]> git.proxmox.com Git - mirror_ubuntu-hirsute-kernel.git/blame - include/linux/netdevice.h
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
[mirror_ubuntu-hirsute-kernel.git] / include / linux / netdevice.h
CommitLineData
1da177e4
LT
1/*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * Definitions for the Interfaces handler.
7 *
8 * Version: @(#)dev.h 1.0.10 08/12/93
9 *
02c30a84 10 * Authors: Ross Biro
1da177e4
LT
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Corey Minyard <wf-rch!minyard@relay.EU.net>
13 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 14 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
15 * Bjorn Ekwall. <bj0rn@blox.se>
16 * Pekka Riikonen <priikone@poseidon.pspt.fi>
17 *
18 * This program is free software; you can redistribute it and/or
19 * modify it under the terms of the GNU General Public License
20 * as published by the Free Software Foundation; either version
21 * 2 of the License, or (at your option) any later version.
22 *
23 * Moved to /usr/include/linux for NET3
24 */
25#ifndef _LINUX_NETDEVICE_H
26#define _LINUX_NETDEVICE_H
27
d7fe0f24 28#include <linux/timer.h>
187f1882 29#include <linux/bug.h>
bea3348e 30#include <linux/delay.h>
60063497 31#include <linux/atomic.h>
53511453 32#include <linux/prefetch.h>
1da177e4
LT
33#include <asm/cache.h>
34#include <asm/byteorder.h>
35
1da177e4 36#include <linux/percpu.h>
4d5b78c0 37#include <linux/rculist.h>
bea3348e 38#include <linux/workqueue.h>
114cf580 39#include <linux/dynamic_queue_limits.h>
1da177e4 40
b1b67dd4 41#include <linux/ethtool.h>
a050c33f 42#include <net/net_namespace.h>
7a6b6f51 43#ifdef CONFIG_DCB
2f90b865
AD
44#include <net/dcbnl.h>
45#endif
5bc1421e 46#include <net/netprio_cgroup.h>
e817f856 47#include <net/xdp.h>
a050c33f 48
a59e2ecb 49#include <linux/netdev_features.h>
77162022 50#include <linux/neighbour.h>
607ca46e 51#include <uapi/linux/netdevice.h>
61bd3857 52#include <uapi/linux/if_bonding.h>
e4c6734e 53#include <uapi/linux/pkt_cls.h>
59cc1f61 54#include <linux/hashtable.h>
a59e2ecb 55
115c1d6e 56struct netpoll_info;
313162d0 57struct device;
c1f19b51 58struct phy_device;
2f657a60 59struct dsa_port;
c6e970a0 60
e679c9c1 61struct sfp_bus;
704232c2
JB
62/* 802.11 specific */
63struct wireless_dev;
98a18b6f
AA
64/* 802.15.4 specific */
65struct wpan_dev;
03c57747 66struct mpls_dev;
7c46a640
AD
67/* UDP Tunnel offloads */
68struct udp_tunnel_info;
a7862b45 69struct bpf_prog;
814abfab 70struct xdp_buff;
1da177e4 71
f629d208
JP
72void netdev_set_default_ethtool_ops(struct net_device *dev,
73 const struct ethtool_ops *ops);
d07d7507 74
9a1654ba
JP
75/* Backlog congestion levels */
76#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
77#define NET_RX_DROP 1 /* packet dropped */
78
572a9d7b
PM
79/*
80 * Transmit return codes: transmit return codes originate from three different
81 * namespaces:
82 *
83 * - qdisc return codes
84 * - driver transmit return codes
85 * - errno values
86 *
87 * Drivers are allowed to return any one of those in their hard_start_xmit()
88 * function. Real network devices commonly used with qdiscs should only return
89 * the driver transmit return codes though - when qdiscs are used, the actual
90 * transmission happens asynchronously, so the value is not propagated to
5e82b4b2
BH
91 * higher layers. Virtual network devices transmit synchronously; in this case
92 * the driver transmit return codes are consumed by dev_queue_xmit(), and all
572a9d7b
PM
93 * others are propagated to higher layers.
94 */
95
96/* qdisc ->enqueue() return codes. */
97#define NET_XMIT_SUCCESS 0x00
9a1654ba
JP
98#define NET_XMIT_DROP 0x01 /* skb dropped */
99#define NET_XMIT_CN 0x02 /* congestion notification */
9a1654ba 100#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
1da177e4 101
b9df3cb8
GR
102/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
103 * indicates that the device will soon be dropping packets, or already drops
104 * some packets of the same priority; prompting us to send less aggressively. */
572a9d7b 105#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
1da177e4
LT
106#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
107
dc1f8bf6 108/* Driver transmit return codes */
9a1654ba 109#define NETDEV_TX_MASK 0xf0
572a9d7b 110
dc1f8bf6 111enum netdev_tx {
572a9d7b 112 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
9a1654ba
JP
113 NETDEV_TX_OK = 0x00, /* driver took care of packet */
114 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
dc1f8bf6
SH
115};
116typedef enum netdev_tx netdev_tx_t;
117
9a1654ba
JP
118/*
119 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
120 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
121 */
122static inline bool dev_xmit_complete(int rc)
123{
124 /*
125 * Positive cases with an skb consumed by a driver:
126 * - successful transmission (rc == NETDEV_TX_OK)
127 * - error while transmitting (rc < 0)
128 * - error while queueing to a different device (rc & NET_XMIT_MASK)
129 */
130 if (likely(rc < NET_XMIT_MASK))
131 return true;
132
133 return false;
134}
135
1da177e4 136/*
5e82b4b2 137 * Compute the worst-case header length according to the protocols
1da177e4
LT
138 * used.
139 */
fe2918b0 140
c0eb4540
KS
141#if defined(CONFIG_HYPERV_NET)
142# define LL_MAX_HEADER 128
143#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
8388e3da
DM
144# if defined(CONFIG_MAC80211_MESH)
145# define LL_MAX_HEADER 128
146# else
147# define LL_MAX_HEADER 96
148# endif
1da177e4 149#else
8388e3da 150# define LL_MAX_HEADER 32
1da177e4
LT
151#endif
152
d11ead75
BH
153#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
154 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
1da177e4
LT
155#define MAX_HEADER LL_MAX_HEADER
156#else
157#define MAX_HEADER (LL_MAX_HEADER + 48)
158#endif
159
160/*
be1f3c2c
BH
161 * Old network device statistics. Fields are native words
162 * (unsigned long) so they can be read and written atomically.
1da177e4 163 */
fe2918b0 164
d94d9fee 165struct net_device_stats {
3cfde79c
BH
166 unsigned long rx_packets;
167 unsigned long tx_packets;
168 unsigned long rx_bytes;
169 unsigned long tx_bytes;
170 unsigned long rx_errors;
171 unsigned long tx_errors;
172 unsigned long rx_dropped;
173 unsigned long tx_dropped;
174 unsigned long multicast;
1da177e4 175 unsigned long collisions;
1da177e4 176 unsigned long rx_length_errors;
3cfde79c
BH
177 unsigned long rx_over_errors;
178 unsigned long rx_crc_errors;
179 unsigned long rx_frame_errors;
180 unsigned long rx_fifo_errors;
181 unsigned long rx_missed_errors;
1da177e4
LT
182 unsigned long tx_aborted_errors;
183 unsigned long tx_carrier_errors;
184 unsigned long tx_fifo_errors;
185 unsigned long tx_heartbeat_errors;
186 unsigned long tx_window_errors;
1da177e4
LT
187 unsigned long rx_compressed;
188 unsigned long tx_compressed;
189};
190
1da177e4
LT
191
192#include <linux/cache.h>
193#include <linux/skbuff.h>
194
adc9300e 195#ifdef CONFIG_RPS
c5905afb
IM
196#include <linux/static_key.h>
197extern struct static_key rps_needed;
13bfff25 198extern struct static_key rfs_needed;
adc9300e
ED
199#endif
200
1da177e4
LT
201struct neighbour;
202struct neigh_parms;
203struct sk_buff;
204
f001fde5
JP
205struct netdev_hw_addr {
206 struct list_head list;
207 unsigned char addr[MAX_ADDR_LEN];
208 unsigned char type;
ccffad25
JP
209#define NETDEV_HW_ADDR_T_LAN 1
210#define NETDEV_HW_ADDR_T_SAN 2
211#define NETDEV_HW_ADDR_T_SLAVE 3
212#define NETDEV_HW_ADDR_T_UNICAST 4
22bedad3 213#define NETDEV_HW_ADDR_T_MULTICAST 5
22bedad3 214 bool global_use;
4cd729b0 215 int sync_cnt;
8f8f103d 216 int refcount;
4543fbef 217 int synced;
f001fde5
JP
218 struct rcu_head rcu_head;
219};
220
31278e71
JP
221struct netdev_hw_addr_list {
222 struct list_head list;
223 int count;
224};
225
22bedad3
JP
226#define netdev_hw_addr_list_count(l) ((l)->count)
227#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
228#define netdev_hw_addr_list_for_each(ha, l) \
229 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 230
22bedad3
JP
231#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
232#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
233#define netdev_for_each_uc_addr(ha, dev) \
234 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
6683ece3 235
22bedad3
JP
236#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
237#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 238#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 239 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
6683ece3 240
d94d9fee 241struct hh_cache {
5b3dc2f3 242 unsigned int hh_len;
3644f0ce 243 seqlock_t hh_lock;
1da177e4
LT
244
245 /* cached hardware header; allow for machine alignment needs. */
246#define HH_DATA_MOD 16
247#define HH_DATA_OFF(__len) \
5ba0eac6 248 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
1da177e4
LT
249#define HH_DATA_ALIGN(__len) \
250 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
251 unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
252};
253
5e82b4b2 254/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
1da177e4
LT
255 * Alternative is:
256 * dev->hard_header_len ? (dev->hard_header_len +
257 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
258 *
259 * We could use other alignment values, but we must maintain the
260 * relationship HH alignment <= LL alignment.
261 */
262#define LL_RESERVED_SPACE(dev) \
f5184d26 263 ((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 264#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
f5184d26 265 ((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 266
3b04ddde
SH
267struct header_ops {
268 int (*create) (struct sk_buff *skb, struct net_device *dev,
269 unsigned short type, const void *daddr,
95c96174 270 const void *saddr, unsigned int len);
3b04ddde 271 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
e69dd336 272 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
3b04ddde
SH
273 void (*cache_update)(struct hh_cache *hh,
274 const struct net_device *dev,
275 const unsigned char *haddr);
2793a23a 276 bool (*validate)(const char *ll_header, unsigned int len);
3b04ddde
SH
277};
278
1da177e4 279/* These flag bits are private to the generic network queueing
5e82b4b2 280 * layer; they may not be explicitly referenced by any other
1da177e4
LT
281 * code.
282 */
283
d94d9fee 284enum netdev_state_t {
1da177e4
LT
285 __LINK_STATE_START,
286 __LINK_STATE_PRESENT,
1da177e4 287 __LINK_STATE_NOCARRIER,
b00055aa
SR
288 __LINK_STATE_LINKWATCH_PENDING,
289 __LINK_STATE_DORMANT,
1da177e4
LT
290};
291
292
293/*
5e82b4b2 294 * This structure holds boot-time configured netdevice settings. They
fe2918b0 295 * are then used in the device probing.
1da177e4
LT
296 */
297struct netdev_boot_setup {
298 char name[IFNAMSIZ];
299 struct ifmap map;
300};
301#define NETDEV_BOOT_SETUP_MAX 8
302
f629d208 303int __init netdev_boot_setup(char *str);
1da177e4 304
6312fe77
LR
305struct gro_list {
306 struct list_head list;
307 int count;
308};
309
bea3348e 310/*
d9f37d01
LR
311 * size of gro hash buckets, must less than bit number of
312 * napi_struct::gro_bitmask
bea3348e 313 */
07d78363 314#define GRO_HASH_BUCKETS 8
d9f37d01
LR
315
316/*
317 * Structure for NAPI scheduling similar to tasklet but with weighting
318 */
bea3348e
SH
319struct napi_struct {
320 /* The poll_list must only be managed by the entity which
321 * changes the state of the NAPI_STATE_SCHED bit. This means
322 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 323 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
324 * can remove from the list right before clearing the bit.
325 */
326 struct list_head poll_list;
327
328 unsigned long state;
329 int weight;
d9f37d01 330 unsigned long gro_bitmask;
bea3348e
SH
331 int (*poll)(struct napi_struct *, int);
332#ifdef CONFIG_NETPOLL
bea3348e 333 int poll_owner;
bea3348e 334#endif
5d38a079 335 struct net_device *dev;
6312fe77 336 struct gro_list gro_hash[GRO_HASH_BUCKETS];
5d38a079 337 struct sk_buff *skb;
3b47d303 338 struct hrtimer timer;
404f7c9e 339 struct list_head dev_list;
af12fa6e
ET
340 struct hlist_node napi_hash_node;
341 unsigned int napi_id;
bea3348e
SH
342};
343
d94d9fee 344enum {
bea3348e 345 NAPI_STATE_SCHED, /* Poll is scheduled */
39e6c820 346 NAPI_STATE_MISSED, /* reschedule a napi */
a0a46196 347 NAPI_STATE_DISABLE, /* Disable pending */
7b363e44 348 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
d64b5e85
ED
349 NAPI_STATE_HASHED, /* In NAPI hash (busy polling possible) */
350 NAPI_STATE_NO_BUSY_POLL,/* Do not add in napi_hash, no busy polling */
217f6974
ED
351 NAPI_STATE_IN_BUSY_POLL,/* sk_busy_loop() owns this NAPI */
352};
353
354enum {
39e6c820
ED
355 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
356 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
357 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
358 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
359 NAPIF_STATE_HASHED = BIT(NAPI_STATE_HASHED),
360 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
361 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
bea3348e
SH
362};
363
5b252f0c 364enum gro_result {
d1c76af9
HX
365 GRO_MERGED,
366 GRO_MERGED_FREE,
367 GRO_HELD,
368 GRO_NORMAL,
369 GRO_DROP,
25393d3f 370 GRO_CONSUMED,
d1c76af9 371};
5b252f0c 372typedef enum gro_result gro_result_t;
d1c76af9 373
8a4eb573
JP
374/*
375 * enum rx_handler_result - Possible return values for rx_handlers.
376 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
377 * further.
378 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
379 * case skb->dev was changed by rx_handler.
380 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 381 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
382 *
383 * rx_handlers are functions called from inside __netif_receive_skb(), to do
384 * special processing of the skb, prior to delivery to protocol handlers.
385 *
386 * Currently, a net_device can only have a single rx_handler registered. Trying
387 * to register a second rx_handler will return -EBUSY.
388 *
389 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
390 * To unregister a rx_handler on a net_device, use
391 * netdev_rx_handler_unregister().
392 *
393 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
394 * do with the skb.
395 *
5e82b4b2 396 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 397 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 398 * the skb to be delivered in some other way.
8a4eb573
JP
399 *
400 * If the rx_handler changed skb->dev, to divert the skb to another
401 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
402 * new device will be called if it exists.
403 *
5e82b4b2 404 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 405 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 406 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 407 *
5e82b4b2 408 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
409 * delivered, it should return RX_HANDLER_PASS.
410 *
411 * A device without a registered rx_handler will behave as if rx_handler
412 * returned RX_HANDLER_PASS.
413 */
414
415enum rx_handler_result {
416 RX_HANDLER_CONSUMED,
417 RX_HANDLER_ANOTHER,
418 RX_HANDLER_EXACT,
419 RX_HANDLER_PASS,
420};
421typedef enum rx_handler_result rx_handler_result_t;
422typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 423
f629d208 424void __napi_schedule(struct napi_struct *n);
bc9ad166 425void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 426
4d29515f 427static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
428{
429 return test_bit(NAPI_STATE_DISABLE, &n->state);
430}
431
39e6c820 432bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
433
434/**
435 * napi_schedule - schedule NAPI poll
5e82b4b2 436 * @n: NAPI context
bea3348e
SH
437 *
438 * Schedule NAPI poll routine to be called if it is not already
439 * running.
440 */
441static inline void napi_schedule(struct napi_struct *n)
442{
443 if (napi_schedule_prep(n))
444 __napi_schedule(n);
445}
446
bc9ad166
ED
447/**
448 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 449 * @n: NAPI context
bc9ad166
ED
450 *
451 * Variant of napi_schedule(), assuming hard irqs are masked.
452 */
453static inline void napi_schedule_irqoff(struct napi_struct *n)
454{
455 if (napi_schedule_prep(n))
456 __napi_schedule_irqoff(n);
457}
458
bfe13f54 459/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 460static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
461{
462 if (napi_schedule_prep(napi)) {
463 __napi_schedule(napi);
4d29515f 464 return true;
bfe13f54 465 }
4d29515f 466 return false;
bfe13f54
RD
467}
468
364b6055 469bool napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
470/**
471 * napi_complete - NAPI processing complete
5e82b4b2 472 * @n: NAPI context
bea3348e
SH
473 *
474 * Mark NAPI processing as complete.
3b47d303 475 * Consider using napi_complete_done() instead.
364b6055 476 * Return false if device should avoid rearming interrupts.
bea3348e 477 */
364b6055 478static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
479{
480 return napi_complete_done(n, 0);
481}
bea3348e 482
af12fa6e
ET
483/**
484 * napi_hash_del - remove a NAPI from global table
5e82b4b2 485 * @napi: NAPI context
af12fa6e 486 *
5e82b4b2 487 * Warning: caller must observe RCU grace period
34cbe27e
ED
488 * before freeing memory containing @napi, if
489 * this function returns true.
93d05d4a 490 * Note: core networking stack automatically calls it
5e82b4b2 491 * from netif_napi_del().
93d05d4a 492 * Drivers might want to call this helper to combine all
5e82b4b2 493 * the needed RCU grace periods into a single one.
af12fa6e 494 */
34cbe27e 495bool napi_hash_del(struct napi_struct *napi);
af12fa6e 496
bea3348e
SH
497/**
498 * napi_disable - prevent NAPI from scheduling
5e82b4b2 499 * @n: NAPI context
bea3348e
SH
500 *
501 * Stop NAPI from being scheduled on this context.
502 * Waits till any outstanding processing completes.
503 */
3b47d303 504void napi_disable(struct napi_struct *n);
bea3348e
SH
505
506/**
507 * napi_enable - enable NAPI scheduling
5e82b4b2 508 * @n: NAPI context
bea3348e
SH
509 *
510 * Resume NAPI from being scheduled on this context.
511 * Must be paired with napi_disable.
512 */
513static inline void napi_enable(struct napi_struct *n)
514{
515 BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
4e857c58 516 smp_mb__before_atomic();
bea3348e 517 clear_bit(NAPI_STATE_SCHED, &n->state);
2d8bff12 518 clear_bit(NAPI_STATE_NPSVC, &n->state);
bea3348e
SH
519}
520
c264c3de
SH
521/**
522 * napi_synchronize - wait until NAPI is not running
5e82b4b2 523 * @n: NAPI context
c264c3de
SH
524 *
525 * Wait until NAPI is done being scheduled on this context.
526 * Waits till any outstanding processing completes but
527 * does not disable future activations.
528 */
529static inline void napi_synchronize(const struct napi_struct *n)
530{
facc432f
AB
531 if (IS_ENABLED(CONFIG_SMP))
532 while (test_bit(NAPI_STATE_SCHED, &n->state))
533 msleep(1);
534 else
535 barrier();
c264c3de 536}
c264c3de 537
6c5c9581
MK
538/**
539 * napi_if_scheduled_mark_missed - if napi is running, set the
540 * NAPIF_STATE_MISSED
541 * @n: NAPI context
542 *
543 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
544 * NAPI is scheduled.
545 **/
546static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
547{
548 unsigned long val, new;
549
550 do {
551 val = READ_ONCE(n->state);
552 if (val & NAPIF_STATE_DISABLE)
553 return true;
554
555 if (!(val & NAPIF_STATE_SCHED))
556 return false;
557
558 new = val | NAPIF_STATE_MISSED;
559 } while (cmpxchg(&n->state, val, new) != val);
560
561 return true;
562}
563
d94d9fee 564enum netdev_queue_state_t {
73466498
TH
565 __QUEUE_STATE_DRV_XOFF,
566 __QUEUE_STATE_STACK_XOFF,
c3f26a26 567 __QUEUE_STATE_FROZEN,
79d16385 568};
8e2f1a63
DB
569
570#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
571#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
572#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
573
574#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
575#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
576 QUEUE_STATE_FROZEN)
577#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
578 QUEUE_STATE_FROZEN)
579
73466498
TH
580/*
581 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
582 * netif_tx_* functions below are used to manipulate this flag. The
583 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
584 * queue independently. The netif_xmit_*stopped functions below are called
585 * to check if the queue has been stopped by the driver or stack (either
586 * of the XOFF bits are set in the state). Drivers should not need to call
587 * netif_xmit*stopped functions, they should only be using netif_tx_*.
588 */
79d16385 589
bb949fbd 590struct netdev_queue {
6a321cb3 591/*
5e82b4b2 592 * read-mostly part
6a321cb3 593 */
bb949fbd 594 struct net_device *dev;
46e5da40 595 struct Qdisc __rcu *qdisc;
b0e1e646 596 struct Qdisc *qdisc_sleeping;
ccf5ff69 597#ifdef CONFIG_SYSFS
1d24eb48
TH
598 struct kobject kobj;
599#endif
f2cd2d3e
ED
600#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
601 int numa_node;
602#endif
c0ef079c
FW
603 unsigned long tx_maxrate;
604 /*
605 * Number of TX timeouts for this queue
606 * (/sys/class/net/DEV/Q/trans_timeout)
607 */
608 unsigned long trans_timeout;
ffcfe25b
AD
609
610 /* Subordinate device that the queue has been assigned to */
611 struct net_device *sb_dev;
661b8d1b
MK
612#ifdef CONFIG_XDP_SOCKETS
613 struct xdp_umem *umem;
614#endif
6a321cb3 615/*
5e82b4b2 616 * write-mostly part
6a321cb3
ED
617 */
618 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
619 int xmit_lock_owner;
9d21493b 620 /*
9b36627a 621 * Time (in jiffies) of last Tx
9d21493b
ED
622 */
623 unsigned long trans_start;
ccf5ff69 624
114cf580
TH
625 unsigned long state;
626
627#ifdef CONFIG_BQL
628 struct dql dql;
629#endif
e8a0464c 630} ____cacheline_aligned_in_smp;
bb949fbd 631
79134e6c
ED
632extern int sysctl_fb_tunnels_only_for_init_net;
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,
16e5cc64
JF
850};
851
f4e63525
JK
852/* These structures hold the attributes of bpf state that are being passed
853 * to the netdevice through the bpf op.
a7862b45 854 */
f4e63525 855enum bpf_netdev_command {
a7862b45
BB
856 /* Set or clear a bpf program used in the earliest stages of packet
857 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
858 * is responsible for calling bpf_prog_put on any old progs that are
859 * stored. In case of error, the callee need not release the new prog
860 * reference, but on success it takes ownership and must bpf_prog_put
861 * when it is no longer used.
862 */
863 XDP_SETUP_PROG,
ee5d032f 864 XDP_SETUP_PROG_HW,
a7862b45 865 XDP_QUERY_PROG,
a25717d2 866 XDP_QUERY_PROG_HW,
ab3f0063
JK
867 /* BPF program for offload callbacks, invoked at program load time. */
868 BPF_OFFLOAD_VERIFIER_PREP,
869 BPF_OFFLOAD_TRANSLATE,
870 BPF_OFFLOAD_DESTROY,
a3884572
JK
871 BPF_OFFLOAD_MAP_ALLOC,
872 BPF_OFFLOAD_MAP_FREE,
74515c57
BT
873 XDP_QUERY_XSK_UMEM,
874 XDP_SETUP_XSK_UMEM,
a7862b45
BB
875};
876
cae1927c 877struct bpf_prog_offload_ops;
ddf9f970 878struct netlink_ext_ack;
74515c57 879struct xdp_umem;
ddf9f970 880
f4e63525
JK
881struct netdev_bpf {
882 enum bpf_netdev_command command;
a7862b45
BB
883 union {
884 /* XDP_SETUP_PROG */
ddf9f970 885 struct {
32d60277 886 u32 flags;
ddf9f970
JK
887 struct bpf_prog *prog;
888 struct netlink_ext_ack *extack;
889 };
a25717d2 890 /* XDP_QUERY_PROG, XDP_QUERY_PROG_HW */
58038695 891 struct {
58038695 892 u32 prog_id;
92f0292b
JK
893 /* flags with which program was installed */
894 u32 prog_flags;
58038695 895 };
ab3f0063
JK
896 /* BPF_OFFLOAD_VERIFIER_PREP */
897 struct {
898 struct bpf_prog *prog;
cae1927c 899 const struct bpf_prog_offload_ops *ops; /* callee set */
ab3f0063
JK
900 } verifier;
901 /* BPF_OFFLOAD_TRANSLATE, BPF_OFFLOAD_DESTROY */
902 struct {
903 struct bpf_prog *prog;
904 } offload;
a3884572
JK
905 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
906 struct {
907 struct bpf_offloaded_map *offmap;
908 };
84c6b868 909 /* XDP_QUERY_XSK_UMEM, XDP_SETUP_XSK_UMEM */
74515c57 910 struct {
84c6b868
JK
911 struct xdp_umem *umem; /* out for query*/
912 u16 queue_id; /* in for query */
74515c57 913 } xsk;
a7862b45
BB
914 };
915};
16e5cc64 916
d77e38e6
SK
917#ifdef CONFIG_XFRM_OFFLOAD
918struct xfrmdev_ops {
919 int (*xdo_dev_state_add) (struct xfrm_state *x);
920 void (*xdo_dev_state_delete) (struct xfrm_state *x);
921 void (*xdo_dev_state_free) (struct xfrm_state *x);
922 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
923 struct xfrm_state *x);
50bd870a 924 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
d77e38e6
SK
925};
926#endif
927
a5c37c63
IL
928#if IS_ENABLED(CONFIG_TLS_DEVICE)
929enum tls_offload_ctx_dir {
930 TLS_OFFLOAD_CTX_DIR_RX,
931 TLS_OFFLOAD_CTX_DIR_TX,
932};
933
934struct tls_crypto_info;
935struct tls_context;
936
937struct tlsdev_ops {
938 int (*tls_dev_add)(struct net_device *netdev, struct sock *sk,
939 enum tls_offload_ctx_dir direction,
940 struct tls_crypto_info *crypto_info,
941 u32 start_offload_tcp_sn);
942 void (*tls_dev_del)(struct net_device *netdev,
943 struct tls_context *ctx,
944 enum tls_offload_ctx_dir direction);
16e4edc2
BP
945 void (*tls_dev_resync_rx)(struct net_device *netdev,
946 struct sock *sk, u32 seq, u64 rcd_sn);
a5c37c63
IL
947};
948#endif
949
6c557001
FW
950struct dev_ifalias {
951 struct rcu_head rcuhead;
952 char ifalias[];
953};
954
d314774c
SH
955/*
956 * This structure defines the management hooks for network devices.
00829823
SH
957 * The following hooks can be defined; unless noted otherwise, they are
958 * optional and can be filled with a null pointer.
d314774c
SH
959 *
960 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
961 * This function is called once when a network device is registered.
962 * The network device can use this for any late stage initialization
963 * or semantic validation. It can fail with an error code which will
964 * be propagated back to register_netdev.
d314774c
SH
965 *
966 * void (*ndo_uninit)(struct net_device *dev);
967 * This function is called when device is unregistered or when registration
968 * fails. It is not called if init fails.
969 *
970 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 971 * This function is called when a network device transitions to the up
d314774c
SH
972 * state.
973 *
974 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 975 * This function is called when a network device transitions to the down
d314774c
SH
976 * state.
977 *
dc1f8bf6
SH
978 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
979 * struct net_device *dev);
00829823 980 * Called when a packet needs to be transmitted.
e79d8429
RR
981 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
982 * the queue before that can happen; it's for obsolete devices and weird
983 * corner cases, but the stack really does a non-trivial amount
984 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 985 * Required; cannot be NULL.
00829823 986 *
1a2a1444
DM
987 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
988 * struct net_device *dev
989 * netdev_features_t features);
990 * Called by core transmit path to determine if device is capable of
991 * performing offload operations on a given packet. This is to give
992 * the device an opportunity to implement any restrictions that cannot
993 * be otherwise expressed by feature flags. The check is called with
994 * the set of features that the stack has calculated and it returns
995 * those the driver believes to be appropriate.
cdba756f 996 *
f663dd9a 997 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
4f49dec9
AD
998 * struct net_device *sb_dev,
999 * select_queue_fallback_t fallback);
5e82b4b2 1000 * Called to decide which queue to use when device supports multiple
00829823
SH
1001 * transmit queues.
1002 *
d314774c
SH
1003 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
1004 * This function is called to allow device receiver to make
5e82b4b2 1005 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
1006 *
1007 * void (*ndo_set_rx_mode)(struct net_device *dev);
1008 * This function is called device changes address list filtering.
01789349 1009 * If driver handles unicast address filtering, it should set
5e82b4b2 1010 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
1011 *
1012 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
1013 * This function is called when the Media Access Control address
37b607c5 1014 * needs to be changed. If this interface is not defined, the
5e82b4b2 1015 * MAC address can not be changed.
d314774c
SH
1016 *
1017 * int (*ndo_validate_addr)(struct net_device *dev);
1018 * Test if Media Access Control address is valid for the device.
1019 *
1020 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
5e82b4b2
BH
1021 * Called when a user requests an ioctl which can't be handled by
1022 * the generic interface code. If not defined ioctls return
d314774c
SH
1023 * not supported error code.
1024 *
1025 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1026 * Used to set network devices bus interface parameters. This interface
5e82b4b2 1027 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1028 * interface (PCI) for low level management.
1029 *
1030 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1031 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1032 * of a device.
d314774c 1033 *
00829823 1034 * void (*ndo_tx_timeout)(struct net_device *dev);
5e82b4b2 1035 * Callback used when the transmitter has not made any progress
d314774c
SH
1036 * for dev->watchdog ticks.
1037 *
bc1f4470 1038 * void (*ndo_get_stats64)(struct net_device *dev,
1039 * struct rtnl_link_stats64 *storage);
d308e38f 1040 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1041 * Called when a user wants to get the network device usage
be1f3c2c 1042 * statistics. Drivers must do one of the following:
3cfde79c
BH
1043 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1044 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1045 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1046 * (which should normally be dev->stats) and return a pointer to
1047 * it. The structure may be changed asynchronously only if each
1048 * field is written atomically.
1049 * 3. Update dev->stats asynchronously and atomically, and define
1050 * neither operation.
d314774c 1051 *
3df5b3c6 1052 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1053 * Return true if this device supports offload stats of this attr_id.
1054 *
1055 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1056 * void *attr_data)
1057 * Get statistics for offload operations by attr_id. Write it into the
1058 * attr_data pointer.
1059 *
5d632cb7 1060 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1061 * If device supports VLAN filtering this function is called when a
80d5c368 1062 * VLAN id is registered.
d314774c 1063 *
5d632cb7 1064 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1065 * If device supports VLAN filtering this function is called when a
80d5c368 1066 * VLAN id is unregistered.
d314774c
SH
1067 *
1068 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1069 *
1070 * SR-IOV management functions.
1071 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1072 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1073 * u8 qos, __be16 proto);
ed616689
SC
1074 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1075 * int max_tx_rate);
5f8444a3 1076 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1077 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1078 * int (*ndo_get_vf_config)(struct net_device *dev,
1079 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1080 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1081 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1082 * struct nlattr *port[]);
01a3d796
VZ
1083 *
1084 * Enable or disable the VF ability to query its RSS Redirection Table and
1085 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1086 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1087 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1088 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1089 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1090 * void *type_data);
6a4bc2b4
FF
1091 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1092 * This is always called from the stack with the rtnl lock held and netif
1093 * tx queues stopped. This allows the netdevice to perform queue
1094 * management safely.
c445477d 1095 *
e9bce845
YZ
1096 * Fiber Channel over Ethernet (FCoE) offload functions.
1097 * int (*ndo_fcoe_enable)(struct net_device *dev);
1098 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1099 * so the underlying device can perform whatever needed configuration or
1100 * initialization to support acceleration of FCoE traffic.
1101 *
1102 * int (*ndo_fcoe_disable)(struct net_device *dev);
1103 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1104 * so the underlying device can perform whatever needed clean-ups to
1105 * stop supporting acceleration of FCoE traffic.
1106 *
1107 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1108 * struct scatterlist *sgl, unsigned int sgc);
1109 * Called when the FCoE Initiator wants to initialize an I/O that
1110 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1111 * perform necessary setup and returns 1 to indicate the device is set up
1112 * successfully to perform DDP on this I/O, otherwise this returns 0.
1113 *
1114 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1115 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1116 * indicated by the FC exchange id 'xid', so the underlying device can
1117 * clean up and reuse resources for later DDP requests.
1118 *
1119 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1120 * struct scatterlist *sgl, unsigned int sgc);
1121 * Called when the FCoE Target wants to initialize an I/O that
1122 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1123 * perform necessary setup and returns 1 to indicate the device is set up
1124 * successfully to perform DDP on this I/O, otherwise this returns 0.
1125 *
68bad94e
NP
1126 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1127 * struct netdev_fcoe_hbainfo *hbainfo);
1128 * Called when the FCoE Protocol stack wants information on the underlying
1129 * device. This information is utilized by the FCoE protocol stack to
1130 * register attributes with Fiber Channel management service as per the
1131 * FC-GS Fabric Device Management Information(FDMI) specification.
1132 *
e9bce845
YZ
1133 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1134 * Called when the underlying device wants to override default World Wide
1135 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1136 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1137 * protocol stack to use.
1138 *
c445477d
BH
1139 * RFS acceleration.
1140 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1141 * u16 rxq_index, u32 flow_id);
1142 * Set hardware filter for RFS. rxq_index is the target queue index;
1143 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1144 * Return the filter ID on success, or a negative error code.
fbaec0ea 1145 *
8b98a70c 1146 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1147 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1148 * Called to make another netdev an underling.
1149 *
1150 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1151 * Called to release previously enslaved netdev.
5455c699
MM
1152 *
1153 * Feature/offload setting functions.
1a2a1444
DM
1154 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1155 * netdev_features_t features);
1156 * Adjusts the requested feature flags according to device-specific
1157 * constraints, and returns the resulting flags. Must not modify
1158 * the device state.
1159 *
c8f44aff 1160 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1161 * Called to update device configuration to new features. Passed
1162 * feature set might be less than what was returned by ndo_fix_features()).
1163 * Must return >0 or -errno if it changed dev->features itself.
1164 *
edc7d573 1165 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1166 * struct net_device *dev,
f6f6424b 1167 * const unsigned char *addr, u16 vid, u16 flags)
77162022 1168 * Adds an FDB entry to dev for addr.
1690be63
VY
1169 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1170 * struct net_device *dev,
f6f6424b 1171 * const unsigned char *addr, u16 vid)
77162022
JF
1172 * Deletes the FDB entry from dev coresponding to addr.
1173 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1174 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1175 * int *idx)
77162022
JF
1176 * Used to add FDB entries to dump requests. Implementers should add
1177 * entries to skb and update idx with the number of entries.
e5a55a89 1178 *
ad41faa8
ND
1179 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
1180 * u16 flags)
e5a55a89 1181 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1182 * struct net_device *dev, u32 filter_mask,
1183 * int nlflags)
ad41faa8
ND
1184 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1185 * u16 flags);
4bf84c35
JP
1186 *
1187 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1188 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1189 * which do not represent real hardware may define this to allow their
1190 * userspace components to manage their virtual carrier state. Devices
1191 * that determine carrier state from physical hardware properties (eg
1192 * network cables) or protocol-dependent mechanisms (eg
1193 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1194 *
1195 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1196 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1197 * Called to get ID of physical port of this device. If driver does
1198 * not implement this, it is assumed that the hw is not able to have
1199 * multiple net devices on single physical port.
53cf5275 1200 *
7c46a640
AD
1201 * void (*ndo_udp_tunnel_add)(struct net_device *dev,
1202 * struct udp_tunnel_info *ti);
1203 * Called by UDP tunnel to notify a driver about the UDP port and socket
1204 * address family that a UDP tunnel is listnening to. It is called only
1205 * when a new port starts listening. The operation is protected by the
1206 * RTNL.
1207 *
1208 * void (*ndo_udp_tunnel_del)(struct net_device *dev,
1209 * struct udp_tunnel_info *ti);
1210 * Called by UDP tunnel to notify the driver about a UDP port and socket
1211 * address family that the UDP tunnel is not listening to anymore. The
1212 * operation is protected by the RTNL.
1213 *
a6cc0cfa
JF
1214 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1215 * struct net_device *dev)
1216 * Called by upper layer devices to accelerate switching or other
1217 * station functionality into hardware. 'pdev is the lowerdev
1218 * to use for the offload and 'dev' is the net device that will
1219 * back the offload. Returns a pointer to the private structure
1220 * the upper layer will maintain.
1221 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1222 * Called by upper layer device to delete the station created
1223 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1224 * the station and priv is the structure returned by the add
1225 * operation.
822b3b2e
JF
1226 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1227 * int queue_index, u32 maxrate);
1228 * Called when a user wants to set a max-rate limitation of specific
1229 * TX queue.
a54acb3a
ND
1230 * int (*ndo_get_iflink)(const struct net_device *dev);
1231 * Called to get the iflink value of this device.
d746d707 1232 * void (*ndo_change_proto_down)(struct net_device *dev,
5e82b4b2 1233 * bool proto_down);
d746d707
AK
1234 * This function is used to pass protocol port error state information
1235 * to the switch driver. The switch driver can react to the proto_down
1236 * by doing a phys down on the associated switch port.
fc4099f1
PS
1237 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1238 * This function is used to get egress tunnel information for given skb.
1239 * This is useful for retrieving outer tunnel header parameters while
1240 * sampling packet.
871b642a
PA
1241 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1242 * This function is used to specify the headroom that the skb must
1243 * consider when allocation skb during packet reception. Setting
1244 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1245 * forward. Setting a negative value resets the rx headroom to the
871b642a 1246 * default value.
f4e63525 1247 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1248 * This function is used to set or query state related to XDP on the
f4e63525
JK
1249 * netdevice and manage BPF offload. See definition of
1250 * enum bpf_netdev_command for details.
42b33468
JDB
1251 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1252 * u32 flags);
735fc405
JDB
1253 * This function is used to submit @n XDP packets for transmit on a
1254 * netdevice. Returns number of frames successfully transmitted, frames
1255 * that got dropped are freed/returned via xdp_return_frame().
1256 * Returns negative number, means general error invoking ndo, meaning
1257 * no frames were xmit'ed and core-caller will free all frames.
d314774c
SH
1258 */
1259struct net_device_ops {
1260 int (*ndo_init)(struct net_device *dev);
1261 void (*ndo_uninit)(struct net_device *dev);
1262 int (*ndo_open)(struct net_device *dev);
1263 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1264 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1265 struct net_device *dev);
1266 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1267 struct net_device *dev,
1268 netdev_features_t features);
00829823 1269 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1270 struct sk_buff *skb,
4f49dec9 1271 struct net_device *sb_dev,
99932d4f 1272 select_queue_fallback_t fallback);
d314774c
SH
1273 void (*ndo_change_rx_flags)(struct net_device *dev,
1274 int flags);
d314774c 1275 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1276 int (*ndo_set_mac_address)(struct net_device *dev,
1277 void *addr);
d314774c 1278 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1279 int (*ndo_do_ioctl)(struct net_device *dev,
1280 struct ifreq *ifr, int cmd);
d314774c
SH
1281 int (*ndo_set_config)(struct net_device *dev,
1282 struct ifmap *map);
00829823
SH
1283 int (*ndo_change_mtu)(struct net_device *dev,
1284 int new_mtu);
1285 int (*ndo_neigh_setup)(struct net_device *dev,
1286 struct neigh_parms *);
d314774c
SH
1287 void (*ndo_tx_timeout) (struct net_device *dev);
1288
bc1f4470 1289 void (*ndo_get_stats64)(struct net_device *dev,
1290 struct rtnl_link_stats64 *storage);
3df5b3c6 1291 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1292 int (*ndo_get_offload_stats)(int attr_id,
1293 const struct net_device *dev,
1294 void *attr_data);
d314774c
SH
1295 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1296
8e586137 1297 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1298 __be16 proto, u16 vid);
8e586137 1299 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1300 __be16 proto, u16 vid);
d314774c 1301#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1302 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1303 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1304 struct netpoll_info *info);
0e34e931 1305 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1306#endif
95c26df8
WM
1307 int (*ndo_set_vf_mac)(struct net_device *dev,
1308 int queue, u8 *mac);
1309 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1310 int queue, u16 vlan,
1311 u8 qos, __be16 proto);
ed616689
SC
1312 int (*ndo_set_vf_rate)(struct net_device *dev,
1313 int vf, int min_tx_rate,
1314 int max_tx_rate);
5f8444a3
GR
1315 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1316 int vf, bool setting);
dd461d6a
HS
1317 int (*ndo_set_vf_trust)(struct net_device *dev,
1318 int vf, bool setting);
95c26df8
WM
1319 int (*ndo_get_vf_config)(struct net_device *dev,
1320 int vf,
1321 struct ifla_vf_info *ivf);
1d8faf48
RE
1322 int (*ndo_set_vf_link_state)(struct net_device *dev,
1323 int vf, int link_state);
3b766cd8
EBE
1324 int (*ndo_get_vf_stats)(struct net_device *dev,
1325 int vf,
1326 struct ifla_vf_stats
1327 *vf_stats);
57b61080
SF
1328 int (*ndo_set_vf_port)(struct net_device *dev,
1329 int vf,
1330 struct nlattr *port[]);
1331 int (*ndo_get_vf_port)(struct net_device *dev,
1332 int vf, struct sk_buff *skb);
cc8e27cc
EC
1333 int (*ndo_set_vf_guid)(struct net_device *dev,
1334 int vf, u64 guid,
1335 int guid_type);
01a3d796
VZ
1336 int (*ndo_set_vf_rss_query_en)(
1337 struct net_device *dev,
1338 int vf, bool setting);
16e5cc64 1339 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1340 enum tc_setup_type type,
de4784ca 1341 void *type_data);
d11ead75 1342#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1343 int (*ndo_fcoe_enable)(struct net_device *dev);
1344 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1345 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1346 u16 xid,
1347 struct scatterlist *sgl,
1348 unsigned int sgc);
1349 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1350 u16 xid);
6247e086
YZ
1351 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1352 u16 xid,
1353 struct scatterlist *sgl,
1354 unsigned int sgc);
68bad94e
NP
1355 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1356 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1357#endif
1358
d11ead75 1359#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1360#define NETDEV_FCOE_WWNN 0
1361#define NETDEV_FCOE_WWPN 1
1362 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1363 u64 *wwn, int type);
4d288d57 1364#endif
3c9c36bc 1365
c445477d
BH
1366#ifdef CONFIG_RFS_ACCEL
1367 int (*ndo_rx_flow_steer)(struct net_device *dev,
1368 const struct sk_buff *skb,
1369 u16 rxq_index,
1370 u32 flow_id);
1371#endif
fbaec0ea 1372 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1373 struct net_device *slave_dev,
1374 struct netlink_ext_ack *extack);
fbaec0ea
JP
1375 int (*ndo_del_slave)(struct net_device *dev,
1376 struct net_device *slave_dev);
c8f44aff
MM
1377 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1378 netdev_features_t features);
5455c699 1379 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1380 netdev_features_t features);
503eebc2
JP
1381 int (*ndo_neigh_construct)(struct net_device *dev,
1382 struct neighbour *n);
1383 void (*ndo_neigh_destroy)(struct net_device *dev,
1384 struct neighbour *n);
77162022
JF
1385
1386 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1387 struct nlattr *tb[],
77162022 1388 struct net_device *dev,
6b6e2725 1389 const unsigned char *addr,
f6f6424b 1390 u16 vid,
77162022
JF
1391 u16 flags);
1392 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1393 struct nlattr *tb[],
77162022 1394 struct net_device *dev,
f6f6424b
JP
1395 const unsigned char *addr,
1396 u16 vid);
77162022
JF
1397 int (*ndo_fdb_dump)(struct sk_buff *skb,
1398 struct netlink_callback *cb,
1399 struct net_device *dev,
5d5eacb3 1400 struct net_device *filter_dev,
d297653d 1401 int *idx);
e5a55a89
JF
1402
1403 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3
RP
1404 struct nlmsghdr *nlh,
1405 u16 flags);
e5a55a89
JF
1406 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1407 u32 pid, u32 seq,
6cbdceeb 1408 struct net_device *dev,
46c264da
ND
1409 u32 filter_mask,
1410 int nlflags);
407af329 1411 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1412 struct nlmsghdr *nlh,
1413 u16 flags);
4bf84c35
JP
1414 int (*ndo_change_carrier)(struct net_device *dev,
1415 bool new_carrier);
66b52b0d 1416 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1417 struct netdev_phys_item_id *ppid);
db24a904
DA
1418 int (*ndo_get_phys_port_name)(struct net_device *dev,
1419 char *name, size_t len);
7c46a640
AD
1420 void (*ndo_udp_tunnel_add)(struct net_device *dev,
1421 struct udp_tunnel_info *ti);
1422 void (*ndo_udp_tunnel_del)(struct net_device *dev,
1423 struct udp_tunnel_info *ti);
a6cc0cfa
JF
1424 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1425 struct net_device *dev);
1426 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1427 void *priv);
1428
25175ba5 1429 int (*ndo_get_lock_subclass)(struct net_device *dev);
822b3b2e
JF
1430 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1431 int queue_index,
1432 u32 maxrate);
a54acb3a 1433 int (*ndo_get_iflink)(const struct net_device *dev);
d746d707
AK
1434 int (*ndo_change_proto_down)(struct net_device *dev,
1435 bool proto_down);
fc4099f1
PS
1436 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1437 struct sk_buff *skb);
871b642a
PA
1438 void (*ndo_set_rx_headroom)(struct net_device *dev,
1439 int needed_headroom);
f4e63525
JK
1440 int (*ndo_bpf)(struct net_device *dev,
1441 struct netdev_bpf *bpf);
735fc405 1442 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1443 struct xdp_frame **xdp,
1444 u32 flags);
e3760c7e
MK
1445 int (*ndo_xsk_async_xmit)(struct net_device *dev,
1446 u32 queue_id);
d314774c
SH
1447};
1448
7aa98047
LR
1449/**
1450 * enum net_device_priv_flags - &struct net_device priv_flags
1451 *
1452 * These are the &struct net_device, they are only set internally
1453 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1454 * userspace; this means that the order of these flags can change
7aa98047
LR
1455 * during any kernel release.
1456 *
1457 * You should have a pretty good reason to be extending these flags.
1458 *
1459 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1460 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1461 * @IFF_BONDING: bonding master or slave
7aa98047 1462 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1463 * @IFF_WAN_HDLC: WAN HDLC device
1464 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1465 * release skb->dst
1466 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1467 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1468 * @IFF_MACVLAN_PORT: device used as macvlan port
1469 * @IFF_BRIDGE_PORT: device used as bridge port
1470 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1471 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1472 * @IFF_UNICAST_FLT: Supports unicast filtering
1473 * @IFF_TEAM_PORT: device used as team port
1474 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1475 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1476 * change when it's running
1477 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1478 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1479 * underlying stacked devices
007979ea 1480 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1481 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1482 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1483 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1484 * @IFF_TEAM: device is a team device
d4ab4286 1485 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1486 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1487 * entity (i.e. the master device for bridged veth)
3c175784 1488 * @IFF_MACSEC: device is a MACsec device
f5426250 1489 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1490 * @IFF_FAILOVER: device is a failover master device
1491 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
7aa98047
LR
1492 */
1493enum netdev_priv_flags {
1494 IFF_802_1Q_VLAN = 1<<0,
1495 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1496 IFF_BONDING = 1<<2,
1497 IFF_ISATAP = 1<<3,
1498 IFF_WAN_HDLC = 1<<4,
1499 IFF_XMIT_DST_RELEASE = 1<<5,
1500 IFF_DONT_BRIDGE = 1<<6,
1501 IFF_DISABLE_NETPOLL = 1<<7,
1502 IFF_MACVLAN_PORT = 1<<8,
1503 IFF_BRIDGE_PORT = 1<<9,
1504 IFF_OVS_DATAPATH = 1<<10,
1505 IFF_TX_SKB_SHARING = 1<<11,
1506 IFF_UNICAST_FLT = 1<<12,
1507 IFF_TEAM_PORT = 1<<13,
1508 IFF_SUPP_NOFCS = 1<<14,
1509 IFF_LIVE_ADDR_CHANGE = 1<<15,
1510 IFF_MACVLAN = 1<<16,
1511 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1512 IFF_L3MDEV_MASTER = 1<<18,
1513 IFF_NO_QUEUE = 1<<19,
1514 IFF_OPENVSWITCH = 1<<20,
1515 IFF_L3MDEV_SLAVE = 1<<21,
1516 IFF_TEAM = 1<<22,
1517 IFF_RXFH_CONFIGURED = 1<<23,
1518 IFF_PHONY_HEADROOM = 1<<24,
1519 IFF_MACSEC = 1<<25,
f5426250 1520 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1521 IFF_FAILOVER = 1<<27,
1522 IFF_FAILOVER_SLAVE = 1<<28,
7aa98047
LR
1523};
1524
1525#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1526#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1527#define IFF_BONDING IFF_BONDING
7aa98047 1528#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1529#define IFF_WAN_HDLC IFF_WAN_HDLC
1530#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1531#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1532#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1533#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1534#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1535#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1536#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1537#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1538#define IFF_TEAM_PORT IFF_TEAM_PORT
1539#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1540#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1541#define IFF_MACVLAN IFF_MACVLAN
02875878 1542#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
007979ea 1543#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1544#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1545#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1546#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1547#define IFF_TEAM IFF_TEAM
d4ab4286 1548#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
3c175784 1549#define IFF_MACSEC IFF_MACSEC
f5426250 1550#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
30c8bd5a
SS
1551#define IFF_FAILOVER IFF_FAILOVER
1552#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
7aa98047 1553
536721b1
KK
1554/**
1555 * struct net_device - The DEVICE structure.
d651983d
MCC
1556 *
1557 * Actually, this whole structure is a big mistake. It mixes I/O
1558 * data with strictly "high-level" data, and it has to know about
1559 * almost every data structure used in the INET module.
536721b1
KK
1560 *
1561 * @name: This is the first field of the "visible" part of this structure
1562 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1563 * of the interface.
536721b1
KK
1564 *
1565 * @name_hlist: Device name hash chain, please keep it close to name[]
1566 * @ifalias: SNMP alias
1567 * @mem_end: Shared memory end
1568 * @mem_start: Shared memory start
1569 * @base_addr: Device I/O address
1570 * @irq: Device IRQ number
1571 *
1572 * @state: Generic network queuing layer state, see netdev_state_t
1573 * @dev_list: The global list of network devices
5e82b4b2
BH
1574 * @napi_list: List entry used for polling NAPI devices
1575 * @unreg_list: List entry when we are unregistering the
1576 * device; see the function unregister_netdev
1577 * @close_list: List entry used when we are closing the device
62d885fe
BP
1578 * @ptype_all: Device-specific packet handlers for all protocols
1579 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1580 *
1581 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1582 * @features: Currently active device features
1583 * @hw_features: User-changeable features
1584 *
1585 * @wanted_features: User-requested features
1586 * @vlan_features: Mask of features inheritable by VLAN devices
1587 *
1588 * @hw_enc_features: Mask of features inherited by encapsulating devices
1589 * This field indicates what encapsulation
1590 * offloads the hardware is capable of doing,
1591 * and drivers will need to set them appropriately.
1592 *
1593 * @mpls_features: Mask of features inheritable by MPLS
1594 *
1595 * @ifindex: interface index
5e82b4b2 1596 * @group: The group the device belongs to
536721b1
KK
1597 *
1598 * @stats: Statistics struct, which was left as a legacy, use
1599 * rtnl_link_stats64 instead
1600 *
1601 * @rx_dropped: Dropped packets by core network,
1602 * do not use this in drivers
1603 * @tx_dropped: Dropped packets by core network,
1604 * do not use this in drivers
6e7333d3
JW
1605 * @rx_nohandler: nohandler dropped packets by core network on
1606 * inactive devices, do not use this in drivers
9e55e5d3
FF
1607 * @carrier_up_count: Number of times the carrier has been up
1608 * @carrier_down_count: Number of times the carrier has been down
536721b1 1609 *
536721b1
KK
1610 * @wireless_handlers: List of functions to handle Wireless Extensions,
1611 * instead of ioctl,
1612 * see <net/iw_handler.h> for details.
1613 * @wireless_data: Instance data managed by the core of wireless extensions
1614 *
1615 * @netdev_ops: Includes several pointers to callbacks,
1616 * if one wants to override the ndo_*() functions
1617 * @ethtool_ops: Management operations
f997c55c
AA
1618 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1619 * discovery handling. Necessary for e.g. 6LoWPAN.
d476059e 1620 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1621 * of Layer 2 headers.
1622 *
1623 * @flags: Interface flags (a la BSD)
1624 * @priv_flags: Like 'flags' but invisible to userspace,
1625 * see if.h for the definitions
1626 * @gflags: Global flags ( kept as legacy )
1627 * @padded: How much padding added by alloc_netdev()
1628 * @operstate: RFC2863 operstate
1629 * @link_mode: Mapping policy to operstate
1630 * @if_port: Selectable AUI, TP, ...
1631 * @dma: DMA channel
1632 * @mtu: Interface MTU value
61e84623
JW
1633 * @min_mtu: Interface Minimum MTU value
1634 * @max_mtu: Interface Maximum MTU value
536721b1 1635 * @type: Interface hardware type
2793a23a 1636 * @hard_header_len: Maximum hardware header length.
217e6fa2 1637 * @min_header_len: Minimum hardware header length
536721b1
KK
1638 *
1639 * @needed_headroom: Extra headroom the hardware may need, but not in all
1640 * cases can this be guaranteed
1641 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1642 * cases can this be guaranteed. Some cases also use
1643 * LL_MAX_HEADER instead to allocate the skb
1644 *
1645 * interface address info:
1646 *
1647 * @perm_addr: Permanent hw address
1648 * @addr_assign_type: Hw address assignment type
1649 * @addr_len: Hardware address length
8626a0c8 1650 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1651 * @dev_id: Used to differentiate devices that share
1652 * the same link layer address
1653 * @dev_port: Used to differentiate devices that share
1654 * the same function
1655 * @addr_list_lock: XXX: need comments on this one
5e82b4b2 1656 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1657 * has been enabled due to the need to listen to
1658 * additional unicast addresses in a device that
1659 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1660 * @uc: unicast mac addresses
1661 * @mc: multicast mac addresses
1662 * @dev_addrs: list of device hw addresses
1663 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1664 * @promiscuity: Number of times the NIC is told to work in
1665 * promiscuous mode; if it becomes 0 the NIC will
1666 * exit promiscuous mode
536721b1
KK
1667 * @allmulti: Counter, enables or disables allmulticast mode
1668 *
1669 * @vlan_info: VLAN info
1670 * @dsa_ptr: dsa specific data
1671 * @tipc_ptr: TIPC specific data
1672 * @atalk_ptr: AppleTalk link
1673 * @ip_ptr: IPv4 specific data
1674 * @dn_ptr: DECnet specific data
1675 * @ip6_ptr: IPv6 specific data
1676 * @ax25_ptr: AX.25 specific data
1677 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
1678 *
536721b1
KK
1679 * @dev_addr: Hw address (before bcast,
1680 * because most packets are unicast)
1681 *
1682 * @_rx: Array of RX queues
1683 * @num_rx_queues: Number of RX queues
1684 * allocated at register_netdev() time
1685 * @real_num_rx_queues: Number of RX queues currently active in device
1686 *
1687 * @rx_handler: handler for received packets
1688 * @rx_handler_data: XXX: need comments on this one
46209401
JP
1689 * @miniq_ingress: ingress/clsact qdisc specific data for
1690 * ingress processing
536721b1
KK
1691 * @ingress_queue: XXX: need comments on this one
1692 * @broadcast: hw bcast address
1693 *
14ffbbb8
TG
1694 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1695 * indexed by RX queue number. Assigned by driver.
1696 * This must only be set if the ndo_rx_flow_steer
1697 * operation is defined
1698 * @index_hlist: Device index hash chain
1699 *
536721b1
KK
1700 * @_tx: Array of TX queues
1701 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1702 * @real_num_tx_queues: Number of TX queues currently active in device
1703 * @qdisc: Root qdisc from userspace point of view
1704 * @tx_queue_len: Max frames per queue allowed
1705 * @tx_global_lock: XXX: need comments on this one
1706 *
1707 * @xps_maps: XXX: need comments on this one
46209401
JP
1708 * @miniq_egress: clsact qdisc specific data for
1709 * egress processing
536721b1 1710 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1711 * the watchdog (see dev_watchdog())
536721b1
KK
1712 * @watchdog_timer: List of timers
1713 *
1714 * @pcpu_refcnt: Number of references to this device
1715 * @todo_list: Delayed register/unregister
536721b1
KK
1716 * @link_watch_list: XXX: need comments on this one
1717 *
1718 * @reg_state: Register/unregister state machine
1719 * @dismantle: Device is going to be freed
1720 * @rtnl_link_state: This enum represents the phases of creating
1721 * a new link
1722 *
cf124db5
DM
1723 * @needs_free_netdev: Should unregister perform free_netdev?
1724 * @priv_destructor: Called from unregister
536721b1
KK
1725 * @npinfo: XXX: need comments on this one
1726 * @nd_net: Network namespace this network device is inside
1727 *
1728 * @ml_priv: Mid-layer private
1729 * @lstats: Loopback statistics
1730 * @tstats: Tunnel statistics
1731 * @dstats: Dummy statistics
1732 * @vstats: Virtual ethernet statistics
1733 *
1734 * @garp_port: GARP
1735 * @mrp_port: MRP
1736 *
1737 * @dev: Class/net/name entry
1738 * @sysfs_groups: Space for optional device, statistics and wireless
1739 * sysfs groups
1740 *
1741 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1742 * @rtnl_link_ops: Rtnl_link_ops
1743 *
1744 * @gso_max_size: Maximum size of generic segmentation offload
1745 * @gso_max_segs: Maximum number of segments that can be passed to the
1746 * NIC for GSO
1747 *
1748 * @dcbnl_ops: Data Center Bridging netlink ops
1749 * @num_tc: Number of traffic classes in the net device
1750 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1751 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
1752 *
1753 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1754 *
1755 * @priomap: XXX: need comments on this one
1756 * @phydev: Physical device may attach itself
1757 * for hardware timestamping
e679c9c1 1758 * @sfp_bus: attached &struct sfp_bus structure.
536721b1 1759 *
123b3652
ED
1760 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
1761 * @qdisc_running_key: lockdep class annotating Qdisc->running seqcount
536721b1 1762 *
d746d707
AK
1763 * @proto_down: protocol port state information can be sent to the
1764 * switch driver and used to set the phys state of the
1765 * switch port.
1766 *
61941143
HK
1767 * @wol_enabled: Wake-on-LAN is enabled
1768 *
1da177e4
LT
1769 * FIXME: cleanup struct net_device such that network protocol info
1770 * moves out.
1771 */
1772
d94d9fee 1773struct net_device {
1da177e4 1774 char name[IFNAMSIZ];
9356b8fc 1775 struct hlist_node name_hlist;
6c557001 1776 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
1777 /*
1778 * I/O specific fields
1779 * FIXME: Merge these and struct ifmap into one
1780 */
536721b1
KK
1781 unsigned long mem_end;
1782 unsigned long mem_start;
1783 unsigned long base_addr;
1784 int irq;
1da177e4
LT
1785
1786 /*
536721b1
KK
1787 * Some hardware also needs these fields (state,dev_list,
1788 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1789 * part of the usual set specified in Space.c.
1790 */
1791
1da177e4
LT
1792 unsigned long state;
1793
7562f876 1794 struct list_head dev_list;
bea3348e 1795 struct list_head napi_list;
44a0873d 1796 struct list_head unreg_list;
5cde2829 1797 struct list_head close_list;
7866a621
SN
1798 struct list_head ptype_all;
1799 struct list_head ptype_specific;
2f268f12 1800
2f268f12
VF
1801 struct {
1802 struct list_head upper;
1803 struct list_head lower;
1804 } adj_list;
1805
c8f44aff 1806 netdev_features_t features;
c8f44aff 1807 netdev_features_t hw_features;
c8f44aff 1808 netdev_features_t wanted_features;
c8f44aff 1809 netdev_features_t vlan_features;
6a674e9c 1810 netdev_features_t hw_enc_features;
0d89d203 1811 netdev_features_t mpls_features;
802ab55a 1812 netdev_features_t gso_partial_features;
04ed3e74 1813
1da177e4 1814 int ifindex;
7a66bbc9 1815 int group;
1da177e4 1816
c45d286e 1817 struct net_device_stats stats;
015f0688 1818
015f0688
ED
1819 atomic_long_t rx_dropped;
1820 atomic_long_t tx_dropped;
6e7333d3 1821 atomic_long_t rx_nohandler;
1da177e4 1822
b2d3bcfa
DD
1823 /* Stats to monitor link on/off, flapping */
1824 atomic_t carrier_up_count;
1825 atomic_t carrier_down_count;
1826
b86e0280 1827#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
1828 const struct iw_handler_def *wireless_handlers;
1829 struct iw_public_data *wireless_data;
b86e0280 1830#endif
d314774c 1831 const struct net_device_ops *netdev_ops;
76fd8593 1832 const struct ethtool_ops *ethtool_ops;
4170604f 1833#ifdef CONFIG_NET_SWITCHDEV
9d47c0a2 1834 const struct switchdev_ops *switchdev_ops;
4170604f 1835#endif
1b69c6d0
DA
1836#ifdef CONFIG_NET_L3_MASTER_DEV
1837 const struct l3mdev_ops *l3mdev_ops;
1838#endif
f997c55c
AA
1839#if IS_ENABLED(CONFIG_IPV6)
1840 const struct ndisc_ops *ndisc_ops;
1841#endif
1da177e4 1842
9cb0d21d 1843#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
1844 const struct xfrmdev_ops *xfrmdev_ops;
1845#endif
1846
a5c37c63
IL
1847#if IS_ENABLED(CONFIG_TLS_DEVICE)
1848 const struct tlsdev_ops *tlsdev_ops;
1849#endif
1850
3b04ddde
SH
1851 const struct header_ops *header_ops;
1852
536721b1
KK
1853 unsigned int flags;
1854 unsigned int priv_flags;
1855
1da177e4 1856 unsigned short gflags;
536721b1 1857 unsigned short padded;
1da177e4 1858
536721b1
KK
1859 unsigned char operstate;
1860 unsigned char link_mode;
b00055aa 1861
536721b1
KK
1862 unsigned char if_port;
1863 unsigned char dma;
bdc220da 1864
536721b1 1865 unsigned int mtu;
61e84623
JW
1866 unsigned int min_mtu;
1867 unsigned int max_mtu;
536721b1
KK
1868 unsigned short type;
1869 unsigned short hard_header_len;
d92be7a4 1870 unsigned char min_header_len;
1da177e4 1871
f5184d26
JB
1872 unsigned short needed_headroom;
1873 unsigned short needed_tailroom;
1874
1da177e4 1875 /* Interface address info. */
536721b1
KK
1876 unsigned char perm_addr[MAX_ADDR_LEN];
1877 unsigned char addr_assign_type;
1878 unsigned char addr_len;
a0a9663d 1879 unsigned short neigh_priv_len;
536721b1
KK
1880 unsigned short dev_id;
1881 unsigned short dev_port;
ccffad25 1882 spinlock_t addr_list_lock;
14ffbbb8
TG
1883 unsigned char name_assign_type;
1884 bool uc_promisc;
536721b1
KK
1885 struct netdev_hw_addr_list uc;
1886 struct netdev_hw_addr_list mc;
1887 struct netdev_hw_addr_list dev_addrs;
1888
4c3d5e7b
ED
1889#ifdef CONFIG_SYSFS
1890 struct kset *queues_kset;
1891#endif
9d45abe1
WC
1892 unsigned int promiscuity;
1893 unsigned int allmulti;
1da177e4 1894
1da177e4 1895
5e82b4b2 1896 /* Protocol-specific pointers */
65ac6a5f 1897
d11ead75 1898#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 1899 struct vlan_info __rcu *vlan_info;
65ac6a5f 1900#endif
34a430d7 1901#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 1902 struct dsa_port *dsa_ptr;
37cb0620
YX
1903#endif
1904#if IS_ENABLED(CONFIG_TIPC)
536721b1 1905 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 1906#endif
89e58148 1907#if IS_ENABLED(CONFIG_IRDA) || IS_ENABLED(CONFIG_ATALK)
536721b1 1908 void *atalk_ptr;
89e58148 1909#endif
536721b1 1910 struct in_device __rcu *ip_ptr;
330c7272 1911#if IS_ENABLED(CONFIG_DECNET)
536721b1 1912 struct dn_dev __rcu *dn_ptr;
330c7272 1913#endif
536721b1 1914 struct inet6_dev __rcu *ip6_ptr;
19ff13f2 1915#if IS_ENABLED(CONFIG_AX25)
536721b1 1916 void *ax25_ptr;
19ff13f2 1917#endif
536721b1 1918 struct wireless_dev *ieee80211_ptr;
98a18b6f 1919 struct wpan_dev *ieee802154_ptr;
03c57747
RS
1920#if IS_ENABLED(CONFIG_MPLS_ROUTING)
1921 struct mpls_dev __rcu *mpls_ptr;
1922#endif
1da177e4 1923
9356b8fc 1924/*
cd13539b 1925 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 1926 */
9356b8fc 1927 /* Interface address info used in eth_type_trans() */
536721b1 1928 unsigned char *dev_addr;
f001fde5 1929
0a9627f2 1930 struct netdev_rx_queue *_rx;
0a9627f2 1931 unsigned int num_rx_queues;
62fe0b40 1932 unsigned int real_num_rx_queues;
0a9627f2 1933
7acedaf5 1934 struct bpf_prog __rcu *xdp_prog;
3b47d303 1935 unsigned long gro_flush_timeout;
61391cde 1936 rx_handler_func_t __rcu *rx_handler;
1937 void __rcu *rx_handler_data;
e8a0464c 1938
4cda01e8 1939#ifdef CONFIG_NET_CLS_ACT
46209401 1940 struct mini_Qdisc __rcu *miniq_ingress;
d2788d34 1941#endif
24824a09 1942 struct netdev_queue __rcu *ingress_queue;
e687ad60 1943#ifdef CONFIG_NETFILTER_INGRESS
960632ec 1944 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 1945#endif
d2788d34 1946
536721b1 1947 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
1948#ifdef CONFIG_RFS_ACCEL
1949 struct cpu_rmap *rx_cpu_rmap;
1950#endif
1951 struct hlist_node index_hlist;
cd13539b
ED
1952
1953/*
1954 * Cache lines mostly used on transmit path
1955 */
e8a0464c
DM
1956 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
1957 unsigned int num_tx_queues;
fd2ea0a7 1958 unsigned int real_num_tx_queues;
af356afa 1959 struct Qdisc *qdisc;
59cc1f61
JK
1960#ifdef CONFIG_NET_SCHED
1961 DECLARE_HASHTABLE (qdisc_hash, 4);
1962#endif
0cd29503 1963 unsigned int tx_queue_len;
c3f26a26 1964 spinlock_t tx_global_lock;
14ffbbb8 1965 int watchdog_timeo;
cd13539b 1966
bf264145 1967#ifdef CONFIG_XPS
80d19669
AN
1968 struct xps_dev_maps __rcu *xps_cpus_map;
1969 struct xps_dev_maps __rcu *xps_rxqs_map;
bf264145 1970#endif
1f211a1b 1971#ifdef CONFIG_NET_CLS_ACT
46209401 1972 struct mini_Qdisc __rcu *miniq_egress;
1f211a1b 1973#endif
0c4f691f 1974
9356b8fc 1975 /* These may be needed for future network-power-down code. */
9356b8fc
ED
1976 struct timer_list watchdog_timer;
1977
29b4433d 1978 int __percpu *pcpu_refcnt;
1da177e4 1979 struct list_head todo_list;
1da177e4 1980
e014debe 1981 struct list_head link_watch_list;
572a103d 1982
1da177e4 1983 enum { NETREG_UNINITIALIZED=0,
b17a7c17 1984 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
1985 NETREG_UNREGISTERING, /* called unregister_netdevice */
1986 NETREG_UNREGISTERED, /* completed unregister todo */
1987 NETREG_RELEASED, /* called free_netdev */
937f1ba5 1988 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
1989 } reg_state:8;
1990
536721b1 1991 bool dismantle;
a2835763
PM
1992
1993 enum {
1994 RTNL_LINK_INITIALIZED,
1995 RTNL_LINK_INITIALIZING,
1996 } rtnl_link_state:16;
1da177e4 1997
cf124db5
DM
1998 bool needs_free_netdev;
1999 void (*priv_destructor)(struct net_device *dev);
1da177e4 2000
1da177e4 2001#ifdef CONFIG_NETPOLL
5fbee843 2002 struct netpoll_info __rcu *npinfo;
1da177e4 2003#endif
eae792b7 2004
0c5c9fb5 2005 possible_net_t nd_net;
4a1c5371 2006
4951704b 2007 /* mid-layer private */
a7855c78 2008 union {
536721b1
KK
2009 void *ml_priv;
2010 struct pcpu_lstats __percpu *lstats;
8f84985f 2011 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2012 struct pcpu_dstats __percpu *dstats;
a7855c78 2013 };
536721b1 2014
fb585b44 2015#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2016 struct garp_port __rcu *garp_port;
fb585b44
TK
2017#endif
2018#if IS_ENABLED(CONFIG_MRP)
febf018d 2019 struct mrp_port __rcu *mrp_port;
fb585b44 2020#endif
1da177e4 2021
5e82b4b2 2022 struct device dev;
0c509a6c 2023 const struct attribute_group *sysfs_groups[4];
a953be53 2024 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2025
38f7b870 2026 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2027
82cc1a7a
PWJ
2028 /* for setting kernel sock attribute on TCP connection setup */
2029#define GSO_MAX_SIZE 65536
2030 unsigned int gso_max_size;
30b678d8
BH
2031#define GSO_MAX_SEGS 65535
2032 u16 gso_max_segs;
743b03a8 2033
7a6b6f51 2034#ifdef CONFIG_DCB
32953543 2035 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2036#endif
ffcfe25b 2037 s16 num_tc;
5e82b4b2
BH
2038 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2039 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2040
d11ead75 2041#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2042 unsigned int fcoe_ddp_xid;
5bc1421e 2043#endif
86f8515f 2044#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2045 struct netprio_map __rcu *priomap;
4d288d57 2046#endif
5e82b4b2 2047 struct phy_device *phydev;
e679c9c1 2048 struct sfp_bus *sfp_bus;
5e82b4b2 2049 struct lock_class_key *qdisc_tx_busylock;
f9eb8aea 2050 struct lock_class_key *qdisc_running_key;
5e82b4b2 2051 bool proto_down;
61941143 2052 unsigned wol_enabled:1;
1da177e4 2053};
43cb76d9 2054#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2055
b5cdae32
DM
2056static inline bool netif_elide_gro(const struct net_device *dev)
2057{
2058 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2059 return true;
2060 return false;
2061}
2062
1da177e4 2063#define NETDEV_ALIGN 32
1da177e4 2064
4f57c087
JF
2065static inline
2066int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2067{
2068 return dev->prio_tc_map[prio & TC_BITMASK];
2069}
2070
2071static inline
2072int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2073{
2074 if (tc >= dev->num_tc)
2075 return -EINVAL;
2076
2077 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2078 return 0;
2079}
2080
8d059b0f 2081int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2082void netdev_reset_tc(struct net_device *dev);
2083int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2084int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2085
2086static inline
2087int netdev_get_num_tc(struct net_device *dev)
2088{
2089 return dev->num_tc;
2090}
2091
ffcfe25b
AD
2092void netdev_unbind_sb_channel(struct net_device *dev,
2093 struct net_device *sb_dev);
2094int netdev_bind_sb_channel_queue(struct net_device *dev,
2095 struct net_device *sb_dev,
2096 u8 tc, u16 count, u16 offset);
2097int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2098static inline int netdev_get_sb_channel(struct net_device *dev)
2099{
2100 return max_t(int, -dev->num_tc, 0);
2101}
2102
e8a0464c
DM
2103static inline
2104struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2105 unsigned int index)
2106{
2107 return &dev->_tx[index];
2108}
2109
10c51b56
DB
2110static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2111 const struct sk_buff *skb)
2112{
2113 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2114}
2115
e8a0464c
DM
2116static inline void netdev_for_each_tx_queue(struct net_device *dev,
2117 void (*f)(struct net_device *,
2118 struct netdev_queue *,
2119 void *),
2120 void *arg)
2121{
2122 unsigned int i;
2123
2124 for (i = 0; i < dev->num_tx_queues; i++)
2125 f(dev, &dev->_tx[i], arg);
2126}
2127
d3fff6c4
ED
2128#define netdev_lockdep_set_classes(dev) \
2129{ \
2130 static struct lock_class_key qdisc_tx_busylock_key; \
2131 static struct lock_class_key qdisc_running_key; \
2132 static struct lock_class_key qdisc_xmit_lock_key; \
2133 static struct lock_class_key dev_addr_list_lock_key; \
2134 unsigned int i; \
2135 \
2136 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
2137 (dev)->qdisc_running_key = &qdisc_running_key; \
2138 lockdep_set_class(&(dev)->addr_list_lock, \
2139 &dev_addr_list_lock_key); \
2140 for (i = 0; i < (dev)->num_tx_queues; i++) \
2141 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2142 &qdisc_xmit_lock_key); \
2143}
2144
f629d208 2145struct netdev_queue *netdev_pick_tx(struct net_device *dev,
f663dd9a 2146 struct sk_buff *skb,
eadec877 2147 struct net_device *sb_dev);
8c4c49df 2148
871b642a
PA
2149/* returns the headroom that the master device needs to take in account
2150 * when forwarding to this dev
2151 */
2152static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2153{
2154 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2155}
2156
2157static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2158{
2159 if (dev->netdev_ops->ndo_set_rx_headroom)
2160 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2161}
2162
2163/* set the device rx headroom to the dev's default */
2164static inline void netdev_reset_rx_headroom(struct net_device *dev)
2165{
2166 netdev_set_rx_headroom(dev, -1);
2167}
2168
c346dca1
YH
2169/*
2170 * Net namespace inlines
2171 */
2172static inline
2173struct net *dev_net(const struct net_device *dev)
2174{
c2d9ba9b 2175 return read_pnet(&dev->nd_net);
c346dca1
YH
2176}
2177
2178static inline
f5aa23fd 2179void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2180{
0c5c9fb5 2181 write_pnet(&dev->nd_net, net);
c346dca1
YH
2182}
2183
bea3348e
SH
2184/**
2185 * netdev_priv - access network device private data
2186 * @dev: network device
2187 *
2188 * Get network device private data
2189 */
6472ce60 2190static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2191{
1ce8e7b5 2192 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2193}
2194
1da177e4
LT
2195/* Set the sysfs physical device reference for the network logical device
2196 * if set prior to registration will cause a symlink during initialization.
2197 */
43cb76d9 2198#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2199
384912ed 2200/* Set the sysfs device type for the network logical device to allow
3f79410c 2201 * fine-grained identification of different network device types. For
5e82b4b2 2202 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2203 */
2204#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2205
82dc3c63
ED
2206/* Default NAPI poll() weight
2207 * Device drivers are strongly advised to not use bigger value
2208 */
2209#define NAPI_POLL_WEIGHT 64
2210
3b582cc1 2211/**
5e82b4b2 2212 * netif_napi_add - initialize a NAPI context
3b582cc1 2213 * @dev: network device
5e82b4b2 2214 * @napi: NAPI context
3b582cc1
SH
2215 * @poll: polling function
2216 * @weight: default weight
2217 *
5e82b4b2
BH
2218 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2219 * *any* of the other NAPI-related functions.
3b582cc1 2220 */
d565b0a1
HX
2221void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2222 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2223
d64b5e85 2224/**
5e82b4b2 2225 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2226 * @dev: network device
5e82b4b2 2227 * @napi: NAPI context
d64b5e85
ED
2228 * @poll: polling function
2229 * @weight: default weight
2230 *
2231 * This variant of netif_napi_add() should be used from drivers using NAPI
2232 * to exclusively poll a TX queue.
2233 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2234 */
2235static inline void netif_tx_napi_add(struct net_device *dev,
2236 struct napi_struct *napi,
2237 int (*poll)(struct napi_struct *, int),
2238 int weight)
2239{
2240 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2241 netif_napi_add(dev, napi, poll, weight);
2242}
2243
d8156534 2244/**
5e82b4b2
BH
2245 * netif_napi_del - remove a NAPI context
2246 * @napi: NAPI context
d8156534 2247 *
5e82b4b2 2248 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2249 */
d565b0a1
HX
2250void netif_napi_del(struct napi_struct *napi);
2251
2252struct napi_gro_cb {
78a478d0 2253 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
5e82b4b2 2254 void *frag0;
78a478d0 2255
7489594c
HX
2256 /* Length of frag0. */
2257 unsigned int frag0_len;
2258
86911732 2259 /* This indicates where we are processing relative to skb->data. */
5e82b4b2 2260 int data_offset;
86911732 2261
d565b0a1 2262 /* This is non-zero if the packet cannot be merged with the new skb. */
bf5a755f
JC
2263 u16 flush;
2264
2265 /* Save the IP ID here and check when we get to the transport layer */
2266 u16 flush_id;
d565b0a1
HX
2267
2268 /* Number of segments aggregated. */
2e71a6f8
ED
2269 u16 count;
2270
15e2396d
TH
2271 /* Start offset for remote checksum offload */
2272 u16 gro_remcsum_start;
2273
2e71a6f8
ED
2274 /* jiffies when first packet was created/queued */
2275 unsigned long age;
86347245 2276
afe93325 2277 /* Used in ipv6_gro_receive() and foo-over-udp */
b582ef09
OG
2278 u16 proto;
2279
baa32ff4
TH
2280 /* This is non-zero if the packet may be of the same flow. */
2281 u8 same_flow:1;
2282
fac8e0f5
JG
2283 /* Used in tunnel GRO receive */
2284 u8 encap_mark:1;
573e8fca
TH
2285
2286 /* GRO checksum is valid */
2287 u8 csum_valid:1;
2288
662880f4
TH
2289 /* Number of checksums via CHECKSUM_UNNECESSARY */
2290 u8 csum_cnt:3;
c3c7c254 2291
baa32ff4
TH
2292 /* Free the skb? */
2293 u8 free:2;
2294#define NAPI_GRO_FREE 1
2295#define NAPI_GRO_FREE_STOLEN_HEAD 2
2296
efc98d08
TH
2297 /* Used in foo-over-udp, set in udp[46]_gro_receive */
2298 u8 is_ipv6:1;
2299
a0ca153f
AD
2300 /* Used in GRE, set in fou/gue_gro_receive */
2301 u8 is_fou:1;
2302
1530545e
AD
2303 /* Used to determine if flush_id can be ignored */
2304 u8 is_atomic:1;
2305
fcd91dd4
SD
2306 /* Number of gro_receive callbacks this packet already went through */
2307 u8 recursion_counter:4;
2308
2309 /* 1 bit hole */
baa32ff4 2310
bf5a755f
JC
2311 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
2312 __wsum csum;
2313
c3c7c254
ED
2314 /* used in skb_gro_receive() slow path */
2315 struct sk_buff *last;
d565b0a1
HX
2316};
2317
2318#define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
d8156534 2319
fcd91dd4
SD
2320#define GRO_RECURSION_LIMIT 15
2321static inline int gro_recursion_inc_test(struct sk_buff *skb)
2322{
2323 return ++NAPI_GRO_CB(skb)->recursion_counter == GRO_RECURSION_LIMIT;
2324}
2325
d4546c25
DM
2326typedef struct sk_buff *(*gro_receive_t)(struct list_head *, struct sk_buff *);
2327static inline struct sk_buff *call_gro_receive(gro_receive_t cb,
2328 struct list_head *head,
2329 struct sk_buff *skb)
fcd91dd4
SD
2330{
2331 if (unlikely(gro_recursion_inc_test(skb))) {
2332 NAPI_GRO_CB(skb)->flush |= 1;
2333 return NULL;
2334 }
2335
2336 return cb(head, skb);
2337}
2338
d4546c25
DM
2339typedef struct sk_buff *(*gro_receive_sk_t)(struct sock *, struct list_head *,
2340 struct sk_buff *);
2341static inline struct sk_buff *call_gro_receive_sk(gro_receive_sk_t cb,
2342 struct sock *sk,
2343 struct list_head *head,
2344 struct sk_buff *skb)
fcd91dd4
SD
2345{
2346 if (unlikely(gro_recursion_inc_test(skb))) {
2347 NAPI_GRO_CB(skb)->flush |= 1;
2348 return NULL;
2349 }
2350
2351 return cb(sk, head, skb);
2352}
2353
1da177e4 2354struct packet_type {
f2ccd8fa 2355 __be16 type; /* This is really htons(ether_type). */
fa788d98 2356 bool ignore_outgoing;
f2ccd8fa
DM
2357 struct net_device *dev; /* NULL is wildcarded here */
2358 int (*func) (struct sk_buff *,
2359 struct net_device *,
2360 struct packet_type *,
2361 struct net_device *);
17266ee9
EC
2362 void (*list_func) (struct list_head *,
2363 struct packet_type *,
2364 struct net_device *);
c0de08d0
EL
2365 bool (*id_match)(struct packet_type *ptype,
2366 struct sock *sk);
1da177e4
LT
2367 void *af_packet_priv;
2368 struct list_head list;
2369};
2370
f191a1d1 2371struct offload_callbacks {
576a30eb 2372 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2373 netdev_features_t features);
d4546c25
DM
2374 struct sk_buff *(*gro_receive)(struct list_head *head,
2375 struct sk_buff *skb);
299603e8 2376 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2377};
2378
2379struct packet_offload {
2380 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2381 u16 priority;
f191a1d1
VY
2382 struct offload_callbacks callbacks;
2383 struct list_head list;
1da177e4
LT
2384};
2385
5e82b4b2 2386/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2387struct pcpu_sw_netstats {
2388 u64 rx_packets;
2389 u64 rx_bytes;
2390 u64 tx_packets;
2391 u64 tx_bytes;
2392 struct u64_stats_sync syncp;
9a5ee462 2393} __aligned(4 * sizeof(u64));
52bb6677
LR
2394
2395struct pcpu_lstats {
2396 u64 packets;
2397 u64 bytes;
2398 struct u64_stats_sync syncp;
9a5ee462 2399} __aligned(2 * sizeof(u64));
8f84985f 2400
aabc92bb
PNA
2401#define __netdev_alloc_pcpu_stats(type, gfp) \
2402({ \
2403 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2404 if (pcpu_stats) { \
2405 int __cpu; \
2406 for_each_possible_cpu(__cpu) { \
2407 typeof(type) *stat; \
2408 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2409 u64_stats_init(&stat->syncp); \
2410 } \
2411 } \
2412 pcpu_stats; \
1c213bd2
WC
2413})
2414
aabc92bb 2415#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2416 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2417
764f5e54
JP
2418enum netdev_lag_tx_type {
2419 NETDEV_LAG_TX_TYPE_UNKNOWN,
2420 NETDEV_LAG_TX_TYPE_RANDOM,
2421 NETDEV_LAG_TX_TYPE_BROADCAST,
2422 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2423 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2424 NETDEV_LAG_TX_TYPE_HASH,
2425};
2426
f44aa9ef
JH
2427enum netdev_lag_hash {
2428 NETDEV_LAG_HASH_NONE,
2429 NETDEV_LAG_HASH_L2,
2430 NETDEV_LAG_HASH_L34,
2431 NETDEV_LAG_HASH_L23,
2432 NETDEV_LAG_HASH_E23,
2433 NETDEV_LAG_HASH_E34,
2434 NETDEV_LAG_HASH_UNKNOWN,
2435};
2436
764f5e54
JP
2437struct netdev_lag_upper_info {
2438 enum netdev_lag_tx_type tx_type;
f44aa9ef 2439 enum netdev_lag_hash hash_type;
764f5e54
JP
2440};
2441
fb1b2e3c
JP
2442struct netdev_lag_lower_state_info {
2443 u8 link_up : 1,
2444 tx_enabled : 1;
2445};
2446
1da177e4
LT
2447#include <linux/notifier.h>
2448
ede2762d
KT
2449/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2450 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2451 * adding new types.
dcfe1421 2452 */
ede2762d
KT
2453enum netdev_cmd {
2454 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2455 NETDEV_DOWN,
2456 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2457 detected a hardware crash and restarted
2458 - we can use this eg to kick tcp sessions
2459 once done */
ede2762d
KT
2460 NETDEV_CHANGE, /* Notify device state change */
2461 NETDEV_REGISTER,
2462 NETDEV_UNREGISTER,
2463 NETDEV_CHANGEMTU, /* notify after mtu change happened */
2464 NETDEV_CHANGEADDR,
2465 NETDEV_GOING_DOWN,
2466 NETDEV_CHANGENAME,
2467 NETDEV_FEAT_CHANGE,
2468 NETDEV_BONDING_FAILOVER,
2469 NETDEV_PRE_UP,
2470 NETDEV_PRE_TYPE_CHANGE,
2471 NETDEV_POST_TYPE_CHANGE,
2472 NETDEV_POST_INIT,
ede2762d
KT
2473 NETDEV_RELEASE,
2474 NETDEV_NOTIFY_PEERS,
2475 NETDEV_JOIN,
2476 NETDEV_CHANGEUPPER,
2477 NETDEV_RESEND_IGMP,
2478 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2479 NETDEV_CHANGEINFODATA,
2480 NETDEV_BONDING_INFO,
2481 NETDEV_PRECHANGEUPPER,
2482 NETDEV_CHANGELOWERSTATE,
2483 NETDEV_UDP_TUNNEL_PUSH_INFO,
2484 NETDEV_UDP_TUNNEL_DROP_INFO,
2485 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2486 NETDEV_CVLAN_FILTER_PUSH_INFO,
2487 NETDEV_CVLAN_FILTER_DROP_INFO,
2488 NETDEV_SVLAN_FILTER_PUSH_INFO,
2489 NETDEV_SVLAN_FILTER_DROP_INFO,
ede2762d
KT
2490};
2491const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2492
f629d208
JP
2493int register_netdevice_notifier(struct notifier_block *nb);
2494int unregister_netdevice_notifier(struct notifier_block *nb);
351638e7
JP
2495
2496struct netdev_notifier_info {
51d0c047
DA
2497 struct net_device *dev;
2498 struct netlink_ext_ack *extack;
351638e7
JP
2499};
2500
af7d6cce
SD
2501struct netdev_notifier_info_ext {
2502 struct netdev_notifier_info info; /* must be first */
2503 union {
2504 u32 mtu;
2505 } ext;
2506};
2507
be9efd36
JP
2508struct netdev_notifier_change_info {
2509 struct netdev_notifier_info info; /* must be first */
2510 unsigned int flags_changed;
2511};
2512
0e4ead9d
JP
2513struct netdev_notifier_changeupper_info {
2514 struct netdev_notifier_info info; /* must be first */
2515 struct net_device *upper_dev; /* new upper dev */
2516 bool master; /* is upper dev master */
5e82b4b2 2517 bool linking; /* is the notification for link or unlink */
29bf24af 2518 void *upper_info; /* upper dev info */
0e4ead9d
JP
2519};
2520
04d48266
JP
2521struct netdev_notifier_changelowerstate_info {
2522 struct netdev_notifier_info info; /* must be first */
2523 void *lower_state_info; /* is lower dev state */
2524};
2525
75538c2b
CW
2526static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2527 struct net_device *dev)
2528{
2529 info->dev = dev;
51d0c047 2530 info->extack = NULL;
75538c2b
CW
2531}
2532
351638e7
JP
2533static inline struct net_device *
2534netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2535{
2536 return info->dev;
2537}
2538
51d0c047
DA
2539static inline struct netlink_ext_ack *
2540netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2541{
2542 return info->extack;
2543}
2544
f629d208 2545int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2546
2547
1da177e4
LT
2548extern rwlock_t dev_base_lock; /* Device list lock */
2549
881d966b
EB
2550#define for_each_netdev(net, d) \
2551 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2552#define for_each_netdev_reverse(net, d) \
2553 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2554#define for_each_netdev_rcu(net, d) \
2555 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2556#define for_each_netdev_safe(net, d, n) \
2557 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2558#define for_each_netdev_continue(net, d) \
2559 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
254245d2 2560#define for_each_netdev_continue_rcu(net, d) \
2561 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2562#define for_each_netdev_in_bond_rcu(bond, slave) \
2563 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2564 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2565#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2566
a050c33f
DL
2567static inline struct net_device *next_net_device(struct net_device *dev)
2568{
2569 struct list_head *lh;
2570 struct net *net;
2571
c346dca1 2572 net = dev_net(dev);
a050c33f
DL
2573 lh = dev->dev_list.next;
2574 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2575}
2576
ce81b76a
ED
2577static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2578{
2579 struct list_head *lh;
2580 struct net *net;
2581
2582 net = dev_net(dev);
ccf43438 2583 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2584 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2585}
2586
a050c33f
DL
2587static inline struct net_device *first_net_device(struct net *net)
2588{
2589 return list_empty(&net->dev_base_head) ? NULL :
2590 net_device_entry(net->dev_base_head.next);
2591}
7562f876 2592
ccf43438
ED
2593static inline struct net_device *first_net_device_rcu(struct net *net)
2594{
2595 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2596
2597 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2598}
2599
f629d208
JP
2600int netdev_boot_setup_check(struct net_device *dev);
2601unsigned long netdev_boot_base(const char *prefix, int unit);
2602struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2603 const char *hwaddr);
2604struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
2605struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type);
2606void dev_add_pack(struct packet_type *pt);
2607void dev_remove_pack(struct packet_type *pt);
2608void __dev_remove_pack(struct packet_type *pt);
2609void dev_add_offload(struct packet_offload *po);
2610void dev_remove_offload(struct packet_offload *po);
f629d208 2611
a54acb3a 2612int dev_get_iflink(const struct net_device *dev);
fc4099f1 2613int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
6c555490
WC
2614struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2615 unsigned short mask);
f629d208
JP
2616struct net_device *dev_get_by_name(struct net *net, const char *name);
2617struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2618struct net_device *__dev_get_by_name(struct net *net, const char *name);
2619int dev_alloc_name(struct net_device *dev, const char *name);
2620int dev_open(struct net_device *dev);
7051b88a 2621void dev_close(struct net_device *dev);
2622void dev_close_many(struct list_head *head, bool unlink);
f629d208 2623void dev_disable_lro(struct net_device *dev);
0c4b51f0 2624int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 2625u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
4f49dec9
AD
2626 struct net_device *sb_dev,
2627 select_queue_fallback_t fallback);
a4ea8a3d 2628u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
4f49dec9
AD
2629 struct net_device *sb_dev,
2630 select_queue_fallback_t fallback);
2b4aa3ce 2631int dev_queue_xmit(struct sk_buff *skb);
eadec877 2632int dev_queue_xmit_accel(struct sk_buff *skb, struct net_device *sb_dev);
865b03f2 2633int dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
f629d208
JP
2634int register_netdevice(struct net_device *dev);
2635void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2636void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2637static inline void unregister_netdevice(struct net_device *dev)
2638{
2639 unregister_netdevice_queue(dev, NULL);
2640}
2641
f629d208
JP
2642int netdev_refcnt_read(const struct net_device *dev);
2643void free_netdev(struct net_device *dev);
74d332c1 2644void netdev_freemem(struct net_device *dev);
f629d208
JP
2645void synchronize_net(void);
2646int init_dummy_netdev(struct net_device *dev);
937f1ba5 2647
f60e5990 2648DECLARE_PER_CPU(int, xmit_recursion);
a70b506e
DB
2649#define XMIT_RECURSION_LIMIT 10
2650
f60e5990 2651static inline int dev_recursion_level(void)
2652{
2653 return this_cpu_read(xmit_recursion);
2654}
2655
f629d208
JP
2656struct net_device *dev_get_by_index(struct net *net, int ifindex);
2657struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2658struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 2659struct net_device *dev_get_by_napi_id(unsigned int napi_id);
f629d208
JP
2660int netdev_get_name(struct net *net, char *name, int ifindex);
2661int dev_restart(struct net_device *dev);
d4546c25 2662int skb_gro_receive(struct sk_buff *p, struct sk_buff *skb);
86911732
HX
2663
2664static inline unsigned int skb_gro_offset(const struct sk_buff *skb)
2665{
2666 return NAPI_GRO_CB(skb)->data_offset;
2667}
2668
2669static inline unsigned int skb_gro_len(const struct sk_buff *skb)
2670{
2671 return skb->len - NAPI_GRO_CB(skb)->data_offset;
2672}
2673
2674static inline void skb_gro_pull(struct sk_buff *skb, unsigned int len)
2675{
2676 NAPI_GRO_CB(skb)->data_offset += len;
2677}
2678
a5b1cf28
HX
2679static inline void *skb_gro_header_fast(struct sk_buff *skb,
2680 unsigned int offset)
86911732 2681{
a5b1cf28
HX
2682 return NAPI_GRO_CB(skb)->frag0 + offset;
2683}
78a478d0 2684
a5b1cf28
HX
2685static inline int skb_gro_header_hard(struct sk_buff *skb, unsigned int hlen)
2686{
2687 return NAPI_GRO_CB(skb)->frag0_len < hlen;
2688}
78a478d0 2689
57ea52a8
HX
2690static inline void skb_gro_frag0_invalidate(struct sk_buff *skb)
2691{
2692 NAPI_GRO_CB(skb)->frag0 = NULL;
2693 NAPI_GRO_CB(skb)->frag0_len = 0;
2694}
2695
a5b1cf28
HX
2696static inline void *skb_gro_header_slow(struct sk_buff *skb, unsigned int hlen,
2697 unsigned int offset)
2698{
17dd759c
HX
2699 if (!pskb_may_pull(skb, hlen))
2700 return NULL;
2701
57ea52a8 2702 skb_gro_frag0_invalidate(skb);
17dd759c 2703 return skb->data + offset;
86911732 2704}
1da177e4 2705
36e7b1b8
HX
2706static inline void *skb_gro_network_header(struct sk_buff *skb)
2707{
78d3fd0b
HX
2708 return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
2709 skb_network_offset(skb);
36e7b1b8
HX
2710}
2711
bf5a755f
JC
2712static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
2713 const void *start, unsigned int len)
2714{
573e8fca 2715 if (NAPI_GRO_CB(skb)->csum_valid)
bf5a755f
JC
2716 NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
2717 csum_partial(start, len, 0));
2718}
2719
573e8fca
TH
2720/* GRO checksum functions. These are logical equivalents of the normal
2721 * checksum functions (in skbuff.h) except that they operate on the GRO
2722 * offsets and fields in sk_buff.
2723 */
2724
2725__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
2726
15e2396d
TH
2727static inline bool skb_at_gro_remcsum_start(struct sk_buff *skb)
2728{
b7fe10e5 2729 return (NAPI_GRO_CB(skb)->gro_remcsum_start == skb_gro_offset(skb));
15e2396d
TH
2730}
2731
573e8fca
TH
2732static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
2733 bool zero_okay,
2734 __sum16 check)
2735{
6edec0e6
TH
2736 return ((skb->ip_summed != CHECKSUM_PARTIAL ||
2737 skb_checksum_start_offset(skb) <
2738 skb_gro_offset(skb)) &&
15e2396d 2739 !skb_at_gro_remcsum_start(skb) &&
662880f4 2740 NAPI_GRO_CB(skb)->csum_cnt == 0 &&
573e8fca
TH
2741 (!zero_okay || check));
2742}
2743
2744static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
2745 __wsum psum)
2746{
2747 if (NAPI_GRO_CB(skb)->csum_valid &&
2748 !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
2749 return 0;
2750
2751 NAPI_GRO_CB(skb)->csum = psum;
2752
2753 return __skb_gro_checksum_complete(skb);
2754}
2755
573e8fca
TH
2756static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
2757{
662880f4
TH
2758 if (NAPI_GRO_CB(skb)->csum_cnt > 0) {
2759 /* Consume a checksum from CHECKSUM_UNNECESSARY */
2760 NAPI_GRO_CB(skb)->csum_cnt--;
2761 } else {
2762 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
2763 * verified a new top level checksum or an encapsulated one
2764 * during GRO. This saves work if we fallback to normal path.
2765 */
2766 __skb_incr_checksum_unnecessary(skb);
573e8fca
TH
2767 }
2768}
2769
2770#define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
2771 compute_pseudo) \
2772({ \
2773 __sum16 __ret = 0; \
2774 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
2775 __ret = __skb_gro_checksum_validate_complete(skb, \
2776 compute_pseudo(skb, proto)); \
219f1d79 2777 if (!__ret) \
573e8fca
TH
2778 skb_gro_incr_csum_unnecessary(skb); \
2779 __ret; \
2780})
2781
2782#define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
2783 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
2784
2785#define skb_gro_checksum_validate_zero_check(skb, proto, check, \
2786 compute_pseudo) \
2787 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
2788
2789#define skb_gro_checksum_simple_validate(skb) \
2790 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
2791
d96535a1
TH
2792static inline bool __skb_gro_checksum_convert_check(struct sk_buff *skb)
2793{
2794 return (NAPI_GRO_CB(skb)->csum_cnt == 0 &&
2795 !NAPI_GRO_CB(skb)->csum_valid);
2796}
2797
2798static inline void __skb_gro_checksum_convert(struct sk_buff *skb,
2799 __sum16 check, __wsum pseudo)
2800{
2801 NAPI_GRO_CB(skb)->csum = ~pseudo;
2802 NAPI_GRO_CB(skb)->csum_valid = 1;
2803}
2804
2805#define skb_gro_checksum_try_convert(skb, proto, check, compute_pseudo) \
2806do { \
2807 if (__skb_gro_checksum_convert_check(skb)) \
2808 __skb_gro_checksum_convert(skb, check, \
2809 compute_pseudo(skb, proto)); \
2810} while (0)
2811
26c4f7da
TH
2812struct gro_remcsum {
2813 int offset;
2814 __wsum delta;
2815};
2816
2817static inline void skb_gro_remcsum_init(struct gro_remcsum *grc)
2818{
846cd667 2819 grc->offset = 0;
26c4f7da
TH
2820 grc->delta = 0;
2821}
2822
b7fe10e5
TH
2823static inline void *skb_gro_remcsum_process(struct sk_buff *skb, void *ptr,
2824 unsigned int off, size_t hdrlen,
2825 int start, int offset,
2826 struct gro_remcsum *grc,
2827 bool nopartial)
dcdc8994
TH
2828{
2829 __wsum delta;
b7fe10e5 2830 size_t plen = hdrlen + max_t(size_t, offset + sizeof(u16), start);
dcdc8994
TH
2831
2832 BUG_ON(!NAPI_GRO_CB(skb)->csum_valid);
2833
15e2396d 2834 if (!nopartial) {
b7fe10e5
TH
2835 NAPI_GRO_CB(skb)->gro_remcsum_start = off + hdrlen + start;
2836 return ptr;
2837 }
2838
2839 ptr = skb_gro_header_fast(skb, off);
2840 if (skb_gro_header_hard(skb, off + plen)) {
2841 ptr = skb_gro_header_slow(skb, off + plen, off);
2842 if (!ptr)
2843 return NULL;
15e2396d
TH
2844 }
2845
b7fe10e5
TH
2846 delta = remcsum_adjust(ptr + hdrlen, NAPI_GRO_CB(skb)->csum,
2847 start, offset);
dcdc8994
TH
2848
2849 /* Adjust skb->csum since we changed the packet */
dcdc8994 2850 NAPI_GRO_CB(skb)->csum = csum_add(NAPI_GRO_CB(skb)->csum, delta);
26c4f7da 2851
b7fe10e5 2852 grc->offset = off + hdrlen + offset;
26c4f7da 2853 grc->delta = delta;
b7fe10e5
TH
2854
2855 return ptr;
dcdc8994
TH
2856}
2857
26c4f7da
TH
2858static inline void skb_gro_remcsum_cleanup(struct sk_buff *skb,
2859 struct gro_remcsum *grc)
2860{
b7fe10e5
TH
2861 void *ptr;
2862 size_t plen = grc->offset + sizeof(u16);
2863
26c4f7da
TH
2864 if (!grc->delta)
2865 return;
2866
b7fe10e5
TH
2867 ptr = skb_gro_header_fast(skb, grc->offset);
2868 if (skb_gro_header_hard(skb, grc->offset + sizeof(u16))) {
2869 ptr = skb_gro_header_slow(skb, plen, grc->offset);
2870 if (!ptr)
2871 return;
2872 }
2873
2874 remcsum_unadjust((__sum16 *)ptr, grc->delta);
26c4f7da 2875}
dcdc8994 2876
25393d3f 2877#ifdef CONFIG_XFRM_OFFLOAD
d4546c25 2878static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
25393d3f
SK
2879{
2880 if (PTR_ERR(pp) != -EINPROGRESS)
2881 NAPI_GRO_CB(skb)->flush |= flush;
2882}
603d4cf8 2883static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 2884 struct sk_buff *pp,
603d4cf8
SD
2885 int flush,
2886 struct gro_remcsum *grc)
2887{
2888 if (PTR_ERR(pp) != -EINPROGRESS) {
2889 NAPI_GRO_CB(skb)->flush |= flush;
2890 skb_gro_remcsum_cleanup(skb, grc);
2891 skb->remcsum_offload = 0;
2892 }
2893}
25393d3f 2894#else
d4546c25 2895static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
5f114163
SK
2896{
2897 NAPI_GRO_CB(skb)->flush |= flush;
2898}
603d4cf8 2899static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 2900 struct sk_buff *pp,
603d4cf8
SD
2901 int flush,
2902 struct gro_remcsum *grc)
2903{
2904 NAPI_GRO_CB(skb)->flush |= flush;
2905 skb_gro_remcsum_cleanup(skb, grc);
2906 skb->remcsum_offload = 0;
2907}
25393d3f 2908#endif
5f114163 2909
0c4e8581
SH
2910static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
2911 unsigned short type,
3b04ddde 2912 const void *daddr, const void *saddr,
95c96174 2913 unsigned int len)
0c4e8581 2914{
f1ecfd5d 2915 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 2916 return 0;
3b04ddde
SH
2917
2918 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
2919}
2920
b95cce35
SH
2921static inline int dev_parse_header(const struct sk_buff *skb,
2922 unsigned char *haddr)
2923{
2924 const struct net_device *dev = skb->dev;
2925
1b83336b 2926 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 2927 return 0;
3b04ddde 2928 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
2929}
2930
2793a23a
WB
2931/* ll_header must have at least hard_header_len allocated */
2932static inline bool dev_validate_header(const struct net_device *dev,
2933 char *ll_header, int len)
2934{
2935 if (likely(len >= dev->hard_header_len))
2936 return true;
217e6fa2
WB
2937 if (len < dev->min_header_len)
2938 return false;
2793a23a
WB
2939
2940 if (capable(CAP_SYS_RAWIO)) {
2941 memset(ll_header + len, 0, dev->hard_header_len - len);
2942 return true;
2943 }
2944
2945 if (dev->header_ops && dev->header_ops->validate)
2946 return dev->header_ops->validate(ll_header, len);
2947
2948 return false;
2949}
2950
36fd633e
AV
2951typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr,
2952 int len, int size);
f629d208 2953int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
1da177e4
LT
2954static inline int unregister_gifconf(unsigned int family)
2955{
2956 return register_gifconf(family, NULL);
2957}
2958
99bbc707 2959#ifdef CONFIG_NET_FLOW_LIMIT
5f121b9a 2960#define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
99bbc707
WB
2961struct sd_flow_limit {
2962 u64 count;
2963 unsigned int num_buckets;
2964 unsigned int history_head;
2965 u16 history[FLOW_LIMIT_HISTORY];
2966 u8 buckets[];
2967};
2968
2969extern int netdev_flow_limit_table_len;
2970#endif /* CONFIG_NET_FLOW_LIMIT */
2971
1da177e4 2972/*
5e82b4b2 2973 * Incoming packets are placed on per-CPU queues
1da177e4 2974 */
d94d9fee 2975struct softnet_data {
1da177e4 2976 struct list_head poll_list;
6e7676c1 2977 struct sk_buff_head process_queue;
1da177e4 2978
dee42870 2979 /* stats */
cd7b5396
DM
2980 unsigned int processed;
2981 unsigned int time_squeeze;
cd7b5396 2982 unsigned int received_rps;
fd793d89 2983#ifdef CONFIG_RPS
88751275 2984 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
2985#endif
2986#ifdef CONFIG_NET_FLOW_LIMIT
2987 struct sd_flow_limit __rcu *flow_limit;
2988#endif
2989 struct Qdisc *output_queue;
2990 struct Qdisc **output_queue_tailp;
2991 struct sk_buff *completion_queue;
f53c7239
SK
2992#ifdef CONFIG_XFRM_OFFLOAD
2993 struct sk_buff_head xfrm_backlog;
2994#endif
4cdb1e2e 2995#ifdef CONFIG_RPS
501e7ef5
ED
2996 /* input_queue_head should be written by cpu owning this struct,
2997 * and only read by other cpus. Worth using a cache line.
2998 */
2999 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3000
3001 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3002 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3003 struct softnet_data *rps_ipi_next;
3004 unsigned int cpu;
76cc8b13 3005 unsigned int input_queue_tail;
1e94d72f 3006#endif
95c96174 3007 unsigned int dropped;
0a9627f2 3008 struct sk_buff_head input_pkt_queue;
bea3348e 3009 struct napi_struct backlog;
99bbc707 3010
1da177e4
LT
3011};
3012
76cc8b13 3013static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3014{
3015#ifdef CONFIG_RPS
76cc8b13
TH
3016 sd->input_queue_head++;
3017#endif
3018}
3019
3020static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3021 unsigned int *qtail)
3022{
3023#ifdef CONFIG_RPS
3024 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3025#endif
3026}
3027
0a9627f2 3028DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3029
f629d208 3030void __netif_schedule(struct Qdisc *q);
46e5da40 3031void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3032
fd2ea0a7
DM
3033static inline void netif_tx_schedule_all(struct net_device *dev)
3034{
3035 unsigned int i;
3036
3037 for (i = 0; i < dev->num_tx_queues; i++)
3038 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3039}
3040
f9a7cbbf 3041static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3042{
73466498 3043 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3044}
3045
bea3348e
SH
3046/**
3047 * netif_start_queue - allow transmit
3048 * @dev: network device
3049 *
3050 * Allow upper layers to call the device hard_start_xmit routine.
3051 */
1da177e4
LT
3052static inline void netif_start_queue(struct net_device *dev)
3053{
e8a0464c 3054 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3055}
3056
fd2ea0a7
DM
3057static inline void netif_tx_start_all_queues(struct net_device *dev)
3058{
3059 unsigned int i;
3060
3061 for (i = 0; i < dev->num_tx_queues; i++) {
3062 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3063 netif_tx_start_queue(txq);
3064 }
3065}
3066
46e5da40 3067void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3068
d29f749e
DJ
3069/**
3070 * netif_wake_queue - restart transmit
3071 * @dev: network device
3072 *
3073 * Allow upper layers to call the device hard_start_xmit routine.
3074 * Used for flow control when transmit resources are available.
3075 */
79d16385
DM
3076static inline void netif_wake_queue(struct net_device *dev)
3077{
e8a0464c 3078 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3079}
3080
fd2ea0a7
DM
3081static inline void netif_tx_wake_all_queues(struct net_device *dev)
3082{
3083 unsigned int i;
3084
3085 for (i = 0; i < dev->num_tx_queues; i++) {
3086 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3087 netif_tx_wake_queue(txq);
3088 }
3089}
3090
f9a7cbbf 3091static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3092{
73466498 3093 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3094}
3095
bea3348e
SH
3096/**
3097 * netif_stop_queue - stop transmitted packets
3098 * @dev: network device
3099 *
3100 * Stop upper layers calling the device hard_start_xmit routine.
3101 * Used for flow control when transmit resources are unavailable.
3102 */
1da177e4
LT
3103static inline void netif_stop_queue(struct net_device *dev)
3104{
e8a0464c 3105 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3106}
3107
a2029240 3108void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 3109
4d29515f 3110static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3111{
73466498 3112 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3113}
3114
bea3348e
SH
3115/**
3116 * netif_queue_stopped - test if transmit queue is flowblocked
3117 * @dev: network device
3118 *
3119 * Test if transmit queue on device is currently unable to send.
3120 */
4d29515f 3121static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3122{
e8a0464c 3123 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3124}
3125
4d29515f 3126static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3127{
73466498
TH
3128 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3129}
3130
8e2f1a63
DB
3131static inline bool
3132netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3133{
3134 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3135}
3136
8e2f1a63
DB
3137static inline bool
3138netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3139{
3140 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3141}
3142
53511453
ED
3143/**
3144 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3145 * @dev_queue: pointer to transmit queue
3146 *
3147 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3148 * to give appropriate hint to the CPU.
53511453
ED
3149 */
3150static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3151{
3152#ifdef CONFIG_BQL
3153 prefetchw(&dev_queue->dql.num_queued);
3154#endif
3155}
3156
3157/**
3158 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3159 * @dev_queue: pointer to transmit queue
3160 *
3161 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3162 * to give appropriate hint to the CPU.
53511453
ED
3163 */
3164static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3165{
3166#ifdef CONFIG_BQL
3167 prefetchw(&dev_queue->dql.limit);
3168#endif
3169}
3170
c5d67bd7
TH
3171static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3172 unsigned int bytes)
3173{
114cf580
TH
3174#ifdef CONFIG_BQL
3175 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3176
3177 if (likely(dql_avail(&dev_queue->dql) >= 0))
3178 return;
3179
3180 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3181
3182 /*
3183 * The XOFF flag must be set before checking the dql_avail below,
3184 * because in netdev_tx_completed_queue we update the dql_completed
3185 * before checking the XOFF flag.
3186 */
3187 smp_mb();
3188
3189 /* check again in case another CPU has just made room avail */
3190 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3191 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3192#endif
c5d67bd7
TH
3193}
3194
3e59020a
ED
3195/* Variant of netdev_tx_sent_queue() for drivers that are aware
3196 * that they should not test BQL status themselves.
3197 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3198 * skb of a batch.
3199 * Returns true if the doorbell must be used to kick the NIC.
3200 */
3201static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3202 unsigned int bytes,
3203 bool xmit_more)
3204{
3205 if (xmit_more) {
3206#ifdef CONFIG_BQL
3207 dql_queued(&dev_queue->dql, bytes);
3208#endif
3209 return netif_tx_queue_stopped(dev_queue);
3210 }
3211 netdev_tx_sent_queue(dev_queue, bytes);
3212 return true;
3213}
3214
0042d0c8
FF
3215/**
3216 * netdev_sent_queue - report the number of bytes queued to hardware
3217 * @dev: network device
3218 * @bytes: number of bytes queued to the hardware device queue
3219 *
3220 * Report the number of bytes queued for sending/completion to the network
3221 * device hardware queue. @bytes should be a good approximation and should
3222 * exactly match netdev_completed_queue() @bytes
3223 */
c5d67bd7
TH
3224static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3225{
3226 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3227}
3228
3229static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3230 unsigned int pkts, unsigned int bytes)
c5d67bd7 3231{
114cf580 3232#ifdef CONFIG_BQL
b37c0fbe
AD
3233 if (unlikely(!bytes))
3234 return;
3235
3236 dql_completed(&dev_queue->dql, bytes);
3237
3238 /*
3239 * Without the memory barrier there is a small possiblity that
3240 * netdev_tx_sent_queue will miss the update and cause the queue to
3241 * be stopped forever
3242 */
3243 smp_mb();
3244
3245 if (dql_avail(&dev_queue->dql) < 0)
3246 return;
3247
3248 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3249 netif_schedule_queue(dev_queue);
114cf580 3250#endif
c5d67bd7
TH
3251}
3252
0042d0c8
FF
3253/**
3254 * netdev_completed_queue - report bytes and packets completed by device
3255 * @dev: network device
3256 * @pkts: actual number of packets sent over the medium
3257 * @bytes: actual number of bytes sent over the medium
3258 *
3259 * Report the number of bytes and packets transmitted by the network device
3260 * hardware queue over the physical medium, @bytes must exactly match the
3261 * @bytes amount passed to netdev_sent_queue()
3262 */
c5d67bd7 3263static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3264 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3265{
3266 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3267}
3268
3269static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3270{
114cf580 3271#ifdef CONFIG_BQL
5c490354 3272 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3273 dql_reset(&q->dql);
3274#endif
c5d67bd7
TH
3275}
3276
0042d0c8
FF
3277/**
3278 * netdev_reset_queue - reset the packets and bytes count of a network device
3279 * @dev_queue: network device
3280 *
3281 * Reset the bytes and packet count of a network device and clear the
3282 * software flow control OFF bit for this network device
3283 */
c5d67bd7
TH
3284static inline void netdev_reset_queue(struct net_device *dev_queue)
3285{
3286 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3287}
3288
b9507bda
DB
3289/**
3290 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3291 * @dev: network device
3292 * @queue_index: given tx queue index
3293 *
3294 * Returns 0 if given tx queue index >= number of device tx queues,
3295 * otherwise returns the originally passed tx queue index.
3296 */
3297static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3298{
3299 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3300 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3301 dev->name, queue_index,
3302 dev->real_num_tx_queues);
3303 return 0;
3304 }
3305
3306 return queue_index;
3307}
3308
bea3348e
SH
3309/**
3310 * netif_running - test if up
3311 * @dev: network device
3312 *
3313 * Test if the device has been brought up.
3314 */
4d29515f 3315static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3316{
3317 return test_bit(__LINK_STATE_START, &dev->state);
3318}
3319
f25f4e44 3320/*
5e82b4b2 3321 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3322 * stop, and a check if it's stopped. All other device management is
3323 * done at the overall netdevice level.
3324 * Also test the device if we're multiqueue.
3325 */
bea3348e
SH
3326
3327/**
3328 * netif_start_subqueue - allow sending packets on subqueue
3329 * @dev: network device
3330 * @queue_index: sub queue index
3331 *
3332 * Start individual transmit queue of a device with multiple transmit queues.
3333 */
f25f4e44
PWJ
3334static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3335{
fd2ea0a7 3336 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3337
3338 netif_tx_start_queue(txq);
f25f4e44
PWJ
3339}
3340
bea3348e
SH
3341/**
3342 * netif_stop_subqueue - stop sending packets on subqueue
3343 * @dev: network device
3344 * @queue_index: sub queue index
3345 *
3346 * Stop individual transmit queue of a device with multiple transmit queues.
3347 */
f25f4e44
PWJ
3348static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3349{
fd2ea0a7 3350 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3351 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3352}
3353
bea3348e
SH
3354/**
3355 * netif_subqueue_stopped - test status of subqueue
3356 * @dev: network device
3357 * @queue_index: sub queue index
3358 *
3359 * Check individual transmit queue of a device with multiple transmit queues.
3360 */
4d29515f
DM
3361static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3362 u16 queue_index)
f25f4e44 3363{
fd2ea0a7 3364 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3365
3366 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3367}
3368
4d29515f
DM
3369static inline bool netif_subqueue_stopped(const struct net_device *dev,
3370 struct sk_buff *skb)
668f895a
PE
3371{
3372 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3373}
bea3348e 3374
738b35cc
FF
3375/**
3376 * netif_wake_subqueue - allow sending packets on subqueue
3377 * @dev: network device
3378 * @queue_index: sub queue index
3379 *
3380 * Resume individual transmit queue of a device with multiple transmit queues.
3381 */
3382static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3383{
3384 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3385
3386 netif_tx_wake_queue(txq);
3387}
f25f4e44 3388
537c00de 3389#ifdef CONFIG_XPS
53af53ae 3390int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3391 u16 index);
80d19669
AN
3392int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
3393 u16 index, bool is_rxqs_map);
3394
3395/**
3396 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3397 * @j: CPU/Rx queue index
3398 * @mask: bitmask of all cpus/rx queues
3399 * @nr_bits: number of bits in the bitmask
3400 *
3401 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3402 */
3403static inline bool netif_attr_test_mask(unsigned long j,
3404 const unsigned long *mask,
3405 unsigned int nr_bits)
3406{
3407 cpu_max_bits_warn(j, nr_bits);
3408 return test_bit(j, mask);
3409}
3410
3411/**
3412 * netif_attr_test_online - Test for online CPU/Rx queue
3413 * @j: CPU/Rx queue index
3414 * @online_mask: bitmask for CPUs/Rx queues that are online
3415 * @nr_bits: number of bits in the bitmask
3416 *
3417 * Returns true if a CPU/Rx queue is online.
3418 */
3419static inline bool netif_attr_test_online(unsigned long j,
3420 const unsigned long *online_mask,
3421 unsigned int nr_bits)
3422{
3423 cpu_max_bits_warn(j, nr_bits);
3424
3425 if (online_mask)
3426 return test_bit(j, online_mask);
3427
3428 return (j < nr_bits);
3429}
3430
3431/**
3432 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3433 * @n: CPU/Rx queue index
3434 * @srcp: the cpumask/Rx queue mask pointer
3435 * @nr_bits: number of bits in the bitmask
3436 *
3437 * Returns >= nr_bits if no further CPUs/Rx queues set.
3438 */
3439static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3440 unsigned int nr_bits)
3441{
3442 /* -1 is a legal arg here. */
3443 if (n != -1)
3444 cpu_max_bits_warn(n, nr_bits);
3445
3446 if (srcp)
3447 return find_next_bit(srcp, nr_bits, n + 1);
3448
3449 return n + 1;
3450}
3451
3452/**
3453 * netif_attrmask_next_and - get the next CPU/Rx queue in *src1p & *src2p
3454 * @n: CPU/Rx queue index
3455 * @src1p: the first CPUs/Rx queues mask pointer
3456 * @src2p: the second CPUs/Rx queues mask pointer
3457 * @nr_bits: number of bits in the bitmask
3458 *
3459 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3460 */
3461static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3462 const unsigned long *src2p,
3463 unsigned int nr_bits)
3464{
3465 /* -1 is a legal arg here. */
3466 if (n != -1)
3467 cpu_max_bits_warn(n, nr_bits);
3468
3469 if (src1p && src2p)
3470 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3471 else if (src1p)
3472 return find_next_bit(src1p, nr_bits, n + 1);
3473 else if (src2p)
3474 return find_next_bit(src2p, nr_bits, n + 1);
3475
3476 return n + 1;
3477}
537c00de
AD
3478#else
3479static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3480 const struct cpumask *mask,
537c00de
AD
3481 u16 index)
3482{
3483 return 0;
3484}
c9fbb2d2
KK
3485
3486static inline int __netif_set_xps_queue(struct net_device *dev,
3487 const unsigned long *mask,
3488 u16 index, bool is_rxqs_map)
3489{
3490 return 0;
3491}
537c00de
AD
3492#endif
3493
bea3348e
SH
3494/**
3495 * netif_is_multiqueue - test if device has multiple transmit queues
3496 * @dev: network device
3497 *
3498 * Check if device has multiple transmit queues
bea3348e 3499 */
4d29515f 3500static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3501{
a02cec21 3502 return dev->num_tx_queues > 1;
f25f4e44 3503}
1da177e4 3504
f629d208 3505int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3506
a953be53 3507#ifdef CONFIG_SYSFS
f629d208 3508int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3509#else
3510static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3511 unsigned int rxqs)
62fe0b40 3512{
c29c2ebd 3513 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3514 return 0;
3515}
3516#endif
3517
65073a67
DB
3518static inline struct netdev_rx_queue *
3519__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
3520{
3521 return dev->_rx + rxq;
3522}
3523
a953be53
MD
3524#ifdef CONFIG_SYSFS
3525static inline unsigned int get_netdev_rx_queue_index(
3526 struct netdev_rx_queue *queue)
3527{
3528 struct net_device *dev = queue->dev;
3529 int index = queue - dev->_rx;
3530
3531 BUG_ON(index >= dev->num_rx_queues);
3532 return index;
3533}
3534#endif
3535
16917b87 3536#define DEFAULT_MAX_NUM_RSS_QUEUES (8)
f629d208 3537int netif_get_num_default_rss_queues(void);
16917b87 3538
e6247027
ED
3539enum skb_free_reason {
3540 SKB_REASON_CONSUMED,
3541 SKB_REASON_DROPPED,
3542};
3543
3544void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3545void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3546
e6247027
ED
3547/*
3548 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3549 * interrupt context or with hardware interrupts being disabled.
3550 * (in_irq() || irqs_disabled())
3551 *
3552 * We provide four helpers that can be used in following contexts :
3553 *
3554 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3555 * replacing kfree_skb(skb)
3556 *
3557 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3558 * Typically used in place of consume_skb(skb) in TX completion path
3559 *
3560 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3561 * replacing kfree_skb(skb)
3562 *
3563 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3564 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3565 */
e6247027
ED
3566static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3567{
3568 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3569}
3570
3571static inline void dev_consume_skb_irq(struct sk_buff *skb)
3572{
3573 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3574}
3575
3576static inline void dev_kfree_skb_any(struct sk_buff *skb)
3577{
3578 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3579}
3580
3581static inline void dev_consume_skb_any(struct sk_buff *skb)
3582{
3583 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3584}
1da177e4 3585
7c497478
JW
3586void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3587int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208
JP
3588int netif_rx(struct sk_buff *skb);
3589int netif_rx_ni(struct sk_buff *skb);
04eb4489 3590int netif_receive_skb(struct sk_buff *skb);
1c601d82 3591int netif_receive_skb_core(struct sk_buff *skb);
f6ad8c1b 3592void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3593gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3594void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3595struct sk_buff *napi_get_frags(struct napi_struct *napi);
3596gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3597struct packet_offload *gro_find_receive_by_type(__be16 type);
3598struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3599
3600static inline void napi_free_frags(struct napi_struct *napi)
3601{
3602 kfree_skb(napi->skb);
3603 napi->skb = NULL;
3604}
3605
24b27fc4 3606bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
3607int netdev_rx_handler_register(struct net_device *dev,
3608 rx_handler_func_t *rx_handler,
3609 void *rx_handler_data);
3610void netdev_rx_handler_unregister(struct net_device *dev);
3611
3612bool dev_valid_name(const char *name);
44c02a2c
AV
3613int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
3614 bool *need_copyout);
36fd633e 3615int dev_ifconf(struct net *net, struct ifconf *, int);
f629d208
JP
3616int dev_ethtool(struct net *net, struct ifreq *);
3617unsigned int dev_get_flags(const struct net_device *);
3618int __dev_change_flags(struct net_device *, unsigned int flags);
3619int dev_change_flags(struct net_device *, unsigned int);
cb178190
DM
3620void __dev_notify_flags(struct net_device *, unsigned int old_flags,
3621 unsigned int gchanges);
f629d208
JP
3622int dev_change_name(struct net_device *, const char *);
3623int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 3624int dev_get_alias(const struct net_device *, char *, size_t);
f629d208 3625int dev_change_net_namespace(struct net_device *, struct net *, const char *);
f51048c3 3626int __dev_set_mtu(struct net_device *, int);
7a4c53be
SH
3627int dev_set_mtu_ext(struct net_device *dev, int mtu,
3628 struct netlink_ext_ack *extack);
f629d208 3629int dev_set_mtu(struct net_device *, int);
6a643ddb 3630int dev_change_tx_queue_len(struct net_device *, unsigned long);
f629d208
JP
3631void dev_set_group(struct net_device *, int);
3632int dev_set_mac_address(struct net_device *, struct sockaddr *);
3633int dev_change_carrier(struct net_device *, bool new_carrier);
3634int dev_get_phys_port_id(struct net_device *dev,
02637fce 3635 struct netdev_phys_item_id *ppid);
db24a904
DA
3636int dev_get_phys_port_name(struct net_device *dev,
3637 char *name, size_t len);
d746d707 3638int dev_change_proto_down(struct net_device *dev, bool proto_down);
f53c7239 3639struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
3640struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3641 struct netdev_queue *txq, int *ret);
d67b9cd2 3642
f4e63525 3643typedef int (*bpf_op_t)(struct net_device *dev, struct netdev_bpf *bpf);
d67b9cd2
DB
3644int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack,
3645 int fd, u32 flags);
a25717d2
JK
3646u32 __dev_xdp_query(struct net_device *dev, bpf_op_t xdp_op,
3647 enum bpf_netdev_command cmd);
84c6b868 3648int xdp_umem_query(struct net_device *dev, u16 queue_id);
d67b9cd2 3649
a0265d28 3650int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 3651int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
3652bool is_skb_forwardable(const struct net_device *dev,
3653 const struct sk_buff *skb);
1da177e4 3654
4e3264d2
MKL
3655static __always_inline int ____dev_forward_skb(struct net_device *dev,
3656 struct sk_buff *skb)
3657{
3658 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
3659 unlikely(!is_skb_forwardable(dev, skb))) {
3660 atomic_long_inc(&dev->rx_dropped);
3661 kfree_skb(skb);
3662 return NET_RX_DROP;
3663 }
3664
3665 skb_scrub_packet(skb, true);
3666 skb->priority = 0;
3667 return 0;
3668}
3669
9f9a742d 3670bool dev_nit_active(struct net_device *dev);
74b20582
DA
3671void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
3672
20380731 3673extern int netdev_budget;
7acf8a1e 3674extern unsigned int netdev_budget_usecs;
1da177e4
LT
3675
3676/* Called by rtnetlink.c:rtnl_unlock() */
f629d208 3677void netdev_run_todo(void);
1da177e4 3678
bea3348e
SH
3679/**
3680 * dev_put - release reference to device
3681 * @dev: network device
3682 *
9ef4429b 3683 * Release reference to device to allow it to be freed.
bea3348e 3684 */
1da177e4
LT
3685static inline void dev_put(struct net_device *dev)
3686{
933393f5 3687 this_cpu_dec(*dev->pcpu_refcnt);
1da177e4
LT
3688}
3689
bea3348e
SH
3690/**
3691 * dev_hold - get reference to device
3692 * @dev: network device
3693 *
9ef4429b 3694 * Hold reference to device to keep it from being freed.
bea3348e 3695 */
15333061
SH
3696static inline void dev_hold(struct net_device *dev)
3697{
933393f5 3698 this_cpu_inc(*dev->pcpu_refcnt);
15333061 3699}
1da177e4
LT
3700
3701/* Carrier loss detection, dial on demand. The functions netif_carrier_on
3702 * and _off may be called from IRQ context, but it is caller
3703 * who is responsible for serialization of these calls.
b00055aa
SR
3704 *
3705 * The name carrier is inappropriate, these functions should really be
3706 * called netif_lowerlayer_*() because they represent the state of any
3707 * kind of lower layer not just hardware media.
1da177e4
LT
3708 */
3709
f629d208
JP
3710void linkwatch_init_dev(struct net_device *dev);
3711void linkwatch_fire_event(struct net_device *dev);
3712void linkwatch_forget_dev(struct net_device *dev);
1da177e4 3713
bea3348e
SH
3714/**
3715 * netif_carrier_ok - test if carrier present
3716 * @dev: network device
3717 *
3718 * Check if carrier is present on device
3719 */
4d29515f 3720static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
3721{
3722 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
3723}
3724
f629d208 3725unsigned long dev_trans_start(struct net_device *dev);
9d21493b 3726
f629d208 3727void __netdev_watchdog_up(struct net_device *dev);
1da177e4 3728
f629d208 3729void netif_carrier_on(struct net_device *dev);
1da177e4 3730
f629d208 3731void netif_carrier_off(struct net_device *dev);
1da177e4 3732
bea3348e
SH
3733/**
3734 * netif_dormant_on - mark device as dormant.
3735 * @dev: network device
3736 *
3737 * Mark device as dormant (as per RFC2863).
3738 *
3739 * The dormant state indicates that the relevant interface is not
3740 * actually in a condition to pass packets (i.e., it is not 'up') but is
3741 * in a "pending" state, waiting for some external event. For "on-
3742 * demand" interfaces, this new state identifies the situation where the
3743 * interface is waiting for events to place it in the up state.
bea3348e 3744 */
b00055aa
SR
3745static inline void netif_dormant_on(struct net_device *dev)
3746{
3747 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
3748 linkwatch_fire_event(dev);
3749}
3750
bea3348e
SH
3751/**
3752 * netif_dormant_off - set device as not dormant.
3753 * @dev: network device
3754 *
3755 * Device is not in dormant state.
3756 */
b00055aa
SR
3757static inline void netif_dormant_off(struct net_device *dev)
3758{
3759 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
3760 linkwatch_fire_event(dev);
3761}
3762
bea3348e 3763/**
8ecbc40a 3764 * netif_dormant - test if device is dormant
bea3348e
SH
3765 * @dev: network device
3766 *
8ecbc40a 3767 * Check if device is dormant.
bea3348e 3768 */
4d29515f 3769static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
3770{
3771 return test_bit(__LINK_STATE_DORMANT, &dev->state);
3772}
3773
3774
bea3348e
SH
3775/**
3776 * netif_oper_up - test if device is operational
3777 * @dev: network device
3778 *
3779 * Check if carrier is operational
3780 */
4d29515f 3781static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 3782{
b00055aa
SR
3783 return (dev->operstate == IF_OPER_UP ||
3784 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
3785}
3786
bea3348e
SH
3787/**
3788 * netif_device_present - is device available or removed
3789 * @dev: network device
3790 *
3791 * Check if device has not been removed from system.
3792 */
4d29515f 3793static inline bool netif_device_present(struct net_device *dev)
1da177e4
LT
3794{
3795 return test_bit(__LINK_STATE_PRESENT, &dev->state);
3796}
3797
f629d208 3798void netif_device_detach(struct net_device *dev);
1da177e4 3799
f629d208 3800void netif_device_attach(struct net_device *dev);
1da177e4
LT
3801
3802/*
3803 * Network interface message level settings
3804 */
1da177e4
LT
3805
3806enum {
3807 NETIF_MSG_DRV = 0x0001,
3808 NETIF_MSG_PROBE = 0x0002,
3809 NETIF_MSG_LINK = 0x0004,
3810 NETIF_MSG_TIMER = 0x0008,
3811 NETIF_MSG_IFDOWN = 0x0010,
3812 NETIF_MSG_IFUP = 0x0020,
3813 NETIF_MSG_RX_ERR = 0x0040,
3814 NETIF_MSG_TX_ERR = 0x0080,
3815 NETIF_MSG_TX_QUEUED = 0x0100,
3816 NETIF_MSG_INTR = 0x0200,
3817 NETIF_MSG_TX_DONE = 0x0400,
3818 NETIF_MSG_RX_STATUS = 0x0800,
3819 NETIF_MSG_PKTDATA = 0x1000,
3820 NETIF_MSG_HW = 0x2000,
3821 NETIF_MSG_WOL = 0x4000,
3822};
3823
3824#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
3825#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
3826#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
3827#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
3828#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
3829#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
3830#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
3831#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
3832#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
3833#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
3834#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
3835#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
3836#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
3837#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
3838#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
3839
3840static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
3841{
3842 /* use default */
3843 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
3844 return default_msg_enable_bits;
3845 if (debug_value == 0) /* no output */
3846 return 0;
3847 /* set low N bits */
3848 return (1 << debug_value) - 1;
3849}
3850
c773e847 3851static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 3852{
c773e847
DM
3853 spin_lock(&txq->_xmit_lock);
3854 txq->xmit_lock_owner = cpu;
22dd7495
JHS
3855}
3856
5a717f4f
MT
3857static inline bool __netif_tx_acquire(struct netdev_queue *txq)
3858{
3859 __acquire(&txq->_xmit_lock);
3860 return true;
3861}
3862
3863static inline void __netif_tx_release(struct netdev_queue *txq)
3864{
3865 __release(&txq->_xmit_lock);
3866}
3867
fd2ea0a7
DM
3868static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
3869{
3870 spin_lock_bh(&txq->_xmit_lock);
3871 txq->xmit_lock_owner = smp_processor_id();
3872}
3873
4d29515f 3874static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 3875{
4d29515f 3876 bool ok = spin_trylock(&txq->_xmit_lock);
c3f26a26
DM
3877 if (likely(ok))
3878 txq->xmit_lock_owner = smp_processor_id();
3879 return ok;
3880}
3881
3882static inline void __netif_tx_unlock(struct netdev_queue *txq)
3883{
3884 txq->xmit_lock_owner = -1;
3885 spin_unlock(&txq->_xmit_lock);
3886}
3887
3888static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
3889{
3890 txq->xmit_lock_owner = -1;
3891 spin_unlock_bh(&txq->_xmit_lock);
3892}
3893
08baf561
ED
3894static inline void txq_trans_update(struct netdev_queue *txq)
3895{
3896 if (txq->xmit_lock_owner != -1)
3897 txq->trans_start = jiffies;
3898}
3899
ba162f8e
FW
3900/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
3901static inline void netif_trans_update(struct net_device *dev)
3902{
9b36627a
FW
3903 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
3904
3905 if (txq->trans_start != jiffies)
3906 txq->trans_start = jiffies;
ba162f8e
FW
3907}
3908
d29f749e
DJ
3909/**
3910 * netif_tx_lock - grab network device transmit lock
3911 * @dev: network device
d29f749e
DJ
3912 *
3913 * Get network device transmit lock
3914 */
22dd7495
JHS
3915static inline void netif_tx_lock(struct net_device *dev)
3916{
e8a0464c 3917 unsigned int i;
c3f26a26 3918 int cpu;
c773e847 3919
c3f26a26
DM
3920 spin_lock(&dev->tx_global_lock);
3921 cpu = smp_processor_id();
e8a0464c
DM
3922 for (i = 0; i < dev->num_tx_queues; i++) {
3923 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
3924
3925 /* We are the only thread of execution doing a
3926 * freeze, but we have to grab the _xmit_lock in
3927 * order to synchronize with threads which are in
3928 * the ->hard_start_xmit() handler and already
3929 * checked the frozen bit.
3930 */
e8a0464c 3931 __netif_tx_lock(txq, cpu);
c3f26a26
DM
3932 set_bit(__QUEUE_STATE_FROZEN, &txq->state);
3933 __netif_tx_unlock(txq);
e8a0464c 3934 }
932ff279
HX
3935}
3936
3937static inline void netif_tx_lock_bh(struct net_device *dev)
3938{
e8a0464c
DM
3939 local_bh_disable();
3940 netif_tx_lock(dev);
932ff279
HX
3941}
3942
932ff279
HX
3943static inline void netif_tx_unlock(struct net_device *dev)
3944{
e8a0464c
DM
3945 unsigned int i;
3946
3947 for (i = 0; i < dev->num_tx_queues; i++) {
3948 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c773e847 3949
c3f26a26
DM
3950 /* No need to grab the _xmit_lock here. If the
3951 * queue is not stopped for another reason, we
3952 * force a schedule.
3953 */
3954 clear_bit(__QUEUE_STATE_FROZEN, &txq->state);
7b3d3e4f 3955 netif_schedule_queue(txq);
c3f26a26
DM
3956 }
3957 spin_unlock(&dev->tx_global_lock);
932ff279
HX
3958}
3959
3960static inline void netif_tx_unlock_bh(struct net_device *dev)
3961{
e8a0464c
DM
3962 netif_tx_unlock(dev);
3963 local_bh_enable();
932ff279
HX
3964}
3965
c773e847 3966#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 3967 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 3968 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
3969 } else { \
3970 __netif_tx_acquire(txq); \
22dd7495
JHS
3971 } \
3972}
3973
5efeac44
EB
3974#define HARD_TX_TRYLOCK(dev, txq) \
3975 (((dev->features & NETIF_F_LLTX) == 0) ? \
3976 __netif_tx_trylock(txq) : \
5a717f4f 3977 __netif_tx_acquire(txq))
5efeac44 3978
c773e847 3979#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 3980 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 3981 __netif_tx_unlock(txq); \
5a717f4f
MT
3982 } else { \
3983 __netif_tx_release(txq); \
22dd7495
JHS
3984 } \
3985}
3986
1da177e4
LT
3987static inline void netif_tx_disable(struct net_device *dev)
3988{
fd2ea0a7 3989 unsigned int i;
c3f26a26 3990 int cpu;
fd2ea0a7 3991
c3f26a26
DM
3992 local_bh_disable();
3993 cpu = smp_processor_id();
fd2ea0a7
DM
3994 for (i = 0; i < dev->num_tx_queues; i++) {
3995 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
3996
3997 __netif_tx_lock(txq, cpu);
fd2ea0a7 3998 netif_tx_stop_queue(txq);
c3f26a26 3999 __netif_tx_unlock(txq);
fd2ea0a7 4000 }
c3f26a26 4001 local_bh_enable();
1da177e4
LT
4002}
4003
e308a5d8
DM
4004static inline void netif_addr_lock(struct net_device *dev)
4005{
4006 spin_lock(&dev->addr_list_lock);
4007}
4008
2429f7ac
JP
4009static inline void netif_addr_lock_nested(struct net_device *dev)
4010{
25175ba5
VY
4011 int subclass = SINGLE_DEPTH_NESTING;
4012
4013 if (dev->netdev_ops->ndo_get_lock_subclass)
4014 subclass = dev->netdev_ops->ndo_get_lock_subclass(dev);
4015
4016 spin_lock_nested(&dev->addr_list_lock, subclass);
2429f7ac
JP
4017}
4018
e308a5d8
DM
4019static inline void netif_addr_lock_bh(struct net_device *dev)
4020{
4021 spin_lock_bh(&dev->addr_list_lock);
4022}
4023
4024static inline void netif_addr_unlock(struct net_device *dev)
4025{
4026 spin_unlock(&dev->addr_list_lock);
4027}
4028
4029static inline void netif_addr_unlock_bh(struct net_device *dev)
4030{
4031 spin_unlock_bh(&dev->addr_list_lock);
4032}
4033
f001fde5 4034/*
31278e71 4035 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4036 * rcu_read_lock held.
4037 */
4038#define for_each_dev_addr(dev, ha) \
31278e71 4039 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4040
1da177e4
LT
4041/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4042
f629d208 4043void ether_setup(struct net_device *dev);
1da177e4
LT
4044
4045/* Support for loadable net-drivers */
f629d208 4046struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4047 unsigned char name_assign_type,
f629d208
JP
4048 void (*setup)(struct net_device *),
4049 unsigned int txqs, unsigned int rxqs);
0ad646c8
CW
4050int dev_get_valid_name(struct net *net, struct net_device *dev,
4051 const char *name);
4052
c835a677
TG
4053#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4054 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4055
c835a677
TG
4056#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4057 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4058 count)
36909ea4 4059
f629d208
JP
4060int register_netdev(struct net_device *dev);
4061void unregister_netdev(struct net_device *dev);
f001fde5 4062
22bedad3 4063/* General hardware address lists handling functions */
f629d208
JP
4064int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4065 struct netdev_hw_addr_list *from_list, int addr_len);
4066void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4067 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4068int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4069 struct net_device *dev,
4070 int (*sync)(struct net_device *, const unsigned char *),
4071 int (*unsync)(struct net_device *,
4072 const unsigned char *));
e7946760
IK
4073int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4074 struct net_device *dev,
4075 int (*sync)(struct net_device *,
4076 const unsigned char *, int),
4077 int (*unsync)(struct net_device *,
4078 const unsigned char *, int));
4079void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4080 struct net_device *dev,
4081 int (*unsync)(struct net_device *,
4082 const unsigned char *, int));
670e5b8e
AD
4083void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4084 struct net_device *dev,
4085 int (*unsync)(struct net_device *,
4086 const unsigned char *));
f629d208 4087void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4088
f001fde5 4089/* Functions used for device addresses handling */
f629d208
JP
4090int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4091 unsigned char addr_type);
4092int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4093 unsigned char addr_type);
f629d208
JP
4094void dev_addr_flush(struct net_device *dev);
4095int dev_addr_init(struct net_device *dev);
a748ee24
JP
4096
4097/* Functions used for unicast addresses handling */
f629d208
JP
4098int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4099int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4100int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4101int dev_uc_sync(struct net_device *to, struct net_device *from);
4102int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4103void dev_uc_unsync(struct net_device *to, struct net_device *from);
4104void dev_uc_flush(struct net_device *dev);
4105void dev_uc_init(struct net_device *dev);
f001fde5 4106
670e5b8e
AD
4107/**
4108 * __dev_uc_sync - Synchonize device's unicast list
4109 * @dev: device to sync
4110 * @sync: function to call if address should be added
4111 * @unsync: function to call if address should be removed
4112 *
4113 * Add newly added addresses to the interface, and release
4114 * addresses that have been deleted.
5e82b4b2 4115 */
670e5b8e
AD
4116static inline int __dev_uc_sync(struct net_device *dev,
4117 int (*sync)(struct net_device *,
4118 const unsigned char *),
4119 int (*unsync)(struct net_device *,
4120 const unsigned char *))
4121{
4122 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4123}
4124
4125/**
e793c0f7 4126 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4127 * @dev: device to sync
4128 * @unsync: function to call if address should be removed
4129 *
4130 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4131 */
670e5b8e
AD
4132static inline void __dev_uc_unsync(struct net_device *dev,
4133 int (*unsync)(struct net_device *,
4134 const unsigned char *))
4135{
4136 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4137}
4138
22bedad3 4139/* Functions used for multicast addresses handling */
f629d208
JP
4140int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4141int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4142int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4143int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4144int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4145int dev_mc_sync(struct net_device *to, struct net_device *from);
4146int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4147void dev_mc_unsync(struct net_device *to, struct net_device *from);
4148void dev_mc_flush(struct net_device *dev);
4149void dev_mc_init(struct net_device *dev);
f001fde5 4150
670e5b8e
AD
4151/**
4152 * __dev_mc_sync - Synchonize device's multicast list
4153 * @dev: device to sync
4154 * @sync: function to call if address should be added
4155 * @unsync: function to call if address should be removed
4156 *
4157 * Add newly added addresses to the interface, and release
4158 * addresses that have been deleted.
5e82b4b2 4159 */
670e5b8e
AD
4160static inline int __dev_mc_sync(struct net_device *dev,
4161 int (*sync)(struct net_device *,
4162 const unsigned char *),
4163 int (*unsync)(struct net_device *,
4164 const unsigned char *))
4165{
4166 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4167}
4168
4169/**
e793c0f7 4170 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4171 * @dev: device to sync
4172 * @unsync: function to call if address should be removed
4173 *
4174 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4175 */
670e5b8e
AD
4176static inline void __dev_mc_unsync(struct net_device *dev,
4177 int (*unsync)(struct net_device *,
4178 const unsigned char *))
4179{
4180 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4181}
4182
4417da66 4183/* Functions used for secondary unicast and multicast support */
f629d208
JP
4184void dev_set_rx_mode(struct net_device *dev);
4185void __dev_set_rx_mode(struct net_device *dev);
4186int dev_set_promiscuity(struct net_device *dev, int inc);
4187int dev_set_allmulti(struct net_device *dev, int inc);
4188void netdev_state_change(struct net_device *dev);
4189void netdev_notify_peers(struct net_device *dev);
4190void netdev_features_change(struct net_device *dev);
1da177e4 4191/* Load a device via the kmod */
f629d208
JP
4192void dev_load(struct net *net, const char *name);
4193struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4194 struct rtnl_link_stats64 *storage);
4195void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4196 const struct net_device_stats *netdev_stats);
eeda3fd6 4197
1da177e4 4198extern int netdev_max_backlog;
3b098e2d 4199extern int netdev_tstamp_prequeue;
1da177e4 4200extern int weight_p;
3d48b53f
MT
4201extern int dev_weight_rx_bias;
4202extern int dev_weight_tx_bias;
4203extern int dev_rx_weight;
4204extern int dev_tx_weight;
9ff162a8 4205
f629d208 4206bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4207struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4208 struct list_head **iter);
f629d208
JP
4209struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
4210 struct list_head **iter);
8b5be856 4211
44a40855
VY
4212/* iterate through upper list, must be called under RCU read lock */
4213#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4214 for (iter = &(dev)->adj_list.upper, \
4215 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4216 updev; \
4217 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4218
1a3f060c
DA
4219int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4220 int (*fn)(struct net_device *upper_dev,
4221 void *data),
4222 void *data);
4223
4224bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4225 struct net_device *upper_dev);
4226
25cc72a3
IS
4227bool netdev_has_any_upper_dev(struct net_device *dev);
4228
f629d208
JP
4229void *netdev_lower_get_next_private(struct net_device *dev,
4230 struct list_head **iter);
4231void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4232 struct list_head **iter);
31088a11
VF
4233
4234#define netdev_for_each_lower_private(dev, priv, iter) \
4235 for (iter = (dev)->adj_list.lower.next, \
4236 priv = netdev_lower_get_next_private(dev, &(iter)); \
4237 priv; \
4238 priv = netdev_lower_get_next_private(dev, &(iter)))
4239
4240#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4241 for (iter = &(dev)->adj_list.lower, \
4242 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4243 priv; \
4244 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4245
4085ebe8
VY
4246void *netdev_lower_get_next(struct net_device *dev,
4247 struct list_head **iter);
7ce856aa 4248
4085ebe8 4249#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4250 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4251 ldev = netdev_lower_get_next(dev, &(iter)); \
4252 ldev; \
4253 ldev = netdev_lower_get_next(dev, &(iter)))
4254
7ce856aa
JP
4255struct net_device *netdev_all_lower_get_next(struct net_device *dev,
4256 struct list_head **iter);
4257struct net_device *netdev_all_lower_get_next_rcu(struct net_device *dev,
4258 struct list_head **iter);
4259
1a3f060c
DA
4260int netdev_walk_all_lower_dev(struct net_device *dev,
4261 int (*fn)(struct net_device *lower_dev,
4262 void *data),
4263 void *data);
4264int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4265 int (*fn)(struct net_device *lower_dev,
4266 void *data),
4267 void *data);
4268
f629d208 4269void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4270void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4271struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4272struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4273int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4274 struct netlink_ext_ack *extack);
f629d208 4275int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4276 struct net_device *upper_dev,
42ab19ee
DA
4277 void *upper_priv, void *upper_info,
4278 struct netlink_ext_ack *extack);
f629d208
JP
4279void netdev_upper_dev_unlink(struct net_device *dev,
4280 struct net_device *upper_dev);
5bb025fa 4281void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4282void *netdev_lower_dev_get_private(struct net_device *dev,
4283 struct net_device *lower_dev);
04d48266
JP
4284void netdev_lower_state_changed(struct net_device *lower_dev,
4285 void *lower_state_info);
960fb622
ED
4286
4287/* RSS keys are 40 or 52 bytes long */
4288#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4289extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4290void netdev_rss_key_fill(void *buffer, size_t len);
4291
952fcfd0 4292int dev_get_nest_level(struct net_device *dev);
f629d208 4293int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4294int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4295int skb_csum_hwoffload_help(struct sk_buff *skb,
4296 const netdev_features_t features);
4297
f629d208
JP
4298struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
4299 netdev_features_t features, bool tx_path);
4300struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
4301 netdev_features_t features);
12b0004d 4302
61bd3857
MS
4303struct netdev_bonding_info {
4304 ifslave slave;
4305 ifbond master;
4306};
4307
4308struct netdev_notifier_bonding_info {
4309 struct netdev_notifier_info info; /* must be first */
4310 struct netdev_bonding_info bonding_info;
4311};
4312
4313void netdev_bonding_info_change(struct net_device *dev,
4314 struct netdev_bonding_info *bonding_info);
4315
12b0004d
CW
4316static inline
4317struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
4318{
4319 return __skb_gso_segment(skb, features, true);
4320}
53d6471c 4321__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4322
4323static inline bool can_checksum_protocol(netdev_features_t features,
4324 __be16 protocol)
4325{
c8cd0989
TH
4326 if (protocol == htons(ETH_P_FCOE))
4327 return !!(features & NETIF_F_FCOE_CRC);
4328
4329 /* Assume this is an IP checksum (not SCTP CRC) */
4330
4331 if (features & NETIF_F_HW_CSUM) {
4332 /* Can checksum everything */
4333 return true;
4334 }
4335
4336 switch (protocol) {
4337 case htons(ETH_P_IP):
4338 return !!(features & NETIF_F_IP_CSUM);
4339 case htons(ETH_P_IPV6):
4340 return !!(features & NETIF_F_IPV6_CSUM);
4341 default:
4342 return false;
4343 }
ec5f0615 4344}
12b0004d 4345
fb286bb2 4346#ifdef CONFIG_BUG
7fe50ac8 4347void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4348#else
7fe50ac8
CW
4349static inline void netdev_rx_csum_fault(struct net_device *dev,
4350 struct sk_buff *skb)
fb286bb2
HX
4351{
4352}
4353#endif
1da177e4 4354/* rx skb timestamps */
f629d208
JP
4355void net_enable_timestamp(void);
4356void net_disable_timestamp(void);
1da177e4 4357
20380731 4358#ifdef CONFIG_PROC_FS
f629d208 4359int __init dev_proc_init(void);
900ff8c6
CW
4360#else
4361#define dev_proc_init() 0
20380731
ACM
4362#endif
4363
4798248e 4364static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4365 struct sk_buff *skb, struct net_device *dev,
4366 bool more)
4798248e 4367{
fa2dbdc2 4368 skb->xmit_more = more ? 1 : 0;
0b725a2c 4369 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4370}
4371
10b3ad8c 4372static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4373 struct netdev_queue *txq, bool more)
4798248e
DM
4374{
4375 const struct net_device_ops *ops = dev->netdev_ops;
10b3ad8c 4376 int rc;
4798248e 4377
fa2dbdc2 4378 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4379 if (rc == NETDEV_TX_OK)
4380 txq_trans_update(txq);
4381
4382 return rc;
4798248e
DM
4383}
4384
b793dc5c 4385int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4386 const void *ns);
b793dc5c 4387void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4388 const void *ns);
58292cbe 4389
b793dc5c 4390static inline int netdev_class_create_file(const struct class_attribute *class_attr)
58292cbe
TH
4391{
4392 return netdev_class_create_file_ns(class_attr, NULL);
4393}
4394
b793dc5c 4395static inline void netdev_class_remove_file(const struct class_attribute *class_attr)
58292cbe
TH
4396{
4397 netdev_class_remove_file_ns(class_attr, NULL);
4398}
b8a9787e 4399
737aec57 4400extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4401
f629d208 4402const char *netdev_drivername(const struct net_device *dev);
6579e57b 4403
f629d208 4404void linkwatch_run_queue(void);
20380731 4405
da08143b
MK
4406static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4407 netdev_features_t f2)
4408{
c8cd0989
TH
4409 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4410 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4411 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4412 else
b6a0e72a 4413 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4414 }
da08143b 4415
c8cd0989 4416 return f1 & f2;
da08143b
MK
4417}
4418
c8f44aff
MM
4419static inline netdev_features_t netdev_get_wanted_features(
4420 struct net_device *dev)
5455c699
MM
4421{
4422 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4423}
c8f44aff
MM
4424netdev_features_t netdev_increment_features(netdev_features_t all,
4425 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4426
4427/* Allow TSO being used on stacked device :
4428 * Performing the GSO segmentation before last device
4429 * is a performance improvement.
4430 */
4431static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4432 netdev_features_t mask)
4433{
4434 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4435}
4436
6cb6a27c 4437int __netdev_update_features(struct net_device *dev);
5455c699 4438void netdev_update_features(struct net_device *dev);
afe12cc8 4439void netdev_change_features(struct net_device *dev);
7f353bf2 4440
fc4a7489
PM
4441void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4442 struct net_device *dev);
4443
e38f3025
TM
4444netdev_features_t passthru_features_check(struct sk_buff *skb,
4445 struct net_device *dev,
4446 netdev_features_t features);
c1e756bf 4447netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 4448
4d29515f 4449static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 4450{
7b748340 4451 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
4452
4453 /* check flags correspondence */
4454 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
4455 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
4456 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 4457 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
4458 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
4459 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4460 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
4461 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
4462 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
4463 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4464 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
4465 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 4466 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 4467 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 4468 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 4469 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 4470 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 4471 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
0345e186 4472
d6b4991a 4473 return (features & feature) == feature;
576a30eb
HX
4474}
4475
4d29515f 4476static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 4477{
278b2513 4478 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 4479 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
4480}
4481
8b86a61d 4482static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 4483 netdev_features_t features)
7967168c 4484{
fc741216 4485 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
4486 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
4487 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
4488}
4489
82cc1a7a
PWJ
4490static inline void netif_set_gso_max_size(struct net_device *dev,
4491 unsigned int size)
4492{
4493 dev->gso_max_size = size;
4494}
4495
7a7ffbab
WCC
4496static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
4497 int pulled_hlen, u16 mac_offset,
4498 int mac_len)
4499{
4500 skb->protocol = protocol;
4501 skb->encapsulation = 1;
4502 skb_push(skb, pulled_hlen);
4503 skb_reset_transport_header(skb);
4504 skb->mac_header = mac_offset;
4505 skb->network_header = skb->mac_header + mac_len;
4506 skb->mac_len = mac_len;
4507}
4508
3c175784
SD
4509static inline bool netif_is_macsec(const struct net_device *dev)
4510{
4511 return dev->priv_flags & IFF_MACSEC;
4512}
4513
b618aaa9 4514static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
4515{
4516 return dev->priv_flags & IFF_MACVLAN;
4517}
4518
b618aaa9 4519static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
4520{
4521 return dev->priv_flags & IFF_MACVLAN_PORT;
4522}
4523
b618aaa9 4524static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 4525{
4526 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
4527}
4528
b618aaa9 4529static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
4530{
4531 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
4532}
4533
3bdc0eba
BG
4534static inline bool netif_supports_nofcs(struct net_device *dev)
4535{
4536 return dev->priv_flags & IFF_SUPP_NOFCS;
4537}
4538
007979ea 4539static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 4540{
007979ea 4541 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
4542}
4543
fee6d4c7
DA
4544static inline bool netif_is_l3_slave(const struct net_device *dev)
4545{
4546 return dev->priv_flags & IFF_L3MDEV_SLAVE;
4547}
4548
0894ae3f
JP
4549static inline bool netif_is_bridge_master(const struct net_device *dev)
4550{
4551 return dev->priv_flags & IFF_EBRIDGE;
4552}
4553
28f9ee22
VY
4554static inline bool netif_is_bridge_port(const struct net_device *dev)
4555{
4556 return dev->priv_flags & IFF_BRIDGE_PORT;
4557}
4558
35d4e172
JP
4559static inline bool netif_is_ovs_master(const struct net_device *dev)
4560{
4561 return dev->priv_flags & IFF_OPENVSWITCH;
4562}
4563
5be66141
JP
4564static inline bool netif_is_ovs_port(const struct net_device *dev)
4565{
4566 return dev->priv_flags & IFF_OVS_DATAPATH;
4567}
4568
b618aaa9 4569static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
4570{
4571 return dev->priv_flags & IFF_TEAM;
4572}
4573
b618aaa9 4574static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
4575{
4576 return dev->priv_flags & IFF_TEAM_PORT;
4577}
4578
b618aaa9 4579static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
4580{
4581 return netif_is_bond_master(dev) || netif_is_team_master(dev);
4582}
4583
b618aaa9 4584static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
4585{
4586 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
4587}
4588
d4ab4286
KJ
4589static inline bool netif_is_rxfh_configured(const struct net_device *dev)
4590{
4591 return dev->priv_flags & IFF_RXFH_CONFIGURED;
4592}
4593
30c8bd5a
SS
4594static inline bool netif_is_failover(const struct net_device *dev)
4595{
4596 return dev->priv_flags & IFF_FAILOVER;
4597}
4598
4599static inline bool netif_is_failover_slave(const struct net_device *dev)
4600{
4601 return dev->priv_flags & IFF_FAILOVER_SLAVE;
4602}
4603
02875878
ED
4604/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
4605static inline void netif_keep_dst(struct net_device *dev)
4606{
4607 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
4608}
4609
18d3df3e
PA
4610/* return true if dev can't cope with mtu frames that need vlan tag insertion */
4611static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
4612{
4613 /* TODO: reserve and use an additional IFF bit, if we get more users */
4614 return dev->priv_flags & IFF_MACSEC;
4615}
4616
505d4f73 4617extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 4618
571ba423
JP
4619/* Logging, debugging and troubleshooting/diagnostic helpers. */
4620
4621/* netdev_printk helpers, similar to dev_printk */
4622
4623static inline const char *netdev_name(const struct net_device *dev)
4624{
c6f854d5
VF
4625 if (!dev->name[0] || strchr(dev->name, '%'))
4626 return "(unnamed net_device)";
571ba423
JP
4627 return dev->name;
4628}
4629
8397ed36
DA
4630static inline bool netdev_unregistering(const struct net_device *dev)
4631{
4632 return dev->reg_state == NETREG_UNREGISTERING;
4633}
4634
ccc7f496
VF
4635static inline const char *netdev_reg_state(const struct net_device *dev)
4636{
4637 switch (dev->reg_state) {
4638 case NETREG_UNINITIALIZED: return " (uninitialized)";
4639 case NETREG_REGISTERED: return "";
4640 case NETREG_UNREGISTERING: return " (unregistering)";
4641 case NETREG_UNREGISTERED: return " (unregistered)";
4642 case NETREG_RELEASED: return " (released)";
4643 case NETREG_DUMMY: return " (dummy)";
4644 }
4645
4646 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
4647 return " (unknown)";
4648}
4649
f629d208 4650__printf(3, 4)
6ea754eb
JP
4651void netdev_printk(const char *level, const struct net_device *dev,
4652 const char *format, ...);
f629d208 4653__printf(2, 3)
6ea754eb 4654void netdev_emerg(const struct net_device *dev, const char *format, ...);
f629d208 4655__printf(2, 3)
6ea754eb 4656void netdev_alert(const struct net_device *dev, const char *format, ...);
f629d208 4657__printf(2, 3)
6ea754eb 4658void netdev_crit(const struct net_device *dev, const char *format, ...);
f629d208 4659__printf(2, 3)
6ea754eb 4660void netdev_err(const struct net_device *dev, const char *format, ...);
f629d208 4661__printf(2, 3)
6ea754eb 4662void netdev_warn(const struct net_device *dev, const char *format, ...);
f629d208 4663__printf(2, 3)
6ea754eb 4664void netdev_notice(const struct net_device *dev, const char *format, ...);
f629d208 4665__printf(2, 3)
6ea754eb 4666void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 4667
375ef2b1
GP
4668#define netdev_level_once(level, dev, fmt, ...) \
4669do { \
4670 static bool __print_once __read_mostly; \
4671 \
4672 if (!__print_once) { \
4673 __print_once = true; \
4674 netdev_printk(level, dev, fmt, ##__VA_ARGS__); \
4675 } \
4676} while (0)
4677
4678#define netdev_emerg_once(dev, fmt, ...) \
4679 netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
4680#define netdev_alert_once(dev, fmt, ...) \
4681 netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
4682#define netdev_crit_once(dev, fmt, ...) \
4683 netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
4684#define netdev_err_once(dev, fmt, ...) \
4685 netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
4686#define netdev_warn_once(dev, fmt, ...) \
4687 netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
4688#define netdev_notice_once(dev, fmt, ...) \
4689 netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
4690#define netdev_info_once(dev, fmt, ...) \
4691 netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
4692
8909c9ad
VK
4693#define MODULE_ALIAS_NETDEV(device) \
4694 MODULE_ALIAS("netdev-" device)
4695
b558c96f 4696#if defined(CONFIG_DYNAMIC_DEBUG)
571ba423
JP
4697#define netdev_dbg(__dev, format, args...) \
4698do { \
ffa10cb4 4699 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 4700} while (0)
b558c96f
JC
4701#elif defined(DEBUG)
4702#define netdev_dbg(__dev, format, args...) \
4703 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
4704#else
4705#define netdev_dbg(__dev, format, args...) \
4706({ \
4707 if (0) \
4708 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
4709})
4710#endif
4711
4712#if defined(VERBOSE_DEBUG)
4713#define netdev_vdbg netdev_dbg
4714#else
4715
4716#define netdev_vdbg(dev, format, args...) \
4717({ \
4718 if (0) \
4719 netdev_printk(KERN_DEBUG, dev, format, ##args); \
4720 0; \
4721})
4722#endif
4723
4724/*
4725 * netdev_WARN() acts like dev_printk(), but with the key difference
4726 * of using a WARN/WARN_ON to get the message out, including the
4727 * file/line information and a backtrace.
4728 */
4729#define netdev_WARN(dev, format, args...) \
e1cfe3d0 4730 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 4731 netdev_reg_state(dev), ##args)
571ba423 4732
72dd831e 4733#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 4734 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
4735 netdev_reg_state(dev), ##args)
4736
b3d95c5c
JP
4737/* netif printk helpers, similar to netdev_printk */
4738
4739#define netif_printk(priv, type, level, dev, fmt, args...) \
4740do { \
4741 if (netif_msg_##type(priv)) \
4742 netdev_printk(level, (dev), fmt, ##args); \
4743} while (0)
4744
f45f4321
JP
4745#define netif_level(level, priv, type, dev, fmt, args...) \
4746do { \
4747 if (netif_msg_##type(priv)) \
4748 netdev_##level(dev, fmt, ##args); \
4749} while (0)
4750
b3d95c5c 4751#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 4752 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 4753#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 4754 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 4755#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 4756 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 4757#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 4758 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 4759#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 4760 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 4761#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 4762 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 4763#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 4764 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 4765
0053ea9c 4766#if defined(CONFIG_DYNAMIC_DEBUG)
b3d95c5c
JP
4767#define netif_dbg(priv, type, netdev, format, args...) \
4768do { \
4769 if (netif_msg_##type(priv)) \
b5fb0a03 4770 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 4771} while (0)
0053ea9c
JP
4772#elif defined(DEBUG)
4773#define netif_dbg(priv, type, dev, format, args...) \
4774 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
4775#else
4776#define netif_dbg(priv, type, dev, format, args...) \
4777({ \
4778 if (0) \
4779 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
4780 0; \
4781})
4782#endif
4783
f617f276
EC
4784/* if @cond then downgrade to debug, else print at @level */
4785#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...) \
4786 do { \
4787 if (cond) \
4788 netif_dbg(priv, type, netdev, fmt, ##args); \
4789 else \
4790 netif_ ## level(priv, type, netdev, fmt, ##args); \
4791 } while (0)
4792
b3d95c5c 4793#if defined(VERBOSE_DEBUG)
bcfcc450 4794#define netif_vdbg netif_dbg
b3d95c5c
JP
4795#else
4796#define netif_vdbg(priv, type, dev, format, args...) \
4797({ \
4798 if (0) \
a4ed89cb 4799 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
4800 0; \
4801})
4802#endif
571ba423 4803
900ff8c6
CW
4804/*
4805 * The list of packet types we will receive (as opposed to discard)
4806 * and the routines to invoke.
4807 *
4808 * Why 16. Because with 16 the only overlap we get on a hash of the
4809 * low nibble of the protocol value is RARP/SNAP/X.25.
4810 *
900ff8c6 4811 * 0800 IP
900ff8c6
CW
4812 * 0001 802.3
4813 * 0002 AX.25
4814 * 0004 802.2
4815 * 8035 RARP
4816 * 0005 SNAP
4817 * 0805 X.25
4818 * 0806 ARP
4819 * 8137 IPX
4820 * 0009 Localtalk
4821 * 86DD IPv6
4822 */
4823#define PTYPE_HASH_SIZE (16)
4824#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
4825
385a154c 4826#endif /* _LINUX_NETDEVICE_H */