *dev)
3637{
3638 spin_lock_bh(&dev->addr_list_lock);
3639}
3640
3641static inline void netif_addr_unlock(struct net_device *dev)
3642{
3643 spin_unlock(&dev->addr_list_lock);
3644}
3645
3646static inline void netif_addr_unlock_bh(struct net_device *dev)
3647{
3648 spin_unlock_bh(&dev->addr_list_lock);
3649}
3650
f001fde5 3651/*
31278e71 3652 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
3653 * rcu_read_lock held.
3654 */
3655#define for_each_dev_addr(dev, ha) \
31278e71 3656 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 3657
1da177e4
LT
3658/* These functions live elsewhere (drivers/net/net_init.c, but related) */
3659
f629d208 3660void ether_setup(struct net_device *dev);
1da177e4
LT
3661
3662/* Support for loadable net-drivers */
f629d208 3663struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 3664 unsigned char name_assign_type,
f629d208
JP
3665 void (*setup)(struct net_device *),
3666 unsigned int txqs, unsigned int rxqs);
c835a677
TG
3667#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
3668 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 3669
c835a677
TG
3670#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
3671 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
3672 count)
36909ea4 3673
f629d208
JP
3674int register_netdev(struct net_device *dev);
3675void unregister_netdev(struct net_device *dev);
f001fde5 3676
22bedad3 3677/* General hardware address lists handling functions */
f629d208
JP
3678int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
3679 struct netdev_hw_addr_list *from_list, int addr_len);
3680void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
3681 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
3682int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
3683 struct net_device *dev,
3684 int (*sync)(struct net_device *, const unsigned char *),
3685 int (*unsync)(struct net_device *,
3686 const unsigned char *));
3687void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
3688 struct net_device *dev,
3689 int (*unsync)(struct net_device *,
3690 const unsigned char *));
f629d208 3691void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 3692
f001fde5 3693/* Functions used for device addresses handling */
f629d208
JP
3694int dev_addr_add(struct net_device *dev, const unsigned char *addr,
3695 unsigned char addr_type);
3696int dev_addr_del(struct net_device *dev, const unsigned char *addr,
3697 unsigned char addr_type);
f629d208
JP
3698void dev_addr_flush(struct net_device *dev);
3699int dev_addr_init(struct net_device *dev);
a748ee24
JP
3700
3701/* Functions used for unicast addresses handling */
f629d208
JP
3702int dev_uc_add(struct net_device *dev, const unsigned char *addr);
3703int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
3704int dev_uc_del(struct net_device *dev, const unsigned char *addr);
3705int dev_uc_sync(struct net_device *to, struct net_device *from);
3706int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
3707void dev_uc_unsync(struct net_device *to, struct net_device *from);
3708void dev_uc_flush(struct net_device *dev);
3709void dev_uc_init(struct net_device *dev);
f001fde5 3710
670e5b8e
AD
3711/**
3712 * __dev_uc_sync - Synchonize device's unicast list
3713 * @dev: device to sync
3714 * @sync: function to call if address should be added
3715 * @unsync: function to call if address should be removed
3716 *
3717 * Add newly added addresses to the interface, and release
3718 * addresses that have been deleted.
5e82b4b2 3719 */
670e5b8e
AD
3720static inline int __dev_uc_sync(struct net_device *dev,
3721 int (*sync)(struct net_device *,
3722 const unsigned char *),
3723 int (*unsync)(struct net_device *,
3724 const unsigned char *))
3725{
3726 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
3727}
3728
3729/**
e793c0f7 3730 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
3731 * @dev: device to sync
3732 * @unsync: function to call if address should be removed
3733 *
3734 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 3735 */
670e5b8e
AD
3736static inline void __dev_uc_unsync(struct net_device *dev,
3737 int (*unsync)(struct net_device *,
3738 const unsigned char *))
3739{
3740 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
3741}
3742
22bedad3 3743/* Functions used for multicast addresses handling */
f629d208
JP
3744int dev_mc_add(struct net_device *dev, const unsigned char *addr);
3745int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
3746int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
3747int dev_mc_del(struct net_device *dev, const unsigned char *addr);
3748int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
3749int dev_mc_sync(struct net_device *to, struct net_device *from);
3750int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
3751void dev_mc_unsync(struct net_device *to, struct net_device *from);
3752void dev_mc_flush(struct net_device *dev);
3753void dev_mc_init(struct net_device *dev);
f001fde5 3754
670e5b8e
AD
3755/**
3756 * __dev_mc_sync - Synchonize device's multicast list
3757 * @dev: device to sync
3758 * @sync: function to call if address should be added
3759 * @unsync: function to call if address should be removed
3760 *
3761 * Add newly added addresses to the interface, and release
3762 * addresses that have been deleted.
5e82b4b2 3763 */
670e5b8e
AD
3764static inline int __dev_mc_sync(struct net_device *dev,
3765 int (*sync)(struct net_device *,
3766 const unsigned char *),
3767 int (*unsync)(struct net_device *,
3768 const unsigned char *))
3769{
3770 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
3771}
3772
3773/**
e793c0f7 3774 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
3775 * @dev: device to sync
3776 * @unsync: function to call if address should be removed
3777 *
3778 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 3779 */
670e5b8e
AD
3780static inline void __dev_mc_unsync(struct net_device *dev,
3781 int (*unsync)(struct net_device *,
3782 const unsigned char *))
3783{
3784 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
3785}
3786
4417da66 3787/* Functions used for secondary unicast and multicast support */
f629d208
JP
3788void dev_set_rx_mode(struct net_device *dev);
3789void __dev_set_rx_mode(struct net_device *dev);
3790int dev_set_promiscuity(struct net_device *dev, int inc);
3791int dev_set_allmulti(struct net_device *dev, int inc);
3792void netdev_state_change(struct net_device *dev);
3793void netdev_notify_peers(struct net_device *dev);
3794void netdev_features_change(struct net_device *dev);
1da177e4 3795/* Load a device via the kmod */
f629d208
JP
3796void dev_load(struct net *net, const char *name);
3797struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
3798 struct rtnl_link_stats64 *storage);
3799void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
3800 const struct net_device_stats *netdev_stats);
eeda3fd6 3801
1da177e4 3802extern int netdev_max_backlog;
3b098e2d 3803extern int netdev_tstamp_prequeue;
1da177e4 3804extern int weight_p;
9ff162a8 3805
f629d208 3806bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
3807struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
3808 struct list_head **iter);
f629d208
JP
3809struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
3810 struct list_head **iter);
8b5be856 3811
44a40855
VY
3812/* iterate through upper list, must be called under RCU read lock */
3813#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
3814 for (iter = &(dev)->adj_list.upper, \
3815 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
3816 updev; \
3817 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
3818
8b5be856 3819/* iterate through upper list, must be called under RCU read lock */
2f268f12
VF
3820#define netdev_for_each_all_upper_dev_rcu(dev, updev, iter) \
3821 for (iter = &(dev)->all_adj_list.upper, \
3822 updev = netdev_all_upper_get_next_dev_rcu(dev, &(iter)); \
3823 updev; \
3824 updev = netdev_all_upper_get_next_dev_rcu(dev, &(iter)))
8b5be856 3825
f629d208
JP
3826void *netdev_lower_get_next_private(struct net_device *dev,
3827 struct list_head **iter);
3828void *netdev_lower_get_next_private_rcu(struct net_device *dev,
3829 struct list_head **iter);
31088a11
VF
3830
3831#define netdev_for_each_lower_private(dev, priv, iter) \
3832 for (iter = (dev)->adj_list.lower.next, \
3833 priv = netdev_lower_get_next_private(dev, &(iter)); \
3834 priv; \
3835 priv = netdev_lower_get_next_private(dev, &(iter)))
3836
3837#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
3838 for (iter = &(dev)->adj_list.lower, \
3839 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
3840 priv; \
3841 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
3842
4085ebe8
VY
3843void *netdev_lower_get_next(struct net_device *dev,
3844 struct list_head **iter);
7ce856aa 3845
4085ebe8 3846#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 3847 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
3848 ldev = netdev_lower_get_next(dev, &(iter)); \
3849 ldev; \
3850 ldev = netdev_lower_get_next(dev, &(iter)))
3851
7ce856aa
JP
3852struct net_device *netdev_all_lower_get_next(struct net_device *dev,
3853 struct list_head **iter);
3854struct net_device *netdev_all_lower_get_next_rcu(struct net_device *dev,
3855 struct list_head **iter);
3856
3857#define netdev_for_each_all_lower_dev(dev, ldev, iter) \
3858 for (iter = (dev)->all_adj_list.lower.next, \
3859 ldev = netdev_all_lower_get_next(dev, &(iter)); \
3860 ldev; \
3861 ldev = netdev_all_lower_get_next(dev, &(iter)))
3862
3863#define netdev_for_each_all_lower_dev_rcu(dev, ldev, iter) \
3864 for (iter = (dev)->all_adj_list.lower.next, \
3865 ldev = netdev_all_lower_get_next_rcu(dev, &(iter)); \
3866 ldev; \
3867 ldev = netdev_all_lower_get_next_rcu(dev, &(iter)))
3868
f629d208 3869void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 3870void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
3871struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
3872struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
3873int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev);
3874int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 3875 struct net_device *upper_dev,
29bf24af 3876 void *upper_priv, void *upper_info);
f629d208
JP
3877void netdev_upper_dev_unlink(struct net_device *dev,
3878 struct net_device *upper_dev);
5bb025fa 3879void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
3880void *netdev_lower_dev_get_private(struct net_device *dev,
3881 struct net_device *lower_dev);
04d48266
JP
3882void netdev_lower_state_changed(struct net_device *lower_dev,
3883 void *lower_state_info);
18bfb924
JP
3884int netdev_default_l2upper_neigh_construct(struct net_device *dev,
3885 struct neighbour *n);
3886void netdev_default_l2upper_neigh_destroy(struct net_device *dev,
3887 struct neighbour *n);
960fb622
ED
3888
3889/* RSS keys are 40 or 52 bytes long */
3890#define NETDEV_RSS_KEY_LEN 52
ba905f5e 3891extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
3892void netdev_rss_key_fill(void *buffer, size_t len);
3893
4085ebe8 3894int dev_get_nest_level(struct net_device *dev,
b618aaa9 3895 bool (*type_check)(const struct net_device *dev));
f629d208
JP
3896int skb_checksum_help(struct sk_buff *skb);
3897struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
3898 netdev_features_t features, bool tx_path);
3899struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
3900 netdev_features_t features);
12b0004d 3901
61bd3857
MS
3902struct netdev_bonding_info {
3903 ifslave slave;
3904 ifbond master;
3905};
3906
3907struct netdev_notifier_bonding_info {
3908 struct netdev_notifier_info info; /* must be first */
3909 struct netdev_bonding_info bonding_info;
3910};
3911
3912void netdev_bonding_info_change(struct net_device *dev,
3913 struct netdev_bonding_info *bonding_info);
3914
12b0004d
CW
3915static inline
3916struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
3917{
3918 return __skb_gso_segment(skb, features, true);
3919}
53d6471c 3920__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
3921
3922static inline bool can_checksum_protocol(netdev_features_t features,
3923 __be16 protocol)
3924{
c8cd0989
TH
3925 if (protocol == htons(ETH_P_FCOE))
3926 return !!(features & NETIF_F_FCOE_CRC);
3927
3928 /* Assume this is an IP checksum (not SCTP CRC) */
3929
3930 if (features & NETIF_F_HW_CSUM) {
3931 /* Can checksum everything */
3932 return true;
3933 }
3934
3935 switch (protocol) {
3936 case htons(ETH_P_IP):
3937 return !!(features & NETIF_F_IP_CSUM);
3938 case htons(ETH_P_IPV6):
3939 return !!(features & NETIF_F_IPV6_CSUM);
3940 default:
3941 return false;
3942 }
ec5f0615 3943}
12b0004d 3944
6ae23ad3
TH
3945/* Map an ethertype into IP protocol if possible */
3946static inline int eproto_to_ipproto(int eproto)
3947{
3948 switch (eproto) {
3949 case htons(ETH_P_IP):
3950 return IPPROTO_IP;
3951 case htons(ETH_P_IPV6):
3952 return IPPROTO_IPV6;
3953 default:
3954 return -1;
3955 }
3956}
3957
fb286bb2 3958#ifdef CONFIG_BUG
f629d208 3959void netdev_rx_csum_fault(struct net_device *dev);
fb286bb2
HX
3960#else
3961static inline void netdev_rx_csum_fault(struct net_device *dev)
3962{
3963}
3964#endif
1da177e4 3965/* rx skb timestamps */
f629d208
JP
3966void net_enable_timestamp(void);
3967void net_disable_timestamp(void);
1da177e4 3968
20380731 3969#ifdef CONFIG_PROC_FS
f629d208 3970int __init dev_proc_init(void);
900ff8c6
CW
3971#else
3972#define dev_proc_init() 0
20380731
ACM
3973#endif
3974
4798248e 3975static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
3976 struct sk_buff *skb, struct net_device *dev,
3977 bool more)
4798248e 3978{
fa2dbdc2 3979 skb->xmit_more = more ? 1 : 0;
0b725a2c 3980 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
3981}
3982
10b3ad8c 3983static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 3984 struct netdev_queue *txq, bool more)
4798248e
DM
3985{
3986 const struct net_device_ops *ops = dev->netdev_ops;
10b3ad8c 3987 int rc;
4798248e 3988
fa2dbdc2 3989 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
3990 if (rc == NETDEV_TX_OK)
3991 txq_trans_update(txq);
3992
3993 return rc;
4798248e
DM
3994}
3995
42a2d923
LT
3996int netdev_class_create_file_ns(struct class_attribute *class_attr,
3997 const void *ns);
3998void netdev_class_remove_file_ns(struct class_attribute *class_attr,
3999 const void *ns);
58292cbe
TH
4000
4001static inline int netdev_class_create_file(struct class_attribute *class_attr)
4002{
4003 return netdev_class_create_file_ns(class_attr, NULL);
4004}
4005
4006static inline void netdev_class_remove_file(struct class_attribute *class_attr)
4007{
4008 netdev_class_remove_file_ns(class_attr, NULL);
4009}
b8a9787e 4010
04600794
JB
4011extern struct kobj_ns_type_operations net_ns_type_operations;
4012
f629d208 4013const char *netdev_drivername(const struct net_device *dev);
6579e57b 4014
f629d208 4015void linkwatch_run_queue(void);
20380731 4016
da08143b
MK
4017static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4018 netdev_features_t f2)
4019{
c8cd0989
TH
4020 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4021 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4022 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4023 else
b6a0e72a 4024 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4025 }
da08143b 4026
c8cd0989 4027 return f1 & f2;
da08143b
MK
4028}
4029
c8f44aff
MM
4030static inline netdev_features_t netdev_get_wanted_features(
4031 struct net_device *dev)
5455c699
MM
4032{
4033 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4034}
c8f44aff
MM
4035netdev_features_t netdev_increment_features(netdev_features_t all,
4036 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4037
4038/* Allow TSO being used on stacked device :
4039 * Performing the GSO segmentation before last device
4040 * is a performance improvement.
4041 */
4042static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4043 netdev_features_t mask)
4044{
4045 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4046}
4047
6cb6a27c 4048int __netdev_update_features(struct net_device *dev);
5455c699 4049void netdev_update_features(struct net_device *dev);
afe12cc8 4050void netdev_change_features(struct net_device *dev);
7f353bf2 4051
fc4a7489
PM
4052void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4053 struct net_device *dev);
4054
e38f3025
TM
4055netdev_features_t passthru_features_check(struct sk_buff *skb,
4056 struct net_device *dev,
4057 netdev_features_t features);
c1e756bf 4058netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 4059
4d29515f 4060static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 4061{
7b748340 4062 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
4063
4064 /* check flags correspondence */
4065 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
4066 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_UFO >> NETIF_F_GSO_SHIFT));
4067 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
4068 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 4069 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
4070 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
4071 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4072 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
4073 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
4074 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
4075 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4076 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
4077 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 4078 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 4079 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 4080 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
0345e186 4081
d6b4991a 4082 return (features & feature) == feature;
576a30eb
HX
4083}
4084
4d29515f 4085static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 4086{
278b2513 4087 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 4088 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
4089}
4090
8b86a61d 4091static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 4092 netdev_features_t features)
7967168c 4093{
fc741216 4094 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
4095 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
4096 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
4097}
4098
82cc1a7a
PWJ
4099static inline void netif_set_gso_max_size(struct net_device *dev,
4100 unsigned int size)
4101{
4102 dev->gso_max_size = size;
4103}
4104
7a7ffbab
WCC
4105static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
4106 int pulled_hlen, u16 mac_offset,
4107 int mac_len)
4108{
4109 skb->protocol = protocol;
4110 skb->encapsulation = 1;
4111 skb_push(skb, pulled_hlen);
4112 skb_reset_transport_header(skb);
4113 skb->mac_header = mac_offset;
4114 skb->network_header = skb->mac_header + mac_len;
4115 skb->mac_len = mac_len;
4116}
4117
3c175784
SD
4118static inline bool netif_is_macsec(const struct net_device *dev)
4119{
4120 return dev->priv_flags & IFF_MACSEC;
4121}
4122
b618aaa9 4123static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
4124{
4125 return dev->priv_flags & IFF_MACVLAN;
4126}
4127
b618aaa9 4128static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
4129{
4130 return dev->priv_flags & IFF_MACVLAN_PORT;
4131}
4132
b618aaa9 4133static inline bool netif_is_ipvlan(const struct net_device *dev)
5933fea7
MB
4134{
4135 return dev->priv_flags & IFF_IPVLAN_SLAVE;
4136}
4137
b618aaa9 4138static inline bool netif_is_ipvlan_port(const struct net_device *dev)
5933fea7
MB
4139{
4140 return dev->priv_flags & IFF_IPVLAN_MASTER;
4141}
4142
b618aaa9 4143static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 4144{
4145 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
4146}
4147
b618aaa9 4148static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
4149{
4150 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
4151}
4152
3bdc0eba
BG
4153static inline bool netif_supports_nofcs(struct net_device *dev)
4154{
4155 return dev->priv_flags & IFF_SUPP_NOFCS;
4156}
4157
007979ea 4158static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 4159{
007979ea 4160 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
4161}
4162
fee6d4c7
DA
4163static inline bool netif_is_l3_slave(const struct net_device *dev)
4164{
4165 return dev->priv_flags & IFF_L3MDEV_SLAVE;
4166}
4167
0894ae3f
JP
4168static inline bool netif_is_bridge_master(const struct net_device *dev)
4169{
4170 return dev->priv_flags & IFF_EBRIDGE;
4171}
4172
28f9ee22
VY
4173static inline bool netif_is_bridge_port(const struct net_device *dev)
4174{
4175 return dev->priv_flags & IFF_BRIDGE_PORT;
4176}
4177
35d4e172
JP
4178static inline bool netif_is_ovs_master(const struct net_device *dev)
4179{
4180 return dev->priv_flags & IFF_OPENVSWITCH;
4181}
4182
b618aaa9 4183static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
4184{
4185 return dev->priv_flags & IFF_TEAM;
4186}
4187
b618aaa9 4188static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
4189{
4190 return dev->priv_flags & IFF_TEAM_PORT;
4191}
4192
b618aaa9 4193static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
4194{
4195 return netif_is_bond_master(dev) || netif_is_team_master(dev);
4196}
4197
b618aaa9 4198static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
4199{
4200 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
4201}
4202
d4ab4286
KJ
4203static inline bool netif_is_rxfh_configured(const struct net_device *dev)
4204{
4205 return dev->priv_flags & IFF_RXFH_CONFIGURED;
4206}
4207
02875878
ED
4208/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
4209static inline void netif_keep_dst(struct net_device *dev)
4210{
4211 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
4212}
4213
18d3df3e
PA
4214/* return true if dev can't cope with mtu frames that need vlan tag insertion */
4215static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
4216{
4217 /* TODO: reserve and use an additional IFF bit, if we get more users */
4218 return dev->priv_flags & IFF_MACSEC;
4219}
4220
505d4f73 4221extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 4222
571ba423
JP
4223/* Logging, debugging and troubleshooting/diagnostic helpers. */
4224
4225/* netdev_printk helpers, similar to dev_printk */
4226
4227static inline const char *netdev_name(const struct net_device *dev)
4228{
c6f854d5
VF
4229 if (!dev->name[0] || strchr(dev->name, '%'))
4230 return "(unnamed net_device)";
571ba423
JP
4231 return dev->name;
4232}
4233
ccc7f496
VF
4234static inline const char *netdev_reg_state(const struct net_device *dev)
4235{
4236 switch (dev->reg_state) {
4237 case NETREG_UNINITIALIZED: return " (uninitialized)";
4238 case NETREG_REGISTERED: return "";
4239 case NETREG_UNREGISTERING: return " (unregistering)";
4240 case NETREG_UNREGISTERED: return " (unregistered)";
4241 case NETREG_RELEASED: return " (released)";
4242 case NETREG_DUMMY: return " (dummy)";
4243 }
4244
4245 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
4246 return " (unknown)";
4247}
4248
f629d208 4249__printf(3, 4)
6ea754eb
JP
4250void netdev_printk(const char *level, const struct net_device *dev,
4251 const char *format, ...);
f629d208 4252__printf(2, 3)
6ea754eb 4253void netdev_emerg(const struct net_device *dev, const char *format, ...);
f629d208 4254__printf(2, 3)
6ea754eb 4255void netdev_alert(const struct net_device *dev, const char *format, ...);
f629d208 4256__printf(2, 3)
6ea754eb 4257void netdev_crit(const struct net_device *dev, const char *format, ...);
f629d208 4258__printf(2, 3)
6ea754eb 4259void netdev_err(const struct net_device *dev, const char *format, ...);
f629d208 4260__printf(2, 3)
6ea754eb 4261void netdev_warn(const struct net_device *dev, const char *format, ...);
f629d208 4262__printf(2, 3)
6ea754eb 4263void netdev_notice(const struct net_device *dev, const char *format, ...);
f629d208 4264__printf(2, 3)
6ea754eb 4265void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 4266
8909c9ad
VK
4267#define MODULE_ALIAS_NETDEV(device) \
4268 MODULE_ALIAS("netdev-" device)
4269
b558c96f 4270#if defined(CONFIG_DYNAMIC_DEBUG)
571ba423
JP
4271#define netdev_dbg(__dev, format, args...) \
4272do { \
ffa10cb4 4273 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 4274} while (0)
b558c96f
JC
4275#elif defined(DEBUG)
4276#define netdev_dbg(__dev, format, args...) \
4277 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
4278#else
4279#define netdev_dbg(__dev, format, args...) \
4280({ \
4281 if (0) \
4282 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
4283})
4284#endif
4285
4286#if defined(VERBOSE_DEBUG)
4287#define netdev_vdbg netdev_dbg
4288#else
4289
4290#define netdev_vdbg(dev, format, args...) \
4291({ \
4292 if (0) \
4293 netdev_printk(KERN_DEBUG, dev, format, ##args); \
4294 0; \
4295})
4296#endif
4297
4298/*
4299 * netdev_WARN() acts like dev_printk(), but with the key difference
4300 * of using a WARN/WARN_ON to get the message out, including the
4301 * file/line information and a backtrace.
4302 */
4303#define netdev_WARN(dev, format, args...) \
ccc7f496
VF
4304 WARN(1, "netdevice: %s%s\n" format, netdev_name(dev), \
4305 netdev_reg_state(dev), ##args)
571ba423 4306
b3d95c5c
JP
4307/* netif printk helpers, similar to netdev_printk */
4308
4309#define netif_printk(priv, type, level, dev, fmt, args...) \
4310do { \
4311 if (netif_msg_##type(priv)) \
4312 netdev_printk(level, (dev), fmt, ##args); \
4313} while (0)
4314
f45f4321
JP
4315#define netif_level(level, priv, type, dev, fmt, args...) \
4316do { \
4317 if (netif_msg_##type(priv)) \
4318 netdev_##level(dev, fmt, ##args); \
4319} while (0)
4320
b3d95c5c 4321#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 4322 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 4323#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 4324 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 4325#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 4326 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 4327#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 4328 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 4329#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 4330 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 4331#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 4332 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 4333#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 4334 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 4335
0053ea9c 4336#if defined(CONFIG_DYNAMIC_DEBUG)
b3d95c5c
JP
4337#define netif_dbg(priv, type, netdev, format, args...) \
4338do { \
4339 if (netif_msg_##type(priv)) \
b5fb0a03 4340 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 4341} while (0)
0053ea9c
JP
4342#elif defined(DEBUG)
4343#define netif_dbg(priv, type, dev, format, args...) \
4344 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
4345#else
4346#define netif_dbg(priv, type, dev, format, args...) \
4347({ \
4348 if (0) \
4349 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
4350 0; \
4351})
4352#endif
4353
4354#if defined(VERBOSE_DEBUG)
bcfcc450 4355#define netif_vdbg netif_dbg
b3d95c5c
JP
4356#else
4357#define netif_vdbg(priv, type, dev, format, args...) \
4358({ \
4359 if (0) \
a4ed89cb 4360 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
4361 0; \
4362})
4363#endif
571ba423 4364
900ff8c6
CW
4365/*
4366 * The list of packet types we will receive (as opposed to discard)
4367 * and the routines to invoke.
4368 *
4369 * Why 16. Because with 16 the only overlap we get on a hash of the
4370 * low nibble of the protocol value is RARP/SNAP/X.25.
4371 *
4372 * NOTE: That is no longer true with the addition of VLAN tags. Not
4373 * sure which should go first, but I bet it won't make much
4374 * difference if we are running VLANs. The good news is that
4375 * this protocol won't be in the list unless compiled in, so
4376 * the average user (w/out VLANs) will not be adversely affected.
4377 * --BLG
4378 *
4379 * 0800 IP
4380 * 8100 802.1Q VLAN
4381 * 0001 802.3
4382 * 0002 AX.25
4383 * 0004 802.2
4384 * 8035 RARP
4385 * 0005 SNAP
4386 * 0805 X.25
4387 * 0806 ARP
4388 * 8137 IPX
4389 * 0009 Localtalk
4390 * 86DD IPv6
4391 */
4392#define PTYPE_HASH_SIZE (16)
4393#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
4394
385a154c 4395#endif /* _LINUX_NETDEVICE_H */