net: provide dev_lstats_read() helper
[linux-2.6-block.git] / include / linux / netdevice.h
CommitLineData
2874c5fd 1/* SPDX-License-Identifier: GPL-2.0-or-later */
1da177e4
LT
2/*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the Interfaces handler.
8 *
9 * Version: @(#)dev.h 1.0.10 08/12/93
10 *
02c30a84 11 * Authors: Ross Biro
1da177e4
LT
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Corey Minyard <wf-rch!minyard@relay.EU.net>
14 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 15 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
16 * Bjorn Ekwall. <bj0rn@blox.se>
17 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 *
1da177e4
LT
19 * Moved to /usr/include/linux for NET3
20 */
21#ifndef _LINUX_NETDEVICE_H
22#define _LINUX_NETDEVICE_H
23
d7fe0f24 24#include <linux/timer.h>
187f1882 25#include <linux/bug.h>
bea3348e 26#include <linux/delay.h>
60063497 27#include <linux/atomic.h>
53511453 28#include <linux/prefetch.h>
1da177e4
LT
29#include <asm/cache.h>
30#include <asm/byteorder.h>
31
1da177e4 32#include <linux/percpu.h>
4d5b78c0 33#include <linux/rculist.h>
bea3348e 34#include <linux/workqueue.h>
114cf580 35#include <linux/dynamic_queue_limits.h>
1da177e4 36
b1b67dd4 37#include <linux/ethtool.h>
a050c33f 38#include <net/net_namespace.h>
7a6b6f51 39#ifdef CONFIG_DCB
2f90b865
AD
40#include <net/dcbnl.h>
41#endif
5bc1421e 42#include <net/netprio_cgroup.h>
e817f856 43#include <net/xdp.h>
a050c33f 44
a59e2ecb 45#include <linux/netdev_features.h>
77162022 46#include <linux/neighbour.h>
607ca46e 47#include <uapi/linux/netdevice.h>
61bd3857 48#include <uapi/linux/if_bonding.h>
e4c6734e 49#include <uapi/linux/pkt_cls.h>
59cc1f61 50#include <linux/hashtable.h>
a59e2ecb 51
115c1d6e 52struct netpoll_info;
313162d0 53struct device;
c1f19b51 54struct phy_device;
2f657a60 55struct dsa_port;
c6e970a0 56
e679c9c1 57struct sfp_bus;
704232c2
JB
58/* 802.11 specific */
59struct wireless_dev;
98a18b6f
AA
60/* 802.15.4 specific */
61struct wpan_dev;
03c57747 62struct mpls_dev;
7c46a640
AD
63/* UDP Tunnel offloads */
64struct udp_tunnel_info;
a7862b45 65struct bpf_prog;
814abfab 66struct xdp_buff;
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 192#include <linux/static_key.h>
dc05360f
ED
193extern struct static_key_false rps_needed;
194extern struct static_key_false rfs_needed;
adc9300e
ED
195#endif
196
1da177e4
LT
197struct neighbour;
198struct neigh_parms;
199struct sk_buff;
200
f001fde5
JP
201struct netdev_hw_addr {
202 struct list_head list;
203 unsigned char addr[MAX_ADDR_LEN];
204 unsigned char type;
ccffad25
JP
205#define NETDEV_HW_ADDR_T_LAN 1
206#define NETDEV_HW_ADDR_T_SAN 2
207#define NETDEV_HW_ADDR_T_SLAVE 3
208#define NETDEV_HW_ADDR_T_UNICAST 4
22bedad3 209#define NETDEV_HW_ADDR_T_MULTICAST 5
22bedad3 210 bool global_use;
4cd729b0 211 int sync_cnt;
8f8f103d 212 int refcount;
4543fbef 213 int synced;
f001fde5
JP
214 struct rcu_head rcu_head;
215};
216
31278e71
JP
217struct netdev_hw_addr_list {
218 struct list_head list;
219 int count;
220};
221
22bedad3
JP
222#define netdev_hw_addr_list_count(l) ((l)->count)
223#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
224#define netdev_hw_addr_list_for_each(ha, l) \
225 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 226
22bedad3
JP
227#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
228#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
229#define netdev_for_each_uc_addr(ha, dev) \
230 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
6683ece3 231
22bedad3
JP
232#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
233#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 234#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 235 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
6683ece3 236
d94d9fee 237struct hh_cache {
5b3dc2f3 238 unsigned int hh_len;
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);
e78b2915 273 __be16 (*parse_protocol)(const struct sk_buff *skb);
3b04ddde
SH
274};
275
1da177e4 276/* These flag bits are private to the generic network queueing
5e82b4b2 277 * layer; they may not be explicitly referenced by any other
1da177e4
LT
278 * code.
279 */
280
d94d9fee 281enum netdev_state_t {
1da177e4
LT
282 __LINK_STATE_START,
283 __LINK_STATE_PRESENT,
1da177e4 284 __LINK_STATE_NOCARRIER,
b00055aa
SR
285 __LINK_STATE_LINKWATCH_PENDING,
286 __LINK_STATE_DORMANT,
1da177e4
LT
287};
288
289
290/*
5e82b4b2 291 * This structure holds boot-time configured netdevice settings. They
fe2918b0 292 * are then used in the device probing.
1da177e4
LT
293 */
294struct netdev_boot_setup {
295 char name[IFNAMSIZ];
296 struct ifmap map;
297};
298#define NETDEV_BOOT_SETUP_MAX 8
299
f629d208 300int __init netdev_boot_setup(char *str);
1da177e4 301
6312fe77
LR
302struct gro_list {
303 struct list_head list;
304 int count;
305};
306
bea3348e 307/*
d9f37d01
LR
308 * size of gro hash buckets, must less than bit number of
309 * napi_struct::gro_bitmask
bea3348e 310 */
07d78363 311#define GRO_HASH_BUCKETS 8
d9f37d01
LR
312
313/*
314 * Structure for NAPI scheduling similar to tasklet but with weighting
315 */
bea3348e
SH
316struct napi_struct {
317 /* The poll_list must only be managed by the entity which
318 * changes the state of the NAPI_STATE_SCHED bit. This means
319 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 320 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
321 * can remove from the list right before clearing the bit.
322 */
323 struct list_head poll_list;
324
325 unsigned long state;
326 int weight;
d9f37d01 327 unsigned long gro_bitmask;
bea3348e
SH
328 int (*poll)(struct napi_struct *, int);
329#ifdef CONFIG_NETPOLL
bea3348e 330 int poll_owner;
bea3348e 331#endif
5d38a079 332 struct net_device *dev;
6312fe77 333 struct gro_list gro_hash[GRO_HASH_BUCKETS];
5d38a079 334 struct sk_buff *skb;
323ebb61
EC
335 struct list_head rx_list; /* Pending GRO_NORMAL skbs */
336 int rx_count; /* length of rx_list */
3b47d303 337 struct hrtimer timer;
404f7c9e 338 struct list_head dev_list;
af12fa6e
ET
339 struct hlist_node napi_hash_node;
340 unsigned int napi_id;
bea3348e
SH
341};
342
d94d9fee 343enum {
bea3348e 344 NAPI_STATE_SCHED, /* Poll is scheduled */
39e6c820 345 NAPI_STATE_MISSED, /* reschedule a napi */
a0a46196 346 NAPI_STATE_DISABLE, /* Disable pending */
7b363e44 347 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
d64b5e85
ED
348 NAPI_STATE_HASHED, /* In NAPI hash (busy polling possible) */
349 NAPI_STATE_NO_BUSY_POLL,/* Do not add in napi_hash, no busy polling */
217f6974
ED
350 NAPI_STATE_IN_BUSY_POLL,/* sk_busy_loop() owns this NAPI */
351};
352
353enum {
39e6c820
ED
354 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
355 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
356 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
357 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
358 NAPIF_STATE_HASHED = BIT(NAPI_STATE_HASHED),
359 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
360 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
bea3348e
SH
361};
362
5b252f0c 363enum gro_result {
d1c76af9
HX
364 GRO_MERGED,
365 GRO_MERGED_FREE,
366 GRO_HELD,
367 GRO_NORMAL,
368 GRO_DROP,
25393d3f 369 GRO_CONSUMED,
d1c76af9 370};
5b252f0c 371typedef enum gro_result gro_result_t;
d1c76af9 372
8a4eb573
JP
373/*
374 * enum rx_handler_result - Possible return values for rx_handlers.
375 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
376 * further.
377 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
378 * case skb->dev was changed by rx_handler.
379 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 380 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
381 *
382 * rx_handlers are functions called from inside __netif_receive_skb(), to do
383 * special processing of the skb, prior to delivery to protocol handlers.
384 *
385 * Currently, a net_device can only have a single rx_handler registered. Trying
386 * to register a second rx_handler will return -EBUSY.
387 *
388 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
389 * To unregister a rx_handler on a net_device, use
390 * netdev_rx_handler_unregister().
391 *
392 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
393 * do with the skb.
394 *
5e82b4b2 395 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 396 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 397 * the skb to be delivered in some other way.
8a4eb573
JP
398 *
399 * If the rx_handler changed skb->dev, to divert the skb to another
400 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
401 * new device will be called if it exists.
402 *
5e82b4b2 403 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 404 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 405 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 406 *
5e82b4b2 407 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
408 * delivered, it should return RX_HANDLER_PASS.
409 *
410 * A device without a registered rx_handler will behave as if rx_handler
411 * returned RX_HANDLER_PASS.
412 */
413
414enum rx_handler_result {
415 RX_HANDLER_CONSUMED,
416 RX_HANDLER_ANOTHER,
417 RX_HANDLER_EXACT,
418 RX_HANDLER_PASS,
419};
420typedef enum rx_handler_result rx_handler_result_t;
421typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 422
f629d208 423void __napi_schedule(struct napi_struct *n);
bc9ad166 424void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 425
4d29515f 426static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
427{
428 return test_bit(NAPI_STATE_DISABLE, &n->state);
429}
430
39e6c820 431bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
432
433/**
434 * napi_schedule - schedule NAPI poll
5e82b4b2 435 * @n: NAPI context
bea3348e
SH
436 *
437 * Schedule NAPI poll routine to be called if it is not already
438 * running.
439 */
440static inline void napi_schedule(struct napi_struct *n)
441{
442 if (napi_schedule_prep(n))
443 __napi_schedule(n);
444}
445
bc9ad166
ED
446/**
447 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 448 * @n: NAPI context
bc9ad166
ED
449 *
450 * Variant of napi_schedule(), assuming hard irqs are masked.
451 */
452static inline void napi_schedule_irqoff(struct napi_struct *n)
453{
454 if (napi_schedule_prep(n))
455 __napi_schedule_irqoff(n);
456}
457
bfe13f54 458/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 459static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
460{
461 if (napi_schedule_prep(napi)) {
462 __napi_schedule(napi);
4d29515f 463 return true;
bfe13f54 464 }
4d29515f 465 return false;
bfe13f54
RD
466}
467
364b6055 468bool napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
469/**
470 * napi_complete - NAPI processing complete
5e82b4b2 471 * @n: NAPI context
bea3348e
SH
472 *
473 * Mark NAPI processing as complete.
3b47d303 474 * Consider using napi_complete_done() instead.
364b6055 475 * Return false if device should avoid rearming interrupts.
bea3348e 476 */
364b6055 477static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
478{
479 return napi_complete_done(n, 0);
480}
bea3348e 481
af12fa6e
ET
482/**
483 * napi_hash_del - remove a NAPI from global table
5e82b4b2 484 * @napi: NAPI context
af12fa6e 485 *
5e82b4b2 486 * Warning: caller must observe RCU grace period
34cbe27e
ED
487 * before freeing memory containing @napi, if
488 * this function returns true.
93d05d4a 489 * Note: core networking stack automatically calls it
5e82b4b2 490 * from netif_napi_del().
93d05d4a 491 * Drivers might want to call this helper to combine all
5e82b4b2 492 * the needed RCU grace periods into a single one.
af12fa6e 493 */
34cbe27e 494bool napi_hash_del(struct napi_struct *napi);
af12fa6e 495
bea3348e
SH
496/**
497 * napi_disable - prevent NAPI from scheduling
5e82b4b2 498 * @n: NAPI context
bea3348e
SH
499 *
500 * Stop NAPI from being scheduled on this context.
501 * Waits till any outstanding processing completes.
502 */
3b47d303 503void napi_disable(struct napi_struct *n);
bea3348e
SH
504
505/**
506 * napi_enable - enable NAPI scheduling
5e82b4b2 507 * @n: NAPI context
bea3348e
SH
508 *
509 * Resume NAPI from being scheduled on this context.
510 * Must be paired with napi_disable.
511 */
512static inline void napi_enable(struct napi_struct *n)
513{
514 BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
4e857c58 515 smp_mb__before_atomic();
bea3348e 516 clear_bit(NAPI_STATE_SCHED, &n->state);
2d8bff12 517 clear_bit(NAPI_STATE_NPSVC, &n->state);
bea3348e
SH
518}
519
c264c3de
SH
520/**
521 * napi_synchronize - wait until NAPI is not running
5e82b4b2 522 * @n: NAPI context
c264c3de
SH
523 *
524 * Wait until NAPI is done being scheduled on this context.
525 * Waits till any outstanding processing completes but
526 * does not disable future activations.
527 */
528static inline void napi_synchronize(const struct napi_struct *n)
529{
facc432f
AB
530 if (IS_ENABLED(CONFIG_SMP))
531 while (test_bit(NAPI_STATE_SCHED, &n->state))
532 msleep(1);
533 else
534 barrier();
c264c3de 535}
c264c3de 536
6c5c9581
MK
537/**
538 * napi_if_scheduled_mark_missed - if napi is running, set the
539 * NAPIF_STATE_MISSED
540 * @n: NAPI context
541 *
542 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
543 * NAPI is scheduled.
544 **/
545static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
546{
547 unsigned long val, new;
548
549 do {
550 val = READ_ONCE(n->state);
551 if (val & NAPIF_STATE_DISABLE)
552 return true;
553
554 if (!(val & NAPIF_STATE_SCHED))
555 return false;
556
557 new = val | NAPIF_STATE_MISSED;
558 } while (cmpxchg(&n->state, val, new) != val);
559
560 return true;
561}
562
d94d9fee 563enum netdev_queue_state_t {
73466498
TH
564 __QUEUE_STATE_DRV_XOFF,
565 __QUEUE_STATE_STACK_XOFF,
c3f26a26 566 __QUEUE_STATE_FROZEN,
79d16385 567};
8e2f1a63
DB
568
569#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
570#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
571#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
572
573#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
574#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
575 QUEUE_STATE_FROZEN)
576#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
577 QUEUE_STATE_FROZEN)
578
73466498
TH
579/*
580 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
581 * netif_tx_* functions below are used to manipulate this flag. The
582 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
583 * queue independently. The netif_xmit_*stopped functions below are called
584 * to check if the queue has been stopped by the driver or stack (either
585 * of the XOFF bits are set in the state). Drivers should not need to call
586 * netif_xmit*stopped functions, they should only be using netif_tx_*.
587 */
79d16385 588
bb949fbd 589struct netdev_queue {
6a321cb3 590/*
5e82b4b2 591 * read-mostly part
6a321cb3 592 */
bb949fbd 593 struct net_device *dev;
46e5da40 594 struct Qdisc __rcu *qdisc;
b0e1e646 595 struct Qdisc *qdisc_sleeping;
ccf5ff69 596#ifdef CONFIG_SYSFS
1d24eb48
TH
597 struct kobject kobj;
598#endif
f2cd2d3e
ED
599#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
600 int numa_node;
601#endif
c0ef079c
FW
602 unsigned long tx_maxrate;
603 /*
604 * Number of TX timeouts for this queue
605 * (/sys/class/net/DEV/Q/trans_timeout)
606 */
607 unsigned long trans_timeout;
ffcfe25b
AD
608
609 /* Subordinate device that the queue has been assigned to */
610 struct net_device *sb_dev;
661b8d1b
MK
611#ifdef CONFIG_XDP_SOCKETS
612 struct xdp_umem *umem;
613#endif
6a321cb3 614/*
5e82b4b2 615 * write-mostly part
6a321cb3
ED
616 */
617 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
618 int xmit_lock_owner;
9d21493b 619 /*
9b36627a 620 * Time (in jiffies) of last Tx
9d21493b
ED
621 */
622 unsigned long trans_start;
ccf5ff69 623
114cf580
TH
624 unsigned long state;
625
626#ifdef CONFIG_BQL
627 struct dql dql;
628#endif
e8a0464c 629} ____cacheline_aligned_in_smp;
bb949fbd 630
79134e6c 631extern int sysctl_fb_tunnels_only_for_init_net;
856c395c 632extern int sysctl_devconf_inherit_init_net;
79134e6c
ED
633
634static inline bool net_has_fallback_tunnels(const struct net *net)
635{
be9fc097
AB
636 return net == &init_net ||
637 !IS_ENABLED(CONFIG_SYSCTL) ||
638 !sysctl_fb_tunnels_only_for_init_net;
79134e6c
ED
639}
640
f2cd2d3e
ED
641static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
642{
643#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
644 return q->numa_node;
645#else
b236da69 646 return NUMA_NO_NODE;
f2cd2d3e
ED
647#endif
648}
649
650static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
651{
652#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
653 q->numa_node = node;
654#endif
655}
656
df334545 657#ifdef CONFIG_RPS
0a9627f2
TH
658/*
659 * This structure holds an RPS map which can be of variable length. The
660 * map is an array of CPUs.
661 */
662struct rps_map {
663 unsigned int len;
664 struct rcu_head rcu;
665 u16 cpus[0];
666};
60b778ce 667#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 668
fec5e652 669/*
c445477d
BH
670 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
671 * tail pointer for that CPU's input queue at the time of last enqueue, and
672 * a hardware filter index.
fec5e652
TH
673 */
674struct rps_dev_flow {
675 u16 cpu;
c445477d 676 u16 filter;
fec5e652
TH
677 unsigned int last_qtail;
678};
c445477d 679#define RPS_NO_FILTER 0xffff
fec5e652
TH
680
681/*
682 * The rps_dev_flow_table structure contains a table of flow mappings.
683 */
684struct rps_dev_flow_table {
685 unsigned int mask;
686 struct rcu_head rcu;
fec5e652
TH
687 struct rps_dev_flow flows[0];
688};
689#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 690 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
691
692/*
693 * The rps_sock_flow_table contains mappings of flows to the last CPU
694 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
695 * Each entry is a 32bit value. Upper part is the high-order bits
696 * of flow hash, lower part is CPU number.
567e4b79 697 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
698 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
699 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 700 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
701 */
702struct rps_sock_flow_table {
567e4b79 703 u32 mask;
93c1af6c
ED
704
705 u32 ents[0] ____cacheline_aligned_in_smp;
fec5e652 706};
567e4b79 707#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
708
709#define RPS_NO_CPU 0xffff
710
567e4b79
ED
711extern u32 rps_cpu_mask;
712extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
713
fec5e652
TH
714static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
715 u32 hash)
716{
717 if (table && hash) {
567e4b79
ED
718 unsigned int index = hash & table->mask;
719 u32 val = hash & ~rps_cpu_mask;
fec5e652 720
5e82b4b2 721 /* We only give a hint, preemption can change CPU under us */
567e4b79 722 val |= raw_smp_processor_id();
fec5e652 723
567e4b79
ED
724 if (table->ents[index] != val)
725 table->ents[index] = val;
fec5e652
TH
726 }
727}
728
c445477d 729#ifdef CONFIG_RFS_ACCEL
f629d208
JP
730bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
731 u16 filter_id);
c445477d 732#endif
a953be53 733#endif /* CONFIG_RPS */
c445477d 734
0a9627f2
TH
735/* This structure contains an instance of an RX queue. */
736struct netdev_rx_queue {
a953be53 737#ifdef CONFIG_RPS
6e3f7faf
ED
738 struct rps_map __rcu *rps_map;
739 struct rps_dev_flow_table __rcu *rps_flow_table;
a953be53 740#endif
6e3f7faf 741 struct kobject kobj;
fe822240 742 struct net_device *dev;
e817f856 743 struct xdp_rxq_info xdp_rxq;
661b8d1b
MK
744#ifdef CONFIG_XDP_SOCKETS
745 struct xdp_umem *umem;
746#endif
0a9627f2 747} ____cacheline_aligned_in_smp;
a953be53
MD
748
749/*
750 * RX queue sysfs structures and functions.
751 */
752struct rx_queue_attribute {
753 struct attribute attr;
718ad681 754 ssize_t (*show)(struct netdev_rx_queue *queue, char *buf);
a953be53 755 ssize_t (*store)(struct netdev_rx_queue *queue,
718ad681 756 const char *buf, size_t len);
a953be53 757};
d314774c 758
bf264145
TH
759#ifdef CONFIG_XPS
760/*
761 * This structure holds an XPS map which can be of variable length. The
762 * map is an array of queues.
763 */
764struct xps_map {
765 unsigned int len;
766 unsigned int alloc_len;
767 struct rcu_head rcu;
768 u16 queues[0];
769};
60b778ce 770#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
771#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
772 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
773
774/*
775 * This structure holds all XPS maps for device. Maps are indexed by CPU.
776 */
777struct xps_dev_maps {
778 struct rcu_head rcu;
80d19669 779 struct xps_map __rcu *attr_map[0]; /* Either CPUs map or RXQs map */
bf264145 780};
80d19669
AN
781
782#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
184c449f 783 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
80d19669
AN
784
785#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
786 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
787
bf264145
TH
788#endif /* CONFIG_XPS */
789
4f57c087
JF
790#define TC_MAX_QUEUE 16
791#define TC_BITMASK 15
792/* HW offloaded queuing disciplines txq count and offset maps */
793struct netdev_tc_txq {
794 u16 count;
795 u16 offset;
796};
797
68bad94e
NP
798#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
799/*
800 * This structure is to hold information about the device
801 * configured to run FCoE protocol stack.
802 */
803struct netdev_fcoe_hbainfo {
804 char manufacturer[64];
805 char serial_number[64];
806 char hardware_version[64];
807 char driver_version[64];
808 char optionrom_version[64];
809 char firmware_version[64];
810 char model[256];
811 char model_description[256];
812};
813#endif
814
02637fce 815#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 816
02637fce
JP
817/* This structure holds a unique identifier to identify some
818 * physical item (port for example) used by a netdevice.
66b52b0d 819 */
02637fce
JP
820struct netdev_phys_item_id {
821 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
822 unsigned char id_len;
823};
824
d754f98b
SF
825static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
826 struct netdev_phys_item_id *b)
827{
828 return a->id_len == b->id_len &&
829 memcmp(a->id, b->id, a->id_len) == 0;
830}
831
99932d4f 832typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
8ec56fc3
AD
833 struct sk_buff *skb,
834 struct net_device *sb_dev);
99932d4f 835
2572ac53 836enum tc_setup_type {
575ed7d3 837 TC_SETUP_QDISC_MQPRIO,
a1b7c5fd 838 TC_SETUP_CLSU32,
5b33f488 839 TC_SETUP_CLSFLOWER,
ade9b658 840 TC_SETUP_CLSMATCHALL,
332ae8e2 841 TC_SETUP_CLSBPF,
8c4083b3 842 TC_SETUP_BLOCK,
8521db4c 843 TC_SETUP_QDISC_CBS,
602f3baf 844 TC_SETUP_QDISC_RED,
7fdb61b4 845 TC_SETUP_QDISC_PRIO,
f971b132 846 TC_SETUP_QDISC_MQ,
25db26a9 847 TC_SETUP_QDISC_ETF,
98b0e5f6 848 TC_SETUP_ROOT_QDISC,
890d8d23 849 TC_SETUP_QDISC_GRED,
9c66d156 850 TC_SETUP_QDISC_TAPRIO,
16e5cc64
JF
851};
852
f4e63525
JK
853/* These structures hold the attributes of bpf state that are being passed
854 * to the netdevice through the bpf op.
a7862b45 855 */
f4e63525 856enum bpf_netdev_command {
a7862b45
BB
857 /* Set or clear a bpf program used in the earliest stages of packet
858 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
859 * is responsible for calling bpf_prog_put on any old progs that are
860 * stored. In case of error, the callee need not release the new prog
861 * reference, but on success it takes ownership and must bpf_prog_put
862 * when it is no longer used.
863 */
864 XDP_SETUP_PROG,
ee5d032f 865 XDP_SETUP_PROG_HW,
a7862b45 866 XDP_QUERY_PROG,
a25717d2 867 XDP_QUERY_PROG_HW,
ab3f0063 868 /* BPF program for offload callbacks, invoked at program load time. */
a3884572
JK
869 BPF_OFFLOAD_MAP_ALLOC,
870 BPF_OFFLOAD_MAP_FREE,
74515c57 871 XDP_SETUP_XSK_UMEM,
a7862b45
BB
872};
873
cae1927c 874struct bpf_prog_offload_ops;
ddf9f970 875struct netlink_ext_ack;
74515c57 876struct xdp_umem;
ddf9f970 877
f4e63525
JK
878struct netdev_bpf {
879 enum bpf_netdev_command command;
a7862b45
BB
880 union {
881 /* XDP_SETUP_PROG */
ddf9f970 882 struct {
32d60277 883 u32 flags;
ddf9f970
JK
884 struct bpf_prog *prog;
885 struct netlink_ext_ack *extack;
886 };
a25717d2 887 /* XDP_QUERY_PROG, XDP_QUERY_PROG_HW */
58038695 888 struct {
58038695 889 u32 prog_id;
92f0292b
JK
890 /* flags with which program was installed */
891 u32 prog_flags;
58038695 892 };
a3884572
JK
893 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
894 struct {
895 struct bpf_offloaded_map *offmap;
896 };
f8ebfaf6 897 /* XDP_SETUP_XSK_UMEM */
74515c57 898 struct {
f8ebfaf6
JS
899 struct xdp_umem *umem;
900 u16 queue_id;
74515c57 901 } xsk;
a7862b45
BB
902 };
903};
16e5cc64 904
9116e5e2
MK
905/* Flags for ndo_xsk_wakeup. */
906#define XDP_WAKEUP_RX (1 << 0)
907#define XDP_WAKEUP_TX (1 << 1)
908
d77e38e6
SK
909#ifdef CONFIG_XFRM_OFFLOAD
910struct xfrmdev_ops {
911 int (*xdo_dev_state_add) (struct xfrm_state *x);
912 void (*xdo_dev_state_delete) (struct xfrm_state *x);
913 void (*xdo_dev_state_free) (struct xfrm_state *x);
914 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
915 struct xfrm_state *x);
50bd870a 916 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
d77e38e6
SK
917};
918#endif
919
6c557001
FW
920struct dev_ifalias {
921 struct rcu_head rcuhead;
922 char ifalias[];
923};
924
b473b0d2 925struct devlink;
da68b4ad 926struct tlsdev_ops;
b473b0d2 927
ff927412
JP
928struct netdev_name_node {
929 struct hlist_node hlist;
36fbf1e5 930 struct list_head list;
ff927412
JP
931 struct net_device *dev;
932 const char *name;
933};
934
36fbf1e5
JP
935int netdev_name_node_alt_create(struct net_device *dev, const char *name);
936int netdev_name_node_alt_destroy(struct net_device *dev, const char *name);
937
d314774c
SH
938/*
939 * This structure defines the management hooks for network devices.
00829823
SH
940 * The following hooks can be defined; unless noted otherwise, they are
941 * optional and can be filled with a null pointer.
d314774c
SH
942 *
943 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
944 * This function is called once when a network device is registered.
945 * The network device can use this for any late stage initialization
946 * or semantic validation. It can fail with an error code which will
947 * be propagated back to register_netdev.
d314774c
SH
948 *
949 * void (*ndo_uninit)(struct net_device *dev);
950 * This function is called when device is unregistered or when registration
951 * fails. It is not called if init fails.
952 *
953 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 954 * This function is called when a network device transitions to the up
d314774c
SH
955 * state.
956 *
957 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 958 * This function is called when a network device transitions to the down
d314774c
SH
959 * state.
960 *
dc1f8bf6
SH
961 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
962 * struct net_device *dev);
00829823 963 * Called when a packet needs to be transmitted.
e79d8429
RR
964 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
965 * the queue before that can happen; it's for obsolete devices and weird
966 * corner cases, but the stack really does a non-trivial amount
967 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 968 * Required; cannot be NULL.
00829823 969 *
1a2a1444
DM
970 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
971 * struct net_device *dev
972 * netdev_features_t features);
973 * Called by core transmit path to determine if device is capable of
974 * performing offload operations on a given packet. This is to give
975 * the device an opportunity to implement any restrictions that cannot
976 * be otherwise expressed by feature flags. The check is called with
977 * the set of features that the stack has calculated and it returns
978 * those the driver believes to be appropriate.
cdba756f 979 *
f663dd9a 980 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
a350ecce 981 * struct net_device *sb_dev);
5e82b4b2 982 * Called to decide which queue to use when device supports multiple
00829823
SH
983 * transmit queues.
984 *
d314774c
SH
985 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
986 * This function is called to allow device receiver to make
5e82b4b2 987 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
988 *
989 * void (*ndo_set_rx_mode)(struct net_device *dev);
990 * This function is called device changes address list filtering.
01789349 991 * If driver handles unicast address filtering, it should set
5e82b4b2 992 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
993 *
994 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
995 * This function is called when the Media Access Control address
37b607c5 996 * needs to be changed. If this interface is not defined, the
5e82b4b2 997 * MAC address can not be changed.
d314774c
SH
998 *
999 * int (*ndo_validate_addr)(struct net_device *dev);
1000 * Test if Media Access Control address is valid for the device.
1001 *
1002 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
5e82b4b2
BH
1003 * Called when a user requests an ioctl which can't be handled by
1004 * the generic interface code. If not defined ioctls return
d314774c
SH
1005 * not supported error code.
1006 *
1007 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1008 * Used to set network devices bus interface parameters. This interface
5e82b4b2 1009 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1010 * interface (PCI) for low level management.
1011 *
1012 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1013 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1014 * of a device.
d314774c 1015 *
00829823 1016 * void (*ndo_tx_timeout)(struct net_device *dev);
5e82b4b2 1017 * Callback used when the transmitter has not made any progress
d314774c
SH
1018 * for dev->watchdog ticks.
1019 *
bc1f4470 1020 * void (*ndo_get_stats64)(struct net_device *dev,
1021 * struct rtnl_link_stats64 *storage);
d308e38f 1022 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1023 * Called when a user wants to get the network device usage
be1f3c2c 1024 * statistics. Drivers must do one of the following:
3cfde79c
BH
1025 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1026 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1027 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1028 * (which should normally be dev->stats) and return a pointer to
1029 * it. The structure may be changed asynchronously only if each
1030 * field is written atomically.
1031 * 3. Update dev->stats asynchronously and atomically, and define
1032 * neither operation.
d314774c 1033 *
3df5b3c6 1034 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1035 * Return true if this device supports offload stats of this attr_id.
1036 *
1037 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1038 * void *attr_data)
1039 * Get statistics for offload operations by attr_id. Write it into the
1040 * attr_data pointer.
1041 *
5d632cb7 1042 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1043 * If device supports VLAN filtering this function is called when a
80d5c368 1044 * VLAN id is registered.
d314774c 1045 *
5d632cb7 1046 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1047 * If device supports VLAN filtering this function is called when a
80d5c368 1048 * VLAN id is unregistered.
d314774c
SH
1049 *
1050 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1051 *
1052 * SR-IOV management functions.
1053 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1054 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1055 * u8 qos, __be16 proto);
ed616689
SC
1056 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1057 * int max_tx_rate);
5f8444a3 1058 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1059 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1060 * int (*ndo_get_vf_config)(struct net_device *dev,
1061 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1062 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1063 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1064 * struct nlattr *port[]);
01a3d796
VZ
1065 *
1066 * Enable or disable the VF ability to query its RSS Redirection Table and
1067 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1068 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1069 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1070 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1071 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1072 * void *type_data);
6a4bc2b4
FF
1073 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1074 * This is always called from the stack with the rtnl lock held and netif
1075 * tx queues stopped. This allows the netdevice to perform queue
1076 * management safely.
c445477d 1077 *
e9bce845
YZ
1078 * Fiber Channel over Ethernet (FCoE) offload functions.
1079 * int (*ndo_fcoe_enable)(struct net_device *dev);
1080 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1081 * so the underlying device can perform whatever needed configuration or
1082 * initialization to support acceleration of FCoE traffic.
1083 *
1084 * int (*ndo_fcoe_disable)(struct net_device *dev);
1085 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1086 * so the underlying device can perform whatever needed clean-ups to
1087 * stop supporting acceleration of FCoE traffic.
1088 *
1089 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1090 * struct scatterlist *sgl, unsigned int sgc);
1091 * Called when the FCoE Initiator wants to initialize an I/O that
1092 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1093 * perform necessary setup and returns 1 to indicate the device is set up
1094 * successfully to perform DDP on this I/O, otherwise this returns 0.
1095 *
1096 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1097 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1098 * indicated by the FC exchange id 'xid', so the underlying device can
1099 * clean up and reuse resources for later DDP requests.
1100 *
1101 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1102 * struct scatterlist *sgl, unsigned int sgc);
1103 * Called when the FCoE Target wants to initialize an I/O that
1104 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1105 * perform necessary setup and returns 1 to indicate the device is set up
1106 * successfully to perform DDP on this I/O, otherwise this returns 0.
1107 *
68bad94e
NP
1108 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1109 * struct netdev_fcoe_hbainfo *hbainfo);
1110 * Called when the FCoE Protocol stack wants information on the underlying
1111 * device. This information is utilized by the FCoE protocol stack to
1112 * register attributes with Fiber Channel management service as per the
1113 * FC-GS Fabric Device Management Information(FDMI) specification.
1114 *
e9bce845
YZ
1115 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1116 * Called when the underlying device wants to override default World Wide
1117 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1118 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1119 * protocol stack to use.
1120 *
c445477d
BH
1121 * RFS acceleration.
1122 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1123 * u16 rxq_index, u32 flow_id);
1124 * Set hardware filter for RFS. rxq_index is the target queue index;
1125 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1126 * Return the filter ID on success, or a negative error code.
fbaec0ea 1127 *
8b98a70c 1128 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1129 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1130 * Called to make another netdev an underling.
1131 *
1132 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1133 * Called to release previously enslaved netdev.
5455c699
MM
1134 *
1135 * Feature/offload setting functions.
1a2a1444
DM
1136 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1137 * netdev_features_t features);
1138 * Adjusts the requested feature flags according to device-specific
1139 * constraints, and returns the resulting flags. Must not modify
1140 * the device state.
1141 *
c8f44aff 1142 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1143 * Called to update device configuration to new features. Passed
1144 * feature set might be less than what was returned by ndo_fix_features()).
1145 * Must return >0 or -errno if it changed dev->features itself.
1146 *
edc7d573 1147 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1148 * struct net_device *dev,
87b0984e
PM
1149 * const unsigned char *addr, u16 vid, u16 flags,
1150 * struct netlink_ext_ack *extack);
77162022 1151 * Adds an FDB entry to dev for addr.
1690be63
VY
1152 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1153 * struct net_device *dev,
f6f6424b 1154 * const unsigned char *addr, u16 vid)
77162022
JF
1155 * Deletes the FDB entry from dev coresponding to addr.
1156 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1157 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1158 * int *idx)
77162022
JF
1159 * Used to add FDB entries to dump requests. Implementers should add
1160 * entries to skb and update idx with the number of entries.
e5a55a89 1161 *
ad41faa8 1162 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
2fd527b7 1163 * u16 flags, struct netlink_ext_ack *extack)
e5a55a89 1164 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1165 * struct net_device *dev, u32 filter_mask,
1166 * int nlflags)
ad41faa8
ND
1167 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1168 * u16 flags);
4bf84c35
JP
1169 *
1170 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1171 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1172 * which do not represent real hardware may define this to allow their
1173 * userspace components to manage their virtual carrier state. Devices
1174 * that determine carrier state from physical hardware properties (eg
1175 * network cables) or protocol-dependent mechanisms (eg
1176 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1177 *
1178 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1179 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1180 * Called to get ID of physical port of this device. If driver does
1181 * not implement this, it is assumed that the hw is not able to have
1182 * multiple net devices on single physical port.
53cf5275 1183 *
d6abc596
FF
1184 * int (*ndo_get_port_parent_id)(struct net_device *dev,
1185 * struct netdev_phys_item_id *ppid)
1186 * Called to get the parent ID of the physical port of this device.
1187 *
7c46a640
AD
1188 * void (*ndo_udp_tunnel_add)(struct net_device *dev,
1189 * struct udp_tunnel_info *ti);
1190 * Called by UDP tunnel to notify a driver about the UDP port and socket
1191 * address family that a UDP tunnel is listnening to. It is called only
1192 * when a new port starts listening. The operation is protected by the
1193 * RTNL.
1194 *
1195 * void (*ndo_udp_tunnel_del)(struct net_device *dev,
1196 * struct udp_tunnel_info *ti);
1197 * Called by UDP tunnel to notify the driver about a UDP port and socket
1198 * address family that the UDP tunnel is not listening to anymore. The
1199 * operation is protected by the RTNL.
1200 *
a6cc0cfa
JF
1201 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1202 * struct net_device *dev)
1203 * Called by upper layer devices to accelerate switching or other
1204 * station functionality into hardware. 'pdev is the lowerdev
1205 * to use for the offload and 'dev' is the net device that will
1206 * back the offload. Returns a pointer to the private structure
1207 * the upper layer will maintain.
1208 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1209 * Called by upper layer device to delete the station created
1210 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1211 * the station and priv is the structure returned by the add
1212 * operation.
822b3b2e
JF
1213 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1214 * int queue_index, u32 maxrate);
1215 * Called when a user wants to set a max-rate limitation of specific
1216 * TX queue.
a54acb3a
ND
1217 * int (*ndo_get_iflink)(const struct net_device *dev);
1218 * Called to get the iflink value of this device.
d746d707 1219 * void (*ndo_change_proto_down)(struct net_device *dev,
5e82b4b2 1220 * bool proto_down);
d746d707
AK
1221 * This function is used to pass protocol port error state information
1222 * to the switch driver. The switch driver can react to the proto_down
1223 * by doing a phys down on the associated switch port.
fc4099f1
PS
1224 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1225 * This function is used to get egress tunnel information for given skb.
1226 * This is useful for retrieving outer tunnel header parameters while
1227 * sampling packet.
871b642a
PA
1228 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1229 * This function is used to specify the headroom that the skb must
1230 * consider when allocation skb during packet reception. Setting
1231 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1232 * forward. Setting a negative value resets the rx headroom to the
871b642a 1233 * default value.
f4e63525 1234 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1235 * This function is used to set or query state related to XDP on the
f4e63525
JK
1236 * netdevice and manage BPF offload. See definition of
1237 * enum bpf_netdev_command for details.
42b33468
JDB
1238 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1239 * u32 flags);
735fc405
JDB
1240 * This function is used to submit @n XDP packets for transmit on a
1241 * netdevice. Returns number of frames successfully transmitted, frames
1242 * that got dropped are freed/returned via xdp_return_frame().
1243 * Returns negative number, means general error invoking ndo, meaning
1244 * no frames were xmit'ed and core-caller will free all frames.
9116e5e2
MK
1245 * int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
1246 * This function is used to wake up the softirq, ksoftirqd or kthread
1247 * responsible for sending and/or receiving packets on a specific
1248 * queue id bound to an AF_XDP socket. The flags field specifies if
1249 * only RX, only Tx, or both should be woken up using the flags
1250 * XDP_WAKEUP_RX and XDP_WAKEUP_TX.
5dc37bb9
JP
1251 * struct devlink_port *(*ndo_get_devlink_port)(struct net_device *dev);
1252 * Get devlink port instance associated with a given netdev.
b473b0d2
JK
1253 * Called with a reference on the netdevice and devlink locks only,
1254 * rtnl_lock is not held.
d314774c
SH
1255 */
1256struct net_device_ops {
1257 int (*ndo_init)(struct net_device *dev);
1258 void (*ndo_uninit)(struct net_device *dev);
1259 int (*ndo_open)(struct net_device *dev);
1260 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1261 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1262 struct net_device *dev);
1263 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1264 struct net_device *dev,
1265 netdev_features_t features);
00829823 1266 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1267 struct sk_buff *skb,
a350ecce 1268 struct net_device *sb_dev);
d314774c
SH
1269 void (*ndo_change_rx_flags)(struct net_device *dev,
1270 int flags);
d314774c 1271 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1272 int (*ndo_set_mac_address)(struct net_device *dev,
1273 void *addr);
d314774c 1274 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1275 int (*ndo_do_ioctl)(struct net_device *dev,
1276 struct ifreq *ifr, int cmd);
d314774c
SH
1277 int (*ndo_set_config)(struct net_device *dev,
1278 struct ifmap *map);
00829823
SH
1279 int (*ndo_change_mtu)(struct net_device *dev,
1280 int new_mtu);
1281 int (*ndo_neigh_setup)(struct net_device *dev,
1282 struct neigh_parms *);
d314774c
SH
1283 void (*ndo_tx_timeout) (struct net_device *dev);
1284
bc1f4470 1285 void (*ndo_get_stats64)(struct net_device *dev,
1286 struct rtnl_link_stats64 *storage);
3df5b3c6 1287 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1288 int (*ndo_get_offload_stats)(int attr_id,
1289 const struct net_device *dev,
1290 void *attr_data);
d314774c
SH
1291 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1292
8e586137 1293 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1294 __be16 proto, u16 vid);
8e586137 1295 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1296 __be16 proto, u16 vid);
d314774c 1297#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1298 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1299 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1300 struct netpoll_info *info);
0e34e931 1301 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1302#endif
95c26df8
WM
1303 int (*ndo_set_vf_mac)(struct net_device *dev,
1304 int queue, u8 *mac);
1305 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1306 int queue, u16 vlan,
1307 u8 qos, __be16 proto);
ed616689
SC
1308 int (*ndo_set_vf_rate)(struct net_device *dev,
1309 int vf, int min_tx_rate,
1310 int max_tx_rate);
5f8444a3
GR
1311 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1312 int vf, bool setting);
dd461d6a
HS
1313 int (*ndo_set_vf_trust)(struct net_device *dev,
1314 int vf, bool setting);
95c26df8
WM
1315 int (*ndo_get_vf_config)(struct net_device *dev,
1316 int vf,
1317 struct ifla_vf_info *ivf);
1d8faf48
RE
1318 int (*ndo_set_vf_link_state)(struct net_device *dev,
1319 int vf, int link_state);
3b766cd8
EBE
1320 int (*ndo_get_vf_stats)(struct net_device *dev,
1321 int vf,
1322 struct ifla_vf_stats
1323 *vf_stats);
57b61080
SF
1324 int (*ndo_set_vf_port)(struct net_device *dev,
1325 int vf,
1326 struct nlattr *port[]);
1327 int (*ndo_get_vf_port)(struct net_device *dev,
1328 int vf, struct sk_buff *skb);
cc8e27cc
EC
1329 int (*ndo_set_vf_guid)(struct net_device *dev,
1330 int vf, u64 guid,
1331 int guid_type);
01a3d796
VZ
1332 int (*ndo_set_vf_rss_query_en)(
1333 struct net_device *dev,
1334 int vf, bool setting);
16e5cc64 1335 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1336 enum tc_setup_type type,
de4784ca 1337 void *type_data);
d11ead75 1338#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1339 int (*ndo_fcoe_enable)(struct net_device *dev);
1340 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1341 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1342 u16 xid,
1343 struct scatterlist *sgl,
1344 unsigned int sgc);
1345 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1346 u16 xid);
6247e086
YZ
1347 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1348 u16 xid,
1349 struct scatterlist *sgl,
1350 unsigned int sgc);
68bad94e
NP
1351 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1352 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1353#endif
1354
d11ead75 1355#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1356#define NETDEV_FCOE_WWNN 0
1357#define NETDEV_FCOE_WWPN 1
1358 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1359 u64 *wwn, int type);
4d288d57 1360#endif
3c9c36bc 1361
c445477d
BH
1362#ifdef CONFIG_RFS_ACCEL
1363 int (*ndo_rx_flow_steer)(struct net_device *dev,
1364 const struct sk_buff *skb,
1365 u16 rxq_index,
1366 u32 flow_id);
1367#endif
fbaec0ea 1368 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1369 struct net_device *slave_dev,
1370 struct netlink_ext_ack *extack);
fbaec0ea
JP
1371 int (*ndo_del_slave)(struct net_device *dev,
1372 struct net_device *slave_dev);
c8f44aff
MM
1373 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1374 netdev_features_t features);
5455c699 1375 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1376 netdev_features_t features);
503eebc2
JP
1377 int (*ndo_neigh_construct)(struct net_device *dev,
1378 struct neighbour *n);
1379 void (*ndo_neigh_destroy)(struct net_device *dev,
1380 struct neighbour *n);
77162022
JF
1381
1382 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1383 struct nlattr *tb[],
77162022 1384 struct net_device *dev,
6b6e2725 1385 const unsigned char *addr,
f6f6424b 1386 u16 vid,
87b0984e
PM
1387 u16 flags,
1388 struct netlink_ext_ack *extack);
77162022 1389 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1390 struct nlattr *tb[],
77162022 1391 struct net_device *dev,
f6f6424b
JP
1392 const unsigned char *addr,
1393 u16 vid);
77162022
JF
1394 int (*ndo_fdb_dump)(struct sk_buff *skb,
1395 struct netlink_callback *cb,
1396 struct net_device *dev,
5d5eacb3 1397 struct net_device *filter_dev,
d297653d 1398 int *idx);
5b2f94b2
RP
1399 int (*ndo_fdb_get)(struct sk_buff *skb,
1400 struct nlattr *tb[],
1401 struct net_device *dev,
1402 const unsigned char *addr,
1403 u16 vid, u32 portid, u32 seq,
1404 struct netlink_ext_ack *extack);
e5a55a89 1405 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3 1406 struct nlmsghdr *nlh,
2fd527b7
PM
1407 u16 flags,
1408 struct netlink_ext_ack *extack);
e5a55a89
JF
1409 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1410 u32 pid, u32 seq,
6cbdceeb 1411 struct net_device *dev,
46c264da
ND
1412 u32 filter_mask,
1413 int nlflags);
407af329 1414 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1415 struct nlmsghdr *nlh,
1416 u16 flags);
4bf84c35
JP
1417 int (*ndo_change_carrier)(struct net_device *dev,
1418 bool new_carrier);
66b52b0d 1419 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1420 struct netdev_phys_item_id *ppid);
d6abc596
FF
1421 int (*ndo_get_port_parent_id)(struct net_device *dev,
1422 struct netdev_phys_item_id *ppid);
db24a904
DA
1423 int (*ndo_get_phys_port_name)(struct net_device *dev,
1424 char *name, size_t len);
7c46a640
AD
1425 void (*ndo_udp_tunnel_add)(struct net_device *dev,
1426 struct udp_tunnel_info *ti);
1427 void (*ndo_udp_tunnel_del)(struct net_device *dev,
1428 struct udp_tunnel_info *ti);
a6cc0cfa
JF
1429 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1430 struct net_device *dev);
1431 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1432 void *priv);
1433
822b3b2e
JF
1434 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1435 int queue_index,
1436 u32 maxrate);
a54acb3a 1437 int (*ndo_get_iflink)(const struct net_device *dev);
d746d707
AK
1438 int (*ndo_change_proto_down)(struct net_device *dev,
1439 bool proto_down);
fc4099f1
PS
1440 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1441 struct sk_buff *skb);
871b642a
PA
1442 void (*ndo_set_rx_headroom)(struct net_device *dev,
1443 int needed_headroom);
f4e63525
JK
1444 int (*ndo_bpf)(struct net_device *dev,
1445 struct netdev_bpf *bpf);
735fc405 1446 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1447 struct xdp_frame **xdp,
1448 u32 flags);
9116e5e2
MK
1449 int (*ndo_xsk_wakeup)(struct net_device *dev,
1450 u32 queue_id, u32 flags);
5dc37bb9 1451 struct devlink_port * (*ndo_get_devlink_port)(struct net_device *dev);
d314774c
SH
1452};
1453
7aa98047
LR
1454/**
1455 * enum net_device_priv_flags - &struct net_device priv_flags
1456 *
1457 * These are the &struct net_device, they are only set internally
1458 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1459 * userspace; this means that the order of these flags can change
7aa98047
LR
1460 * during any kernel release.
1461 *
1462 * You should have a pretty good reason to be extending these flags.
1463 *
1464 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1465 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1466 * @IFF_BONDING: bonding master or slave
7aa98047 1467 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1468 * @IFF_WAN_HDLC: WAN HDLC device
1469 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1470 * release skb->dst
1471 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1472 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1473 * @IFF_MACVLAN_PORT: device used as macvlan port
1474 * @IFF_BRIDGE_PORT: device used as bridge port
1475 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1476 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1477 * @IFF_UNICAST_FLT: Supports unicast filtering
1478 * @IFF_TEAM_PORT: device used as team port
1479 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1480 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1481 * change when it's running
1482 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1483 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1484 * underlying stacked devices
007979ea 1485 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1486 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1487 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1488 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1489 * @IFF_TEAM: device is a team device
d4ab4286 1490 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1491 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1492 * entity (i.e. the master device for bridged veth)
3c175784 1493 * @IFF_MACSEC: device is a MACsec device
f5426250 1494 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1495 * @IFF_FAILOVER: device is a failover master device
1496 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
d5256083 1497 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
8065a779 1498 * @IFF_LIVE_RENAME_OK: rename is allowed while device is up and running
7aa98047
LR
1499 */
1500enum netdev_priv_flags {
1501 IFF_802_1Q_VLAN = 1<<0,
1502 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1503 IFF_BONDING = 1<<2,
1504 IFF_ISATAP = 1<<3,
1505 IFF_WAN_HDLC = 1<<4,
1506 IFF_XMIT_DST_RELEASE = 1<<5,
1507 IFF_DONT_BRIDGE = 1<<6,
1508 IFF_DISABLE_NETPOLL = 1<<7,
1509 IFF_MACVLAN_PORT = 1<<8,
1510 IFF_BRIDGE_PORT = 1<<9,
1511 IFF_OVS_DATAPATH = 1<<10,
1512 IFF_TX_SKB_SHARING = 1<<11,
1513 IFF_UNICAST_FLT = 1<<12,
1514 IFF_TEAM_PORT = 1<<13,
1515 IFF_SUPP_NOFCS = 1<<14,
1516 IFF_LIVE_ADDR_CHANGE = 1<<15,
1517 IFF_MACVLAN = 1<<16,
1518 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1519 IFF_L3MDEV_MASTER = 1<<18,
1520 IFF_NO_QUEUE = 1<<19,
1521 IFF_OPENVSWITCH = 1<<20,
1522 IFF_L3MDEV_SLAVE = 1<<21,
1523 IFF_TEAM = 1<<22,
1524 IFF_RXFH_CONFIGURED = 1<<23,
1525 IFF_PHONY_HEADROOM = 1<<24,
1526 IFF_MACSEC = 1<<25,
f5426250 1527 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1528 IFF_FAILOVER = 1<<27,
1529 IFF_FAILOVER_SLAVE = 1<<28,
d5256083 1530 IFF_L3MDEV_RX_HANDLER = 1<<29,
8065a779 1531 IFF_LIVE_RENAME_OK = 1<<30,
7aa98047
LR
1532};
1533
1534#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1535#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1536#define IFF_BONDING IFF_BONDING
7aa98047 1537#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1538#define IFF_WAN_HDLC IFF_WAN_HDLC
1539#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1540#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1541#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1542#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1543#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1544#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1545#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1546#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1547#define IFF_TEAM_PORT IFF_TEAM_PORT
1548#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1549#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1550#define IFF_MACVLAN IFF_MACVLAN
02875878 1551#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
007979ea 1552#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1553#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1554#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1555#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1556#define IFF_TEAM IFF_TEAM
d4ab4286 1557#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
3c175784 1558#define IFF_MACSEC IFF_MACSEC
f5426250 1559#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
30c8bd5a
SS
1560#define IFF_FAILOVER IFF_FAILOVER
1561#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
d5256083 1562#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
8065a779 1563#define IFF_LIVE_RENAME_OK IFF_LIVE_RENAME_OK
7aa98047 1564
536721b1
KK
1565/**
1566 * struct net_device - The DEVICE structure.
d651983d
MCC
1567 *
1568 * Actually, this whole structure is a big mistake. It mixes I/O
1569 * data with strictly "high-level" data, and it has to know about
1570 * almost every data structure used in the INET module.
536721b1
KK
1571 *
1572 * @name: This is the first field of the "visible" part of this structure
1573 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1574 * of the interface.
536721b1 1575 *
ff927412 1576 * @name_node: Name hashlist node
536721b1
KK
1577 * @ifalias: SNMP alias
1578 * @mem_end: Shared memory end
1579 * @mem_start: Shared memory start
1580 * @base_addr: Device I/O address
1581 * @irq: Device IRQ number
1582 *
1583 * @state: Generic network queuing layer state, see netdev_state_t
1584 * @dev_list: The global list of network devices
5e82b4b2
BH
1585 * @napi_list: List entry used for polling NAPI devices
1586 * @unreg_list: List entry when we are unregistering the
1587 * device; see the function unregister_netdev
1588 * @close_list: List entry used when we are closing the device
62d885fe
BP
1589 * @ptype_all: Device-specific packet handlers for all protocols
1590 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1591 *
1592 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1593 * @features: Currently active device features
1594 * @hw_features: User-changeable features
1595 *
1596 * @wanted_features: User-requested features
1597 * @vlan_features: Mask of features inheritable by VLAN devices
1598 *
1599 * @hw_enc_features: Mask of features inherited by encapsulating devices
1600 * This field indicates what encapsulation
1601 * offloads the hardware is capable of doing,
1602 * and drivers will need to set them appropriately.
1603 *
1604 * @mpls_features: Mask of features inheritable by MPLS
1605 *
1606 * @ifindex: interface index
5e82b4b2 1607 * @group: The group the device belongs to
536721b1
KK
1608 *
1609 * @stats: Statistics struct, which was left as a legacy, use
1610 * rtnl_link_stats64 instead
1611 *
1612 * @rx_dropped: Dropped packets by core network,
1613 * do not use this in drivers
1614 * @tx_dropped: Dropped packets by core network,
1615 * do not use this in drivers
6e7333d3
JW
1616 * @rx_nohandler: nohandler dropped packets by core network on
1617 * inactive devices, do not use this in drivers
9e55e5d3
FF
1618 * @carrier_up_count: Number of times the carrier has been up
1619 * @carrier_down_count: Number of times the carrier has been down
536721b1 1620 *
536721b1
KK
1621 * @wireless_handlers: List of functions to handle Wireless Extensions,
1622 * instead of ioctl,
1623 * see <net/iw_handler.h> for details.
1624 * @wireless_data: Instance data managed by the core of wireless extensions
1625 *
1626 * @netdev_ops: Includes several pointers to callbacks,
1627 * if one wants to override the ndo_*() functions
1628 * @ethtool_ops: Management operations
f997c55c
AA
1629 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1630 * discovery handling. Necessary for e.g. 6LoWPAN.
d476059e 1631 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1632 * of Layer 2 headers.
1633 *
1634 * @flags: Interface flags (a la BSD)
1635 * @priv_flags: Like 'flags' but invisible to userspace,
1636 * see if.h for the definitions
1637 * @gflags: Global flags ( kept as legacy )
1638 * @padded: How much padding added by alloc_netdev()
1639 * @operstate: RFC2863 operstate
1640 * @link_mode: Mapping policy to operstate
1641 * @if_port: Selectable AUI, TP, ...
1642 * @dma: DMA channel
1643 * @mtu: Interface MTU value
61e84623
JW
1644 * @min_mtu: Interface Minimum MTU value
1645 * @max_mtu: Interface Maximum MTU value
536721b1 1646 * @type: Interface hardware type
2793a23a 1647 * @hard_header_len: Maximum hardware header length.
217e6fa2 1648 * @min_header_len: Minimum hardware header length
536721b1
KK
1649 *
1650 * @needed_headroom: Extra headroom the hardware may need, but not in all
1651 * cases can this be guaranteed
1652 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1653 * cases can this be guaranteed. Some cases also use
1654 * LL_MAX_HEADER instead to allocate the skb
1655 *
1656 * interface address info:
1657 *
1658 * @perm_addr: Permanent hw address
1659 * @addr_assign_type: Hw address assignment type
1660 * @addr_len: Hardware address length
5343da4c
TY
1661 * @upper_level: Maximum depth level of upper devices.
1662 * @lower_level: Maximum depth level of lower devices.
8626a0c8 1663 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1664 * @dev_id: Used to differentiate devices that share
1665 * the same link layer address
1666 * @dev_port: Used to differentiate devices that share
1667 * the same function
1668 * @addr_list_lock: XXX: need comments on this one
5e82b4b2 1669 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1670 * has been enabled due to the need to listen to
1671 * additional unicast addresses in a device that
1672 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1673 * @uc: unicast mac addresses
1674 * @mc: multicast mac addresses
1675 * @dev_addrs: list of device hw addresses
1676 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1677 * @promiscuity: Number of times the NIC is told to work in
1678 * promiscuous mode; if it becomes 0 the NIC will
1679 * exit promiscuous mode
536721b1
KK
1680 * @allmulti: Counter, enables or disables allmulticast mode
1681 *
1682 * @vlan_info: VLAN info
1683 * @dsa_ptr: dsa specific data
1684 * @tipc_ptr: TIPC specific data
1685 * @atalk_ptr: AppleTalk link
1686 * @ip_ptr: IPv4 specific data
1687 * @dn_ptr: DECnet specific data
1688 * @ip6_ptr: IPv6 specific data
1689 * @ax25_ptr: AX.25 specific data
1690 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
1691 *
536721b1
KK
1692 * @dev_addr: Hw address (before bcast,
1693 * because most packets are unicast)
1694 *
1695 * @_rx: Array of RX queues
1696 * @num_rx_queues: Number of RX queues
1697 * allocated at register_netdev() time
1698 * @real_num_rx_queues: Number of RX queues currently active in device
1699 *
1700 * @rx_handler: handler for received packets
1701 * @rx_handler_data: XXX: need comments on this one
46209401
JP
1702 * @miniq_ingress: ingress/clsact qdisc specific data for
1703 * ingress processing
536721b1
KK
1704 * @ingress_queue: XXX: need comments on this one
1705 * @broadcast: hw bcast address
1706 *
14ffbbb8
TG
1707 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1708 * indexed by RX queue number. Assigned by driver.
1709 * This must only be set if the ndo_rx_flow_steer
1710 * operation is defined
1711 * @index_hlist: Device index hash chain
1712 *
536721b1
KK
1713 * @_tx: Array of TX queues
1714 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1715 * @real_num_tx_queues: Number of TX queues currently active in device
1716 * @qdisc: Root qdisc from userspace point of view
1717 * @tx_queue_len: Max frames per queue allowed
1718 * @tx_global_lock: XXX: need comments on this one
1719 *
1720 * @xps_maps: XXX: need comments on this one
46209401
JP
1721 * @miniq_egress: clsact qdisc specific data for
1722 * egress processing
536721b1 1723 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1724 * the watchdog (see dev_watchdog())
536721b1
KK
1725 * @watchdog_timer: List of timers
1726 *
1727 * @pcpu_refcnt: Number of references to this device
1728 * @todo_list: Delayed register/unregister
536721b1
KK
1729 * @link_watch_list: XXX: need comments on this one
1730 *
1731 * @reg_state: Register/unregister state machine
1732 * @dismantle: Device is going to be freed
1733 * @rtnl_link_state: This enum represents the phases of creating
1734 * a new link
1735 *
cf124db5
DM
1736 * @needs_free_netdev: Should unregister perform free_netdev?
1737 * @priv_destructor: Called from unregister
536721b1
KK
1738 * @npinfo: XXX: need comments on this one
1739 * @nd_net: Network namespace this network device is inside
1740 *
1741 * @ml_priv: Mid-layer private
1742 * @lstats: Loopback statistics
1743 * @tstats: Tunnel statistics
1744 * @dstats: Dummy statistics
1745 * @vstats: Virtual ethernet statistics
1746 *
1747 * @garp_port: GARP
1748 * @mrp_port: MRP
1749 *
1750 * @dev: Class/net/name entry
1751 * @sysfs_groups: Space for optional device, statistics and wireless
1752 * sysfs groups
1753 *
1754 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1755 * @rtnl_link_ops: Rtnl_link_ops
1756 *
1757 * @gso_max_size: Maximum size of generic segmentation offload
1758 * @gso_max_segs: Maximum number of segments that can be passed to the
1759 * NIC for GSO
1760 *
1761 * @dcbnl_ops: Data Center Bridging netlink ops
1762 * @num_tc: Number of traffic classes in the net device
1763 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1764 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
1765 *
1766 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1767 *
1768 * @priomap: XXX: need comments on this one
1769 * @phydev: Physical device may attach itself
1770 * for hardware timestamping
e679c9c1 1771 * @sfp_bus: attached &struct sfp_bus structure.
ab92d68f
TY
1772 * @qdisc_tx_busylock_key: lockdep class annotating Qdisc->busylock
1773 spinlock
1774 * @qdisc_running_key: lockdep class annotating Qdisc->running seqcount
1775 * @qdisc_xmit_lock_key: lockdep class annotating
1776 * netdev_queue->_xmit_lock spinlock
1777 * @addr_list_lock_key: lockdep class annotating
1778 * net_device->addr_list_lock spinlock
536721b1 1779 *
d746d707
AK
1780 * @proto_down: protocol port state information can be sent to the
1781 * switch driver and used to set the phys state of the
1782 * switch port.
1783 *
61941143
HK
1784 * @wol_enabled: Wake-on-LAN is enabled
1785 *
1da177e4
LT
1786 * FIXME: cleanup struct net_device such that network protocol info
1787 * moves out.
1788 */
1789
d94d9fee 1790struct net_device {
1da177e4 1791 char name[IFNAMSIZ];
ff927412 1792 struct netdev_name_node *name_node;
6c557001 1793 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
1794 /*
1795 * I/O specific fields
1796 * FIXME: Merge these and struct ifmap into one
1797 */
536721b1
KK
1798 unsigned long mem_end;
1799 unsigned long mem_start;
1800 unsigned long base_addr;
1801 int irq;
1da177e4
LT
1802
1803 /*
536721b1
KK
1804 * Some hardware also needs these fields (state,dev_list,
1805 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1806 * part of the usual set specified in Space.c.
1807 */
1808
1da177e4
LT
1809 unsigned long state;
1810
7562f876 1811 struct list_head dev_list;
bea3348e 1812 struct list_head napi_list;
44a0873d 1813 struct list_head unreg_list;
5cde2829 1814 struct list_head close_list;
7866a621
SN
1815 struct list_head ptype_all;
1816 struct list_head ptype_specific;
2f268f12 1817
2f268f12
VF
1818 struct {
1819 struct list_head upper;
1820 struct list_head lower;
1821 } adj_list;
1822
c8f44aff 1823 netdev_features_t features;
c8f44aff 1824 netdev_features_t hw_features;
c8f44aff 1825 netdev_features_t wanted_features;
c8f44aff 1826 netdev_features_t vlan_features;
6a674e9c 1827 netdev_features_t hw_enc_features;
0d89d203 1828 netdev_features_t mpls_features;
802ab55a 1829 netdev_features_t gso_partial_features;
04ed3e74 1830
1da177e4 1831 int ifindex;
7a66bbc9 1832 int group;
1da177e4 1833
c45d286e 1834 struct net_device_stats stats;
015f0688 1835
015f0688
ED
1836 atomic_long_t rx_dropped;
1837 atomic_long_t tx_dropped;
6e7333d3 1838 atomic_long_t rx_nohandler;
1da177e4 1839
b2d3bcfa
DD
1840 /* Stats to monitor link on/off, flapping */
1841 atomic_t carrier_up_count;
1842 atomic_t carrier_down_count;
1843
b86e0280 1844#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
1845 const struct iw_handler_def *wireless_handlers;
1846 struct iw_public_data *wireless_data;
b86e0280 1847#endif
d314774c 1848 const struct net_device_ops *netdev_ops;
76fd8593 1849 const struct ethtool_ops *ethtool_ops;
1b69c6d0
DA
1850#ifdef CONFIG_NET_L3_MASTER_DEV
1851 const struct l3mdev_ops *l3mdev_ops;
1852#endif
f997c55c
AA
1853#if IS_ENABLED(CONFIG_IPV6)
1854 const struct ndisc_ops *ndisc_ops;
1855#endif
1da177e4 1856
9cb0d21d 1857#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
1858 const struct xfrmdev_ops *xfrmdev_ops;
1859#endif
1860
a5c37c63
IL
1861#if IS_ENABLED(CONFIG_TLS_DEVICE)
1862 const struct tlsdev_ops *tlsdev_ops;
1863#endif
1864
3b04ddde
SH
1865 const struct header_ops *header_ops;
1866
536721b1
KK
1867 unsigned int flags;
1868 unsigned int priv_flags;
1869
1da177e4 1870 unsigned short gflags;
536721b1 1871 unsigned short padded;
1da177e4 1872
536721b1
KK
1873 unsigned char operstate;
1874 unsigned char link_mode;
b00055aa 1875
536721b1
KK
1876 unsigned char if_port;
1877 unsigned char dma;
bdc220da 1878
536721b1 1879 unsigned int mtu;
61e84623
JW
1880 unsigned int min_mtu;
1881 unsigned int max_mtu;
536721b1
KK
1882 unsigned short type;
1883 unsigned short hard_header_len;
d92be7a4 1884 unsigned char min_header_len;
1da177e4 1885
f5184d26
JB
1886 unsigned short needed_headroom;
1887 unsigned short needed_tailroom;
1888
1da177e4 1889 /* Interface address info. */
536721b1
KK
1890 unsigned char perm_addr[MAX_ADDR_LEN];
1891 unsigned char addr_assign_type;
1892 unsigned char addr_len;
5343da4c
TY
1893 unsigned char upper_level;
1894 unsigned char lower_level;
a0a9663d 1895 unsigned short neigh_priv_len;
536721b1
KK
1896 unsigned short dev_id;
1897 unsigned short dev_port;
ccffad25 1898 spinlock_t addr_list_lock;
14ffbbb8
TG
1899 unsigned char name_assign_type;
1900 bool uc_promisc;
536721b1
KK
1901 struct netdev_hw_addr_list uc;
1902 struct netdev_hw_addr_list mc;
1903 struct netdev_hw_addr_list dev_addrs;
1904
4c3d5e7b
ED
1905#ifdef CONFIG_SYSFS
1906 struct kset *queues_kset;
1907#endif
9d45abe1
WC
1908 unsigned int promiscuity;
1909 unsigned int allmulti;
1da177e4 1910
1da177e4 1911
5e82b4b2 1912 /* Protocol-specific pointers */
65ac6a5f 1913
d11ead75 1914#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 1915 struct vlan_info __rcu *vlan_info;
65ac6a5f 1916#endif
34a430d7 1917#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 1918 struct dsa_port *dsa_ptr;
37cb0620
YX
1919#endif
1920#if IS_ENABLED(CONFIG_TIPC)
536721b1 1921 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 1922#endif
89e58148 1923#if IS_ENABLED(CONFIG_IRDA) || IS_ENABLED(CONFIG_ATALK)
536721b1 1924 void *atalk_ptr;
89e58148 1925#endif
536721b1 1926 struct in_device __rcu *ip_ptr;
330c7272 1927#if IS_ENABLED(CONFIG_DECNET)
536721b1 1928 struct dn_dev __rcu *dn_ptr;
330c7272 1929#endif
536721b1 1930 struct inet6_dev __rcu *ip6_ptr;
19ff13f2 1931#if IS_ENABLED(CONFIG_AX25)
536721b1 1932 void *ax25_ptr;
19ff13f2 1933#endif
536721b1 1934 struct wireless_dev *ieee80211_ptr;
98a18b6f 1935 struct wpan_dev *ieee802154_ptr;
03c57747
RS
1936#if IS_ENABLED(CONFIG_MPLS_ROUTING)
1937 struct mpls_dev __rcu *mpls_ptr;
1938#endif
1da177e4 1939
9356b8fc 1940/*
cd13539b 1941 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 1942 */
9356b8fc 1943 /* Interface address info used in eth_type_trans() */
536721b1 1944 unsigned char *dev_addr;
f001fde5 1945
0a9627f2 1946 struct netdev_rx_queue *_rx;
0a9627f2 1947 unsigned int num_rx_queues;
62fe0b40 1948 unsigned int real_num_rx_queues;
0a9627f2 1949
7acedaf5 1950 struct bpf_prog __rcu *xdp_prog;
3b47d303 1951 unsigned long gro_flush_timeout;
61391cde 1952 rx_handler_func_t __rcu *rx_handler;
1953 void __rcu *rx_handler_data;
e8a0464c 1954
4cda01e8 1955#ifdef CONFIG_NET_CLS_ACT
46209401 1956 struct mini_Qdisc __rcu *miniq_ingress;
d2788d34 1957#endif
24824a09 1958 struct netdev_queue __rcu *ingress_queue;
e687ad60 1959#ifdef CONFIG_NETFILTER_INGRESS
960632ec 1960 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 1961#endif
d2788d34 1962
536721b1 1963 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
1964#ifdef CONFIG_RFS_ACCEL
1965 struct cpu_rmap *rx_cpu_rmap;
1966#endif
1967 struct hlist_node index_hlist;
cd13539b
ED
1968
1969/*
1970 * Cache lines mostly used on transmit path
1971 */
e8a0464c
DM
1972 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
1973 unsigned int num_tx_queues;
fd2ea0a7 1974 unsigned int real_num_tx_queues;
af356afa 1975 struct Qdisc *qdisc;
59cc1f61
JK
1976#ifdef CONFIG_NET_SCHED
1977 DECLARE_HASHTABLE (qdisc_hash, 4);
1978#endif
0cd29503 1979 unsigned int tx_queue_len;
c3f26a26 1980 spinlock_t tx_global_lock;
14ffbbb8 1981 int watchdog_timeo;
cd13539b 1982
bf264145 1983#ifdef CONFIG_XPS
80d19669
AN
1984 struct xps_dev_maps __rcu *xps_cpus_map;
1985 struct xps_dev_maps __rcu *xps_rxqs_map;
bf264145 1986#endif
1f211a1b 1987#ifdef CONFIG_NET_CLS_ACT
46209401 1988 struct mini_Qdisc __rcu *miniq_egress;
1f211a1b 1989#endif
0c4f691f 1990
9356b8fc 1991 /* These may be needed for future network-power-down code. */
9356b8fc
ED
1992 struct timer_list watchdog_timer;
1993
29b4433d 1994 int __percpu *pcpu_refcnt;
1da177e4 1995 struct list_head todo_list;
1da177e4 1996
e014debe 1997 struct list_head link_watch_list;
572a103d 1998
1da177e4 1999 enum { NETREG_UNINITIALIZED=0,
b17a7c17 2000 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
2001 NETREG_UNREGISTERING, /* called unregister_netdevice */
2002 NETREG_UNREGISTERED, /* completed unregister todo */
2003 NETREG_RELEASED, /* called free_netdev */
937f1ba5 2004 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
2005 } reg_state:8;
2006
536721b1 2007 bool dismantle;
a2835763
PM
2008
2009 enum {
2010 RTNL_LINK_INITIALIZED,
2011 RTNL_LINK_INITIALIZING,
2012 } rtnl_link_state:16;
1da177e4 2013
cf124db5
DM
2014 bool needs_free_netdev;
2015 void (*priv_destructor)(struct net_device *dev);
1da177e4 2016
1da177e4 2017#ifdef CONFIG_NETPOLL
5fbee843 2018 struct netpoll_info __rcu *npinfo;
1da177e4 2019#endif
eae792b7 2020
0c5c9fb5 2021 possible_net_t nd_net;
4a1c5371 2022
4951704b 2023 /* mid-layer private */
a7855c78 2024 union {
536721b1
KK
2025 void *ml_priv;
2026 struct pcpu_lstats __percpu *lstats;
8f84985f 2027 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2028 struct pcpu_dstats __percpu *dstats;
a7855c78 2029 };
536721b1 2030
fb585b44 2031#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2032 struct garp_port __rcu *garp_port;
fb585b44
TK
2033#endif
2034#if IS_ENABLED(CONFIG_MRP)
febf018d 2035 struct mrp_port __rcu *mrp_port;
fb585b44 2036#endif
1da177e4 2037
5e82b4b2 2038 struct device dev;
0c509a6c 2039 const struct attribute_group *sysfs_groups[4];
a953be53 2040 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2041
38f7b870 2042 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2043
82cc1a7a
PWJ
2044 /* for setting kernel sock attribute on TCP connection setup */
2045#define GSO_MAX_SIZE 65536
2046 unsigned int gso_max_size;
30b678d8
BH
2047#define GSO_MAX_SEGS 65535
2048 u16 gso_max_segs;
743b03a8 2049
7a6b6f51 2050#ifdef CONFIG_DCB
32953543 2051 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2052#endif
ffcfe25b 2053 s16 num_tc;
5e82b4b2
BH
2054 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2055 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2056
d11ead75 2057#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2058 unsigned int fcoe_ddp_xid;
5bc1421e 2059#endif
86f8515f 2060#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2061 struct netprio_map __rcu *priomap;
4d288d57 2062#endif
5e82b4b2 2063 struct phy_device *phydev;
e679c9c1 2064 struct sfp_bus *sfp_bus;
ab92d68f
TY
2065 struct lock_class_key qdisc_tx_busylock_key;
2066 struct lock_class_key qdisc_running_key;
2067 struct lock_class_key qdisc_xmit_lock_key;
2068 struct lock_class_key addr_list_lock_key;
5e82b4b2 2069 bool proto_down;
61941143 2070 unsigned wol_enabled:1;
1da177e4 2071};
43cb76d9 2072#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2073
b5cdae32
DM
2074static inline bool netif_elide_gro(const struct net_device *dev)
2075{
2076 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2077 return true;
2078 return false;
2079}
2080
1da177e4 2081#define NETDEV_ALIGN 32
1da177e4 2082
4f57c087
JF
2083static inline
2084int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2085{
2086 return dev->prio_tc_map[prio & TC_BITMASK];
2087}
2088
2089static inline
2090int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2091{
2092 if (tc >= dev->num_tc)
2093 return -EINVAL;
2094
2095 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2096 return 0;
2097}
2098
8d059b0f 2099int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2100void netdev_reset_tc(struct net_device *dev);
2101int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2102int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2103
2104static inline
2105int netdev_get_num_tc(struct net_device *dev)
2106{
2107 return dev->num_tc;
2108}
2109
ffcfe25b
AD
2110void netdev_unbind_sb_channel(struct net_device *dev,
2111 struct net_device *sb_dev);
2112int netdev_bind_sb_channel_queue(struct net_device *dev,
2113 struct net_device *sb_dev,
2114 u8 tc, u16 count, u16 offset);
2115int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2116static inline int netdev_get_sb_channel(struct net_device *dev)
2117{
2118 return max_t(int, -dev->num_tc, 0);
2119}
2120
e8a0464c
DM
2121static inline
2122struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2123 unsigned int index)
2124{
2125 return &dev->_tx[index];
2126}
2127
10c51b56
DB
2128static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2129 const struct sk_buff *skb)
2130{
2131 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2132}
2133
e8a0464c
DM
2134static inline void netdev_for_each_tx_queue(struct net_device *dev,
2135 void (*f)(struct net_device *,
2136 struct netdev_queue *,
2137 void *),
2138 void *arg)
2139{
2140 unsigned int i;
2141
2142 for (i = 0; i < dev->num_tx_queues; i++)
2143 f(dev, &dev->_tx[i], arg);
2144}
2145
b71b5837
PA
2146u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2147 struct net_device *sb_dev);
4bd97d51
PA
2148struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2149 struct sk_buff *skb,
2150 struct net_device *sb_dev);
8c4c49df 2151
871b642a
PA
2152/* returns the headroom that the master device needs to take in account
2153 * when forwarding to this dev
2154 */
2155static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2156{
2157 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2158}
2159
2160static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2161{
2162 if (dev->netdev_ops->ndo_set_rx_headroom)
2163 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2164}
2165
2166/* set the device rx headroom to the dev's default */
2167static inline void netdev_reset_rx_headroom(struct net_device *dev)
2168{
2169 netdev_set_rx_headroom(dev, -1);
2170}
2171
c346dca1
YH
2172/*
2173 * Net namespace inlines
2174 */
2175static inline
2176struct net *dev_net(const struct net_device *dev)
2177{
c2d9ba9b 2178 return read_pnet(&dev->nd_net);
c346dca1
YH
2179}
2180
2181static inline
f5aa23fd 2182void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2183{
0c5c9fb5 2184 write_pnet(&dev->nd_net, net);
c346dca1
YH
2185}
2186
bea3348e
SH
2187/**
2188 * netdev_priv - access network device private data
2189 * @dev: network device
2190 *
2191 * Get network device private data
2192 */
6472ce60 2193static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2194{
1ce8e7b5 2195 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2196}
2197
1da177e4
LT
2198/* Set the sysfs physical device reference for the network logical device
2199 * if set prior to registration will cause a symlink during initialization.
2200 */
43cb76d9 2201#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2202
384912ed 2203/* Set the sysfs device type for the network logical device to allow
3f79410c 2204 * fine-grained identification of different network device types. For
5e82b4b2 2205 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2206 */
2207#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2208
82dc3c63
ED
2209/* Default NAPI poll() weight
2210 * Device drivers are strongly advised to not use bigger value
2211 */
2212#define NAPI_POLL_WEIGHT 64
2213
3b582cc1 2214/**
5e82b4b2 2215 * netif_napi_add - initialize a NAPI context
3b582cc1 2216 * @dev: network device
5e82b4b2 2217 * @napi: NAPI context
3b582cc1
SH
2218 * @poll: polling function
2219 * @weight: default weight
2220 *
5e82b4b2
BH
2221 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2222 * *any* of the other NAPI-related functions.
3b582cc1 2223 */
d565b0a1
HX
2224void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2225 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2226
d64b5e85 2227/**
5e82b4b2 2228 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2229 * @dev: network device
5e82b4b2 2230 * @napi: NAPI context
d64b5e85
ED
2231 * @poll: polling function
2232 * @weight: default weight
2233 *
2234 * This variant of netif_napi_add() should be used from drivers using NAPI
2235 * to exclusively poll a TX queue.
2236 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2237 */
2238static inline void netif_tx_napi_add(struct net_device *dev,
2239 struct napi_struct *napi,
2240 int (*poll)(struct napi_struct *, int),
2241 int weight)
2242{
2243 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2244 netif_napi_add(dev, napi, poll, weight);
2245}
2246
d8156534 2247/**
5e82b4b2
BH
2248 * netif_napi_del - remove a NAPI context
2249 * @napi: NAPI context
d8156534 2250 *
5e82b4b2 2251 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2252 */
d565b0a1
HX
2253void netif_napi_del(struct napi_struct *napi);
2254
2255struct napi_gro_cb {
78a478d0 2256 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
5e82b4b2 2257 void *frag0;
78a478d0 2258
7489594c
HX
2259 /* Length of frag0. */
2260 unsigned int frag0_len;
2261
86911732 2262 /* This indicates where we are processing relative to skb->data. */
5e82b4b2 2263 int data_offset;
86911732 2264
d565b0a1 2265 /* This is non-zero if the packet cannot be merged with the new skb. */
bf5a755f
JC
2266 u16 flush;
2267
2268 /* Save the IP ID here and check when we get to the transport layer */
2269 u16 flush_id;
d565b0a1
HX
2270
2271 /* Number of segments aggregated. */
2e71a6f8
ED
2272 u16 count;
2273
15e2396d
TH
2274 /* Start offset for remote checksum offload */
2275 u16 gro_remcsum_start;
2276
2e71a6f8
ED
2277 /* jiffies when first packet was created/queued */
2278 unsigned long age;
86347245 2279
afe93325 2280 /* Used in ipv6_gro_receive() and foo-over-udp */
b582ef09
OG
2281 u16 proto;
2282
baa32ff4
TH
2283 /* This is non-zero if the packet may be of the same flow. */
2284 u8 same_flow:1;
2285
fac8e0f5
JG
2286 /* Used in tunnel GRO receive */
2287 u8 encap_mark:1;
573e8fca
TH
2288
2289 /* GRO checksum is valid */
2290 u8 csum_valid:1;
2291
662880f4
TH
2292 /* Number of checksums via CHECKSUM_UNNECESSARY */
2293 u8 csum_cnt:3;
c3c7c254 2294
baa32ff4
TH
2295 /* Free the skb? */
2296 u8 free:2;
2297#define NAPI_GRO_FREE 1
2298#define NAPI_GRO_FREE_STOLEN_HEAD 2
2299
efc98d08
TH
2300 /* Used in foo-over-udp, set in udp[46]_gro_receive */
2301 u8 is_ipv6:1;
2302
a0ca153f
AD
2303 /* Used in GRE, set in fou/gue_gro_receive */
2304 u8 is_fou:1;
2305
1530545e
AD
2306 /* Used to determine if flush_id can be ignored */
2307 u8 is_atomic:1;
2308
fcd91dd4
SD
2309 /* Number of gro_receive callbacks this packet already went through */
2310 u8 recursion_counter:4;
2311
2312 /* 1 bit hole */
baa32ff4 2313
bf5a755f
JC
2314 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
2315 __wsum csum;
2316
c3c7c254
ED
2317 /* used in skb_gro_receive() slow path */
2318 struct sk_buff *last;
d565b0a1
HX
2319};
2320
2321#define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
d8156534 2322
fcd91dd4
SD
2323#define GRO_RECURSION_LIMIT 15
2324static inline int gro_recursion_inc_test(struct sk_buff *skb)
2325{
2326 return ++NAPI_GRO_CB(skb)->recursion_counter == GRO_RECURSION_LIMIT;
2327}
2328
d4546c25
DM
2329typedef struct sk_buff *(*gro_receive_t)(struct list_head *, struct sk_buff *);
2330static inline struct sk_buff *call_gro_receive(gro_receive_t cb,
2331 struct list_head *head,
2332 struct sk_buff *skb)
fcd91dd4
SD
2333{
2334 if (unlikely(gro_recursion_inc_test(skb))) {
2335 NAPI_GRO_CB(skb)->flush |= 1;
2336 return NULL;
2337 }
2338
2339 return cb(head, skb);
2340}
2341
d4546c25
DM
2342typedef struct sk_buff *(*gro_receive_sk_t)(struct sock *, struct list_head *,
2343 struct sk_buff *);
2344static inline struct sk_buff *call_gro_receive_sk(gro_receive_sk_t cb,
2345 struct sock *sk,
2346 struct list_head *head,
2347 struct sk_buff *skb)
fcd91dd4
SD
2348{
2349 if (unlikely(gro_recursion_inc_test(skb))) {
2350 NAPI_GRO_CB(skb)->flush |= 1;
2351 return NULL;
2352 }
2353
2354 return cb(sk, head, skb);
2355}
2356
1da177e4 2357struct packet_type {
f2ccd8fa 2358 __be16 type; /* This is really htons(ether_type). */
fa788d98 2359 bool ignore_outgoing;
f2ccd8fa
DM
2360 struct net_device *dev; /* NULL is wildcarded here */
2361 int (*func) (struct sk_buff *,
2362 struct net_device *,
2363 struct packet_type *,
2364 struct net_device *);
17266ee9
EC
2365 void (*list_func) (struct list_head *,
2366 struct packet_type *,
2367 struct net_device *);
c0de08d0
EL
2368 bool (*id_match)(struct packet_type *ptype,
2369 struct sock *sk);
1da177e4
LT
2370 void *af_packet_priv;
2371 struct list_head list;
2372};
2373
f191a1d1 2374struct offload_callbacks {
576a30eb 2375 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2376 netdev_features_t features);
d4546c25
DM
2377 struct sk_buff *(*gro_receive)(struct list_head *head,
2378 struct sk_buff *skb);
299603e8 2379 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2380};
2381
2382struct packet_offload {
2383 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2384 u16 priority;
f191a1d1
VY
2385 struct offload_callbacks callbacks;
2386 struct list_head list;
1da177e4
LT
2387};
2388
5e82b4b2 2389/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2390struct pcpu_sw_netstats {
2391 u64 rx_packets;
2392 u64 rx_bytes;
2393 u64 tx_packets;
2394 u64 tx_bytes;
2395 struct u64_stats_sync syncp;
9a5ee462 2396} __aligned(4 * sizeof(u64));
52bb6677
LR
2397
2398struct pcpu_lstats {
2399 u64 packets;
2400 u64 bytes;
2401 struct u64_stats_sync syncp;
9a5ee462 2402} __aligned(2 * sizeof(u64));
8f84985f 2403
de7d5084
ED
2404void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
2405
aabc92bb
PNA
2406#define __netdev_alloc_pcpu_stats(type, gfp) \
2407({ \
2408 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2409 if (pcpu_stats) { \
2410 int __cpu; \
2411 for_each_possible_cpu(__cpu) { \
2412 typeof(type) *stat; \
2413 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2414 u64_stats_init(&stat->syncp); \
2415 } \
2416 } \
2417 pcpu_stats; \
1c213bd2
WC
2418})
2419
aabc92bb 2420#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2421 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2422
764f5e54
JP
2423enum netdev_lag_tx_type {
2424 NETDEV_LAG_TX_TYPE_UNKNOWN,
2425 NETDEV_LAG_TX_TYPE_RANDOM,
2426 NETDEV_LAG_TX_TYPE_BROADCAST,
2427 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2428 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2429 NETDEV_LAG_TX_TYPE_HASH,
2430};
2431
f44aa9ef
JH
2432enum netdev_lag_hash {
2433 NETDEV_LAG_HASH_NONE,
2434 NETDEV_LAG_HASH_L2,
2435 NETDEV_LAG_HASH_L34,
2436 NETDEV_LAG_HASH_L23,
2437 NETDEV_LAG_HASH_E23,
2438 NETDEV_LAG_HASH_E34,
2439 NETDEV_LAG_HASH_UNKNOWN,
2440};
2441
764f5e54
JP
2442struct netdev_lag_upper_info {
2443 enum netdev_lag_tx_type tx_type;
f44aa9ef 2444 enum netdev_lag_hash hash_type;
764f5e54
JP
2445};
2446
fb1b2e3c
JP
2447struct netdev_lag_lower_state_info {
2448 u8 link_up : 1,
2449 tx_enabled : 1;
2450};
2451
1da177e4
LT
2452#include <linux/notifier.h>
2453
ede2762d
KT
2454/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2455 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2456 * adding new types.
dcfe1421 2457 */
ede2762d
KT
2458enum netdev_cmd {
2459 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2460 NETDEV_DOWN,
2461 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2462 detected a hardware crash and restarted
2463 - we can use this eg to kick tcp sessions
2464 once done */
ede2762d
KT
2465 NETDEV_CHANGE, /* Notify device state change */
2466 NETDEV_REGISTER,
2467 NETDEV_UNREGISTER,
2468 NETDEV_CHANGEMTU, /* notify after mtu change happened */
1570415f
PM
2469 NETDEV_CHANGEADDR, /* notify after the address change */
2470 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
ede2762d
KT
2471 NETDEV_GOING_DOWN,
2472 NETDEV_CHANGENAME,
2473 NETDEV_FEAT_CHANGE,
2474 NETDEV_BONDING_FAILOVER,
2475 NETDEV_PRE_UP,
2476 NETDEV_PRE_TYPE_CHANGE,
2477 NETDEV_POST_TYPE_CHANGE,
2478 NETDEV_POST_INIT,
ede2762d
KT
2479 NETDEV_RELEASE,
2480 NETDEV_NOTIFY_PEERS,
2481 NETDEV_JOIN,
2482 NETDEV_CHANGEUPPER,
2483 NETDEV_RESEND_IGMP,
2484 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2485 NETDEV_CHANGEINFODATA,
2486 NETDEV_BONDING_INFO,
2487 NETDEV_PRECHANGEUPPER,
2488 NETDEV_CHANGELOWERSTATE,
2489 NETDEV_UDP_TUNNEL_PUSH_INFO,
2490 NETDEV_UDP_TUNNEL_DROP_INFO,
2491 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2492 NETDEV_CVLAN_FILTER_PUSH_INFO,
2493 NETDEV_CVLAN_FILTER_DROP_INFO,
2494 NETDEV_SVLAN_FILTER_PUSH_INFO,
2495 NETDEV_SVLAN_FILTER_DROP_INFO,
ede2762d
KT
2496};
2497const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2498
f629d208
JP
2499int register_netdevice_notifier(struct notifier_block *nb);
2500int unregister_netdevice_notifier(struct notifier_block *nb);
a30c7b42
JP
2501int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
2502int unregister_netdevice_notifier_net(struct net *net,
2503 struct notifier_block *nb);
351638e7
JP
2504
2505struct netdev_notifier_info {
51d0c047
DA
2506 struct net_device *dev;
2507 struct netlink_ext_ack *extack;
351638e7
JP
2508};
2509
af7d6cce
SD
2510struct netdev_notifier_info_ext {
2511 struct netdev_notifier_info info; /* must be first */
2512 union {
2513 u32 mtu;
2514 } ext;
2515};
2516
be9efd36
JP
2517struct netdev_notifier_change_info {
2518 struct netdev_notifier_info info; /* must be first */
2519 unsigned int flags_changed;
2520};
2521
0e4ead9d
JP
2522struct netdev_notifier_changeupper_info {
2523 struct netdev_notifier_info info; /* must be first */
2524 struct net_device *upper_dev; /* new upper dev */
2525 bool master; /* is upper dev master */
5e82b4b2 2526 bool linking; /* is the notification for link or unlink */
29bf24af 2527 void *upper_info; /* upper dev info */
0e4ead9d
JP
2528};
2529
04d48266
JP
2530struct netdev_notifier_changelowerstate_info {
2531 struct netdev_notifier_info info; /* must be first */
2532 void *lower_state_info; /* is lower dev state */
2533};
2534
1570415f
PM
2535struct netdev_notifier_pre_changeaddr_info {
2536 struct netdev_notifier_info info; /* must be first */
2537 const unsigned char *dev_addr;
2538};
2539
75538c2b
CW
2540static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2541 struct net_device *dev)
2542{
2543 info->dev = dev;
51d0c047 2544 info->extack = NULL;
75538c2b
CW
2545}
2546
351638e7
JP
2547static inline struct net_device *
2548netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2549{
2550 return info->dev;
2551}
2552
51d0c047
DA
2553static inline struct netlink_ext_ack *
2554netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2555{
2556 return info->extack;
2557}
2558
f629d208 2559int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2560
2561
1da177e4
LT
2562extern rwlock_t dev_base_lock; /* Device list lock */
2563
881d966b
EB
2564#define for_each_netdev(net, d) \
2565 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2566#define for_each_netdev_reverse(net, d) \
2567 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2568#define for_each_netdev_rcu(net, d) \
2569 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2570#define for_each_netdev_safe(net, d, n) \
2571 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2572#define for_each_netdev_continue(net, d) \
2573 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
afa0df59
JP
2574#define for_each_netdev_continue_reverse(net, d) \
2575 list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
2576 dev_list)
254245d2 2577#define for_each_netdev_continue_rcu(net, d) \
2578 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2579#define for_each_netdev_in_bond_rcu(bond, slave) \
2580 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2581 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2582#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2583
a050c33f
DL
2584static inline struct net_device *next_net_device(struct net_device *dev)
2585{
2586 struct list_head *lh;
2587 struct net *net;
2588
c346dca1 2589 net = dev_net(dev);
a050c33f
DL
2590 lh = dev->dev_list.next;
2591 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2592}
2593
ce81b76a
ED
2594static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2595{
2596 struct list_head *lh;
2597 struct net *net;
2598
2599 net = dev_net(dev);
ccf43438 2600 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2601 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2602}
2603
a050c33f
DL
2604static inline struct net_device *first_net_device(struct net *net)
2605{
2606 return list_empty(&net->dev_base_head) ? NULL :
2607 net_device_entry(net->dev_base_head.next);
2608}
7562f876 2609
ccf43438
ED
2610static inline struct net_device *first_net_device_rcu(struct net *net)
2611{
2612 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2613
2614 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2615}
2616
f629d208
JP
2617int netdev_boot_setup_check(struct net_device *dev);
2618unsigned long netdev_boot_base(const char *prefix, int unit);
2619struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2620 const char *hwaddr);
2621struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
2622struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type);
2623void dev_add_pack(struct packet_type *pt);
2624void dev_remove_pack(struct packet_type *pt);
2625void __dev_remove_pack(struct packet_type *pt);
2626void dev_add_offload(struct packet_offload *po);
2627void dev_remove_offload(struct packet_offload *po);
f629d208 2628
a54acb3a 2629int dev_get_iflink(const struct net_device *dev);
fc4099f1 2630int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
6c555490
WC
2631struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2632 unsigned short mask);
f629d208
JP
2633struct net_device *dev_get_by_name(struct net *net, const char *name);
2634struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2635struct net_device *__dev_get_by_name(struct net *net, const char *name);
2636int dev_alloc_name(struct net_device *dev, const char *name);
00f54e68 2637int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
7051b88a 2638void dev_close(struct net_device *dev);
2639void dev_close_many(struct list_head *head, bool unlink);
f629d208 2640void dev_disable_lro(struct net_device *dev);
0c4b51f0 2641int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 2642u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
a350ecce 2643 struct net_device *sb_dev);
a4ea8a3d 2644u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
a350ecce 2645 struct net_device *sb_dev);
2b4aa3ce 2646int dev_queue_xmit(struct sk_buff *skb);
eadec877 2647int dev_queue_xmit_accel(struct sk_buff *skb, struct net_device *sb_dev);
865b03f2 2648int dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
f629d208
JP
2649int register_netdevice(struct net_device *dev);
2650void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2651void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2652static inline void unregister_netdevice(struct net_device *dev)
2653{
2654 unregister_netdevice_queue(dev, NULL);
2655}
2656
f629d208
JP
2657int netdev_refcnt_read(const struct net_device *dev);
2658void free_netdev(struct net_device *dev);
74d332c1 2659void netdev_freemem(struct net_device *dev);
f629d208
JP
2660void synchronize_net(void);
2661int init_dummy_netdev(struct net_device *dev);
937f1ba5 2662
f629d208
JP
2663struct net_device *dev_get_by_index(struct net *net, int ifindex);
2664struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2665struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 2666struct net_device *dev_get_by_napi_id(unsigned int napi_id);
f629d208
JP
2667int netdev_get_name(struct net *net, char *name, int ifindex);
2668int dev_restart(struct net_device *dev);
d4546c25 2669int skb_gro_receive(struct sk_buff *p, struct sk_buff *skb);
86911732
HX
2670
2671static inline unsigned int skb_gro_offset(const struct sk_buff *skb)
2672{
2673 return NAPI_GRO_CB(skb)->data_offset;
2674}
2675
2676static inline unsigned int skb_gro_len(const struct sk_buff *skb)
2677{
2678 return skb->len - NAPI_GRO_CB(skb)->data_offset;
2679}
2680
2681static inline void skb_gro_pull(struct sk_buff *skb, unsigned int len)
2682{
2683 NAPI_GRO_CB(skb)->data_offset += len;
2684}
2685
a5b1cf28
HX
2686static inline void *skb_gro_header_fast(struct sk_buff *skb,
2687 unsigned int offset)
86911732 2688{
a5b1cf28
HX
2689 return NAPI_GRO_CB(skb)->frag0 + offset;
2690}
78a478d0 2691
a5b1cf28
HX
2692static inline int skb_gro_header_hard(struct sk_buff *skb, unsigned int hlen)
2693{
2694 return NAPI_GRO_CB(skb)->frag0_len < hlen;
2695}
78a478d0 2696
57ea52a8
HX
2697static inline void skb_gro_frag0_invalidate(struct sk_buff *skb)
2698{
2699 NAPI_GRO_CB(skb)->frag0 = NULL;
2700 NAPI_GRO_CB(skb)->frag0_len = 0;
2701}
2702
a5b1cf28
HX
2703static inline void *skb_gro_header_slow(struct sk_buff *skb, unsigned int hlen,
2704 unsigned int offset)
2705{
17dd759c
HX
2706 if (!pskb_may_pull(skb, hlen))
2707 return NULL;
2708
57ea52a8 2709 skb_gro_frag0_invalidate(skb);
17dd759c 2710 return skb->data + offset;
86911732 2711}
1da177e4 2712
36e7b1b8
HX
2713static inline void *skb_gro_network_header(struct sk_buff *skb)
2714{
78d3fd0b
HX
2715 return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
2716 skb_network_offset(skb);
36e7b1b8
HX
2717}
2718
bf5a755f
JC
2719static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
2720 const void *start, unsigned int len)
2721{
573e8fca 2722 if (NAPI_GRO_CB(skb)->csum_valid)
bf5a755f
JC
2723 NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
2724 csum_partial(start, len, 0));
2725}
2726
573e8fca
TH
2727/* GRO checksum functions. These are logical equivalents of the normal
2728 * checksum functions (in skbuff.h) except that they operate on the GRO
2729 * offsets and fields in sk_buff.
2730 */
2731
2732__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
2733
15e2396d
TH
2734static inline bool skb_at_gro_remcsum_start(struct sk_buff *skb)
2735{
b7fe10e5 2736 return (NAPI_GRO_CB(skb)->gro_remcsum_start == skb_gro_offset(skb));
15e2396d
TH
2737}
2738
573e8fca
TH
2739static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
2740 bool zero_okay,
2741 __sum16 check)
2742{
6edec0e6
TH
2743 return ((skb->ip_summed != CHECKSUM_PARTIAL ||
2744 skb_checksum_start_offset(skb) <
2745 skb_gro_offset(skb)) &&
15e2396d 2746 !skb_at_gro_remcsum_start(skb) &&
662880f4 2747 NAPI_GRO_CB(skb)->csum_cnt == 0 &&
573e8fca
TH
2748 (!zero_okay || check));
2749}
2750
2751static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
2752 __wsum psum)
2753{
2754 if (NAPI_GRO_CB(skb)->csum_valid &&
2755 !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
2756 return 0;
2757
2758 NAPI_GRO_CB(skb)->csum = psum;
2759
2760 return __skb_gro_checksum_complete(skb);
2761}
2762
573e8fca
TH
2763static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
2764{
662880f4
TH
2765 if (NAPI_GRO_CB(skb)->csum_cnt > 0) {
2766 /* Consume a checksum from CHECKSUM_UNNECESSARY */
2767 NAPI_GRO_CB(skb)->csum_cnt--;
2768 } else {
2769 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
2770 * verified a new top level checksum or an encapsulated one
2771 * during GRO. This saves work if we fallback to normal path.
2772 */
2773 __skb_incr_checksum_unnecessary(skb);
573e8fca
TH
2774 }
2775}
2776
2777#define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
2778 compute_pseudo) \
2779({ \
2780 __sum16 __ret = 0; \
2781 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
2782 __ret = __skb_gro_checksum_validate_complete(skb, \
2783 compute_pseudo(skb, proto)); \
219f1d79 2784 if (!__ret) \
573e8fca
TH
2785 skb_gro_incr_csum_unnecessary(skb); \
2786 __ret; \
2787})
2788
2789#define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
2790 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
2791
2792#define skb_gro_checksum_validate_zero_check(skb, proto, check, \
2793 compute_pseudo) \
2794 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
2795
2796#define skb_gro_checksum_simple_validate(skb) \
2797 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
2798
d96535a1
TH
2799static inline bool __skb_gro_checksum_convert_check(struct sk_buff *skb)
2800{
2801 return (NAPI_GRO_CB(skb)->csum_cnt == 0 &&
2802 !NAPI_GRO_CB(skb)->csum_valid);
2803}
2804
2805static inline void __skb_gro_checksum_convert(struct sk_buff *skb,
2806 __sum16 check, __wsum pseudo)
2807{
2808 NAPI_GRO_CB(skb)->csum = ~pseudo;
2809 NAPI_GRO_CB(skb)->csum_valid = 1;
2810}
2811
2812#define skb_gro_checksum_try_convert(skb, proto, check, compute_pseudo) \
2813do { \
2814 if (__skb_gro_checksum_convert_check(skb)) \
2815 __skb_gro_checksum_convert(skb, check, \
2816 compute_pseudo(skb, proto)); \
2817} while (0)
2818
26c4f7da
TH
2819struct gro_remcsum {
2820 int offset;
2821 __wsum delta;
2822};
2823
2824static inline void skb_gro_remcsum_init(struct gro_remcsum *grc)
2825{
846cd667 2826 grc->offset = 0;
26c4f7da
TH
2827 grc->delta = 0;
2828}
2829
b7fe10e5
TH
2830static inline void *skb_gro_remcsum_process(struct sk_buff *skb, void *ptr,
2831 unsigned int off, size_t hdrlen,
2832 int start, int offset,
2833 struct gro_remcsum *grc,
2834 bool nopartial)
dcdc8994
TH
2835{
2836 __wsum delta;
b7fe10e5 2837 size_t plen = hdrlen + max_t(size_t, offset + sizeof(u16), start);
dcdc8994
TH
2838
2839 BUG_ON(!NAPI_GRO_CB(skb)->csum_valid);
2840
15e2396d 2841 if (!nopartial) {
b7fe10e5
TH
2842 NAPI_GRO_CB(skb)->gro_remcsum_start = off + hdrlen + start;
2843 return ptr;
2844 }
2845
2846 ptr = skb_gro_header_fast(skb, off);
2847 if (skb_gro_header_hard(skb, off + plen)) {
2848 ptr = skb_gro_header_slow(skb, off + plen, off);
2849 if (!ptr)
2850 return NULL;
15e2396d
TH
2851 }
2852
b7fe10e5
TH
2853 delta = remcsum_adjust(ptr + hdrlen, NAPI_GRO_CB(skb)->csum,
2854 start, offset);
dcdc8994
TH
2855
2856 /* Adjust skb->csum since we changed the packet */
dcdc8994 2857 NAPI_GRO_CB(skb)->csum = csum_add(NAPI_GRO_CB(skb)->csum, delta);
26c4f7da 2858
b7fe10e5 2859 grc->offset = off + hdrlen + offset;
26c4f7da 2860 grc->delta = delta;
b7fe10e5
TH
2861
2862 return ptr;
dcdc8994
TH
2863}
2864
26c4f7da
TH
2865static inline void skb_gro_remcsum_cleanup(struct sk_buff *skb,
2866 struct gro_remcsum *grc)
2867{
b7fe10e5
TH
2868 void *ptr;
2869 size_t plen = grc->offset + sizeof(u16);
2870
26c4f7da
TH
2871 if (!grc->delta)
2872 return;
2873
b7fe10e5
TH
2874 ptr = skb_gro_header_fast(skb, grc->offset);
2875 if (skb_gro_header_hard(skb, grc->offset + sizeof(u16))) {
2876 ptr = skb_gro_header_slow(skb, plen, grc->offset);
2877 if (!ptr)
2878 return;
2879 }
2880
2881 remcsum_unadjust((__sum16 *)ptr, grc->delta);
26c4f7da 2882}
dcdc8994 2883
25393d3f 2884#ifdef CONFIG_XFRM_OFFLOAD
d4546c25 2885static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
25393d3f
SK
2886{
2887 if (PTR_ERR(pp) != -EINPROGRESS)
2888 NAPI_GRO_CB(skb)->flush |= flush;
2889}
603d4cf8 2890static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 2891 struct sk_buff *pp,
603d4cf8
SD
2892 int flush,
2893 struct gro_remcsum *grc)
2894{
2895 if (PTR_ERR(pp) != -EINPROGRESS) {
2896 NAPI_GRO_CB(skb)->flush |= flush;
2897 skb_gro_remcsum_cleanup(skb, grc);
2898 skb->remcsum_offload = 0;
2899 }
2900}
25393d3f 2901#else
d4546c25 2902static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
5f114163
SK
2903{
2904 NAPI_GRO_CB(skb)->flush |= flush;
2905}
603d4cf8 2906static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 2907 struct sk_buff *pp,
603d4cf8
SD
2908 int flush,
2909 struct gro_remcsum *grc)
2910{
2911 NAPI_GRO_CB(skb)->flush |= flush;
2912 skb_gro_remcsum_cleanup(skb, grc);
2913 skb->remcsum_offload = 0;
2914}
25393d3f 2915#endif
5f114163 2916
0c4e8581
SH
2917static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
2918 unsigned short type,
3b04ddde 2919 const void *daddr, const void *saddr,
95c96174 2920 unsigned int len)
0c4e8581 2921{
f1ecfd5d 2922 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 2923 return 0;
3b04ddde
SH
2924
2925 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
2926}
2927
b95cce35
SH
2928static inline int dev_parse_header(const struct sk_buff *skb,
2929 unsigned char *haddr)
2930{
2931 const struct net_device *dev = skb->dev;
2932
1b83336b 2933 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 2934 return 0;
3b04ddde 2935 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
2936}
2937
e78b2915
MM
2938static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
2939{
2940 const struct net_device *dev = skb->dev;
2941
2942 if (!dev->header_ops || !dev->header_ops->parse_protocol)
2943 return 0;
2944 return dev->header_ops->parse_protocol(skb);
2945}
2946
2793a23a
WB
2947/* ll_header must have at least hard_header_len allocated */
2948static inline bool dev_validate_header(const struct net_device *dev,
2949 char *ll_header, int len)
2950{
2951 if (likely(len >= dev->hard_header_len))
2952 return true;
217e6fa2
WB
2953 if (len < dev->min_header_len)
2954 return false;
2793a23a
WB
2955
2956 if (capable(CAP_SYS_RAWIO)) {
2957 memset(ll_header + len, 0, dev->hard_header_len - len);
2958 return true;
2959 }
2960
2961 if (dev->header_ops && dev->header_ops->validate)
2962 return dev->header_ops->validate(ll_header, len);
2963
2964 return false;
2965}
2966
36fd633e
AV
2967typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr,
2968 int len, int size);
f629d208 2969int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
1da177e4
LT
2970static inline int unregister_gifconf(unsigned int family)
2971{
2972 return register_gifconf(family, NULL);
2973}
2974
99bbc707 2975#ifdef CONFIG_NET_FLOW_LIMIT
5f121b9a 2976#define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
99bbc707
WB
2977struct sd_flow_limit {
2978 u64 count;
2979 unsigned int num_buckets;
2980 unsigned int history_head;
2981 u16 history[FLOW_LIMIT_HISTORY];
2982 u8 buckets[];
2983};
2984
2985extern int netdev_flow_limit_table_len;
2986#endif /* CONFIG_NET_FLOW_LIMIT */
2987
1da177e4 2988/*
5e82b4b2 2989 * Incoming packets are placed on per-CPU queues
1da177e4 2990 */
d94d9fee 2991struct softnet_data {
1da177e4 2992 struct list_head poll_list;
6e7676c1 2993 struct sk_buff_head process_queue;
1da177e4 2994
dee42870 2995 /* stats */
cd7b5396
DM
2996 unsigned int processed;
2997 unsigned int time_squeeze;
cd7b5396 2998 unsigned int received_rps;
fd793d89 2999#ifdef CONFIG_RPS
88751275 3000 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
3001#endif
3002#ifdef CONFIG_NET_FLOW_LIMIT
3003 struct sd_flow_limit __rcu *flow_limit;
3004#endif
3005 struct Qdisc *output_queue;
3006 struct Qdisc **output_queue_tailp;
3007 struct sk_buff *completion_queue;
f53c7239
SK
3008#ifdef CONFIG_XFRM_OFFLOAD
3009 struct sk_buff_head xfrm_backlog;
3010#endif
97cdcf37
FW
3011 /* written and read only by owning cpu: */
3012 struct {
3013 u16 recursion;
3014 u8 more;
3015 } xmit;
4cdb1e2e 3016#ifdef CONFIG_RPS
501e7ef5
ED
3017 /* input_queue_head should be written by cpu owning this struct,
3018 * and only read by other cpus. Worth using a cache line.
3019 */
3020 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3021
3022 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3023 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3024 struct softnet_data *rps_ipi_next;
3025 unsigned int cpu;
76cc8b13 3026 unsigned int input_queue_tail;
1e94d72f 3027#endif
95c96174 3028 unsigned int dropped;
0a9627f2 3029 struct sk_buff_head input_pkt_queue;
bea3348e 3030 struct napi_struct backlog;
99bbc707 3031
1da177e4
LT
3032};
3033
76cc8b13 3034static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3035{
3036#ifdef CONFIG_RPS
76cc8b13
TH
3037 sd->input_queue_head++;
3038#endif
3039}
3040
3041static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3042 unsigned int *qtail)
3043{
3044#ifdef CONFIG_RPS
3045 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3046#endif
3047}
3048
0a9627f2 3049DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3050
97cdcf37
FW
3051static inline int dev_recursion_level(void)
3052{
28b05b92 3053 return this_cpu_read(softnet_data.xmit.recursion);
97cdcf37
FW
3054}
3055
3056#define XMIT_RECURSION_LIMIT 10
3057static inline bool dev_xmit_recursion(void)
3058{
3059 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3060 XMIT_RECURSION_LIMIT);
3061}
3062
3063static inline void dev_xmit_recursion_inc(void)
3064{
3065 __this_cpu_inc(softnet_data.xmit.recursion);
3066}
3067
3068static inline void dev_xmit_recursion_dec(void)
3069{
3070 __this_cpu_dec(softnet_data.xmit.recursion);
3071}
3072
f629d208 3073void __netif_schedule(struct Qdisc *q);
46e5da40 3074void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3075
fd2ea0a7
DM
3076static inline void netif_tx_schedule_all(struct net_device *dev)
3077{
3078 unsigned int i;
3079
3080 for (i = 0; i < dev->num_tx_queues; i++)
3081 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3082}
3083
f9a7cbbf 3084static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3085{
73466498 3086 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3087}
3088
bea3348e
SH
3089/**
3090 * netif_start_queue - allow transmit
3091 * @dev: network device
3092 *
3093 * Allow upper layers to call the device hard_start_xmit routine.
3094 */
1da177e4
LT
3095static inline void netif_start_queue(struct net_device *dev)
3096{
e8a0464c 3097 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3098}
3099
fd2ea0a7
DM
3100static inline void netif_tx_start_all_queues(struct net_device *dev)
3101{
3102 unsigned int i;
3103
3104 for (i = 0; i < dev->num_tx_queues; i++) {
3105 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3106 netif_tx_start_queue(txq);
3107 }
3108}
3109
46e5da40 3110void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3111
d29f749e
DJ
3112/**
3113 * netif_wake_queue - restart transmit
3114 * @dev: network device
3115 *
3116 * Allow upper layers to call the device hard_start_xmit routine.
3117 * Used for flow control when transmit resources are available.
3118 */
79d16385
DM
3119static inline void netif_wake_queue(struct net_device *dev)
3120{
e8a0464c 3121 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3122}
3123
fd2ea0a7
DM
3124static inline void netif_tx_wake_all_queues(struct net_device *dev)
3125{
3126 unsigned int i;
3127
3128 for (i = 0; i < dev->num_tx_queues; i++) {
3129 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3130 netif_tx_wake_queue(txq);
3131 }
3132}
3133
f9a7cbbf 3134static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3135{
73466498 3136 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3137}
3138
bea3348e
SH
3139/**
3140 * netif_stop_queue - stop transmitted packets
3141 * @dev: network device
3142 *
3143 * Stop upper layers calling the device hard_start_xmit routine.
3144 * Used for flow control when transmit resources are unavailable.
3145 */
1da177e4
LT
3146static inline void netif_stop_queue(struct net_device *dev)
3147{
e8a0464c 3148 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3149}
3150
a2029240 3151void netif_tx_stop_all_queues(struct net_device *dev);
ab92d68f 3152void netdev_update_lockdep_key(struct net_device *dev);
fd2ea0a7 3153
4d29515f 3154static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3155{
73466498 3156 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3157}
3158
bea3348e
SH
3159/**
3160 * netif_queue_stopped - test if transmit queue is flowblocked
3161 * @dev: network device
3162 *
3163 * Test if transmit queue on device is currently unable to send.
3164 */
4d29515f 3165static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3166{
e8a0464c 3167 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3168}
3169
4d29515f 3170static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3171{
73466498
TH
3172 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3173}
3174
8e2f1a63
DB
3175static inline bool
3176netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3177{
3178 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3179}
3180
8e2f1a63
DB
3181static inline bool
3182netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3183{
3184 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3185}
3186
53511453
ED
3187/**
3188 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3189 * @dev_queue: pointer to transmit queue
3190 *
3191 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3192 * to give appropriate hint to the CPU.
53511453
ED
3193 */
3194static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3195{
3196#ifdef CONFIG_BQL
3197 prefetchw(&dev_queue->dql.num_queued);
3198#endif
3199}
3200
3201/**
3202 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3203 * @dev_queue: pointer to transmit queue
3204 *
3205 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3206 * to give appropriate hint to the CPU.
53511453
ED
3207 */
3208static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3209{
3210#ifdef CONFIG_BQL
3211 prefetchw(&dev_queue->dql.limit);
3212#endif
3213}
3214
c5d67bd7
TH
3215static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3216 unsigned int bytes)
3217{
114cf580
TH
3218#ifdef CONFIG_BQL
3219 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3220
3221 if (likely(dql_avail(&dev_queue->dql) >= 0))
3222 return;
3223
3224 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3225
3226 /*
3227 * The XOFF flag must be set before checking the dql_avail below,
3228 * because in netdev_tx_completed_queue we update the dql_completed
3229 * before checking the XOFF flag.
3230 */
3231 smp_mb();
3232
3233 /* check again in case another CPU has just made room avail */
3234 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3235 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3236#endif
c5d67bd7
TH
3237}
3238
3e59020a
ED
3239/* Variant of netdev_tx_sent_queue() for drivers that are aware
3240 * that they should not test BQL status themselves.
3241 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3242 * skb of a batch.
3243 * Returns true if the doorbell must be used to kick the NIC.
3244 */
3245static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3246 unsigned int bytes,
3247 bool xmit_more)
3248{
3249 if (xmit_more) {
3250#ifdef CONFIG_BQL
3251 dql_queued(&dev_queue->dql, bytes);
3252#endif
3253 return netif_tx_queue_stopped(dev_queue);
3254 }
3255 netdev_tx_sent_queue(dev_queue, bytes);
3256 return true;
3257}
3258
0042d0c8
FF
3259/**
3260 * netdev_sent_queue - report the number of bytes queued to hardware
3261 * @dev: network device
3262 * @bytes: number of bytes queued to the hardware device queue
3263 *
3264 * Report the number of bytes queued for sending/completion to the network
3265 * device hardware queue. @bytes should be a good approximation and should
3266 * exactly match netdev_completed_queue() @bytes
3267 */
c5d67bd7
TH
3268static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3269{
3270 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3271}
3272
620344c4
HK
3273static inline bool __netdev_sent_queue(struct net_device *dev,
3274 unsigned int bytes,
3275 bool xmit_more)
3276{
3277 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3278 xmit_more);
3279}
3280
c5d67bd7 3281static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3282 unsigned int pkts, unsigned int bytes)
c5d67bd7 3283{
114cf580 3284#ifdef CONFIG_BQL
b37c0fbe
AD
3285 if (unlikely(!bytes))
3286 return;
3287
3288 dql_completed(&dev_queue->dql, bytes);
3289
3290 /*
3291 * Without the memory barrier there is a small possiblity that
3292 * netdev_tx_sent_queue will miss the update and cause the queue to
3293 * be stopped forever
3294 */
3295 smp_mb();
3296
f3acd33d 3297 if (unlikely(dql_avail(&dev_queue->dql) < 0))
b37c0fbe
AD
3298 return;
3299
3300 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3301 netif_schedule_queue(dev_queue);
114cf580 3302#endif
c5d67bd7
TH
3303}
3304
0042d0c8
FF
3305/**
3306 * netdev_completed_queue - report bytes and packets completed by device
3307 * @dev: network device
3308 * @pkts: actual number of packets sent over the medium
3309 * @bytes: actual number of bytes sent over the medium
3310 *
3311 * Report the number of bytes and packets transmitted by the network device
3312 * hardware queue over the physical medium, @bytes must exactly match the
3313 * @bytes amount passed to netdev_sent_queue()
3314 */
c5d67bd7 3315static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3316 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3317{
3318 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3319}
3320
3321static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3322{
114cf580 3323#ifdef CONFIG_BQL
5c490354 3324 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3325 dql_reset(&q->dql);
3326#endif
c5d67bd7
TH
3327}
3328
0042d0c8
FF
3329/**
3330 * netdev_reset_queue - reset the packets and bytes count of a network device
3331 * @dev_queue: network device
3332 *
3333 * Reset the bytes and packet count of a network device and clear the
3334 * software flow control OFF bit for this network device
3335 */
c5d67bd7
TH
3336static inline void netdev_reset_queue(struct net_device *dev_queue)
3337{
3338 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3339}
3340
b9507bda
DB
3341/**
3342 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3343 * @dev: network device
3344 * @queue_index: given tx queue index
3345 *
3346 * Returns 0 if given tx queue index >= number of device tx queues,
3347 * otherwise returns the originally passed tx queue index.
3348 */
3349static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3350{
3351 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3352 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3353 dev->name, queue_index,
3354 dev->real_num_tx_queues);
3355 return 0;
3356 }
3357
3358 return queue_index;
3359}
3360
bea3348e
SH
3361/**
3362 * netif_running - test if up
3363 * @dev: network device
3364 *
3365 * Test if the device has been brought up.
3366 */
4d29515f 3367static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3368{
3369 return test_bit(__LINK_STATE_START, &dev->state);
3370}
3371
f25f4e44 3372/*
5e82b4b2 3373 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3374 * stop, and a check if it's stopped. All other device management is
3375 * done at the overall netdevice level.
3376 * Also test the device if we're multiqueue.
3377 */
bea3348e
SH
3378
3379/**
3380 * netif_start_subqueue - allow sending packets on subqueue
3381 * @dev: network device
3382 * @queue_index: sub queue index
3383 *
3384 * Start individual transmit queue of a device with multiple transmit queues.
3385 */
f25f4e44
PWJ
3386static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3387{
fd2ea0a7 3388 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3389
3390 netif_tx_start_queue(txq);
f25f4e44
PWJ
3391}
3392
bea3348e
SH
3393/**
3394 * netif_stop_subqueue - stop sending packets on subqueue
3395 * @dev: network device
3396 * @queue_index: sub queue index
3397 *
3398 * Stop individual transmit queue of a device with multiple transmit queues.
3399 */
f25f4e44
PWJ
3400static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3401{
fd2ea0a7 3402 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3403 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3404}
3405
bea3348e
SH
3406/**
3407 * netif_subqueue_stopped - test status of subqueue
3408 * @dev: network device
3409 * @queue_index: sub queue index
3410 *
3411 * Check individual transmit queue of a device with multiple transmit queues.
3412 */
4d29515f
DM
3413static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3414 u16 queue_index)
f25f4e44 3415{
fd2ea0a7 3416 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3417
3418 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3419}
3420
4d29515f
DM
3421static inline bool netif_subqueue_stopped(const struct net_device *dev,
3422 struct sk_buff *skb)
668f895a
PE
3423{
3424 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3425}
bea3348e 3426
738b35cc
FF
3427/**
3428 * netif_wake_subqueue - allow sending packets on subqueue
3429 * @dev: network device
3430 * @queue_index: sub queue index
3431 *
3432 * Resume individual transmit queue of a device with multiple transmit queues.
3433 */
3434static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3435{
3436 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3437
3438 netif_tx_wake_queue(txq);
3439}
f25f4e44 3440
537c00de 3441#ifdef CONFIG_XPS
53af53ae 3442int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3443 u16 index);
80d19669
AN
3444int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
3445 u16 index, bool is_rxqs_map);
3446
3447/**
3448 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3449 * @j: CPU/Rx queue index
3450 * @mask: bitmask of all cpus/rx queues
3451 * @nr_bits: number of bits in the bitmask
3452 *
3453 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3454 */
3455static inline bool netif_attr_test_mask(unsigned long j,
3456 const unsigned long *mask,
3457 unsigned int nr_bits)
3458{
3459 cpu_max_bits_warn(j, nr_bits);
3460 return test_bit(j, mask);
3461}
3462
3463/**
3464 * netif_attr_test_online - Test for online CPU/Rx queue
3465 * @j: CPU/Rx queue index
3466 * @online_mask: bitmask for CPUs/Rx queues that are online
3467 * @nr_bits: number of bits in the bitmask
3468 *
3469 * Returns true if a CPU/Rx queue is online.
3470 */
3471static inline bool netif_attr_test_online(unsigned long j,
3472 const unsigned long *online_mask,
3473 unsigned int nr_bits)
3474{
3475 cpu_max_bits_warn(j, nr_bits);
3476
3477 if (online_mask)
3478 return test_bit(j, online_mask);
3479
3480 return (j < nr_bits);
3481}
3482
3483/**
3484 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3485 * @n: CPU/Rx queue index
3486 * @srcp: the cpumask/Rx queue mask pointer
3487 * @nr_bits: number of bits in the bitmask
3488 *
3489 * Returns >= nr_bits if no further CPUs/Rx queues set.
3490 */
3491static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3492 unsigned int nr_bits)
3493{
3494 /* -1 is a legal arg here. */
3495 if (n != -1)
3496 cpu_max_bits_warn(n, nr_bits);
3497
3498 if (srcp)
3499 return find_next_bit(srcp, nr_bits, n + 1);
3500
3501 return n + 1;
3502}
3503
3504/**
3505 * netif_attrmask_next_and - get the next CPU/Rx queue in *src1p & *src2p
3506 * @n: CPU/Rx queue index
3507 * @src1p: the first CPUs/Rx queues mask pointer
3508 * @src2p: the second CPUs/Rx queues mask pointer
3509 * @nr_bits: number of bits in the bitmask
3510 *
3511 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3512 */
3513static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3514 const unsigned long *src2p,
3515 unsigned int nr_bits)
3516{
3517 /* -1 is a legal arg here. */
3518 if (n != -1)
3519 cpu_max_bits_warn(n, nr_bits);
3520
3521 if (src1p && src2p)
3522 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3523 else if (src1p)
3524 return find_next_bit(src1p, nr_bits, n + 1);
3525 else if (src2p)
3526 return find_next_bit(src2p, nr_bits, n + 1);
3527
3528 return n + 1;
3529}
537c00de
AD
3530#else
3531static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3532 const struct cpumask *mask,
537c00de
AD
3533 u16 index)
3534{
3535 return 0;
3536}
c9fbb2d2
KK
3537
3538static inline int __netif_set_xps_queue(struct net_device *dev,
3539 const unsigned long *mask,
3540 u16 index, bool is_rxqs_map)
3541{
3542 return 0;
3543}
537c00de
AD
3544#endif
3545
bea3348e
SH
3546/**
3547 * netif_is_multiqueue - test if device has multiple transmit queues
3548 * @dev: network device
3549 *
3550 * Check if device has multiple transmit queues
bea3348e 3551 */
4d29515f 3552static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3553{
a02cec21 3554 return dev->num_tx_queues > 1;
f25f4e44 3555}
1da177e4 3556
f629d208 3557int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3558
a953be53 3559#ifdef CONFIG_SYSFS
f629d208 3560int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3561#else
3562static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3563 unsigned int rxqs)
62fe0b40 3564{
c29c2ebd 3565 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3566 return 0;
3567}
3568#endif
3569
65073a67
DB
3570static inline struct netdev_rx_queue *
3571__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
3572{
3573 return dev->_rx + rxq;
3574}
3575
a953be53
MD
3576#ifdef CONFIG_SYSFS
3577static inline unsigned int get_netdev_rx_queue_index(
3578 struct netdev_rx_queue *queue)
3579{
3580 struct net_device *dev = queue->dev;
3581 int index = queue - dev->_rx;
3582
3583 BUG_ON(index >= dev->num_rx_queues);
3584 return index;
3585}
3586#endif
3587
16917b87 3588#define DEFAULT_MAX_NUM_RSS_QUEUES (8)
f629d208 3589int netif_get_num_default_rss_queues(void);
16917b87 3590
e6247027
ED
3591enum skb_free_reason {
3592 SKB_REASON_CONSUMED,
3593 SKB_REASON_DROPPED,
3594};
3595
3596void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3597void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3598
e6247027
ED
3599/*
3600 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3601 * interrupt context or with hardware interrupts being disabled.
3602 * (in_irq() || irqs_disabled())
3603 *
3604 * We provide four helpers that can be used in following contexts :
3605 *
3606 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3607 * replacing kfree_skb(skb)
3608 *
3609 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3610 * Typically used in place of consume_skb(skb) in TX completion path
3611 *
3612 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3613 * replacing kfree_skb(skb)
3614 *
3615 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3616 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3617 */
e6247027
ED
3618static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3619{
3620 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3621}
3622
3623static inline void dev_consume_skb_irq(struct sk_buff *skb)
3624{
3625 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3626}
3627
3628static inline void dev_kfree_skb_any(struct sk_buff *skb)
3629{
3630 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3631}
3632
3633static inline void dev_consume_skb_any(struct sk_buff *skb)
3634{
3635 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3636}
1da177e4 3637
7c497478
JW
3638void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3639int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208
JP
3640int netif_rx(struct sk_buff *skb);
3641int netif_rx_ni(struct sk_buff *skb);
04eb4489 3642int netif_receive_skb(struct sk_buff *skb);
1c601d82 3643int netif_receive_skb_core(struct sk_buff *skb);
f6ad8c1b 3644void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3645gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3646void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3647struct sk_buff *napi_get_frags(struct napi_struct *napi);
3648gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3649struct packet_offload *gro_find_receive_by_type(__be16 type);
3650struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3651
3652static inline void napi_free_frags(struct napi_struct *napi)
3653{
3654 kfree_skb(napi->skb);
3655 napi->skb = NULL;
3656}
3657
24b27fc4 3658bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
3659int netdev_rx_handler_register(struct net_device *dev,
3660 rx_handler_func_t *rx_handler,
3661 void *rx_handler_data);
3662void netdev_rx_handler_unregister(struct net_device *dev);
3663
3664bool dev_valid_name(const char *name);
44c02a2c
AV
3665int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
3666 bool *need_copyout);
36fd633e 3667int dev_ifconf(struct net *net, struct ifconf *, int);
f629d208
JP
3668int dev_ethtool(struct net *net, struct ifreq *);
3669unsigned int dev_get_flags(const struct net_device *);
6d040321
PM
3670int __dev_change_flags(struct net_device *dev, unsigned int flags,
3671 struct netlink_ext_ack *extack);
567c5e13
PM
3672int dev_change_flags(struct net_device *dev, unsigned int flags,
3673 struct netlink_ext_ack *extack);
cb178190
DM
3674void __dev_notify_flags(struct net_device *, unsigned int old_flags,
3675 unsigned int gchanges);
f629d208
JP
3676int dev_change_name(struct net_device *, const char *);
3677int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 3678int dev_get_alias(const struct net_device *, char *, size_t);
f629d208 3679int dev_change_net_namespace(struct net_device *, struct net *, const char *);
f51048c3 3680int __dev_set_mtu(struct net_device *, int);
7a4c53be
SH
3681int dev_set_mtu_ext(struct net_device *dev, int mtu,
3682 struct netlink_ext_ack *extack);
f629d208 3683int dev_set_mtu(struct net_device *, int);
6a643ddb 3684int dev_change_tx_queue_len(struct net_device *, unsigned long);
f629d208 3685void dev_set_group(struct net_device *, int);
d59cdf94
PM
3686int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
3687 struct netlink_ext_ack *extack);
3a37a963
PM
3688int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
3689 struct netlink_ext_ack *extack);
f629d208
JP
3690int dev_change_carrier(struct net_device *, bool new_carrier);
3691int dev_get_phys_port_id(struct net_device *dev,
02637fce 3692 struct netdev_phys_item_id *ppid);
db24a904
DA
3693int dev_get_phys_port_name(struct net_device *dev,
3694 char *name, size_t len);
d6abc596
FF
3695int dev_get_port_parent_id(struct net_device *dev,
3696 struct netdev_phys_item_id *ppid, bool recurse);
3697bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
d746d707 3698int dev_change_proto_down(struct net_device *dev, bool proto_down);
b5899679 3699int dev_change_proto_down_generic(struct net_device *dev, bool proto_down);
f53c7239 3700struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
3701struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3702 struct netdev_queue *txq, int *ret);
d67b9cd2 3703
f4e63525 3704typedef int (*bpf_op_t)(struct net_device *dev, struct netdev_bpf *bpf);
d67b9cd2
DB
3705int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack,
3706 int fd, u32 flags);
a25717d2
JK
3707u32 __dev_xdp_query(struct net_device *dev, bpf_op_t xdp_op,
3708 enum bpf_netdev_command cmd);
84c6b868 3709int xdp_umem_query(struct net_device *dev, u16 queue_id);
d67b9cd2 3710
a0265d28 3711int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 3712int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
3713bool is_skb_forwardable(const struct net_device *dev,
3714 const struct sk_buff *skb);
1da177e4 3715
4e3264d2
MKL
3716static __always_inline int ____dev_forward_skb(struct net_device *dev,
3717 struct sk_buff *skb)
3718{
3719 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
3720 unlikely(!is_skb_forwardable(dev, skb))) {
3721 atomic_long_inc(&dev->rx_dropped);
3722 kfree_skb(skb);
3723 return NET_RX_DROP;
3724 }
3725
3726 skb_scrub_packet(skb, true);
3727 skb->priority = 0;
3728 return 0;
3729}
3730
9f9a742d 3731bool dev_nit_active(struct net_device *dev);
74b20582
DA
3732void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
3733
20380731 3734extern int netdev_budget;
7acf8a1e 3735extern unsigned int netdev_budget_usecs;
1da177e4
LT
3736
3737/* Called by rtnetlink.c:rtnl_unlock() */
f629d208 3738void netdev_run_todo(void);
1da177e4 3739
bea3348e
SH
3740/**
3741 * dev_put - release reference to device
3742 * @dev: network device
3743 *
9ef4429b 3744 * Release reference to device to allow it to be freed.
bea3348e 3745 */
1da177e4
LT
3746static inline void dev_put(struct net_device *dev)
3747{
933393f5 3748 this_cpu_dec(*dev->pcpu_refcnt);
1da177e4
LT
3749}
3750
bea3348e
SH
3751/**
3752 * dev_hold - get reference to device
3753 * @dev: network device
3754 *
9ef4429b 3755 * Hold reference to device to keep it from being freed.
bea3348e 3756 */
15333061
SH
3757static inline void dev_hold(struct net_device *dev)
3758{
933393f5 3759 this_cpu_inc(*dev->pcpu_refcnt);
15333061 3760}
1da177e4
LT
3761
3762/* Carrier loss detection, dial on demand. The functions netif_carrier_on
3763 * and _off may be called from IRQ context, but it is caller
3764 * who is responsible for serialization of these calls.
b00055aa
SR
3765 *
3766 * The name carrier is inappropriate, these functions should really be
3767 * called netif_lowerlayer_*() because they represent the state of any
3768 * kind of lower layer not just hardware media.
1da177e4
LT
3769 */
3770
f629d208
JP
3771void linkwatch_init_dev(struct net_device *dev);
3772void linkwatch_fire_event(struct net_device *dev);
3773void linkwatch_forget_dev(struct net_device *dev);
1da177e4 3774
bea3348e
SH
3775/**
3776 * netif_carrier_ok - test if carrier present
3777 * @dev: network device
3778 *
3779 * Check if carrier is present on device
3780 */
4d29515f 3781static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
3782{
3783 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
3784}
3785
f629d208 3786unsigned long dev_trans_start(struct net_device *dev);
9d21493b 3787
f629d208 3788void __netdev_watchdog_up(struct net_device *dev);
1da177e4 3789
f629d208 3790void netif_carrier_on(struct net_device *dev);
1da177e4 3791
f629d208 3792void netif_carrier_off(struct net_device *dev);
1da177e4 3793
bea3348e
SH
3794/**
3795 * netif_dormant_on - mark device as dormant.
3796 * @dev: network device
3797 *
3798 * Mark device as dormant (as per RFC2863).
3799 *
3800 * The dormant state indicates that the relevant interface is not
3801 * actually in a condition to pass packets (i.e., it is not 'up') but is
3802 * in a "pending" state, waiting for some external event. For "on-
3803 * demand" interfaces, this new state identifies the situation where the
3804 * interface is waiting for events to place it in the up state.
bea3348e 3805 */
b00055aa
SR
3806static inline void netif_dormant_on(struct net_device *dev)
3807{
3808 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
3809 linkwatch_fire_event(dev);
3810}
3811
bea3348e
SH
3812/**
3813 * netif_dormant_off - set device as not dormant.
3814 * @dev: network device
3815 *
3816 * Device is not in dormant state.
3817 */
b00055aa
SR
3818static inline void netif_dormant_off(struct net_device *dev)
3819{
3820 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
3821 linkwatch_fire_event(dev);
3822}
3823
bea3348e 3824/**
8ecbc40a 3825 * netif_dormant - test if device is dormant
bea3348e
SH
3826 * @dev: network device
3827 *
8ecbc40a 3828 * Check if device is dormant.
bea3348e 3829 */
4d29515f 3830static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
3831{
3832 return test_bit(__LINK_STATE_DORMANT, &dev->state);
3833}
3834
3835
bea3348e
SH
3836/**
3837 * netif_oper_up - test if device is operational
3838 * @dev: network device
3839 *
3840 * Check if carrier is operational
3841 */
4d29515f 3842static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 3843{
b00055aa
SR
3844 return (dev->operstate == IF_OPER_UP ||
3845 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
3846}
3847
bea3348e
SH
3848/**
3849 * netif_device_present - is device available or removed
3850 * @dev: network device
3851 *
3852 * Check if device has not been removed from system.
3853 */
4d29515f 3854static inline bool netif_device_present(struct net_device *dev)
1da177e4
LT
3855{
3856 return test_bit(__LINK_STATE_PRESENT, &dev->state);
3857}
3858
f629d208 3859void netif_device_detach(struct net_device *dev);
1da177e4 3860
f629d208 3861void netif_device_attach(struct net_device *dev);
1da177e4
LT
3862
3863/*
3864 * Network interface message level settings
3865 */
1da177e4
LT
3866
3867enum {
3868 NETIF_MSG_DRV = 0x0001,
3869 NETIF_MSG_PROBE = 0x0002,
3870 NETIF_MSG_LINK = 0x0004,
3871 NETIF_MSG_TIMER = 0x0008,
3872 NETIF_MSG_IFDOWN = 0x0010,
3873 NETIF_MSG_IFUP = 0x0020,
3874 NETIF_MSG_RX_ERR = 0x0040,
3875 NETIF_MSG_TX_ERR = 0x0080,
3876 NETIF_MSG_TX_QUEUED = 0x0100,
3877 NETIF_MSG_INTR = 0x0200,
3878 NETIF_MSG_TX_DONE = 0x0400,
3879 NETIF_MSG_RX_STATUS = 0x0800,
3880 NETIF_MSG_PKTDATA = 0x1000,
3881 NETIF_MSG_HW = 0x2000,
3882 NETIF_MSG_WOL = 0x4000,
3883};
3884
3885#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
3886#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
3887#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
3888#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
3889#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
3890#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
3891#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
3892#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
3893#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
3894#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
3895#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
3896#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
3897#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
3898#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
3899#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
3900
3901static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
3902{
3903 /* use default */
3904 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
3905 return default_msg_enable_bits;
3906 if (debug_value == 0) /* no output */
3907 return 0;
3908 /* set low N bits */
f4d7b3e2 3909 return (1U << debug_value) - 1;
1da177e4
LT
3910}
3911
c773e847 3912static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 3913{
c773e847
DM
3914 spin_lock(&txq->_xmit_lock);
3915 txq->xmit_lock_owner = cpu;
22dd7495
JHS
3916}
3917
5a717f4f
MT
3918static inline bool __netif_tx_acquire(struct netdev_queue *txq)
3919{
3920 __acquire(&txq->_xmit_lock);
3921 return true;
3922}
3923
3924static inline void __netif_tx_release(struct netdev_queue *txq)
3925{
3926 __release(&txq->_xmit_lock);
3927}
3928
fd2ea0a7
DM
3929static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
3930{
3931 spin_lock_bh(&txq->_xmit_lock);
3932 txq->xmit_lock_owner = smp_processor_id();
3933}
3934
4d29515f 3935static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 3936{
4d29515f 3937 bool ok = spin_trylock(&txq->_xmit_lock);
c3f26a26
DM
3938 if (likely(ok))
3939 txq->xmit_lock_owner = smp_processor_id();
3940 return ok;
3941}
3942
3943static inline void __netif_tx_unlock(struct netdev_queue *txq)
3944{
3945 txq->xmit_lock_owner = -1;
3946 spin_unlock(&txq->_xmit_lock);
3947}
3948
3949static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
3950{
3951 txq->xmit_lock_owner = -1;
3952 spin_unlock_bh(&txq->_xmit_lock);
3953}
3954
08baf561
ED
3955static inline void txq_trans_update(struct netdev_queue *txq)
3956{
3957 if (txq->xmit_lock_owner != -1)
3958 txq->trans_start = jiffies;
3959}
3960
ba162f8e
FW
3961/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
3962static inline void netif_trans_update(struct net_device *dev)
3963{
9b36627a
FW
3964 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
3965
3966 if (txq->trans_start != jiffies)
3967 txq->trans_start = jiffies;
ba162f8e
FW
3968}
3969
d29f749e
DJ
3970/**
3971 * netif_tx_lock - grab network device transmit lock
3972 * @dev: network device
d29f749e
DJ
3973 *
3974 * Get network device transmit lock
3975 */
22dd7495
JHS
3976static inline void netif_tx_lock(struct net_device *dev)
3977{
e8a0464c 3978 unsigned int i;
c3f26a26 3979 int cpu;
c773e847 3980
c3f26a26
DM
3981 spin_lock(&dev->tx_global_lock);
3982 cpu = smp_processor_id();
e8a0464c
DM
3983 for (i = 0; i < dev->num_tx_queues; i++) {
3984 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
3985
3986 /* We are the only thread of execution doing a
3987 * freeze, but we have to grab the _xmit_lock in
3988 * order to synchronize with threads which are in
3989 * the ->hard_start_xmit() handler and already
3990 * checked the frozen bit.
3991 */
e8a0464c 3992 __netif_tx_lock(txq, cpu);
c3f26a26
DM
3993 set_bit(__QUEUE_STATE_FROZEN, &txq->state);
3994 __netif_tx_unlock(txq);
e8a0464c 3995 }
932ff279
HX
3996}
3997
3998static inline void netif_tx_lock_bh(struct net_device *dev)
3999{
e8a0464c
DM
4000 local_bh_disable();
4001 netif_tx_lock(dev);
932ff279
HX
4002}
4003
932ff279
HX
4004static inline void netif_tx_unlock(struct net_device *dev)
4005{
e8a0464c
DM
4006 unsigned int i;
4007
4008 for (i = 0; i < dev->num_tx_queues; i++) {
4009 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c773e847 4010
c3f26a26
DM
4011 /* No need to grab the _xmit_lock here. If the
4012 * queue is not stopped for another reason, we
4013 * force a schedule.
4014 */
4015 clear_bit(__QUEUE_STATE_FROZEN, &txq->state);
7b3d3e4f 4016 netif_schedule_queue(txq);
c3f26a26
DM
4017 }
4018 spin_unlock(&dev->tx_global_lock);
932ff279
HX
4019}
4020
4021static inline void netif_tx_unlock_bh(struct net_device *dev)
4022{
e8a0464c
DM
4023 netif_tx_unlock(dev);
4024 local_bh_enable();
932ff279
HX
4025}
4026
c773e847 4027#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 4028 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4029 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
4030 } else { \
4031 __netif_tx_acquire(txq); \
22dd7495
JHS
4032 } \
4033}
4034
5efeac44
EB
4035#define HARD_TX_TRYLOCK(dev, txq) \
4036 (((dev->features & NETIF_F_LLTX) == 0) ? \
4037 __netif_tx_trylock(txq) : \
5a717f4f 4038 __netif_tx_acquire(txq))
5efeac44 4039
c773e847 4040#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 4041 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4042 __netif_tx_unlock(txq); \
5a717f4f
MT
4043 } else { \
4044 __netif_tx_release(txq); \
22dd7495
JHS
4045 } \
4046}
4047
1da177e4
LT
4048static inline void netif_tx_disable(struct net_device *dev)
4049{
fd2ea0a7 4050 unsigned int i;
c3f26a26 4051 int cpu;
fd2ea0a7 4052
c3f26a26
DM
4053 local_bh_disable();
4054 cpu = smp_processor_id();
fd2ea0a7
DM
4055 for (i = 0; i < dev->num_tx_queues; i++) {
4056 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4057
4058 __netif_tx_lock(txq, cpu);
fd2ea0a7 4059 netif_tx_stop_queue(txq);
c3f26a26 4060 __netif_tx_unlock(txq);
fd2ea0a7 4061 }
c3f26a26 4062 local_bh_enable();
1da177e4
LT
4063}
4064
e308a5d8
DM
4065static inline void netif_addr_lock(struct net_device *dev)
4066{
4067 spin_lock(&dev->addr_list_lock);
4068}
4069
4070static inline void netif_addr_lock_bh(struct net_device *dev)
4071{
4072 spin_lock_bh(&dev->addr_list_lock);
4073}
4074
4075static inline void netif_addr_unlock(struct net_device *dev)
4076{
4077 spin_unlock(&dev->addr_list_lock);
4078}
4079
4080static inline void netif_addr_unlock_bh(struct net_device *dev)
4081{
4082 spin_unlock_bh(&dev->addr_list_lock);
4083}
4084
f001fde5 4085/*
31278e71 4086 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4087 * rcu_read_lock held.
4088 */
4089#define for_each_dev_addr(dev, ha) \
31278e71 4090 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4091
1da177e4
LT
4092/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4093
f629d208 4094void ether_setup(struct net_device *dev);
1da177e4
LT
4095
4096/* Support for loadable net-drivers */
f629d208 4097struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4098 unsigned char name_assign_type,
f629d208
JP
4099 void (*setup)(struct net_device *),
4100 unsigned int txqs, unsigned int rxqs);
c835a677
TG
4101#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4102 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4103
c835a677
TG
4104#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4105 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4106 count)
36909ea4 4107
f629d208
JP
4108int register_netdev(struct net_device *dev);
4109void unregister_netdev(struct net_device *dev);
f001fde5 4110
22bedad3 4111/* General hardware address lists handling functions */
f629d208
JP
4112int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4113 struct netdev_hw_addr_list *from_list, int addr_len);
4114void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4115 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4116int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4117 struct net_device *dev,
4118 int (*sync)(struct net_device *, const unsigned char *),
4119 int (*unsync)(struct net_device *,
4120 const unsigned char *));
e7946760
IK
4121int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4122 struct net_device *dev,
4123 int (*sync)(struct net_device *,
4124 const unsigned char *, int),
4125 int (*unsync)(struct net_device *,
4126 const unsigned char *, int));
4127void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4128 struct net_device *dev,
4129 int (*unsync)(struct net_device *,
4130 const unsigned char *, int));
670e5b8e
AD
4131void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4132 struct net_device *dev,
4133 int (*unsync)(struct net_device *,
4134 const unsigned char *));
f629d208 4135void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4136
f001fde5 4137/* Functions used for device addresses handling */
f629d208
JP
4138int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4139 unsigned char addr_type);
4140int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4141 unsigned char addr_type);
f629d208
JP
4142void dev_addr_flush(struct net_device *dev);
4143int dev_addr_init(struct net_device *dev);
a748ee24
JP
4144
4145/* Functions used for unicast addresses handling */
f629d208
JP
4146int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4147int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4148int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4149int dev_uc_sync(struct net_device *to, struct net_device *from);
4150int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4151void dev_uc_unsync(struct net_device *to, struct net_device *from);
4152void dev_uc_flush(struct net_device *dev);
4153void dev_uc_init(struct net_device *dev);
f001fde5 4154
670e5b8e
AD
4155/**
4156 * __dev_uc_sync - Synchonize device's unicast list
4157 * @dev: device to sync
4158 * @sync: function to call if address should be added
4159 * @unsync: function to call if address should be removed
4160 *
4161 * Add newly added addresses to the interface, and release
4162 * addresses that have been deleted.
5e82b4b2 4163 */
670e5b8e
AD
4164static inline int __dev_uc_sync(struct net_device *dev,
4165 int (*sync)(struct net_device *,
4166 const unsigned char *),
4167 int (*unsync)(struct net_device *,
4168 const unsigned char *))
4169{
4170 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4171}
4172
4173/**
e793c0f7 4174 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4175 * @dev: device to sync
4176 * @unsync: function to call if address should be removed
4177 *
4178 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4179 */
670e5b8e
AD
4180static inline void __dev_uc_unsync(struct net_device *dev,
4181 int (*unsync)(struct net_device *,
4182 const unsigned char *))
4183{
4184 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4185}
4186
22bedad3 4187/* Functions used for multicast addresses handling */
f629d208
JP
4188int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4189int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4190int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4191int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4192int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4193int dev_mc_sync(struct net_device *to, struct net_device *from);
4194int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4195void dev_mc_unsync(struct net_device *to, struct net_device *from);
4196void dev_mc_flush(struct net_device *dev);
4197void dev_mc_init(struct net_device *dev);
f001fde5 4198
670e5b8e
AD
4199/**
4200 * __dev_mc_sync - Synchonize device's multicast list
4201 * @dev: device to sync
4202 * @sync: function to call if address should be added
4203 * @unsync: function to call if address should be removed
4204 *
4205 * Add newly added addresses to the interface, and release
4206 * addresses that have been deleted.
5e82b4b2 4207 */
670e5b8e
AD
4208static inline int __dev_mc_sync(struct net_device *dev,
4209 int (*sync)(struct net_device *,
4210 const unsigned char *),
4211 int (*unsync)(struct net_device *,
4212 const unsigned char *))
4213{
4214 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4215}
4216
4217/**
e793c0f7 4218 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4219 * @dev: device to sync
4220 * @unsync: function to call if address should be removed
4221 *
4222 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4223 */
670e5b8e
AD
4224static inline void __dev_mc_unsync(struct net_device *dev,
4225 int (*unsync)(struct net_device *,
4226 const unsigned char *))
4227{
4228 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4229}
4230
4417da66 4231/* Functions used for secondary unicast and multicast support */
f629d208
JP
4232void dev_set_rx_mode(struct net_device *dev);
4233void __dev_set_rx_mode(struct net_device *dev);
4234int dev_set_promiscuity(struct net_device *dev, int inc);
4235int dev_set_allmulti(struct net_device *dev, int inc);
4236void netdev_state_change(struct net_device *dev);
4237void netdev_notify_peers(struct net_device *dev);
4238void netdev_features_change(struct net_device *dev);
1da177e4 4239/* Load a device via the kmod */
f629d208
JP
4240void dev_load(struct net *net, const char *name);
4241struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4242 struct rtnl_link_stats64 *storage);
4243void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4244 const struct net_device_stats *netdev_stats);
eeda3fd6 4245
1da177e4 4246extern int netdev_max_backlog;
3b098e2d 4247extern int netdev_tstamp_prequeue;
1da177e4 4248extern int weight_p;
3d48b53f
MT
4249extern int dev_weight_rx_bias;
4250extern int dev_weight_tx_bias;
4251extern int dev_rx_weight;
4252extern int dev_tx_weight;
323ebb61 4253extern int gro_normal_batch;
9ff162a8 4254
f629d208 4255bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4256struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4257 struct list_head **iter);
f629d208
JP
4258struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
4259 struct list_head **iter);
8b5be856 4260
44a40855
VY
4261/* iterate through upper list, must be called under RCU read lock */
4262#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4263 for (iter = &(dev)->adj_list.upper, \
4264 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4265 updev; \
4266 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4267
1a3f060c
DA
4268int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4269 int (*fn)(struct net_device *upper_dev,
4270 void *data),
4271 void *data);
4272
4273bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4274 struct net_device *upper_dev);
4275
25cc72a3
IS
4276bool netdev_has_any_upper_dev(struct net_device *dev);
4277
f629d208
JP
4278void *netdev_lower_get_next_private(struct net_device *dev,
4279 struct list_head **iter);
4280void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4281 struct list_head **iter);
31088a11
VF
4282
4283#define netdev_for_each_lower_private(dev, priv, iter) \
4284 for (iter = (dev)->adj_list.lower.next, \
4285 priv = netdev_lower_get_next_private(dev, &(iter)); \
4286 priv; \
4287 priv = netdev_lower_get_next_private(dev, &(iter)))
4288
4289#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4290 for (iter = &(dev)->adj_list.lower, \
4291 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4292 priv; \
4293 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4294
4085ebe8
VY
4295void *netdev_lower_get_next(struct net_device *dev,
4296 struct list_head **iter);
7ce856aa 4297
4085ebe8 4298#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4299 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4300 ldev = netdev_lower_get_next(dev, &(iter)); \
4301 ldev; \
4302 ldev = netdev_lower_get_next(dev, &(iter)))
4303
7ce856aa
JP
4304struct net_device *netdev_all_lower_get_next(struct net_device *dev,
4305 struct list_head **iter);
4306struct net_device *netdev_all_lower_get_next_rcu(struct net_device *dev,
4307 struct list_head **iter);
4308
1a3f060c
DA
4309int netdev_walk_all_lower_dev(struct net_device *dev,
4310 int (*fn)(struct net_device *lower_dev,
4311 void *data),
4312 void *data);
4313int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4314 int (*fn)(struct net_device *lower_dev,
4315 void *data),
4316 void *data);
4317
f629d208 4318void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4319void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4320struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4321struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4322int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4323 struct netlink_ext_ack *extack);
f629d208 4324int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4325 struct net_device *upper_dev,
42ab19ee
DA
4326 void *upper_priv, void *upper_info,
4327 struct netlink_ext_ack *extack);
f629d208
JP
4328void netdev_upper_dev_unlink(struct net_device *dev,
4329 struct net_device *upper_dev);
32b6d34f
TY
4330int netdev_adjacent_change_prepare(struct net_device *old_dev,
4331 struct net_device *new_dev,
4332 struct net_device *dev,
4333 struct netlink_ext_ack *extack);
4334void netdev_adjacent_change_commit(struct net_device *old_dev,
4335 struct net_device *new_dev,
4336 struct net_device *dev);
4337void netdev_adjacent_change_abort(struct net_device *old_dev,
4338 struct net_device *new_dev,
4339 struct net_device *dev);
5bb025fa 4340void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4341void *netdev_lower_dev_get_private(struct net_device *dev,
4342 struct net_device *lower_dev);
04d48266
JP
4343void netdev_lower_state_changed(struct net_device *lower_dev,
4344 void *lower_state_info);
960fb622
ED
4345
4346/* RSS keys are 40 or 52 bytes long */
4347#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4348extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4349void netdev_rss_key_fill(void *buffer, size_t len);
4350
f629d208 4351int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4352int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4353int skb_csum_hwoffload_help(struct sk_buff *skb,
4354 const netdev_features_t features);
4355
f629d208
JP
4356struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
4357 netdev_features_t features, bool tx_path);
4358struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
4359 netdev_features_t features);
12b0004d 4360
61bd3857
MS
4361struct netdev_bonding_info {
4362 ifslave slave;
4363 ifbond master;
4364};
4365
4366struct netdev_notifier_bonding_info {
4367 struct netdev_notifier_info info; /* must be first */
4368 struct netdev_bonding_info bonding_info;
4369};
4370
4371void netdev_bonding_info_change(struct net_device *dev,
4372 struct netdev_bonding_info *bonding_info);
4373
12b0004d
CW
4374static inline
4375struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
4376{
4377 return __skb_gso_segment(skb, features, true);
4378}
53d6471c 4379__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4380
4381static inline bool can_checksum_protocol(netdev_features_t features,
4382 __be16 protocol)
4383{
c8cd0989
TH
4384 if (protocol == htons(ETH_P_FCOE))
4385 return !!(features & NETIF_F_FCOE_CRC);
4386
4387 /* Assume this is an IP checksum (not SCTP CRC) */
4388
4389 if (features & NETIF_F_HW_CSUM) {
4390 /* Can checksum everything */
4391 return true;
4392 }
4393
4394 switch (protocol) {
4395 case htons(ETH_P_IP):
4396 return !!(features & NETIF_F_IP_CSUM);
4397 case htons(ETH_P_IPV6):
4398 return !!(features & NETIF_F_IPV6_CSUM);
4399 default:
4400 return false;
4401 }
ec5f0615 4402}
12b0004d 4403
fb286bb2 4404#ifdef CONFIG_BUG
7fe50ac8 4405void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4406#else
7fe50ac8
CW
4407static inline void netdev_rx_csum_fault(struct net_device *dev,
4408 struct sk_buff *skb)
fb286bb2
HX
4409{
4410}
4411#endif
1da177e4 4412/* rx skb timestamps */
f629d208
JP
4413void net_enable_timestamp(void);
4414void net_disable_timestamp(void);
1da177e4 4415
20380731 4416#ifdef CONFIG_PROC_FS
f629d208 4417int __init dev_proc_init(void);
900ff8c6
CW
4418#else
4419#define dev_proc_init() 0
20380731
ACM
4420#endif
4421
4798248e 4422static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4423 struct sk_buff *skb, struct net_device *dev,
4424 bool more)
4798248e 4425{
6b16f9ee 4426 __this_cpu_write(softnet_data.xmit.more, more);
0b725a2c 4427 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4428}
4429
97cdcf37
FW
4430static inline bool netdev_xmit_more(void)
4431{
4432 return __this_cpu_read(softnet_data.xmit.more);
4433}
4434
10b3ad8c 4435static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4436 struct netdev_queue *txq, bool more)
4798248e
DM
4437{
4438 const struct net_device_ops *ops = dev->netdev_ops;
2183435c 4439 netdev_tx_t rc;
4798248e 4440
fa2dbdc2 4441 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4442 if (rc == NETDEV_TX_OK)
4443 txq_trans_update(txq);
4444
4445 return rc;
4798248e
DM
4446}
4447
b793dc5c 4448int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4449 const void *ns);
b793dc5c 4450void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4451 const void *ns);
58292cbe 4452
b793dc5c 4453static inline int netdev_class_create_file(const struct class_attribute *class_attr)
58292cbe
TH
4454{
4455 return netdev_class_create_file_ns(class_attr, NULL);
4456}
4457
b793dc5c 4458static inline void netdev_class_remove_file(const struct class_attribute *class_attr)
58292cbe
TH
4459{
4460 netdev_class_remove_file_ns(class_attr, NULL);
4461}
b8a9787e 4462
737aec57 4463extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4464
f629d208 4465const char *netdev_drivername(const struct net_device *dev);
6579e57b 4466
f629d208 4467void linkwatch_run_queue(void);
20380731 4468
da08143b
MK
4469static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4470 netdev_features_t f2)
4471{
c8cd0989
TH
4472 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4473 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4474 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4475 else
b6a0e72a 4476 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4477 }
da08143b 4478
c8cd0989 4479 return f1 & f2;
da08143b
MK
4480}
4481
c8f44aff
MM
4482static inline netdev_features_t netdev_get_wanted_features(
4483 struct net_device *dev)
5455c699
MM
4484{
4485 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4486}
c8f44aff
MM
4487netdev_features_t netdev_increment_features(netdev_features_t all,
4488 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4489
4490/* Allow TSO being used on stacked device :
4491 * Performing the GSO segmentation before last device
4492 * is a performance improvement.
4493 */
4494static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4495 netdev_features_t mask)
4496{
4497 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4498}
4499
6cb6a27c 4500int __netdev_update_features(struct net_device *dev);
5455c699 4501void netdev_update_features(struct net_device *dev);
afe12cc8 4502void netdev_change_features(struct net_device *dev);
7f353bf2 4503
fc4a7489
PM
4504void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4505 struct net_device *dev);
4506
e38f3025
TM
4507netdev_features_t passthru_features_check(struct sk_buff *skb,
4508 struct net_device *dev,
4509 netdev_features_t features);
c1e756bf 4510netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 4511
4d29515f 4512static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 4513{
7b748340 4514 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
4515
4516 /* check flags correspondence */
4517 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
4518 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
4519 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 4520 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
4521 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
4522 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4523 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
4524 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
4525 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
4526 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4527 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
4528 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 4529 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 4530 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 4531 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 4532 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 4533 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 4534 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
0345e186 4535
d6b4991a 4536 return (features & feature) == feature;
576a30eb
HX
4537}
4538
4d29515f 4539static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 4540{
278b2513 4541 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 4542 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
4543}
4544
8b86a61d 4545static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 4546 netdev_features_t features)
7967168c 4547{
fc741216 4548 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
4549 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
4550 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
4551}
4552
82cc1a7a
PWJ
4553static inline void netif_set_gso_max_size(struct net_device *dev,
4554 unsigned int size)
4555{
4556 dev->gso_max_size = size;
4557}
4558
7a7ffbab
WCC
4559static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
4560 int pulled_hlen, u16 mac_offset,
4561 int mac_len)
4562{
4563 skb->protocol = protocol;
4564 skb->encapsulation = 1;
4565 skb_push(skb, pulled_hlen);
4566 skb_reset_transport_header(skb);
4567 skb->mac_header = mac_offset;
4568 skb->network_header = skb->mac_header + mac_len;
4569 skb->mac_len = mac_len;
4570}
4571
3c175784
SD
4572static inline bool netif_is_macsec(const struct net_device *dev)
4573{
4574 return dev->priv_flags & IFF_MACSEC;
4575}
4576
b618aaa9 4577static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
4578{
4579 return dev->priv_flags & IFF_MACVLAN;
4580}
4581
b618aaa9 4582static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
4583{
4584 return dev->priv_flags & IFF_MACVLAN_PORT;
4585}
4586
b618aaa9 4587static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 4588{
4589 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
4590}
4591
b618aaa9 4592static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
4593{
4594 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
4595}
4596
3bdc0eba
BG
4597static inline bool netif_supports_nofcs(struct net_device *dev)
4598{
4599 return dev->priv_flags & IFF_SUPP_NOFCS;
4600}
4601
d5256083
DB
4602static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
4603{
4604 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
4605}
4606
007979ea 4607static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 4608{
007979ea 4609 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
4610}
4611
fee6d4c7
DA
4612static inline bool netif_is_l3_slave(const struct net_device *dev)
4613{
4614 return dev->priv_flags & IFF_L3MDEV_SLAVE;
4615}
4616
0894ae3f
JP
4617static inline bool netif_is_bridge_master(const struct net_device *dev)
4618{
4619 return dev->priv_flags & IFF_EBRIDGE;
4620}
4621
28f9ee22
VY
4622static inline bool netif_is_bridge_port(const struct net_device *dev)
4623{
4624 return dev->priv_flags & IFF_BRIDGE_PORT;
4625}
4626
35d4e172
JP
4627static inline bool netif_is_ovs_master(const struct net_device *dev)
4628{
4629 return dev->priv_flags & IFF_OPENVSWITCH;
4630}
4631
5be66141
JP
4632static inline bool netif_is_ovs_port(const struct net_device *dev)
4633{
4634 return dev->priv_flags & IFF_OVS_DATAPATH;
4635}
4636
b618aaa9 4637static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
4638{
4639 return dev->priv_flags & IFF_TEAM;
4640}
4641
b618aaa9 4642static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
4643{
4644 return dev->priv_flags & IFF_TEAM_PORT;
4645}
4646
b618aaa9 4647static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
4648{
4649 return netif_is_bond_master(dev) || netif_is_team_master(dev);
4650}
4651
b618aaa9 4652static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
4653{
4654 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
4655}
4656
d4ab4286
KJ
4657static inline bool netif_is_rxfh_configured(const struct net_device *dev)
4658{
4659 return dev->priv_flags & IFF_RXFH_CONFIGURED;
4660}
4661
30c8bd5a
SS
4662static inline bool netif_is_failover(const struct net_device *dev)
4663{
4664 return dev->priv_flags & IFF_FAILOVER;
4665}
4666
4667static inline bool netif_is_failover_slave(const struct net_device *dev)
4668{
4669 return dev->priv_flags & IFF_FAILOVER_SLAVE;
4670}
4671
02875878
ED
4672/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
4673static inline void netif_keep_dst(struct net_device *dev)
4674{
4675 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
4676}
4677
18d3df3e
PA
4678/* return true if dev can't cope with mtu frames that need vlan tag insertion */
4679static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
4680{
4681 /* TODO: reserve and use an additional IFF bit, if we get more users */
4682 return dev->priv_flags & IFF_MACSEC;
4683}
4684
505d4f73 4685extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 4686
571ba423
JP
4687/* Logging, debugging and troubleshooting/diagnostic helpers. */
4688
4689/* netdev_printk helpers, similar to dev_printk */
4690
4691static inline const char *netdev_name(const struct net_device *dev)
4692{
c6f854d5
VF
4693 if (!dev->name[0] || strchr(dev->name, '%'))
4694 return "(unnamed net_device)";
571ba423
JP
4695 return dev->name;
4696}
4697
8397ed36
DA
4698static inline bool netdev_unregistering(const struct net_device *dev)
4699{
4700 return dev->reg_state == NETREG_UNREGISTERING;
4701}
4702
ccc7f496
VF
4703static inline const char *netdev_reg_state(const struct net_device *dev)
4704{
4705 switch (dev->reg_state) {
4706 case NETREG_UNINITIALIZED: return " (uninitialized)";
4707 case NETREG_REGISTERED: return "";
4708 case NETREG_UNREGISTERING: return " (unregistering)";
4709 case NETREG_UNREGISTERED: return " (unregistered)";
4710 case NETREG_RELEASED: return " (released)";
4711 case NETREG_DUMMY: return " (dummy)";
4712 }
4713
4714 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
4715 return " (unknown)";
4716}
4717
ce3fdb69 4718__printf(3, 4) __cold
6ea754eb
JP
4719void netdev_printk(const char *level, const struct net_device *dev,
4720 const char *format, ...);
ce3fdb69 4721__printf(2, 3) __cold
6ea754eb 4722void netdev_emerg(const struct net_device *dev, const char *format, ...);
ce3fdb69 4723__printf(2, 3) __cold
6ea754eb 4724void netdev_alert(const struct net_device *dev, const char *format, ...);
ce3fdb69 4725__printf(2, 3) __cold
6ea754eb 4726void netdev_crit(const struct net_device *dev, const char *format, ...);
ce3fdb69 4727__printf(2, 3) __cold
6ea754eb 4728void netdev_err(const struct net_device *dev, const char *format, ...);
ce3fdb69 4729__printf(2, 3) __cold
6ea754eb 4730void netdev_warn(const struct net_device *dev, const char *format, ...);
ce3fdb69 4731__printf(2, 3) __cold
6ea754eb 4732void netdev_notice(const struct net_device *dev, const char *format, ...);
ce3fdb69 4733__printf(2, 3) __cold
6ea754eb 4734void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 4735
375ef2b1
GP
4736#define netdev_level_once(level, dev, fmt, ...) \
4737do { \
4738 static bool __print_once __read_mostly; \
4739 \
4740 if (!__print_once) { \
4741 __print_once = true; \
4742 netdev_printk(level, dev, fmt, ##__VA_ARGS__); \
4743 } \
4744} while (0)
4745
4746#define netdev_emerg_once(dev, fmt, ...) \
4747 netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
4748#define netdev_alert_once(dev, fmt, ...) \
4749 netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
4750#define netdev_crit_once(dev, fmt, ...) \
4751 netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
4752#define netdev_err_once(dev, fmt, ...) \
4753 netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
4754#define netdev_warn_once(dev, fmt, ...) \
4755 netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
4756#define netdev_notice_once(dev, fmt, ...) \
4757 netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
4758#define netdev_info_once(dev, fmt, ...) \
4759 netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
4760
8909c9ad
VK
4761#define MODULE_ALIAS_NETDEV(device) \
4762 MODULE_ALIAS("netdev-" device)
4763
b558c96f 4764#if defined(CONFIG_DYNAMIC_DEBUG)
571ba423
JP
4765#define netdev_dbg(__dev, format, args...) \
4766do { \
ffa10cb4 4767 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 4768} while (0)
b558c96f
JC
4769#elif defined(DEBUG)
4770#define netdev_dbg(__dev, format, args...) \
4771 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
4772#else
4773#define netdev_dbg(__dev, format, args...) \
4774({ \
4775 if (0) \
4776 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
4777})
4778#endif
4779
4780#if defined(VERBOSE_DEBUG)
4781#define netdev_vdbg netdev_dbg
4782#else
4783
4784#define netdev_vdbg(dev, format, args...) \
4785({ \
4786 if (0) \
4787 netdev_printk(KERN_DEBUG, dev, format, ##args); \
4788 0; \
4789})
4790#endif
4791
4792/*
4793 * netdev_WARN() acts like dev_printk(), but with the key difference
4794 * of using a WARN/WARN_ON to get the message out, including the
4795 * file/line information and a backtrace.
4796 */
4797#define netdev_WARN(dev, format, args...) \
e1cfe3d0 4798 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 4799 netdev_reg_state(dev), ##args)
571ba423 4800
72dd831e 4801#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 4802 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
4803 netdev_reg_state(dev), ##args)
4804
b3d95c5c
JP
4805/* netif printk helpers, similar to netdev_printk */
4806
4807#define netif_printk(priv, type, level, dev, fmt, args...) \
4808do { \
4809 if (netif_msg_##type(priv)) \
4810 netdev_printk(level, (dev), fmt, ##args); \
4811} while (0)
4812
f45f4321
JP
4813#define netif_level(level, priv, type, dev, fmt, args...) \
4814do { \
4815 if (netif_msg_##type(priv)) \
4816 netdev_##level(dev, fmt, ##args); \
4817} while (0)
4818
b3d95c5c 4819#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 4820 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 4821#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 4822 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 4823#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 4824 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 4825#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 4826 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 4827#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 4828 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 4829#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 4830 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 4831#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 4832 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 4833
0053ea9c 4834#if defined(CONFIG_DYNAMIC_DEBUG)
b3d95c5c
JP
4835#define netif_dbg(priv, type, netdev, format, args...) \
4836do { \
4837 if (netif_msg_##type(priv)) \
b5fb0a03 4838 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 4839} while (0)
0053ea9c
JP
4840#elif defined(DEBUG)
4841#define netif_dbg(priv, type, dev, format, args...) \
4842 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
4843#else
4844#define netif_dbg(priv, type, dev, format, args...) \
4845({ \
4846 if (0) \
4847 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
4848 0; \
4849})
4850#endif
4851
f617f276
EC
4852/* if @cond then downgrade to debug, else print at @level */
4853#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...) \
4854 do { \
4855 if (cond) \
4856 netif_dbg(priv, type, netdev, fmt, ##args); \
4857 else \
4858 netif_ ## level(priv, type, netdev, fmt, ##args); \
4859 } while (0)
4860
b3d95c5c 4861#if defined(VERBOSE_DEBUG)
bcfcc450 4862#define netif_vdbg netif_dbg
b3d95c5c
JP
4863#else
4864#define netif_vdbg(priv, type, dev, format, args...) \
4865({ \
4866 if (0) \
a4ed89cb 4867 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
4868 0; \
4869})
4870#endif
571ba423 4871
900ff8c6
CW
4872/*
4873 * The list of packet types we will receive (as opposed to discard)
4874 * and the routines to invoke.
4875 *
4876 * Why 16. Because with 16 the only overlap we get on a hash of the
4877 * low nibble of the protocol value is RARP/SNAP/X.25.
4878 *
900ff8c6 4879 * 0800 IP
900ff8c6
CW
4880 * 0001 802.3
4881 * 0002 AX.25
4882 * 0004 802.2
4883 * 8035 RARP
4884 * 0005 SNAP
4885 * 0805 X.25
4886 * 0806 ARP
4887 * 8137 IPX
4888 * 0009 Localtalk
4889 * 86DD IPv6
4890 */
4891#define PTYPE_HASH_SIZE (16)
4892#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
4893
4de83b88
MB
4894extern struct net_device *blackhole_netdev;
4895
385a154c 4896#endif /* _LINUX_NETDEVICE_H */