]> git.proxmox.com Git - mirror_ubuntu-hirsute-kernel.git/blame - include/linux/netdevice.h
net: procfs: use index hashlist instead of name hashlist
[mirror_ubuntu-hirsute-kernel.git] / include / linux / netdevice.h
CommitLineData
2874c5fd 1/* SPDX-License-Identifier: GPL-2.0-or-later */
1da177e4
LT
2/*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the Interfaces handler.
8 *
9 * Version: @(#)dev.h 1.0.10 08/12/93
10 *
02c30a84 11 * Authors: Ross Biro
1da177e4
LT
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Corey Minyard <wf-rch!minyard@relay.EU.net>
14 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 15 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
16 * Bjorn Ekwall. <bj0rn@blox.se>
17 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 *
1da177e4
LT
19 * Moved to /usr/include/linux for NET3
20 */
21#ifndef _LINUX_NETDEVICE_H
22#define _LINUX_NETDEVICE_H
23
d7fe0f24 24#include <linux/timer.h>
187f1882 25#include <linux/bug.h>
bea3348e 26#include <linux/delay.h>
60063497 27#include <linux/atomic.h>
53511453 28#include <linux/prefetch.h>
1da177e4
LT
29#include <asm/cache.h>
30#include <asm/byteorder.h>
31
1da177e4 32#include <linux/percpu.h>
4d5b78c0 33#include <linux/rculist.h>
bea3348e 34#include <linux/workqueue.h>
114cf580 35#include <linux/dynamic_queue_limits.h>
1da177e4 36
b1b67dd4 37#include <linux/ethtool.h>
a050c33f 38#include <net/net_namespace.h>
7a6b6f51 39#ifdef CONFIG_DCB
2f90b865
AD
40#include <net/dcbnl.h>
41#endif
5bc1421e 42#include <net/netprio_cgroup.h>
e817f856 43#include <net/xdp.h>
a050c33f 44
a59e2ecb 45#include <linux/netdev_features.h>
77162022 46#include <linux/neighbour.h>
607ca46e 47#include <uapi/linux/netdevice.h>
61bd3857 48#include <uapi/linux/if_bonding.h>
e4c6734e 49#include <uapi/linux/pkt_cls.h>
59cc1f61 50#include <linux/hashtable.h>
a59e2ecb 51
115c1d6e 52struct netpoll_info;
313162d0 53struct device;
c1f19b51 54struct phy_device;
2f657a60 55struct dsa_port;
c6e970a0 56
e679c9c1 57struct sfp_bus;
704232c2
JB
58/* 802.11 specific */
59struct wireless_dev;
98a18b6f
AA
60/* 802.15.4 specific */
61struct wpan_dev;
03c57747 62struct mpls_dev;
7c46a640
AD
63/* UDP Tunnel offloads */
64struct udp_tunnel_info;
a7862b45 65struct bpf_prog;
814abfab 66struct xdp_buff;
1da177e4 67
f629d208
JP
68void netdev_set_default_ethtool_ops(struct net_device *dev,
69 const struct ethtool_ops *ops);
d07d7507 70
9a1654ba
JP
71/* Backlog congestion levels */
72#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
73#define NET_RX_DROP 1 /* packet dropped */
74
572a9d7b
PM
75/*
76 * Transmit return codes: transmit return codes originate from three different
77 * namespaces:
78 *
79 * - qdisc return codes
80 * - driver transmit return codes
81 * - errno values
82 *
83 * Drivers are allowed to return any one of those in their hard_start_xmit()
84 * function. Real network devices commonly used with qdiscs should only return
85 * the driver transmit return codes though - when qdiscs are used, the actual
86 * transmission happens asynchronously, so the value is not propagated to
5e82b4b2
BH
87 * higher layers. Virtual network devices transmit synchronously; in this case
88 * the driver transmit return codes are consumed by dev_queue_xmit(), and all
572a9d7b
PM
89 * others are propagated to higher layers.
90 */
91
92/* qdisc ->enqueue() return codes. */
93#define NET_XMIT_SUCCESS 0x00
9a1654ba
JP
94#define NET_XMIT_DROP 0x01 /* skb dropped */
95#define NET_XMIT_CN 0x02 /* congestion notification */
9a1654ba 96#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
1da177e4 97
b9df3cb8
GR
98/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
99 * indicates that the device will soon be dropping packets, or already drops
100 * some packets of the same priority; prompting us to send less aggressively. */
572a9d7b 101#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
1da177e4
LT
102#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
103
dc1f8bf6 104/* Driver transmit return codes */
9a1654ba 105#define NETDEV_TX_MASK 0xf0
572a9d7b 106
dc1f8bf6 107enum netdev_tx {
572a9d7b 108 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
9a1654ba
JP
109 NETDEV_TX_OK = 0x00, /* driver took care of packet */
110 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
dc1f8bf6
SH
111};
112typedef enum netdev_tx netdev_tx_t;
113
9a1654ba
JP
114/*
115 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
116 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
117 */
118static inline bool dev_xmit_complete(int rc)
119{
120 /*
121 * Positive cases with an skb consumed by a driver:
122 * - successful transmission (rc == NETDEV_TX_OK)
123 * - error while transmitting (rc < 0)
124 * - error while queueing to a different device (rc & NET_XMIT_MASK)
125 */
126 if (likely(rc < NET_XMIT_MASK))
127 return true;
128
129 return false;
130}
131
1da177e4 132/*
5e82b4b2 133 * Compute the worst-case header length according to the protocols
1da177e4
LT
134 * used.
135 */
fe2918b0 136
c0eb4540
KS
137#if defined(CONFIG_HYPERV_NET)
138# define LL_MAX_HEADER 128
139#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
8388e3da
DM
140# if defined(CONFIG_MAC80211_MESH)
141# define LL_MAX_HEADER 128
142# else
143# define LL_MAX_HEADER 96
144# endif
1da177e4 145#else
8388e3da 146# define LL_MAX_HEADER 32
1da177e4
LT
147#endif
148
d11ead75
BH
149#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
150 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
1da177e4
LT
151#define MAX_HEADER LL_MAX_HEADER
152#else
153#define MAX_HEADER (LL_MAX_HEADER + 48)
154#endif
155
156/*
be1f3c2c
BH
157 * Old network device statistics. Fields are native words
158 * (unsigned long) so they can be read and written atomically.
1da177e4 159 */
fe2918b0 160
d94d9fee 161struct net_device_stats {
3cfde79c
BH
162 unsigned long rx_packets;
163 unsigned long tx_packets;
164 unsigned long rx_bytes;
165 unsigned long tx_bytes;
166 unsigned long rx_errors;
167 unsigned long tx_errors;
168 unsigned long rx_dropped;
169 unsigned long tx_dropped;
170 unsigned long multicast;
1da177e4 171 unsigned long collisions;
1da177e4 172 unsigned long rx_length_errors;
3cfde79c
BH
173 unsigned long rx_over_errors;
174 unsigned long rx_crc_errors;
175 unsigned long rx_frame_errors;
176 unsigned long rx_fifo_errors;
177 unsigned long rx_missed_errors;
1da177e4
LT
178 unsigned long tx_aborted_errors;
179 unsigned long tx_carrier_errors;
180 unsigned long tx_fifo_errors;
181 unsigned long tx_heartbeat_errors;
182 unsigned long tx_window_errors;
1da177e4
LT
183 unsigned long rx_compressed;
184 unsigned long tx_compressed;
185};
186
1da177e4
LT
187
188#include <linux/cache.h>
189#include <linux/skbuff.h>
190
adc9300e 191#ifdef CONFIG_RPS
c5905afb 192#include <linux/static_key.h>
dc05360f
ED
193extern struct static_key_false rps_needed;
194extern struct static_key_false rfs_needed;
adc9300e
ED
195#endif
196
1da177e4
LT
197struct neighbour;
198struct neigh_parms;
199struct sk_buff;
200
f001fde5
JP
201struct netdev_hw_addr {
202 struct list_head list;
203 unsigned char addr[MAX_ADDR_LEN];
204 unsigned char type;
ccffad25
JP
205#define NETDEV_HW_ADDR_T_LAN 1
206#define NETDEV_HW_ADDR_T_SAN 2
207#define NETDEV_HW_ADDR_T_SLAVE 3
208#define NETDEV_HW_ADDR_T_UNICAST 4
22bedad3 209#define NETDEV_HW_ADDR_T_MULTICAST 5
22bedad3 210 bool global_use;
4cd729b0 211 int sync_cnt;
8f8f103d 212 int refcount;
4543fbef 213 int synced;
f001fde5
JP
214 struct rcu_head rcu_head;
215};
216
31278e71
JP
217struct netdev_hw_addr_list {
218 struct list_head list;
219 int count;
220};
221
22bedad3
JP
222#define netdev_hw_addr_list_count(l) ((l)->count)
223#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
224#define netdev_hw_addr_list_for_each(ha, l) \
225 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 226
22bedad3
JP
227#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
228#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
229#define netdev_for_each_uc_addr(ha, dev) \
230 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
6683ece3 231
22bedad3
JP
232#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
233#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 234#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 235 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
6683ece3 236
d94d9fee 237struct hh_cache {
5b3dc2f3 238 unsigned int hh_len;
3644f0ce 239 seqlock_t hh_lock;
1da177e4
LT
240
241 /* cached hardware header; allow for machine alignment needs. */
242#define HH_DATA_MOD 16
243#define HH_DATA_OFF(__len) \
5ba0eac6 244 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
1da177e4
LT
245#define HH_DATA_ALIGN(__len) \
246 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
247 unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
248};
249
5e82b4b2 250/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
1da177e4
LT
251 * Alternative is:
252 * dev->hard_header_len ? (dev->hard_header_len +
253 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
254 *
255 * We could use other alignment values, but we must maintain the
256 * relationship HH alignment <= LL alignment.
257 */
258#define LL_RESERVED_SPACE(dev) \
f5184d26 259 ((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 260#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
f5184d26 261 ((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 262
3b04ddde
SH
263struct header_ops {
264 int (*create) (struct sk_buff *skb, struct net_device *dev,
265 unsigned short type, const void *daddr,
95c96174 266 const void *saddr, unsigned int len);
3b04ddde 267 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
e69dd336 268 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
3b04ddde
SH
269 void (*cache_update)(struct hh_cache *hh,
270 const struct net_device *dev,
271 const unsigned char *haddr);
2793a23a 272 bool (*validate)(const char *ll_header, unsigned int len);
e78b2915 273 __be16 (*parse_protocol)(const struct sk_buff *skb);
3b04ddde
SH
274};
275
1da177e4 276/* These flag bits are private to the generic network queueing
5e82b4b2 277 * layer; they may not be explicitly referenced by any other
1da177e4
LT
278 * code.
279 */
280
d94d9fee 281enum netdev_state_t {
1da177e4
LT
282 __LINK_STATE_START,
283 __LINK_STATE_PRESENT,
1da177e4 284 __LINK_STATE_NOCARRIER,
b00055aa
SR
285 __LINK_STATE_LINKWATCH_PENDING,
286 __LINK_STATE_DORMANT,
1da177e4
LT
287};
288
289
290/*
5e82b4b2 291 * This structure holds boot-time configured netdevice settings. They
fe2918b0 292 * are then used in the device probing.
1da177e4
LT
293 */
294struct netdev_boot_setup {
295 char name[IFNAMSIZ];
296 struct ifmap map;
297};
298#define NETDEV_BOOT_SETUP_MAX 8
299
f629d208 300int __init netdev_boot_setup(char *str);
1da177e4 301
6312fe77
LR
302struct gro_list {
303 struct list_head list;
304 int count;
305};
306
bea3348e 307/*
d9f37d01
LR
308 * size of gro hash buckets, must less than bit number of
309 * napi_struct::gro_bitmask
bea3348e 310 */
07d78363 311#define GRO_HASH_BUCKETS 8
d9f37d01
LR
312
313/*
314 * Structure for NAPI scheduling similar to tasklet but with weighting
315 */
bea3348e
SH
316struct napi_struct {
317 /* The poll_list must only be managed by the entity which
318 * changes the state of the NAPI_STATE_SCHED bit. This means
319 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 320 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
321 * can remove from the list right before clearing the bit.
322 */
323 struct list_head poll_list;
324
325 unsigned long state;
326 int weight;
d9f37d01 327 unsigned long gro_bitmask;
bea3348e
SH
328 int (*poll)(struct napi_struct *, int);
329#ifdef CONFIG_NETPOLL
bea3348e 330 int poll_owner;
bea3348e 331#endif
5d38a079 332 struct net_device *dev;
6312fe77 333 struct gro_list gro_hash[GRO_HASH_BUCKETS];
5d38a079 334 struct sk_buff *skb;
323ebb61
EC
335 struct list_head rx_list; /* Pending GRO_NORMAL skbs */
336 int rx_count; /* length of rx_list */
3b47d303 337 struct hrtimer timer;
404f7c9e 338 struct list_head dev_list;
af12fa6e
ET
339 struct hlist_node napi_hash_node;
340 unsigned int napi_id;
bea3348e
SH
341};
342
d94d9fee 343enum {
bea3348e 344 NAPI_STATE_SCHED, /* Poll is scheduled */
39e6c820 345 NAPI_STATE_MISSED, /* reschedule a napi */
a0a46196 346 NAPI_STATE_DISABLE, /* Disable pending */
7b363e44 347 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
d64b5e85
ED
348 NAPI_STATE_HASHED, /* In NAPI hash (busy polling possible) */
349 NAPI_STATE_NO_BUSY_POLL,/* Do not add in napi_hash, no busy polling */
217f6974
ED
350 NAPI_STATE_IN_BUSY_POLL,/* sk_busy_loop() owns this NAPI */
351};
352
353enum {
39e6c820
ED
354 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
355 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
356 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
357 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
358 NAPIF_STATE_HASHED = BIT(NAPI_STATE_HASHED),
359 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
360 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
bea3348e
SH
361};
362
5b252f0c 363enum gro_result {
d1c76af9
HX
364 GRO_MERGED,
365 GRO_MERGED_FREE,
366 GRO_HELD,
367 GRO_NORMAL,
368 GRO_DROP,
25393d3f 369 GRO_CONSUMED,
d1c76af9 370};
5b252f0c 371typedef enum gro_result gro_result_t;
d1c76af9 372
8a4eb573
JP
373/*
374 * enum rx_handler_result - Possible return values for rx_handlers.
375 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
376 * further.
377 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
378 * case skb->dev was changed by rx_handler.
379 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 380 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
381 *
382 * rx_handlers are functions called from inside __netif_receive_skb(), to do
383 * special processing of the skb, prior to delivery to protocol handlers.
384 *
385 * Currently, a net_device can only have a single rx_handler registered. Trying
386 * to register a second rx_handler will return -EBUSY.
387 *
388 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
389 * To unregister a rx_handler on a net_device, use
390 * netdev_rx_handler_unregister().
391 *
392 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
393 * do with the skb.
394 *
5e82b4b2 395 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 396 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 397 * the skb to be delivered in some other way.
8a4eb573
JP
398 *
399 * If the rx_handler changed skb->dev, to divert the skb to another
400 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
401 * new device will be called if it exists.
402 *
5e82b4b2 403 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 404 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 405 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 406 *
5e82b4b2 407 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
408 * delivered, it should return RX_HANDLER_PASS.
409 *
410 * A device without a registered rx_handler will behave as if rx_handler
411 * returned RX_HANDLER_PASS.
412 */
413
414enum rx_handler_result {
415 RX_HANDLER_CONSUMED,
416 RX_HANDLER_ANOTHER,
417 RX_HANDLER_EXACT,
418 RX_HANDLER_PASS,
419};
420typedef enum rx_handler_result rx_handler_result_t;
421typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 422
f629d208 423void __napi_schedule(struct napi_struct *n);
bc9ad166 424void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 425
4d29515f 426static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
427{
428 return test_bit(NAPI_STATE_DISABLE, &n->state);
429}
430
39e6c820 431bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
432
433/**
434 * napi_schedule - schedule NAPI poll
5e82b4b2 435 * @n: NAPI context
bea3348e
SH
436 *
437 * Schedule NAPI poll routine to be called if it is not already
438 * running.
439 */
440static inline void napi_schedule(struct napi_struct *n)
441{
442 if (napi_schedule_prep(n))
443 __napi_schedule(n);
444}
445
bc9ad166
ED
446/**
447 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 448 * @n: NAPI context
bc9ad166
ED
449 *
450 * Variant of napi_schedule(), assuming hard irqs are masked.
451 */
452static inline void napi_schedule_irqoff(struct napi_struct *n)
453{
454 if (napi_schedule_prep(n))
455 __napi_schedule_irqoff(n);
456}
457
bfe13f54 458/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 459static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
460{
461 if (napi_schedule_prep(napi)) {
462 __napi_schedule(napi);
4d29515f 463 return true;
bfe13f54 464 }
4d29515f 465 return false;
bfe13f54
RD
466}
467
364b6055 468bool napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
469/**
470 * napi_complete - NAPI processing complete
5e82b4b2 471 * @n: NAPI context
bea3348e
SH
472 *
473 * Mark NAPI processing as complete.
3b47d303 474 * Consider using napi_complete_done() instead.
364b6055 475 * Return false if device should avoid rearming interrupts.
bea3348e 476 */
364b6055 477static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
478{
479 return napi_complete_done(n, 0);
480}
bea3348e 481
af12fa6e
ET
482/**
483 * napi_hash_del - remove a NAPI from global table
5e82b4b2 484 * @napi: NAPI context
af12fa6e 485 *
5e82b4b2 486 * Warning: caller must observe RCU grace period
34cbe27e
ED
487 * before freeing memory containing @napi, if
488 * this function returns true.
93d05d4a 489 * Note: core networking stack automatically calls it
5e82b4b2 490 * from netif_napi_del().
93d05d4a 491 * Drivers might want to call this helper to combine all
5e82b4b2 492 * the needed RCU grace periods into a single one.
af12fa6e 493 */
34cbe27e 494bool napi_hash_del(struct napi_struct *napi);
af12fa6e 495
bea3348e
SH
496/**
497 * napi_disable - prevent NAPI from scheduling
5e82b4b2 498 * @n: NAPI context
bea3348e
SH
499 *
500 * Stop NAPI from being scheduled on this context.
501 * Waits till any outstanding processing completes.
502 */
3b47d303 503void napi_disable(struct napi_struct *n);
bea3348e
SH
504
505/**
506 * napi_enable - enable NAPI scheduling
5e82b4b2 507 * @n: NAPI context
bea3348e
SH
508 *
509 * Resume NAPI from being scheduled on this context.
510 * Must be paired with napi_disable.
511 */
512static inline void napi_enable(struct napi_struct *n)
513{
514 BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
4e857c58 515 smp_mb__before_atomic();
bea3348e 516 clear_bit(NAPI_STATE_SCHED, &n->state);
2d8bff12 517 clear_bit(NAPI_STATE_NPSVC, &n->state);
bea3348e
SH
518}
519
c264c3de
SH
520/**
521 * napi_synchronize - wait until NAPI is not running
5e82b4b2 522 * @n: NAPI context
c264c3de
SH
523 *
524 * Wait until NAPI is done being scheduled on this context.
525 * Waits till any outstanding processing completes but
526 * does not disable future activations.
527 */
528static inline void napi_synchronize(const struct napi_struct *n)
529{
facc432f
AB
530 if (IS_ENABLED(CONFIG_SMP))
531 while (test_bit(NAPI_STATE_SCHED, &n->state))
532 msleep(1);
533 else
534 barrier();
c264c3de 535}
c264c3de 536
6c5c9581
MK
537/**
538 * napi_if_scheduled_mark_missed - if napi is running, set the
539 * NAPIF_STATE_MISSED
540 * @n: NAPI context
541 *
542 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
543 * NAPI is scheduled.
544 **/
545static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
546{
547 unsigned long val, new;
548
549 do {
550 val = READ_ONCE(n->state);
551 if (val & NAPIF_STATE_DISABLE)
552 return true;
553
554 if (!(val & NAPIF_STATE_SCHED))
555 return false;
556
557 new = val | NAPIF_STATE_MISSED;
558 } while (cmpxchg(&n->state, val, new) != val);
559
560 return true;
561}
562
d94d9fee 563enum netdev_queue_state_t {
73466498
TH
564 __QUEUE_STATE_DRV_XOFF,
565 __QUEUE_STATE_STACK_XOFF,
c3f26a26 566 __QUEUE_STATE_FROZEN,
79d16385 567};
8e2f1a63
DB
568
569#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
570#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
571#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
572
573#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
574#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
575 QUEUE_STATE_FROZEN)
576#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
577 QUEUE_STATE_FROZEN)
578
73466498
TH
579/*
580 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
581 * netif_tx_* functions below are used to manipulate this flag. The
582 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
583 * queue independently. The netif_xmit_*stopped functions below are called
584 * to check if the queue has been stopped by the driver or stack (either
585 * of the XOFF bits are set in the state). Drivers should not need to call
586 * netif_xmit*stopped functions, they should only be using netif_tx_*.
587 */
79d16385 588
bb949fbd 589struct netdev_queue {
6a321cb3 590/*
5e82b4b2 591 * read-mostly part
6a321cb3 592 */
bb949fbd 593 struct net_device *dev;
46e5da40 594 struct Qdisc __rcu *qdisc;
b0e1e646 595 struct Qdisc *qdisc_sleeping;
ccf5ff69 596#ifdef CONFIG_SYSFS
1d24eb48
TH
597 struct kobject kobj;
598#endif
f2cd2d3e
ED
599#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
600 int numa_node;
601#endif
c0ef079c
FW
602 unsigned long tx_maxrate;
603 /*
604 * Number of TX timeouts for this queue
605 * (/sys/class/net/DEV/Q/trans_timeout)
606 */
607 unsigned long trans_timeout;
ffcfe25b
AD
608
609 /* Subordinate device that the queue has been assigned to */
610 struct net_device *sb_dev;
661b8d1b
MK
611#ifdef CONFIG_XDP_SOCKETS
612 struct xdp_umem *umem;
613#endif
6a321cb3 614/*
5e82b4b2 615 * write-mostly part
6a321cb3
ED
616 */
617 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
618 int xmit_lock_owner;
9d21493b 619 /*
9b36627a 620 * Time (in jiffies) of last Tx
9d21493b
ED
621 */
622 unsigned long trans_start;
ccf5ff69 623
114cf580
TH
624 unsigned long state;
625
626#ifdef CONFIG_BQL
627 struct dql dql;
628#endif
e8a0464c 629} ____cacheline_aligned_in_smp;
bb949fbd 630
79134e6c 631extern int sysctl_fb_tunnels_only_for_init_net;
856c395c 632extern int sysctl_devconf_inherit_init_net;
79134e6c
ED
633
634static inline bool net_has_fallback_tunnels(const struct net *net)
635{
be9fc097
AB
636 return net == &init_net ||
637 !IS_ENABLED(CONFIG_SYSCTL) ||
638 !sysctl_fb_tunnels_only_for_init_net;
79134e6c
ED
639}
640
f2cd2d3e
ED
641static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
642{
643#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
644 return q->numa_node;
645#else
b236da69 646 return NUMA_NO_NODE;
f2cd2d3e
ED
647#endif
648}
649
650static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
651{
652#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
653 q->numa_node = node;
654#endif
655}
656
df334545 657#ifdef CONFIG_RPS
0a9627f2
TH
658/*
659 * This structure holds an RPS map which can be of variable length. The
660 * map is an array of CPUs.
661 */
662struct rps_map {
663 unsigned int len;
664 struct rcu_head rcu;
665 u16 cpus[0];
666};
60b778ce 667#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 668
fec5e652 669/*
c445477d
BH
670 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
671 * tail pointer for that CPU's input queue at the time of last enqueue, and
672 * a hardware filter index.
fec5e652
TH
673 */
674struct rps_dev_flow {
675 u16 cpu;
c445477d 676 u16 filter;
fec5e652
TH
677 unsigned int last_qtail;
678};
c445477d 679#define RPS_NO_FILTER 0xffff
fec5e652
TH
680
681/*
682 * The rps_dev_flow_table structure contains a table of flow mappings.
683 */
684struct rps_dev_flow_table {
685 unsigned int mask;
686 struct rcu_head rcu;
fec5e652
TH
687 struct rps_dev_flow flows[0];
688};
689#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 690 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
691
692/*
693 * The rps_sock_flow_table contains mappings of flows to the last CPU
694 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
695 * Each entry is a 32bit value. Upper part is the high-order bits
696 * of flow hash, lower part is CPU number.
567e4b79 697 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
698 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
699 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 700 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
701 */
702struct rps_sock_flow_table {
567e4b79 703 u32 mask;
93c1af6c
ED
704
705 u32 ents[0] ____cacheline_aligned_in_smp;
fec5e652 706};
567e4b79 707#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
708
709#define RPS_NO_CPU 0xffff
710
567e4b79
ED
711extern u32 rps_cpu_mask;
712extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
713
fec5e652
TH
714static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
715 u32 hash)
716{
717 if (table && hash) {
567e4b79
ED
718 unsigned int index = hash & table->mask;
719 u32 val = hash & ~rps_cpu_mask;
fec5e652 720
5e82b4b2 721 /* We only give a hint, preemption can change CPU under us */
567e4b79 722 val |= raw_smp_processor_id();
fec5e652 723
567e4b79
ED
724 if (table->ents[index] != val)
725 table->ents[index] = val;
fec5e652
TH
726 }
727}
728
c445477d 729#ifdef CONFIG_RFS_ACCEL
f629d208
JP
730bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
731 u16 filter_id);
c445477d 732#endif
a953be53 733#endif /* CONFIG_RPS */
c445477d 734
0a9627f2
TH
735/* This structure contains an instance of an RX queue. */
736struct netdev_rx_queue {
a953be53 737#ifdef CONFIG_RPS
6e3f7faf
ED
738 struct rps_map __rcu *rps_map;
739 struct rps_dev_flow_table __rcu *rps_flow_table;
a953be53 740#endif
6e3f7faf 741 struct kobject kobj;
fe822240 742 struct net_device *dev;
e817f856 743 struct xdp_rxq_info xdp_rxq;
661b8d1b
MK
744#ifdef CONFIG_XDP_SOCKETS
745 struct xdp_umem *umem;
746#endif
0a9627f2 747} ____cacheline_aligned_in_smp;
a953be53
MD
748
749/*
750 * RX queue sysfs structures and functions.
751 */
752struct rx_queue_attribute {
753 struct attribute attr;
718ad681 754 ssize_t (*show)(struct netdev_rx_queue *queue, char *buf);
a953be53 755 ssize_t (*store)(struct netdev_rx_queue *queue,
718ad681 756 const char *buf, size_t len);
a953be53 757};
d314774c 758
bf264145
TH
759#ifdef CONFIG_XPS
760/*
761 * This structure holds an XPS map which can be of variable length. The
762 * map is an array of queues.
763 */
764struct xps_map {
765 unsigned int len;
766 unsigned int alloc_len;
767 struct rcu_head rcu;
768 u16 queues[0];
769};
60b778ce 770#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
771#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
772 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
773
774/*
775 * This structure holds all XPS maps for device. Maps are indexed by CPU.
776 */
777struct xps_dev_maps {
778 struct rcu_head rcu;
80d19669 779 struct xps_map __rcu *attr_map[0]; /* Either CPUs map or RXQs map */
bf264145 780};
80d19669
AN
781
782#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
184c449f 783 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
80d19669
AN
784
785#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
786 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
787
bf264145
TH
788#endif /* CONFIG_XPS */
789
4f57c087
JF
790#define TC_MAX_QUEUE 16
791#define TC_BITMASK 15
792/* HW offloaded queuing disciplines txq count and offset maps */
793struct netdev_tc_txq {
794 u16 count;
795 u16 offset;
796};
797
68bad94e
NP
798#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
799/*
800 * This structure is to hold information about the device
801 * configured to run FCoE protocol stack.
802 */
803struct netdev_fcoe_hbainfo {
804 char manufacturer[64];
805 char serial_number[64];
806 char hardware_version[64];
807 char driver_version[64];
808 char optionrom_version[64];
809 char firmware_version[64];
810 char model[256];
811 char model_description[256];
812};
813#endif
814
02637fce 815#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 816
02637fce
JP
817/* This structure holds a unique identifier to identify some
818 * physical item (port for example) used by a netdevice.
66b52b0d 819 */
02637fce
JP
820struct netdev_phys_item_id {
821 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
822 unsigned char id_len;
823};
824
d754f98b
SF
825static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
826 struct netdev_phys_item_id *b)
827{
828 return a->id_len == b->id_len &&
829 memcmp(a->id, b->id, a->id_len) == 0;
830}
831
99932d4f 832typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
8ec56fc3
AD
833 struct sk_buff *skb,
834 struct net_device *sb_dev);
99932d4f 835
2572ac53 836enum tc_setup_type {
575ed7d3 837 TC_SETUP_QDISC_MQPRIO,
a1b7c5fd 838 TC_SETUP_CLSU32,
5b33f488 839 TC_SETUP_CLSFLOWER,
ade9b658 840 TC_SETUP_CLSMATCHALL,
332ae8e2 841 TC_SETUP_CLSBPF,
8c4083b3 842 TC_SETUP_BLOCK,
8521db4c 843 TC_SETUP_QDISC_CBS,
602f3baf 844 TC_SETUP_QDISC_RED,
7fdb61b4 845 TC_SETUP_QDISC_PRIO,
f971b132 846 TC_SETUP_QDISC_MQ,
25db26a9 847 TC_SETUP_QDISC_ETF,
98b0e5f6 848 TC_SETUP_ROOT_QDISC,
890d8d23 849 TC_SETUP_QDISC_GRED,
9c66d156 850 TC_SETUP_QDISC_TAPRIO,
16e5cc64
JF
851};
852
f4e63525
JK
853/* These structures hold the attributes of bpf state that are being passed
854 * to the netdevice through the bpf op.
a7862b45 855 */
f4e63525 856enum bpf_netdev_command {
a7862b45
BB
857 /* Set or clear a bpf program used in the earliest stages of packet
858 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
859 * is responsible for calling bpf_prog_put on any old progs that are
860 * stored. In case of error, the callee need not release the new prog
861 * reference, but on success it takes ownership and must bpf_prog_put
862 * when it is no longer used.
863 */
864 XDP_SETUP_PROG,
ee5d032f 865 XDP_SETUP_PROG_HW,
a7862b45 866 XDP_QUERY_PROG,
a25717d2 867 XDP_QUERY_PROG_HW,
ab3f0063 868 /* BPF program for offload callbacks, invoked at program load time. */
a3884572
JK
869 BPF_OFFLOAD_MAP_ALLOC,
870 BPF_OFFLOAD_MAP_FREE,
74515c57 871 XDP_SETUP_XSK_UMEM,
a7862b45
BB
872};
873
cae1927c 874struct bpf_prog_offload_ops;
ddf9f970 875struct netlink_ext_ack;
74515c57 876struct xdp_umem;
ddf9f970 877
f4e63525
JK
878struct netdev_bpf {
879 enum bpf_netdev_command command;
a7862b45
BB
880 union {
881 /* XDP_SETUP_PROG */
ddf9f970 882 struct {
32d60277 883 u32 flags;
ddf9f970
JK
884 struct bpf_prog *prog;
885 struct netlink_ext_ack *extack;
886 };
a25717d2 887 /* XDP_QUERY_PROG, XDP_QUERY_PROG_HW */
58038695 888 struct {
58038695 889 u32 prog_id;
92f0292b
JK
890 /* flags with which program was installed */
891 u32 prog_flags;
58038695 892 };
a3884572
JK
893 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
894 struct {
895 struct bpf_offloaded_map *offmap;
896 };
f8ebfaf6 897 /* XDP_SETUP_XSK_UMEM */
74515c57 898 struct {
f8ebfaf6
JS
899 struct xdp_umem *umem;
900 u16 queue_id;
74515c57 901 } xsk;
a7862b45
BB
902 };
903};
16e5cc64 904
9116e5e2
MK
905/* Flags for ndo_xsk_wakeup. */
906#define XDP_WAKEUP_RX (1 << 0)
907#define XDP_WAKEUP_TX (1 << 1)
908
d77e38e6
SK
909#ifdef CONFIG_XFRM_OFFLOAD
910struct xfrmdev_ops {
911 int (*xdo_dev_state_add) (struct xfrm_state *x);
912 void (*xdo_dev_state_delete) (struct xfrm_state *x);
913 void (*xdo_dev_state_free) (struct xfrm_state *x);
914 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
915 struct xfrm_state *x);
50bd870a 916 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
d77e38e6
SK
917};
918#endif
919
6c557001
FW
920struct dev_ifalias {
921 struct rcu_head rcuhead;
922 char ifalias[];
923};
924
b473b0d2 925struct devlink;
da68b4ad 926struct tlsdev_ops;
b473b0d2 927
d314774c
SH
928/*
929 * This structure defines the management hooks for network devices.
00829823
SH
930 * The following hooks can be defined; unless noted otherwise, they are
931 * optional and can be filled with a null pointer.
d314774c
SH
932 *
933 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
934 * This function is called once when a network device is registered.
935 * The network device can use this for any late stage initialization
936 * or semantic validation. It can fail with an error code which will
937 * be propagated back to register_netdev.
d314774c
SH
938 *
939 * void (*ndo_uninit)(struct net_device *dev);
940 * This function is called when device is unregistered or when registration
941 * fails. It is not called if init fails.
942 *
943 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 944 * This function is called when a network device transitions to the up
d314774c
SH
945 * state.
946 *
947 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 948 * This function is called when a network device transitions to the down
d314774c
SH
949 * state.
950 *
dc1f8bf6
SH
951 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
952 * struct net_device *dev);
00829823 953 * Called when a packet needs to be transmitted.
e79d8429
RR
954 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
955 * the queue before that can happen; it's for obsolete devices and weird
956 * corner cases, but the stack really does a non-trivial amount
957 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 958 * Required; cannot be NULL.
00829823 959 *
1a2a1444
DM
960 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
961 * struct net_device *dev
962 * netdev_features_t features);
963 * Called by core transmit path to determine if device is capable of
964 * performing offload operations on a given packet. This is to give
965 * the device an opportunity to implement any restrictions that cannot
966 * be otherwise expressed by feature flags. The check is called with
967 * the set of features that the stack has calculated and it returns
968 * those the driver believes to be appropriate.
cdba756f 969 *
f663dd9a 970 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
a350ecce 971 * struct net_device *sb_dev);
5e82b4b2 972 * Called to decide which queue to use when device supports multiple
00829823
SH
973 * transmit queues.
974 *
d314774c
SH
975 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
976 * This function is called to allow device receiver to make
5e82b4b2 977 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
978 *
979 * void (*ndo_set_rx_mode)(struct net_device *dev);
980 * This function is called device changes address list filtering.
01789349 981 * If driver handles unicast address filtering, it should set
5e82b4b2 982 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
983 *
984 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
985 * This function is called when the Media Access Control address
37b607c5 986 * needs to be changed. If this interface is not defined, the
5e82b4b2 987 * MAC address can not be changed.
d314774c
SH
988 *
989 * int (*ndo_validate_addr)(struct net_device *dev);
990 * Test if Media Access Control address is valid for the device.
991 *
992 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
5e82b4b2
BH
993 * Called when a user requests an ioctl which can't be handled by
994 * the generic interface code. If not defined ioctls return
d314774c
SH
995 * not supported error code.
996 *
997 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
998 * Used to set network devices bus interface parameters. This interface
5e82b4b2 999 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1000 * interface (PCI) for low level management.
1001 *
1002 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1003 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1004 * of a device.
d314774c 1005 *
00829823 1006 * void (*ndo_tx_timeout)(struct net_device *dev);
5e82b4b2 1007 * Callback used when the transmitter has not made any progress
d314774c
SH
1008 * for dev->watchdog ticks.
1009 *
bc1f4470 1010 * void (*ndo_get_stats64)(struct net_device *dev,
1011 * struct rtnl_link_stats64 *storage);
d308e38f 1012 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1013 * Called when a user wants to get the network device usage
be1f3c2c 1014 * statistics. Drivers must do one of the following:
3cfde79c
BH
1015 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1016 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1017 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1018 * (which should normally be dev->stats) and return a pointer to
1019 * it. The structure may be changed asynchronously only if each
1020 * field is written atomically.
1021 * 3. Update dev->stats asynchronously and atomically, and define
1022 * neither operation.
d314774c 1023 *
3df5b3c6 1024 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1025 * Return true if this device supports offload stats of this attr_id.
1026 *
1027 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1028 * void *attr_data)
1029 * Get statistics for offload operations by attr_id. Write it into the
1030 * attr_data pointer.
1031 *
5d632cb7 1032 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1033 * If device supports VLAN filtering this function is called when a
80d5c368 1034 * VLAN id is registered.
d314774c 1035 *
5d632cb7 1036 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1037 * If device supports VLAN filtering this function is called when a
80d5c368 1038 * VLAN id is unregistered.
d314774c
SH
1039 *
1040 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1041 *
1042 * SR-IOV management functions.
1043 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1044 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1045 * u8 qos, __be16 proto);
ed616689
SC
1046 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1047 * int max_tx_rate);
5f8444a3 1048 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1049 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1050 * int (*ndo_get_vf_config)(struct net_device *dev,
1051 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1052 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1053 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1054 * struct nlattr *port[]);
01a3d796
VZ
1055 *
1056 * Enable or disable the VF ability to query its RSS Redirection Table and
1057 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1058 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1059 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1060 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1061 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1062 * void *type_data);
6a4bc2b4
FF
1063 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1064 * This is always called from the stack with the rtnl lock held and netif
1065 * tx queues stopped. This allows the netdevice to perform queue
1066 * management safely.
c445477d 1067 *
e9bce845
YZ
1068 * Fiber Channel over Ethernet (FCoE) offload functions.
1069 * int (*ndo_fcoe_enable)(struct net_device *dev);
1070 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1071 * so the underlying device can perform whatever needed configuration or
1072 * initialization to support acceleration of FCoE traffic.
1073 *
1074 * int (*ndo_fcoe_disable)(struct net_device *dev);
1075 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1076 * so the underlying device can perform whatever needed clean-ups to
1077 * stop supporting acceleration of FCoE traffic.
1078 *
1079 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1080 * struct scatterlist *sgl, unsigned int sgc);
1081 * Called when the FCoE Initiator wants to initialize an I/O that
1082 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1083 * perform necessary setup and returns 1 to indicate the device is set up
1084 * successfully to perform DDP on this I/O, otherwise this returns 0.
1085 *
1086 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1087 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1088 * indicated by the FC exchange id 'xid', so the underlying device can
1089 * clean up and reuse resources for later DDP requests.
1090 *
1091 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1092 * struct scatterlist *sgl, unsigned int sgc);
1093 * Called when the FCoE Target wants to initialize an I/O that
1094 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1095 * perform necessary setup and returns 1 to indicate the device is set up
1096 * successfully to perform DDP on this I/O, otherwise this returns 0.
1097 *
68bad94e
NP
1098 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1099 * struct netdev_fcoe_hbainfo *hbainfo);
1100 * Called when the FCoE Protocol stack wants information on the underlying
1101 * device. This information is utilized by the FCoE protocol stack to
1102 * register attributes with Fiber Channel management service as per the
1103 * FC-GS Fabric Device Management Information(FDMI) specification.
1104 *
e9bce845
YZ
1105 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1106 * Called when the underlying device wants to override default World Wide
1107 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1108 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1109 * protocol stack to use.
1110 *
c445477d
BH
1111 * RFS acceleration.
1112 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1113 * u16 rxq_index, u32 flow_id);
1114 * Set hardware filter for RFS. rxq_index is the target queue index;
1115 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1116 * Return the filter ID on success, or a negative error code.
fbaec0ea 1117 *
8b98a70c 1118 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1119 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1120 * Called to make another netdev an underling.
1121 *
1122 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1123 * Called to release previously enslaved netdev.
5455c699
MM
1124 *
1125 * Feature/offload setting functions.
1a2a1444
DM
1126 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1127 * netdev_features_t features);
1128 * Adjusts the requested feature flags according to device-specific
1129 * constraints, and returns the resulting flags. Must not modify
1130 * the device state.
1131 *
c8f44aff 1132 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1133 * Called to update device configuration to new features. Passed
1134 * feature set might be less than what was returned by ndo_fix_features()).
1135 * Must return >0 or -errno if it changed dev->features itself.
1136 *
edc7d573 1137 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1138 * struct net_device *dev,
87b0984e
PM
1139 * const unsigned char *addr, u16 vid, u16 flags,
1140 * struct netlink_ext_ack *extack);
77162022 1141 * Adds an FDB entry to dev for addr.
1690be63
VY
1142 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1143 * struct net_device *dev,
f6f6424b 1144 * const unsigned char *addr, u16 vid)
77162022
JF
1145 * Deletes the FDB entry from dev coresponding to addr.
1146 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1147 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1148 * int *idx)
77162022
JF
1149 * Used to add FDB entries to dump requests. Implementers should add
1150 * entries to skb and update idx with the number of entries.
e5a55a89 1151 *
ad41faa8 1152 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
2fd527b7 1153 * u16 flags, struct netlink_ext_ack *extack)
e5a55a89 1154 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1155 * struct net_device *dev, u32 filter_mask,
1156 * int nlflags)
ad41faa8
ND
1157 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1158 * u16 flags);
4bf84c35
JP
1159 *
1160 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1161 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1162 * which do not represent real hardware may define this to allow their
1163 * userspace components to manage their virtual carrier state. Devices
1164 * that determine carrier state from physical hardware properties (eg
1165 * network cables) or protocol-dependent mechanisms (eg
1166 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1167 *
1168 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1169 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1170 * Called to get ID of physical port of this device. If driver does
1171 * not implement this, it is assumed that the hw is not able to have
1172 * multiple net devices on single physical port.
53cf5275 1173 *
d6abc596
FF
1174 * int (*ndo_get_port_parent_id)(struct net_device *dev,
1175 * struct netdev_phys_item_id *ppid)
1176 * Called to get the parent ID of the physical port of this device.
1177 *
7c46a640
AD
1178 * void (*ndo_udp_tunnel_add)(struct net_device *dev,
1179 * struct udp_tunnel_info *ti);
1180 * Called by UDP tunnel to notify a driver about the UDP port and socket
1181 * address family that a UDP tunnel is listnening to. It is called only
1182 * when a new port starts listening. The operation is protected by the
1183 * RTNL.
1184 *
1185 * void (*ndo_udp_tunnel_del)(struct net_device *dev,
1186 * struct udp_tunnel_info *ti);
1187 * Called by UDP tunnel to notify the driver about a UDP port and socket
1188 * address family that the UDP tunnel is not listening to anymore. The
1189 * operation is protected by the RTNL.
1190 *
a6cc0cfa
JF
1191 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1192 * struct net_device *dev)
1193 * Called by upper layer devices to accelerate switching or other
1194 * station functionality into hardware. 'pdev is the lowerdev
1195 * to use for the offload and 'dev' is the net device that will
1196 * back the offload. Returns a pointer to the private structure
1197 * the upper layer will maintain.
1198 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1199 * Called by upper layer device to delete the station created
1200 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1201 * the station and priv is the structure returned by the add
1202 * operation.
822b3b2e
JF
1203 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1204 * int queue_index, u32 maxrate);
1205 * Called when a user wants to set a max-rate limitation of specific
1206 * TX queue.
a54acb3a
ND
1207 * int (*ndo_get_iflink)(const struct net_device *dev);
1208 * Called to get the iflink value of this device.
d746d707 1209 * void (*ndo_change_proto_down)(struct net_device *dev,
5e82b4b2 1210 * bool proto_down);
d746d707
AK
1211 * This function is used to pass protocol port error state information
1212 * to the switch driver. The switch driver can react to the proto_down
1213 * by doing a phys down on the associated switch port.
fc4099f1
PS
1214 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1215 * This function is used to get egress tunnel information for given skb.
1216 * This is useful for retrieving outer tunnel header parameters while
1217 * sampling packet.
871b642a
PA
1218 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1219 * This function is used to specify the headroom that the skb must
1220 * consider when allocation skb during packet reception. Setting
1221 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1222 * forward. Setting a negative value resets the rx headroom to the
871b642a 1223 * default value.
f4e63525 1224 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1225 * This function is used to set or query state related to XDP on the
f4e63525
JK
1226 * netdevice and manage BPF offload. See definition of
1227 * enum bpf_netdev_command for details.
42b33468
JDB
1228 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1229 * u32 flags);
735fc405
JDB
1230 * This function is used to submit @n XDP packets for transmit on a
1231 * netdevice. Returns number of frames successfully transmitted, frames
1232 * that got dropped are freed/returned via xdp_return_frame().
1233 * Returns negative number, means general error invoking ndo, meaning
1234 * no frames were xmit'ed and core-caller will free all frames.
9116e5e2
MK
1235 * int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
1236 * This function is used to wake up the softirq, ksoftirqd or kthread
1237 * responsible for sending and/or receiving packets on a specific
1238 * queue id bound to an AF_XDP socket. The flags field specifies if
1239 * only RX, only Tx, or both should be woken up using the flags
1240 * XDP_WAKEUP_RX and XDP_WAKEUP_TX.
5dc37bb9
JP
1241 * struct devlink_port *(*ndo_get_devlink_port)(struct net_device *dev);
1242 * Get devlink port instance associated with a given netdev.
b473b0d2
JK
1243 * Called with a reference on the netdevice and devlink locks only,
1244 * rtnl_lock is not held.
d314774c
SH
1245 */
1246struct net_device_ops {
1247 int (*ndo_init)(struct net_device *dev);
1248 void (*ndo_uninit)(struct net_device *dev);
1249 int (*ndo_open)(struct net_device *dev);
1250 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1251 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1252 struct net_device *dev);
1253 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1254 struct net_device *dev,
1255 netdev_features_t features);
00829823 1256 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1257 struct sk_buff *skb,
a350ecce 1258 struct net_device *sb_dev);
d314774c
SH
1259 void (*ndo_change_rx_flags)(struct net_device *dev,
1260 int flags);
d314774c 1261 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1262 int (*ndo_set_mac_address)(struct net_device *dev,
1263 void *addr);
d314774c 1264 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1265 int (*ndo_do_ioctl)(struct net_device *dev,
1266 struct ifreq *ifr, int cmd);
d314774c
SH
1267 int (*ndo_set_config)(struct net_device *dev,
1268 struct ifmap *map);
00829823
SH
1269 int (*ndo_change_mtu)(struct net_device *dev,
1270 int new_mtu);
1271 int (*ndo_neigh_setup)(struct net_device *dev,
1272 struct neigh_parms *);
d314774c
SH
1273 void (*ndo_tx_timeout) (struct net_device *dev);
1274
bc1f4470 1275 void (*ndo_get_stats64)(struct net_device *dev,
1276 struct rtnl_link_stats64 *storage);
3df5b3c6 1277 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1278 int (*ndo_get_offload_stats)(int attr_id,
1279 const struct net_device *dev,
1280 void *attr_data);
d314774c
SH
1281 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1282
8e586137 1283 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1284 __be16 proto, u16 vid);
8e586137 1285 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1286 __be16 proto, u16 vid);
d314774c 1287#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1288 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1289 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1290 struct netpoll_info *info);
0e34e931 1291 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1292#endif
95c26df8
WM
1293 int (*ndo_set_vf_mac)(struct net_device *dev,
1294 int queue, u8 *mac);
1295 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1296 int queue, u16 vlan,
1297 u8 qos, __be16 proto);
ed616689
SC
1298 int (*ndo_set_vf_rate)(struct net_device *dev,
1299 int vf, int min_tx_rate,
1300 int max_tx_rate);
5f8444a3
GR
1301 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1302 int vf, bool setting);
dd461d6a
HS
1303 int (*ndo_set_vf_trust)(struct net_device *dev,
1304 int vf, bool setting);
95c26df8
WM
1305 int (*ndo_get_vf_config)(struct net_device *dev,
1306 int vf,
1307 struct ifla_vf_info *ivf);
1d8faf48
RE
1308 int (*ndo_set_vf_link_state)(struct net_device *dev,
1309 int vf, int link_state);
3b766cd8
EBE
1310 int (*ndo_get_vf_stats)(struct net_device *dev,
1311 int vf,
1312 struct ifla_vf_stats
1313 *vf_stats);
57b61080
SF
1314 int (*ndo_set_vf_port)(struct net_device *dev,
1315 int vf,
1316 struct nlattr *port[]);
1317 int (*ndo_get_vf_port)(struct net_device *dev,
1318 int vf, struct sk_buff *skb);
cc8e27cc
EC
1319 int (*ndo_set_vf_guid)(struct net_device *dev,
1320 int vf, u64 guid,
1321 int guid_type);
01a3d796
VZ
1322 int (*ndo_set_vf_rss_query_en)(
1323 struct net_device *dev,
1324 int vf, bool setting);
16e5cc64 1325 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1326 enum tc_setup_type type,
de4784ca 1327 void *type_data);
d11ead75 1328#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1329 int (*ndo_fcoe_enable)(struct net_device *dev);
1330 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1331 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1332 u16 xid,
1333 struct scatterlist *sgl,
1334 unsigned int sgc);
1335 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1336 u16 xid);
6247e086
YZ
1337 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1338 u16 xid,
1339 struct scatterlist *sgl,
1340 unsigned int sgc);
68bad94e
NP
1341 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1342 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1343#endif
1344
d11ead75 1345#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1346#define NETDEV_FCOE_WWNN 0
1347#define NETDEV_FCOE_WWPN 1
1348 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1349 u64 *wwn, int type);
4d288d57 1350#endif
3c9c36bc 1351
c445477d
BH
1352#ifdef CONFIG_RFS_ACCEL
1353 int (*ndo_rx_flow_steer)(struct net_device *dev,
1354 const struct sk_buff *skb,
1355 u16 rxq_index,
1356 u32 flow_id);
1357#endif
fbaec0ea 1358 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1359 struct net_device *slave_dev,
1360 struct netlink_ext_ack *extack);
fbaec0ea
JP
1361 int (*ndo_del_slave)(struct net_device *dev,
1362 struct net_device *slave_dev);
c8f44aff
MM
1363 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1364 netdev_features_t features);
5455c699 1365 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1366 netdev_features_t features);
503eebc2
JP
1367 int (*ndo_neigh_construct)(struct net_device *dev,
1368 struct neighbour *n);
1369 void (*ndo_neigh_destroy)(struct net_device *dev,
1370 struct neighbour *n);
77162022
JF
1371
1372 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1373 struct nlattr *tb[],
77162022 1374 struct net_device *dev,
6b6e2725 1375 const unsigned char *addr,
f6f6424b 1376 u16 vid,
87b0984e
PM
1377 u16 flags,
1378 struct netlink_ext_ack *extack);
77162022 1379 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1380 struct nlattr *tb[],
77162022 1381 struct net_device *dev,
f6f6424b
JP
1382 const unsigned char *addr,
1383 u16 vid);
77162022
JF
1384 int (*ndo_fdb_dump)(struct sk_buff *skb,
1385 struct netlink_callback *cb,
1386 struct net_device *dev,
5d5eacb3 1387 struct net_device *filter_dev,
d297653d 1388 int *idx);
5b2f94b2
RP
1389 int (*ndo_fdb_get)(struct sk_buff *skb,
1390 struct nlattr *tb[],
1391 struct net_device *dev,
1392 const unsigned char *addr,
1393 u16 vid, u32 portid, u32 seq,
1394 struct netlink_ext_ack *extack);
e5a55a89 1395 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3 1396 struct nlmsghdr *nlh,
2fd527b7
PM
1397 u16 flags,
1398 struct netlink_ext_ack *extack);
e5a55a89
JF
1399 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1400 u32 pid, u32 seq,
6cbdceeb 1401 struct net_device *dev,
46c264da
ND
1402 u32 filter_mask,
1403 int nlflags);
407af329 1404 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1405 struct nlmsghdr *nlh,
1406 u16 flags);
4bf84c35
JP
1407 int (*ndo_change_carrier)(struct net_device *dev,
1408 bool new_carrier);
66b52b0d 1409 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1410 struct netdev_phys_item_id *ppid);
d6abc596
FF
1411 int (*ndo_get_port_parent_id)(struct net_device *dev,
1412 struct netdev_phys_item_id *ppid);
db24a904
DA
1413 int (*ndo_get_phys_port_name)(struct net_device *dev,
1414 char *name, size_t len);
7c46a640
AD
1415 void (*ndo_udp_tunnel_add)(struct net_device *dev,
1416 struct udp_tunnel_info *ti);
1417 void (*ndo_udp_tunnel_del)(struct net_device *dev,
1418 struct udp_tunnel_info *ti);
a6cc0cfa
JF
1419 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1420 struct net_device *dev);
1421 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1422 void *priv);
1423
25175ba5 1424 int (*ndo_get_lock_subclass)(struct net_device *dev);
822b3b2e
JF
1425 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1426 int queue_index,
1427 u32 maxrate);
a54acb3a 1428 int (*ndo_get_iflink)(const struct net_device *dev);
d746d707
AK
1429 int (*ndo_change_proto_down)(struct net_device *dev,
1430 bool proto_down);
fc4099f1
PS
1431 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1432 struct sk_buff *skb);
871b642a
PA
1433 void (*ndo_set_rx_headroom)(struct net_device *dev,
1434 int needed_headroom);
f4e63525
JK
1435 int (*ndo_bpf)(struct net_device *dev,
1436 struct netdev_bpf *bpf);
735fc405 1437 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1438 struct xdp_frame **xdp,
1439 u32 flags);
9116e5e2
MK
1440 int (*ndo_xsk_wakeup)(struct net_device *dev,
1441 u32 queue_id, u32 flags);
5dc37bb9 1442 struct devlink_port * (*ndo_get_devlink_port)(struct net_device *dev);
d314774c
SH
1443};
1444
7aa98047
LR
1445/**
1446 * enum net_device_priv_flags - &struct net_device priv_flags
1447 *
1448 * These are the &struct net_device, they are only set internally
1449 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1450 * userspace; this means that the order of these flags can change
7aa98047
LR
1451 * during any kernel release.
1452 *
1453 * You should have a pretty good reason to be extending these flags.
1454 *
1455 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1456 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1457 * @IFF_BONDING: bonding master or slave
7aa98047 1458 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1459 * @IFF_WAN_HDLC: WAN HDLC device
1460 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1461 * release skb->dst
1462 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1463 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1464 * @IFF_MACVLAN_PORT: device used as macvlan port
1465 * @IFF_BRIDGE_PORT: device used as bridge port
1466 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1467 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1468 * @IFF_UNICAST_FLT: Supports unicast filtering
1469 * @IFF_TEAM_PORT: device used as team port
1470 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1471 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1472 * change when it's running
1473 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1474 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1475 * underlying stacked devices
007979ea 1476 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1477 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1478 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1479 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1480 * @IFF_TEAM: device is a team device
d4ab4286 1481 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1482 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1483 * entity (i.e. the master device for bridged veth)
3c175784 1484 * @IFF_MACSEC: device is a MACsec device
f5426250 1485 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1486 * @IFF_FAILOVER: device is a failover master device
1487 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
d5256083 1488 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
8065a779 1489 * @IFF_LIVE_RENAME_OK: rename is allowed while device is up and running
7aa98047
LR
1490 */
1491enum netdev_priv_flags {
1492 IFF_802_1Q_VLAN = 1<<0,
1493 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1494 IFF_BONDING = 1<<2,
1495 IFF_ISATAP = 1<<3,
1496 IFF_WAN_HDLC = 1<<4,
1497 IFF_XMIT_DST_RELEASE = 1<<5,
1498 IFF_DONT_BRIDGE = 1<<6,
1499 IFF_DISABLE_NETPOLL = 1<<7,
1500 IFF_MACVLAN_PORT = 1<<8,
1501 IFF_BRIDGE_PORT = 1<<9,
1502 IFF_OVS_DATAPATH = 1<<10,
1503 IFF_TX_SKB_SHARING = 1<<11,
1504 IFF_UNICAST_FLT = 1<<12,
1505 IFF_TEAM_PORT = 1<<13,
1506 IFF_SUPP_NOFCS = 1<<14,
1507 IFF_LIVE_ADDR_CHANGE = 1<<15,
1508 IFF_MACVLAN = 1<<16,
1509 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1510 IFF_L3MDEV_MASTER = 1<<18,
1511 IFF_NO_QUEUE = 1<<19,
1512 IFF_OPENVSWITCH = 1<<20,
1513 IFF_L3MDEV_SLAVE = 1<<21,
1514 IFF_TEAM = 1<<22,
1515 IFF_RXFH_CONFIGURED = 1<<23,
1516 IFF_PHONY_HEADROOM = 1<<24,
1517 IFF_MACSEC = 1<<25,
f5426250 1518 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1519 IFF_FAILOVER = 1<<27,
1520 IFF_FAILOVER_SLAVE = 1<<28,
d5256083 1521 IFF_L3MDEV_RX_HANDLER = 1<<29,
8065a779 1522 IFF_LIVE_RENAME_OK = 1<<30,
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
d5256083 1553#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
8065a779 1554#define IFF_LIVE_RENAME_OK IFF_LIVE_RENAME_OK
7aa98047 1555
536721b1
KK
1556/**
1557 * struct net_device - The DEVICE structure.
d651983d
MCC
1558 *
1559 * Actually, this whole structure is a big mistake. It mixes I/O
1560 * data with strictly "high-level" data, and it has to know about
1561 * almost every data structure used in the INET module.
536721b1
KK
1562 *
1563 * @name: This is the first field of the "visible" part of this structure
1564 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1565 * of the interface.
536721b1
KK
1566 *
1567 * @name_hlist: Device name hash chain, please keep it close to name[]
1568 * @ifalias: SNMP alias
1569 * @mem_end: Shared memory end
1570 * @mem_start: Shared memory start
1571 * @base_addr: Device I/O address
1572 * @irq: Device IRQ number
1573 *
1574 * @state: Generic network queuing layer state, see netdev_state_t
1575 * @dev_list: The global list of network devices
5e82b4b2
BH
1576 * @napi_list: List entry used for polling NAPI devices
1577 * @unreg_list: List entry when we are unregistering the
1578 * device; see the function unregister_netdev
1579 * @close_list: List entry used when we are closing the device
62d885fe
BP
1580 * @ptype_all: Device-specific packet handlers for all protocols
1581 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1582 *
1583 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1584 * @features: Currently active device features
1585 * @hw_features: User-changeable features
1586 *
1587 * @wanted_features: User-requested features
1588 * @vlan_features: Mask of features inheritable by VLAN devices
1589 *
1590 * @hw_enc_features: Mask of features inherited by encapsulating devices
1591 * This field indicates what encapsulation
1592 * offloads the hardware is capable of doing,
1593 * and drivers will need to set them appropriately.
1594 *
1595 * @mpls_features: Mask of features inheritable by MPLS
1596 *
1597 * @ifindex: interface index
5e82b4b2 1598 * @group: The group the device belongs to
536721b1
KK
1599 *
1600 * @stats: Statistics struct, which was left as a legacy, use
1601 * rtnl_link_stats64 instead
1602 *
1603 * @rx_dropped: Dropped packets by core network,
1604 * do not use this in drivers
1605 * @tx_dropped: Dropped packets by core network,
1606 * do not use this in drivers
6e7333d3
JW
1607 * @rx_nohandler: nohandler dropped packets by core network on
1608 * inactive devices, do not use this in drivers
9e55e5d3
FF
1609 * @carrier_up_count: Number of times the carrier has been up
1610 * @carrier_down_count: Number of times the carrier has been down
536721b1 1611 *
536721b1
KK
1612 * @wireless_handlers: List of functions to handle Wireless Extensions,
1613 * instead of ioctl,
1614 * see <net/iw_handler.h> for details.
1615 * @wireless_data: Instance data managed by the core of wireless extensions
1616 *
1617 * @netdev_ops: Includes several pointers to callbacks,
1618 * if one wants to override the ndo_*() functions
1619 * @ethtool_ops: Management operations
f997c55c
AA
1620 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1621 * discovery handling. Necessary for e.g. 6LoWPAN.
d476059e 1622 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1623 * of Layer 2 headers.
1624 *
1625 * @flags: Interface flags (a la BSD)
1626 * @priv_flags: Like 'flags' but invisible to userspace,
1627 * see if.h for the definitions
1628 * @gflags: Global flags ( kept as legacy )
1629 * @padded: How much padding added by alloc_netdev()
1630 * @operstate: RFC2863 operstate
1631 * @link_mode: Mapping policy to operstate
1632 * @if_port: Selectable AUI, TP, ...
1633 * @dma: DMA channel
1634 * @mtu: Interface MTU value
61e84623
JW
1635 * @min_mtu: Interface Minimum MTU value
1636 * @max_mtu: Interface Maximum MTU value
536721b1 1637 * @type: Interface hardware type
2793a23a 1638 * @hard_header_len: Maximum hardware header length.
217e6fa2 1639 * @min_header_len: Minimum hardware header length
536721b1
KK
1640 *
1641 * @needed_headroom: Extra headroom the hardware may need, but not in all
1642 * cases can this be guaranteed
1643 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1644 * cases can this be guaranteed. Some cases also use
1645 * LL_MAX_HEADER instead to allocate the skb
1646 *
1647 * interface address info:
1648 *
1649 * @perm_addr: Permanent hw address
1650 * @addr_assign_type: Hw address assignment type
1651 * @addr_len: Hardware address length
8626a0c8 1652 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1653 * @dev_id: Used to differentiate devices that share
1654 * the same link layer address
1655 * @dev_port: Used to differentiate devices that share
1656 * the same function
1657 * @addr_list_lock: XXX: need comments on this one
5e82b4b2 1658 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1659 * has been enabled due to the need to listen to
1660 * additional unicast addresses in a device that
1661 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1662 * @uc: unicast mac addresses
1663 * @mc: multicast mac addresses
1664 * @dev_addrs: list of device hw addresses
1665 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1666 * @promiscuity: Number of times the NIC is told to work in
1667 * promiscuous mode; if it becomes 0 the NIC will
1668 * exit promiscuous mode
536721b1
KK
1669 * @allmulti: Counter, enables or disables allmulticast mode
1670 *
1671 * @vlan_info: VLAN info
1672 * @dsa_ptr: dsa specific data
1673 * @tipc_ptr: TIPC specific data
1674 * @atalk_ptr: AppleTalk link
1675 * @ip_ptr: IPv4 specific data
1676 * @dn_ptr: DECnet specific data
1677 * @ip6_ptr: IPv6 specific data
1678 * @ax25_ptr: AX.25 specific data
1679 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
1680 *
536721b1
KK
1681 * @dev_addr: Hw address (before bcast,
1682 * because most packets are unicast)
1683 *
1684 * @_rx: Array of RX queues
1685 * @num_rx_queues: Number of RX queues
1686 * allocated at register_netdev() time
1687 * @real_num_rx_queues: Number of RX queues currently active in device
1688 *
1689 * @rx_handler: handler for received packets
1690 * @rx_handler_data: XXX: need comments on this one
46209401
JP
1691 * @miniq_ingress: ingress/clsact qdisc specific data for
1692 * ingress processing
536721b1
KK
1693 * @ingress_queue: XXX: need comments on this one
1694 * @broadcast: hw bcast address
1695 *
14ffbbb8
TG
1696 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1697 * indexed by RX queue number. Assigned by driver.
1698 * This must only be set if the ndo_rx_flow_steer
1699 * operation is defined
1700 * @index_hlist: Device index hash chain
1701 *
536721b1
KK
1702 * @_tx: Array of TX queues
1703 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1704 * @real_num_tx_queues: Number of TX queues currently active in device
1705 * @qdisc: Root qdisc from userspace point of view
1706 * @tx_queue_len: Max frames per queue allowed
1707 * @tx_global_lock: XXX: need comments on this one
1708 *
1709 * @xps_maps: XXX: need comments on this one
46209401
JP
1710 * @miniq_egress: clsact qdisc specific data for
1711 * egress processing
536721b1 1712 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1713 * the watchdog (see dev_watchdog())
536721b1
KK
1714 * @watchdog_timer: List of timers
1715 *
1716 * @pcpu_refcnt: Number of references to this device
1717 * @todo_list: Delayed register/unregister
536721b1
KK
1718 * @link_watch_list: XXX: need comments on this one
1719 *
1720 * @reg_state: Register/unregister state machine
1721 * @dismantle: Device is going to be freed
1722 * @rtnl_link_state: This enum represents the phases of creating
1723 * a new link
1724 *
cf124db5
DM
1725 * @needs_free_netdev: Should unregister perform free_netdev?
1726 * @priv_destructor: Called from unregister
536721b1
KK
1727 * @npinfo: XXX: need comments on this one
1728 * @nd_net: Network namespace this network device is inside
1729 *
1730 * @ml_priv: Mid-layer private
1731 * @lstats: Loopback statistics
1732 * @tstats: Tunnel statistics
1733 * @dstats: Dummy statistics
1734 * @vstats: Virtual ethernet statistics
1735 *
1736 * @garp_port: GARP
1737 * @mrp_port: MRP
1738 *
1739 * @dev: Class/net/name entry
1740 * @sysfs_groups: Space for optional device, statistics and wireless
1741 * sysfs groups
1742 *
1743 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1744 * @rtnl_link_ops: Rtnl_link_ops
1745 *
1746 * @gso_max_size: Maximum size of generic segmentation offload
1747 * @gso_max_segs: Maximum number of segments that can be passed to the
1748 * NIC for GSO
1749 *
1750 * @dcbnl_ops: Data Center Bridging netlink ops
1751 * @num_tc: Number of traffic classes in the net device
1752 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1753 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
1754 *
1755 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1756 *
1757 * @priomap: XXX: need comments on this one
1758 * @phydev: Physical device may attach itself
1759 * for hardware timestamping
e679c9c1 1760 * @sfp_bus: attached &struct sfp_bus structure.
536721b1 1761 *
123b3652
ED
1762 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
1763 * @qdisc_running_key: lockdep class annotating Qdisc->running seqcount
536721b1 1764 *
d746d707
AK
1765 * @proto_down: protocol port state information can be sent to the
1766 * switch driver and used to set the phys state of the
1767 * switch port.
1768 *
61941143
HK
1769 * @wol_enabled: Wake-on-LAN is enabled
1770 *
1da177e4
LT
1771 * FIXME: cleanup struct net_device such that network protocol info
1772 * moves out.
1773 */
1774
d94d9fee 1775struct net_device {
1da177e4 1776 char name[IFNAMSIZ];
9356b8fc 1777 struct hlist_node name_hlist;
6c557001 1778 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
1779 /*
1780 * I/O specific fields
1781 * FIXME: Merge these and struct ifmap into one
1782 */
536721b1
KK
1783 unsigned long mem_end;
1784 unsigned long mem_start;
1785 unsigned long base_addr;
1786 int irq;
1da177e4
LT
1787
1788 /*
536721b1
KK
1789 * Some hardware also needs these fields (state,dev_list,
1790 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1791 * part of the usual set specified in Space.c.
1792 */
1793
1da177e4
LT
1794 unsigned long state;
1795
7562f876 1796 struct list_head dev_list;
bea3348e 1797 struct list_head napi_list;
44a0873d 1798 struct list_head unreg_list;
5cde2829 1799 struct list_head close_list;
7866a621
SN
1800 struct list_head ptype_all;
1801 struct list_head ptype_specific;
2f268f12 1802
2f268f12
VF
1803 struct {
1804 struct list_head upper;
1805 struct list_head lower;
1806 } adj_list;
1807
c8f44aff 1808 netdev_features_t features;
c8f44aff 1809 netdev_features_t hw_features;
c8f44aff 1810 netdev_features_t wanted_features;
c8f44aff 1811 netdev_features_t vlan_features;
6a674e9c 1812 netdev_features_t hw_enc_features;
0d89d203 1813 netdev_features_t mpls_features;
802ab55a 1814 netdev_features_t gso_partial_features;
04ed3e74 1815
1da177e4 1816 int ifindex;
7a66bbc9 1817 int group;
1da177e4 1818
c45d286e 1819 struct net_device_stats stats;
015f0688 1820
015f0688
ED
1821 atomic_long_t rx_dropped;
1822 atomic_long_t tx_dropped;
6e7333d3 1823 atomic_long_t rx_nohandler;
1da177e4 1824
b2d3bcfa
DD
1825 /* Stats to monitor link on/off, flapping */
1826 atomic_t carrier_up_count;
1827 atomic_t carrier_down_count;
1828
b86e0280 1829#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
1830 const struct iw_handler_def *wireless_handlers;
1831 struct iw_public_data *wireless_data;
b86e0280 1832#endif
d314774c 1833 const struct net_device_ops *netdev_ops;
76fd8593 1834 const struct ethtool_ops *ethtool_ops;
1b69c6d0
DA
1835#ifdef CONFIG_NET_L3_MASTER_DEV
1836 const struct l3mdev_ops *l3mdev_ops;
1837#endif
f997c55c
AA
1838#if IS_ENABLED(CONFIG_IPV6)
1839 const struct ndisc_ops *ndisc_ops;
1840#endif
1da177e4 1841
9cb0d21d 1842#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
1843 const struct xfrmdev_ops *xfrmdev_ops;
1844#endif
1845
a5c37c63
IL
1846#if IS_ENABLED(CONFIG_TLS_DEVICE)
1847 const struct tlsdev_ops *tlsdev_ops;
1848#endif
1849
3b04ddde
SH
1850 const struct header_ops *header_ops;
1851
536721b1
KK
1852 unsigned int flags;
1853 unsigned int priv_flags;
1854
1da177e4 1855 unsigned short gflags;
536721b1 1856 unsigned short padded;
1da177e4 1857
536721b1
KK
1858 unsigned char operstate;
1859 unsigned char link_mode;
b00055aa 1860
536721b1
KK
1861 unsigned char if_port;
1862 unsigned char dma;
bdc220da 1863
536721b1 1864 unsigned int mtu;
61e84623
JW
1865 unsigned int min_mtu;
1866 unsigned int max_mtu;
536721b1
KK
1867 unsigned short type;
1868 unsigned short hard_header_len;
d92be7a4 1869 unsigned char min_header_len;
1da177e4 1870
f5184d26
JB
1871 unsigned short needed_headroom;
1872 unsigned short needed_tailroom;
1873
1da177e4 1874 /* Interface address info. */
536721b1
KK
1875 unsigned char perm_addr[MAX_ADDR_LEN];
1876 unsigned char addr_assign_type;
1877 unsigned char addr_len;
a0a9663d 1878 unsigned short neigh_priv_len;
536721b1
KK
1879 unsigned short dev_id;
1880 unsigned short dev_port;
ccffad25 1881 spinlock_t addr_list_lock;
14ffbbb8
TG
1882 unsigned char name_assign_type;
1883 bool uc_promisc;
536721b1
KK
1884 struct netdev_hw_addr_list uc;
1885 struct netdev_hw_addr_list mc;
1886 struct netdev_hw_addr_list dev_addrs;
1887
4c3d5e7b
ED
1888#ifdef CONFIG_SYSFS
1889 struct kset *queues_kset;
1890#endif
9d45abe1
WC
1891 unsigned int promiscuity;
1892 unsigned int allmulti;
1da177e4 1893
1da177e4 1894
5e82b4b2 1895 /* Protocol-specific pointers */
65ac6a5f 1896
d11ead75 1897#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 1898 struct vlan_info __rcu *vlan_info;
65ac6a5f 1899#endif
34a430d7 1900#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 1901 struct dsa_port *dsa_ptr;
37cb0620
YX
1902#endif
1903#if IS_ENABLED(CONFIG_TIPC)
536721b1 1904 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 1905#endif
89e58148 1906#if IS_ENABLED(CONFIG_IRDA) || IS_ENABLED(CONFIG_ATALK)
536721b1 1907 void *atalk_ptr;
89e58148 1908#endif
536721b1 1909 struct in_device __rcu *ip_ptr;
330c7272 1910#if IS_ENABLED(CONFIG_DECNET)
536721b1 1911 struct dn_dev __rcu *dn_ptr;
330c7272 1912#endif
536721b1 1913 struct inet6_dev __rcu *ip6_ptr;
19ff13f2 1914#if IS_ENABLED(CONFIG_AX25)
536721b1 1915 void *ax25_ptr;
19ff13f2 1916#endif
536721b1 1917 struct wireless_dev *ieee80211_ptr;
98a18b6f 1918 struct wpan_dev *ieee802154_ptr;
03c57747
RS
1919#if IS_ENABLED(CONFIG_MPLS_ROUTING)
1920 struct mpls_dev __rcu *mpls_ptr;
1921#endif
1da177e4 1922
9356b8fc 1923/*
cd13539b 1924 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 1925 */
9356b8fc 1926 /* Interface address info used in eth_type_trans() */
536721b1 1927 unsigned char *dev_addr;
f001fde5 1928
0a9627f2 1929 struct netdev_rx_queue *_rx;
0a9627f2 1930 unsigned int num_rx_queues;
62fe0b40 1931 unsigned int real_num_rx_queues;
0a9627f2 1932
7acedaf5 1933 struct bpf_prog __rcu *xdp_prog;
3b47d303 1934 unsigned long gro_flush_timeout;
61391cde 1935 rx_handler_func_t __rcu *rx_handler;
1936 void __rcu *rx_handler_data;
e8a0464c 1937
4cda01e8 1938#ifdef CONFIG_NET_CLS_ACT
46209401 1939 struct mini_Qdisc __rcu *miniq_ingress;
d2788d34 1940#endif
24824a09 1941 struct netdev_queue __rcu *ingress_queue;
e687ad60 1942#ifdef CONFIG_NETFILTER_INGRESS
960632ec 1943 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 1944#endif
d2788d34 1945
536721b1 1946 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
1947#ifdef CONFIG_RFS_ACCEL
1948 struct cpu_rmap *rx_cpu_rmap;
1949#endif
1950 struct hlist_node index_hlist;
cd13539b
ED
1951
1952/*
1953 * Cache lines mostly used on transmit path
1954 */
e8a0464c
DM
1955 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
1956 unsigned int num_tx_queues;
fd2ea0a7 1957 unsigned int real_num_tx_queues;
af356afa 1958 struct Qdisc *qdisc;
59cc1f61
JK
1959#ifdef CONFIG_NET_SCHED
1960 DECLARE_HASHTABLE (qdisc_hash, 4);
1961#endif
0cd29503 1962 unsigned int tx_queue_len;
c3f26a26 1963 spinlock_t tx_global_lock;
14ffbbb8 1964 int watchdog_timeo;
cd13539b 1965
bf264145 1966#ifdef CONFIG_XPS
80d19669
AN
1967 struct xps_dev_maps __rcu *xps_cpus_map;
1968 struct xps_dev_maps __rcu *xps_rxqs_map;
bf264145 1969#endif
1f211a1b 1970#ifdef CONFIG_NET_CLS_ACT
46209401 1971 struct mini_Qdisc __rcu *miniq_egress;
1f211a1b 1972#endif
0c4f691f 1973
9356b8fc 1974 /* These may be needed for future network-power-down code. */
9356b8fc
ED
1975 struct timer_list watchdog_timer;
1976
29b4433d 1977 int __percpu *pcpu_refcnt;
1da177e4 1978 struct list_head todo_list;
1da177e4 1979
e014debe 1980 struct list_head link_watch_list;
572a103d 1981
1da177e4 1982 enum { NETREG_UNINITIALIZED=0,
b17a7c17 1983 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
1984 NETREG_UNREGISTERING, /* called unregister_netdevice */
1985 NETREG_UNREGISTERED, /* completed unregister todo */
1986 NETREG_RELEASED, /* called free_netdev */
937f1ba5 1987 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
1988 } reg_state:8;
1989
536721b1 1990 bool dismantle;
a2835763
PM
1991
1992 enum {
1993 RTNL_LINK_INITIALIZED,
1994 RTNL_LINK_INITIALIZING,
1995 } rtnl_link_state:16;
1da177e4 1996
cf124db5
DM
1997 bool needs_free_netdev;
1998 void (*priv_destructor)(struct net_device *dev);
1da177e4 1999
1da177e4 2000#ifdef CONFIG_NETPOLL
5fbee843 2001 struct netpoll_info __rcu *npinfo;
1da177e4 2002#endif
eae792b7 2003
0c5c9fb5 2004 possible_net_t nd_net;
4a1c5371 2005
4951704b 2006 /* mid-layer private */
a7855c78 2007 union {
536721b1
KK
2008 void *ml_priv;
2009 struct pcpu_lstats __percpu *lstats;
8f84985f 2010 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2011 struct pcpu_dstats __percpu *dstats;
a7855c78 2012 };
536721b1 2013
fb585b44 2014#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2015 struct garp_port __rcu *garp_port;
fb585b44
TK
2016#endif
2017#if IS_ENABLED(CONFIG_MRP)
febf018d 2018 struct mrp_port __rcu *mrp_port;
fb585b44 2019#endif
1da177e4 2020
5e82b4b2 2021 struct device dev;
0c509a6c 2022 const struct attribute_group *sysfs_groups[4];
a953be53 2023 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2024
38f7b870 2025 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2026
82cc1a7a
PWJ
2027 /* for setting kernel sock attribute on TCP connection setup */
2028#define GSO_MAX_SIZE 65536
2029 unsigned int gso_max_size;
30b678d8
BH
2030#define GSO_MAX_SEGS 65535
2031 u16 gso_max_segs;
743b03a8 2032
7a6b6f51 2033#ifdef CONFIG_DCB
32953543 2034 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2035#endif
ffcfe25b 2036 s16 num_tc;
5e82b4b2
BH
2037 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2038 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2039
d11ead75 2040#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2041 unsigned int fcoe_ddp_xid;
5bc1421e 2042#endif
86f8515f 2043#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2044 struct netprio_map __rcu *priomap;
4d288d57 2045#endif
5e82b4b2 2046 struct phy_device *phydev;
e679c9c1 2047 struct sfp_bus *sfp_bus;
5e82b4b2 2048 struct lock_class_key *qdisc_tx_busylock;
f9eb8aea 2049 struct lock_class_key *qdisc_running_key;
5e82b4b2 2050 bool proto_down;
61941143 2051 unsigned wol_enabled:1;
1da177e4 2052};
43cb76d9 2053#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2054
b5cdae32
DM
2055static inline bool netif_elide_gro(const struct net_device *dev)
2056{
2057 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2058 return true;
2059 return false;
2060}
2061
1da177e4 2062#define NETDEV_ALIGN 32
1da177e4 2063
4f57c087
JF
2064static inline
2065int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2066{
2067 return dev->prio_tc_map[prio & TC_BITMASK];
2068}
2069
2070static inline
2071int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2072{
2073 if (tc >= dev->num_tc)
2074 return -EINVAL;
2075
2076 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2077 return 0;
2078}
2079
8d059b0f 2080int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2081void netdev_reset_tc(struct net_device *dev);
2082int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2083int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2084
2085static inline
2086int netdev_get_num_tc(struct net_device *dev)
2087{
2088 return dev->num_tc;
2089}
2090
ffcfe25b
AD
2091void netdev_unbind_sb_channel(struct net_device *dev,
2092 struct net_device *sb_dev);
2093int netdev_bind_sb_channel_queue(struct net_device *dev,
2094 struct net_device *sb_dev,
2095 u8 tc, u16 count, u16 offset);
2096int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2097static inline int netdev_get_sb_channel(struct net_device *dev)
2098{
2099 return max_t(int, -dev->num_tc, 0);
2100}
2101
e8a0464c
DM
2102static inline
2103struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2104 unsigned int index)
2105{
2106 return &dev->_tx[index];
2107}
2108
10c51b56
DB
2109static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2110 const struct sk_buff *skb)
2111{
2112 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2113}
2114
e8a0464c
DM
2115static inline void netdev_for_each_tx_queue(struct net_device *dev,
2116 void (*f)(struct net_device *,
2117 struct netdev_queue *,
2118 void *),
2119 void *arg)
2120{
2121 unsigned int i;
2122
2123 for (i = 0; i < dev->num_tx_queues; i++)
2124 f(dev, &dev->_tx[i], arg);
2125}
2126
d3fff6c4
ED
2127#define netdev_lockdep_set_classes(dev) \
2128{ \
2129 static struct lock_class_key qdisc_tx_busylock_key; \
2130 static struct lock_class_key qdisc_running_key; \
2131 static struct lock_class_key qdisc_xmit_lock_key; \
2132 static struct lock_class_key dev_addr_list_lock_key; \
2133 unsigned int i; \
2134 \
2135 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
2136 (dev)->qdisc_running_key = &qdisc_running_key; \
2137 lockdep_set_class(&(dev)->addr_list_lock, \
2138 &dev_addr_list_lock_key); \
2139 for (i = 0; i < (dev)->num_tx_queues; i++) \
2140 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2141 &qdisc_xmit_lock_key); \
2142}
2143
b71b5837
PA
2144u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2145 struct net_device *sb_dev);
4bd97d51
PA
2146struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2147 struct sk_buff *skb,
2148 struct net_device *sb_dev);
8c4c49df 2149
871b642a
PA
2150/* returns the headroom that the master device needs to take in account
2151 * when forwarding to this dev
2152 */
2153static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2154{
2155 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2156}
2157
2158static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2159{
2160 if (dev->netdev_ops->ndo_set_rx_headroom)
2161 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2162}
2163
2164/* set the device rx headroom to the dev's default */
2165static inline void netdev_reset_rx_headroom(struct net_device *dev)
2166{
2167 netdev_set_rx_headroom(dev, -1);
2168}
2169
c346dca1
YH
2170/*
2171 * Net namespace inlines
2172 */
2173static inline
2174struct net *dev_net(const struct net_device *dev)
2175{
c2d9ba9b 2176 return read_pnet(&dev->nd_net);
c346dca1
YH
2177}
2178
2179static inline
f5aa23fd 2180void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2181{
0c5c9fb5 2182 write_pnet(&dev->nd_net, net);
c346dca1
YH
2183}
2184
bea3348e
SH
2185/**
2186 * netdev_priv - access network device private data
2187 * @dev: network device
2188 *
2189 * Get network device private data
2190 */
6472ce60 2191static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2192{
1ce8e7b5 2193 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2194}
2195
1da177e4
LT
2196/* Set the sysfs physical device reference for the network logical device
2197 * if set prior to registration will cause a symlink during initialization.
2198 */
43cb76d9 2199#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2200
384912ed 2201/* Set the sysfs device type for the network logical device to allow
3f79410c 2202 * fine-grained identification of different network device types. For
5e82b4b2 2203 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2204 */
2205#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2206
82dc3c63
ED
2207/* Default NAPI poll() weight
2208 * Device drivers are strongly advised to not use bigger value
2209 */
2210#define NAPI_POLL_WEIGHT 64
2211
3b582cc1 2212/**
5e82b4b2 2213 * netif_napi_add - initialize a NAPI context
3b582cc1 2214 * @dev: network device
5e82b4b2 2215 * @napi: NAPI context
3b582cc1
SH
2216 * @poll: polling function
2217 * @weight: default weight
2218 *
5e82b4b2
BH
2219 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2220 * *any* of the other NAPI-related functions.
3b582cc1 2221 */
d565b0a1
HX
2222void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2223 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2224
d64b5e85 2225/**
5e82b4b2 2226 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2227 * @dev: network device
5e82b4b2 2228 * @napi: NAPI context
d64b5e85
ED
2229 * @poll: polling function
2230 * @weight: default weight
2231 *
2232 * This variant of netif_napi_add() should be used from drivers using NAPI
2233 * to exclusively poll a TX queue.
2234 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2235 */
2236static inline void netif_tx_napi_add(struct net_device *dev,
2237 struct napi_struct *napi,
2238 int (*poll)(struct napi_struct *, int),
2239 int weight)
2240{
2241 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2242 netif_napi_add(dev, napi, poll, weight);
2243}
2244
d8156534 2245/**
5e82b4b2
BH
2246 * netif_napi_del - remove a NAPI context
2247 * @napi: NAPI context
d8156534 2248 *
5e82b4b2 2249 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2250 */
d565b0a1
HX
2251void netif_napi_del(struct napi_struct *napi);
2252
2253struct napi_gro_cb {
78a478d0 2254 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
5e82b4b2 2255 void *frag0;
78a478d0 2256
7489594c
HX
2257 /* Length of frag0. */
2258 unsigned int frag0_len;
2259
86911732 2260 /* This indicates where we are processing relative to skb->data. */
5e82b4b2 2261 int data_offset;
86911732 2262
d565b0a1 2263 /* This is non-zero if the packet cannot be merged with the new skb. */
bf5a755f
JC
2264 u16 flush;
2265
2266 /* Save the IP ID here and check when we get to the transport layer */
2267 u16 flush_id;
d565b0a1
HX
2268
2269 /* Number of segments aggregated. */
2e71a6f8
ED
2270 u16 count;
2271
15e2396d
TH
2272 /* Start offset for remote checksum offload */
2273 u16 gro_remcsum_start;
2274
2e71a6f8
ED
2275 /* jiffies when first packet was created/queued */
2276 unsigned long age;
86347245 2277
afe93325 2278 /* Used in ipv6_gro_receive() and foo-over-udp */
b582ef09
OG
2279 u16 proto;
2280
baa32ff4
TH
2281 /* This is non-zero if the packet may be of the same flow. */
2282 u8 same_flow:1;
2283
fac8e0f5
JG
2284 /* Used in tunnel GRO receive */
2285 u8 encap_mark:1;
573e8fca
TH
2286
2287 /* GRO checksum is valid */
2288 u8 csum_valid:1;
2289
662880f4
TH
2290 /* Number of checksums via CHECKSUM_UNNECESSARY */
2291 u8 csum_cnt:3;
c3c7c254 2292
baa32ff4
TH
2293 /* Free the skb? */
2294 u8 free:2;
2295#define NAPI_GRO_FREE 1
2296#define NAPI_GRO_FREE_STOLEN_HEAD 2
2297
efc98d08
TH
2298 /* Used in foo-over-udp, set in udp[46]_gro_receive */
2299 u8 is_ipv6:1;
2300
a0ca153f
AD
2301 /* Used in GRE, set in fou/gue_gro_receive */
2302 u8 is_fou:1;
2303
1530545e
AD
2304 /* Used to determine if flush_id can be ignored */
2305 u8 is_atomic:1;
2306
fcd91dd4
SD
2307 /* Number of gro_receive callbacks this packet already went through */
2308 u8 recursion_counter:4;
2309
2310 /* 1 bit hole */
baa32ff4 2311
bf5a755f
JC
2312 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
2313 __wsum csum;
2314
c3c7c254
ED
2315 /* used in skb_gro_receive() slow path */
2316 struct sk_buff *last;
d565b0a1
HX
2317};
2318
2319#define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
d8156534 2320
fcd91dd4
SD
2321#define GRO_RECURSION_LIMIT 15
2322static inline int gro_recursion_inc_test(struct sk_buff *skb)
2323{
2324 return ++NAPI_GRO_CB(skb)->recursion_counter == GRO_RECURSION_LIMIT;
2325}
2326
d4546c25
DM
2327typedef struct sk_buff *(*gro_receive_t)(struct list_head *, struct sk_buff *);
2328static inline struct sk_buff *call_gro_receive(gro_receive_t cb,
2329 struct list_head *head,
2330 struct sk_buff *skb)
fcd91dd4
SD
2331{
2332 if (unlikely(gro_recursion_inc_test(skb))) {
2333 NAPI_GRO_CB(skb)->flush |= 1;
2334 return NULL;
2335 }
2336
2337 return cb(head, skb);
2338}
2339
d4546c25
DM
2340typedef struct sk_buff *(*gro_receive_sk_t)(struct sock *, struct list_head *,
2341 struct sk_buff *);
2342static inline struct sk_buff *call_gro_receive_sk(gro_receive_sk_t cb,
2343 struct sock *sk,
2344 struct list_head *head,
2345 struct sk_buff *skb)
fcd91dd4
SD
2346{
2347 if (unlikely(gro_recursion_inc_test(skb))) {
2348 NAPI_GRO_CB(skb)->flush |= 1;
2349 return NULL;
2350 }
2351
2352 return cb(sk, head, skb);
2353}
2354
1da177e4 2355struct packet_type {
f2ccd8fa 2356 __be16 type; /* This is really htons(ether_type). */
fa788d98 2357 bool ignore_outgoing;
f2ccd8fa
DM
2358 struct net_device *dev; /* NULL is wildcarded here */
2359 int (*func) (struct sk_buff *,
2360 struct net_device *,
2361 struct packet_type *,
2362 struct net_device *);
17266ee9
EC
2363 void (*list_func) (struct list_head *,
2364 struct packet_type *,
2365 struct net_device *);
c0de08d0
EL
2366 bool (*id_match)(struct packet_type *ptype,
2367 struct sock *sk);
1da177e4
LT
2368 void *af_packet_priv;
2369 struct list_head list;
2370};
2371
f191a1d1 2372struct offload_callbacks {
576a30eb 2373 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2374 netdev_features_t features);
d4546c25
DM
2375 struct sk_buff *(*gro_receive)(struct list_head *head,
2376 struct sk_buff *skb);
299603e8 2377 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2378};
2379
2380struct packet_offload {
2381 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2382 u16 priority;
f191a1d1
VY
2383 struct offload_callbacks callbacks;
2384 struct list_head list;
1da177e4
LT
2385};
2386
5e82b4b2 2387/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2388struct pcpu_sw_netstats {
2389 u64 rx_packets;
2390 u64 rx_bytes;
2391 u64 tx_packets;
2392 u64 tx_bytes;
2393 struct u64_stats_sync syncp;
9a5ee462 2394} __aligned(4 * sizeof(u64));
52bb6677
LR
2395
2396struct pcpu_lstats {
2397 u64 packets;
2398 u64 bytes;
2399 struct u64_stats_sync syncp;
9a5ee462 2400} __aligned(2 * sizeof(u64));
8f84985f 2401
aabc92bb
PNA
2402#define __netdev_alloc_pcpu_stats(type, gfp) \
2403({ \
2404 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2405 if (pcpu_stats) { \
2406 int __cpu; \
2407 for_each_possible_cpu(__cpu) { \
2408 typeof(type) *stat; \
2409 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2410 u64_stats_init(&stat->syncp); \
2411 } \
2412 } \
2413 pcpu_stats; \
1c213bd2
WC
2414})
2415
aabc92bb 2416#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2417 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2418
764f5e54
JP
2419enum netdev_lag_tx_type {
2420 NETDEV_LAG_TX_TYPE_UNKNOWN,
2421 NETDEV_LAG_TX_TYPE_RANDOM,
2422 NETDEV_LAG_TX_TYPE_BROADCAST,
2423 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2424 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2425 NETDEV_LAG_TX_TYPE_HASH,
2426};
2427
f44aa9ef
JH
2428enum netdev_lag_hash {
2429 NETDEV_LAG_HASH_NONE,
2430 NETDEV_LAG_HASH_L2,
2431 NETDEV_LAG_HASH_L34,
2432 NETDEV_LAG_HASH_L23,
2433 NETDEV_LAG_HASH_E23,
2434 NETDEV_LAG_HASH_E34,
2435 NETDEV_LAG_HASH_UNKNOWN,
2436};
2437
764f5e54
JP
2438struct netdev_lag_upper_info {
2439 enum netdev_lag_tx_type tx_type;
f44aa9ef 2440 enum netdev_lag_hash hash_type;
764f5e54
JP
2441};
2442
fb1b2e3c
JP
2443struct netdev_lag_lower_state_info {
2444 u8 link_up : 1,
2445 tx_enabled : 1;
2446};
2447
1da177e4
LT
2448#include <linux/notifier.h>
2449
ede2762d
KT
2450/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2451 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2452 * adding new types.
dcfe1421 2453 */
ede2762d
KT
2454enum netdev_cmd {
2455 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2456 NETDEV_DOWN,
2457 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2458 detected a hardware crash and restarted
2459 - we can use this eg to kick tcp sessions
2460 once done */
ede2762d
KT
2461 NETDEV_CHANGE, /* Notify device state change */
2462 NETDEV_REGISTER,
2463 NETDEV_UNREGISTER,
2464 NETDEV_CHANGEMTU, /* notify after mtu change happened */
1570415f
PM
2465 NETDEV_CHANGEADDR, /* notify after the address change */
2466 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
ede2762d
KT
2467 NETDEV_GOING_DOWN,
2468 NETDEV_CHANGENAME,
2469 NETDEV_FEAT_CHANGE,
2470 NETDEV_BONDING_FAILOVER,
2471 NETDEV_PRE_UP,
2472 NETDEV_PRE_TYPE_CHANGE,
2473 NETDEV_POST_TYPE_CHANGE,
2474 NETDEV_POST_INIT,
ede2762d
KT
2475 NETDEV_RELEASE,
2476 NETDEV_NOTIFY_PEERS,
2477 NETDEV_JOIN,
2478 NETDEV_CHANGEUPPER,
2479 NETDEV_RESEND_IGMP,
2480 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2481 NETDEV_CHANGEINFODATA,
2482 NETDEV_BONDING_INFO,
2483 NETDEV_PRECHANGEUPPER,
2484 NETDEV_CHANGELOWERSTATE,
2485 NETDEV_UDP_TUNNEL_PUSH_INFO,
2486 NETDEV_UDP_TUNNEL_DROP_INFO,
2487 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2488 NETDEV_CVLAN_FILTER_PUSH_INFO,
2489 NETDEV_CVLAN_FILTER_DROP_INFO,
2490 NETDEV_SVLAN_FILTER_PUSH_INFO,
2491 NETDEV_SVLAN_FILTER_DROP_INFO,
ede2762d
KT
2492};
2493const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2494
f629d208
JP
2495int register_netdevice_notifier(struct notifier_block *nb);
2496int unregister_netdevice_notifier(struct notifier_block *nb);
351638e7
JP
2497
2498struct netdev_notifier_info {
51d0c047
DA
2499 struct net_device *dev;
2500 struct netlink_ext_ack *extack;
351638e7
JP
2501};
2502
af7d6cce
SD
2503struct netdev_notifier_info_ext {
2504 struct netdev_notifier_info info; /* must be first */
2505 union {
2506 u32 mtu;
2507 } ext;
2508};
2509
be9efd36
JP
2510struct netdev_notifier_change_info {
2511 struct netdev_notifier_info info; /* must be first */
2512 unsigned int flags_changed;
2513};
2514
0e4ead9d
JP
2515struct netdev_notifier_changeupper_info {
2516 struct netdev_notifier_info info; /* must be first */
2517 struct net_device *upper_dev; /* new upper dev */
2518 bool master; /* is upper dev master */
5e82b4b2 2519 bool linking; /* is the notification for link or unlink */
29bf24af 2520 void *upper_info; /* upper dev info */
0e4ead9d
JP
2521};
2522
04d48266
JP
2523struct netdev_notifier_changelowerstate_info {
2524 struct netdev_notifier_info info; /* must be first */
2525 void *lower_state_info; /* is lower dev state */
2526};
2527
1570415f
PM
2528struct netdev_notifier_pre_changeaddr_info {
2529 struct netdev_notifier_info info; /* must be first */
2530 const unsigned char *dev_addr;
2531};
2532
75538c2b
CW
2533static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2534 struct net_device *dev)
2535{
2536 info->dev = dev;
51d0c047 2537 info->extack = NULL;
75538c2b
CW
2538}
2539
351638e7
JP
2540static inline struct net_device *
2541netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2542{
2543 return info->dev;
2544}
2545
51d0c047
DA
2546static inline struct netlink_ext_ack *
2547netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2548{
2549 return info->extack;
2550}
2551
f629d208 2552int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2553
2554
1da177e4
LT
2555extern rwlock_t dev_base_lock; /* Device list lock */
2556
881d966b
EB
2557#define for_each_netdev(net, d) \
2558 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2559#define for_each_netdev_reverse(net, d) \
2560 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2561#define for_each_netdev_rcu(net, d) \
2562 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2563#define for_each_netdev_safe(net, d, n) \
2564 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2565#define for_each_netdev_continue(net, d) \
2566 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
254245d2 2567#define for_each_netdev_continue_rcu(net, d) \
2568 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2569#define for_each_netdev_in_bond_rcu(bond, slave) \
2570 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2571 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2572#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2573
a050c33f
DL
2574static inline struct net_device *next_net_device(struct net_device *dev)
2575{
2576 struct list_head *lh;
2577 struct net *net;
2578
c346dca1 2579 net = dev_net(dev);
a050c33f
DL
2580 lh = dev->dev_list.next;
2581 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2582}
2583
ce81b76a
ED
2584static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2585{
2586 struct list_head *lh;
2587 struct net *net;
2588
2589 net = dev_net(dev);
ccf43438 2590 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2591 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2592}
2593
a050c33f
DL
2594static inline struct net_device *first_net_device(struct net *net)
2595{
2596 return list_empty(&net->dev_base_head) ? NULL :
2597 net_device_entry(net->dev_base_head.next);
2598}
7562f876 2599
ccf43438
ED
2600static inline struct net_device *first_net_device_rcu(struct net *net)
2601{
2602 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2603
2604 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2605}
2606
f629d208
JP
2607int netdev_boot_setup_check(struct net_device *dev);
2608unsigned long netdev_boot_base(const char *prefix, int unit);
2609struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2610 const char *hwaddr);
2611struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
2612struct net_device *__dev_getfirstbyhwtype(struct net *net, unsigned short type);
2613void dev_add_pack(struct packet_type *pt);
2614void dev_remove_pack(struct packet_type *pt);
2615void __dev_remove_pack(struct packet_type *pt);
2616void dev_add_offload(struct packet_offload *po);
2617void dev_remove_offload(struct packet_offload *po);
f629d208 2618
a54acb3a 2619int dev_get_iflink(const struct net_device *dev);
fc4099f1 2620int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
6c555490
WC
2621struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2622 unsigned short mask);
f629d208
JP
2623struct net_device *dev_get_by_name(struct net *net, const char *name);
2624struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2625struct net_device *__dev_get_by_name(struct net *net, const char *name);
2626int dev_alloc_name(struct net_device *dev, const char *name);
00f54e68 2627int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
7051b88a 2628void dev_close(struct net_device *dev);
2629void dev_close_many(struct list_head *head, bool unlink);
f629d208 2630void dev_disable_lro(struct net_device *dev);
0c4b51f0 2631int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 2632u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
a350ecce 2633 struct net_device *sb_dev);
a4ea8a3d 2634u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
a350ecce 2635 struct net_device *sb_dev);
2b4aa3ce 2636int dev_queue_xmit(struct sk_buff *skb);
eadec877 2637int dev_queue_xmit_accel(struct sk_buff *skb, struct net_device *sb_dev);
865b03f2 2638int dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
f629d208
JP
2639int register_netdevice(struct net_device *dev);
2640void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2641void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2642static inline void unregister_netdevice(struct net_device *dev)
2643{
2644 unregister_netdevice_queue(dev, NULL);
2645}
2646
f629d208
JP
2647int netdev_refcnt_read(const struct net_device *dev);
2648void free_netdev(struct net_device *dev);
74d332c1 2649void netdev_freemem(struct net_device *dev);
f629d208
JP
2650void synchronize_net(void);
2651int init_dummy_netdev(struct net_device *dev);
937f1ba5 2652
f629d208
JP
2653struct net_device *dev_get_by_index(struct net *net, int ifindex);
2654struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2655struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 2656struct net_device *dev_get_by_napi_id(unsigned int napi_id);
f629d208
JP
2657int netdev_get_name(struct net *net, char *name, int ifindex);
2658int dev_restart(struct net_device *dev);
d4546c25 2659int skb_gro_receive(struct sk_buff *p, struct sk_buff *skb);
86911732
HX
2660
2661static inline unsigned int skb_gro_offset(const struct sk_buff *skb)
2662{
2663 return NAPI_GRO_CB(skb)->data_offset;
2664}
2665
2666static inline unsigned int skb_gro_len(const struct sk_buff *skb)
2667{
2668 return skb->len - NAPI_GRO_CB(skb)->data_offset;
2669}
2670
2671static inline void skb_gro_pull(struct sk_buff *skb, unsigned int len)
2672{
2673 NAPI_GRO_CB(skb)->data_offset += len;
2674}
2675
a5b1cf28
HX
2676static inline void *skb_gro_header_fast(struct sk_buff *skb,
2677 unsigned int offset)
86911732 2678{
a5b1cf28
HX
2679 return NAPI_GRO_CB(skb)->frag0 + offset;
2680}
78a478d0 2681
a5b1cf28
HX
2682static inline int skb_gro_header_hard(struct sk_buff *skb, unsigned int hlen)
2683{
2684 return NAPI_GRO_CB(skb)->frag0_len < hlen;
2685}
78a478d0 2686
57ea52a8
HX
2687static inline void skb_gro_frag0_invalidate(struct sk_buff *skb)
2688{
2689 NAPI_GRO_CB(skb)->frag0 = NULL;
2690 NAPI_GRO_CB(skb)->frag0_len = 0;
2691}
2692
a5b1cf28
HX
2693static inline void *skb_gro_header_slow(struct sk_buff *skb, unsigned int hlen,
2694 unsigned int offset)
2695{
17dd759c
HX
2696 if (!pskb_may_pull(skb, hlen))
2697 return NULL;
2698
57ea52a8 2699 skb_gro_frag0_invalidate(skb);
17dd759c 2700 return skb->data + offset;
86911732 2701}
1da177e4 2702
36e7b1b8
HX
2703static inline void *skb_gro_network_header(struct sk_buff *skb)
2704{
78d3fd0b
HX
2705 return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
2706 skb_network_offset(skb);
36e7b1b8
HX
2707}
2708
bf5a755f
JC
2709static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
2710 const void *start, unsigned int len)
2711{
573e8fca 2712 if (NAPI_GRO_CB(skb)->csum_valid)
bf5a755f
JC
2713 NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
2714 csum_partial(start, len, 0));
2715}
2716
573e8fca
TH
2717/* GRO checksum functions. These are logical equivalents of the normal
2718 * checksum functions (in skbuff.h) except that they operate on the GRO
2719 * offsets and fields in sk_buff.
2720 */
2721
2722__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
2723
15e2396d
TH
2724static inline bool skb_at_gro_remcsum_start(struct sk_buff *skb)
2725{
b7fe10e5 2726 return (NAPI_GRO_CB(skb)->gro_remcsum_start == skb_gro_offset(skb));
15e2396d
TH
2727}
2728
573e8fca
TH
2729static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
2730 bool zero_okay,
2731 __sum16 check)
2732{
6edec0e6
TH
2733 return ((skb->ip_summed != CHECKSUM_PARTIAL ||
2734 skb_checksum_start_offset(skb) <
2735 skb_gro_offset(skb)) &&
15e2396d 2736 !skb_at_gro_remcsum_start(skb) &&
662880f4 2737 NAPI_GRO_CB(skb)->csum_cnt == 0 &&
573e8fca
TH
2738 (!zero_okay || check));
2739}
2740
2741static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
2742 __wsum psum)
2743{
2744 if (NAPI_GRO_CB(skb)->csum_valid &&
2745 !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
2746 return 0;
2747
2748 NAPI_GRO_CB(skb)->csum = psum;
2749
2750 return __skb_gro_checksum_complete(skb);
2751}
2752
573e8fca
TH
2753static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
2754{
662880f4
TH
2755 if (NAPI_GRO_CB(skb)->csum_cnt > 0) {
2756 /* Consume a checksum from CHECKSUM_UNNECESSARY */
2757 NAPI_GRO_CB(skb)->csum_cnt--;
2758 } else {
2759 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
2760 * verified a new top level checksum or an encapsulated one
2761 * during GRO. This saves work if we fallback to normal path.
2762 */
2763 __skb_incr_checksum_unnecessary(skb);
573e8fca
TH
2764 }
2765}
2766
2767#define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
2768 compute_pseudo) \
2769({ \
2770 __sum16 __ret = 0; \
2771 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
2772 __ret = __skb_gro_checksum_validate_complete(skb, \
2773 compute_pseudo(skb, proto)); \
219f1d79 2774 if (!__ret) \
573e8fca
TH
2775 skb_gro_incr_csum_unnecessary(skb); \
2776 __ret; \
2777})
2778
2779#define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
2780 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
2781
2782#define skb_gro_checksum_validate_zero_check(skb, proto, check, \
2783 compute_pseudo) \
2784 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
2785
2786#define skb_gro_checksum_simple_validate(skb) \
2787 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
2788
d96535a1
TH
2789static inline bool __skb_gro_checksum_convert_check(struct sk_buff *skb)
2790{
2791 return (NAPI_GRO_CB(skb)->csum_cnt == 0 &&
2792 !NAPI_GRO_CB(skb)->csum_valid);
2793}
2794
2795static inline void __skb_gro_checksum_convert(struct sk_buff *skb,
2796 __sum16 check, __wsum pseudo)
2797{
2798 NAPI_GRO_CB(skb)->csum = ~pseudo;
2799 NAPI_GRO_CB(skb)->csum_valid = 1;
2800}
2801
2802#define skb_gro_checksum_try_convert(skb, proto, check, compute_pseudo) \
2803do { \
2804 if (__skb_gro_checksum_convert_check(skb)) \
2805 __skb_gro_checksum_convert(skb, check, \
2806 compute_pseudo(skb, proto)); \
2807} while (0)
2808
26c4f7da
TH
2809struct gro_remcsum {
2810 int offset;
2811 __wsum delta;
2812};
2813
2814static inline void skb_gro_remcsum_init(struct gro_remcsum *grc)
2815{
846cd667 2816 grc->offset = 0;
26c4f7da
TH
2817 grc->delta = 0;
2818}
2819
b7fe10e5
TH
2820static inline void *skb_gro_remcsum_process(struct sk_buff *skb, void *ptr,
2821 unsigned int off, size_t hdrlen,
2822 int start, int offset,
2823 struct gro_remcsum *grc,
2824 bool nopartial)
dcdc8994
TH
2825{
2826 __wsum delta;
b7fe10e5 2827 size_t plen = hdrlen + max_t(size_t, offset + sizeof(u16), start);
dcdc8994
TH
2828
2829 BUG_ON(!NAPI_GRO_CB(skb)->csum_valid);
2830
15e2396d 2831 if (!nopartial) {
b7fe10e5
TH
2832 NAPI_GRO_CB(skb)->gro_remcsum_start = off + hdrlen + start;
2833 return ptr;
2834 }
2835
2836 ptr = skb_gro_header_fast(skb, off);
2837 if (skb_gro_header_hard(skb, off + plen)) {
2838 ptr = skb_gro_header_slow(skb, off + plen, off);
2839 if (!ptr)
2840 return NULL;
15e2396d
TH
2841 }
2842
b7fe10e5
TH
2843 delta = remcsum_adjust(ptr + hdrlen, NAPI_GRO_CB(skb)->csum,
2844 start, offset);
dcdc8994
TH
2845
2846 /* Adjust skb->csum since we changed the packet */
dcdc8994 2847 NAPI_GRO_CB(skb)->csum = csum_add(NAPI_GRO_CB(skb)->csum, delta);
26c4f7da 2848
b7fe10e5 2849 grc->offset = off + hdrlen + offset;
26c4f7da 2850 grc->delta = delta;
b7fe10e5
TH
2851
2852 return ptr;
dcdc8994
TH
2853}
2854
26c4f7da
TH
2855static inline void skb_gro_remcsum_cleanup(struct sk_buff *skb,
2856 struct gro_remcsum *grc)
2857{
b7fe10e5
TH
2858 void *ptr;
2859 size_t plen = grc->offset + sizeof(u16);
2860
26c4f7da
TH
2861 if (!grc->delta)
2862 return;
2863
b7fe10e5
TH
2864 ptr = skb_gro_header_fast(skb, grc->offset);
2865 if (skb_gro_header_hard(skb, grc->offset + sizeof(u16))) {
2866 ptr = skb_gro_header_slow(skb, plen, grc->offset);
2867 if (!ptr)
2868 return;
2869 }
2870
2871 remcsum_unadjust((__sum16 *)ptr, grc->delta);
26c4f7da 2872}
dcdc8994 2873
25393d3f 2874#ifdef CONFIG_XFRM_OFFLOAD
d4546c25 2875static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
25393d3f
SK
2876{
2877 if (PTR_ERR(pp) != -EINPROGRESS)
2878 NAPI_GRO_CB(skb)->flush |= flush;
2879}
603d4cf8 2880static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 2881 struct sk_buff *pp,
603d4cf8
SD
2882 int flush,
2883 struct gro_remcsum *grc)
2884{
2885 if (PTR_ERR(pp) != -EINPROGRESS) {
2886 NAPI_GRO_CB(skb)->flush |= flush;
2887 skb_gro_remcsum_cleanup(skb, grc);
2888 skb->remcsum_offload = 0;
2889 }
2890}
25393d3f 2891#else
d4546c25 2892static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
5f114163
SK
2893{
2894 NAPI_GRO_CB(skb)->flush |= flush;
2895}
603d4cf8 2896static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 2897 struct sk_buff *pp,
603d4cf8
SD
2898 int flush,
2899 struct gro_remcsum *grc)
2900{
2901 NAPI_GRO_CB(skb)->flush |= flush;
2902 skb_gro_remcsum_cleanup(skb, grc);
2903 skb->remcsum_offload = 0;
2904}
25393d3f 2905#endif
5f114163 2906
0c4e8581
SH
2907static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
2908 unsigned short type,
3b04ddde 2909 const void *daddr, const void *saddr,
95c96174 2910 unsigned int len)
0c4e8581 2911{
f1ecfd5d 2912 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 2913 return 0;
3b04ddde
SH
2914
2915 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
2916}
2917
b95cce35
SH
2918static inline int dev_parse_header(const struct sk_buff *skb,
2919 unsigned char *haddr)
2920{
2921 const struct net_device *dev = skb->dev;
2922
1b83336b 2923 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 2924 return 0;
3b04ddde 2925 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
2926}
2927
e78b2915
MM
2928static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
2929{
2930 const struct net_device *dev = skb->dev;
2931
2932 if (!dev->header_ops || !dev->header_ops->parse_protocol)
2933 return 0;
2934 return dev->header_ops->parse_protocol(skb);
2935}
2936
2793a23a
WB
2937/* ll_header must have at least hard_header_len allocated */
2938static inline bool dev_validate_header(const struct net_device *dev,
2939 char *ll_header, int len)
2940{
2941 if (likely(len >= dev->hard_header_len))
2942 return true;
217e6fa2
WB
2943 if (len < dev->min_header_len)
2944 return false;
2793a23a
WB
2945
2946 if (capable(CAP_SYS_RAWIO)) {
2947 memset(ll_header + len, 0, dev->hard_header_len - len);
2948 return true;
2949 }
2950
2951 if (dev->header_ops && dev->header_ops->validate)
2952 return dev->header_ops->validate(ll_header, len);
2953
2954 return false;
2955}
2956
36fd633e
AV
2957typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr,
2958 int len, int size);
f629d208 2959int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
1da177e4
LT
2960static inline int unregister_gifconf(unsigned int family)
2961{
2962 return register_gifconf(family, NULL);
2963}
2964
99bbc707 2965#ifdef CONFIG_NET_FLOW_LIMIT
5f121b9a 2966#define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
99bbc707
WB
2967struct sd_flow_limit {
2968 u64 count;
2969 unsigned int num_buckets;
2970 unsigned int history_head;
2971 u16 history[FLOW_LIMIT_HISTORY];
2972 u8 buckets[];
2973};
2974
2975extern int netdev_flow_limit_table_len;
2976#endif /* CONFIG_NET_FLOW_LIMIT */
2977
1da177e4 2978/*
5e82b4b2 2979 * Incoming packets are placed on per-CPU queues
1da177e4 2980 */
d94d9fee 2981struct softnet_data {
1da177e4 2982 struct list_head poll_list;
6e7676c1 2983 struct sk_buff_head process_queue;
1da177e4 2984
dee42870 2985 /* stats */
cd7b5396
DM
2986 unsigned int processed;
2987 unsigned int time_squeeze;
cd7b5396 2988 unsigned int received_rps;
fd793d89 2989#ifdef CONFIG_RPS
88751275 2990 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
2991#endif
2992#ifdef CONFIG_NET_FLOW_LIMIT
2993 struct sd_flow_limit __rcu *flow_limit;
2994#endif
2995 struct Qdisc *output_queue;
2996 struct Qdisc **output_queue_tailp;
2997 struct sk_buff *completion_queue;
f53c7239
SK
2998#ifdef CONFIG_XFRM_OFFLOAD
2999 struct sk_buff_head xfrm_backlog;
3000#endif
97cdcf37
FW
3001 /* written and read only by owning cpu: */
3002 struct {
3003 u16 recursion;
3004 u8 more;
3005 } xmit;
4cdb1e2e 3006#ifdef CONFIG_RPS
501e7ef5
ED
3007 /* input_queue_head should be written by cpu owning this struct,
3008 * and only read by other cpus. Worth using a cache line.
3009 */
3010 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3011
3012 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3013 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3014 struct softnet_data *rps_ipi_next;
3015 unsigned int cpu;
76cc8b13 3016 unsigned int input_queue_tail;
1e94d72f 3017#endif
95c96174 3018 unsigned int dropped;
0a9627f2 3019 struct sk_buff_head input_pkt_queue;
bea3348e 3020 struct napi_struct backlog;
99bbc707 3021
1da177e4
LT
3022};
3023
76cc8b13 3024static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3025{
3026#ifdef CONFIG_RPS
76cc8b13
TH
3027 sd->input_queue_head++;
3028#endif
3029}
3030
3031static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3032 unsigned int *qtail)
3033{
3034#ifdef CONFIG_RPS
3035 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3036#endif
3037}
3038
0a9627f2 3039DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3040
97cdcf37
FW
3041static inline int dev_recursion_level(void)
3042{
28b05b92 3043 return this_cpu_read(softnet_data.xmit.recursion);
97cdcf37
FW
3044}
3045
3046#define XMIT_RECURSION_LIMIT 10
3047static inline bool dev_xmit_recursion(void)
3048{
3049 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3050 XMIT_RECURSION_LIMIT);
3051}
3052
3053static inline void dev_xmit_recursion_inc(void)
3054{
3055 __this_cpu_inc(softnet_data.xmit.recursion);
3056}
3057
3058static inline void dev_xmit_recursion_dec(void)
3059{
3060 __this_cpu_dec(softnet_data.xmit.recursion);
3061}
3062
f629d208 3063void __netif_schedule(struct Qdisc *q);
46e5da40 3064void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3065
fd2ea0a7
DM
3066static inline void netif_tx_schedule_all(struct net_device *dev)
3067{
3068 unsigned int i;
3069
3070 for (i = 0; i < dev->num_tx_queues; i++)
3071 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3072}
3073
f9a7cbbf 3074static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3075{
73466498 3076 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3077}
3078
bea3348e
SH
3079/**
3080 * netif_start_queue - allow transmit
3081 * @dev: network device
3082 *
3083 * Allow upper layers to call the device hard_start_xmit routine.
3084 */
1da177e4
LT
3085static inline void netif_start_queue(struct net_device *dev)
3086{
e8a0464c 3087 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3088}
3089
fd2ea0a7
DM
3090static inline void netif_tx_start_all_queues(struct net_device *dev)
3091{
3092 unsigned int i;
3093
3094 for (i = 0; i < dev->num_tx_queues; i++) {
3095 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3096 netif_tx_start_queue(txq);
3097 }
3098}
3099
46e5da40 3100void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3101
d29f749e
DJ
3102/**
3103 * netif_wake_queue - restart transmit
3104 * @dev: network device
3105 *
3106 * Allow upper layers to call the device hard_start_xmit routine.
3107 * Used for flow control when transmit resources are available.
3108 */
79d16385
DM
3109static inline void netif_wake_queue(struct net_device *dev)
3110{
e8a0464c 3111 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3112}
3113
fd2ea0a7
DM
3114static inline void netif_tx_wake_all_queues(struct net_device *dev)
3115{
3116 unsigned int i;
3117
3118 for (i = 0; i < dev->num_tx_queues; i++) {
3119 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3120 netif_tx_wake_queue(txq);
3121 }
3122}
3123
f9a7cbbf 3124static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3125{
73466498 3126 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3127}
3128
bea3348e
SH
3129/**
3130 * netif_stop_queue - stop transmitted packets
3131 * @dev: network device
3132 *
3133 * Stop upper layers calling the device hard_start_xmit routine.
3134 * Used for flow control when transmit resources are unavailable.
3135 */
1da177e4
LT
3136static inline void netif_stop_queue(struct net_device *dev)
3137{
e8a0464c 3138 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3139}
3140
a2029240 3141void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 3142
4d29515f 3143static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3144{
73466498 3145 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3146}
3147
bea3348e
SH
3148/**
3149 * netif_queue_stopped - test if transmit queue is flowblocked
3150 * @dev: network device
3151 *
3152 * Test if transmit queue on device is currently unable to send.
3153 */
4d29515f 3154static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3155{
e8a0464c 3156 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3157}
3158
4d29515f 3159static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3160{
73466498
TH
3161 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3162}
3163
8e2f1a63
DB
3164static inline bool
3165netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3166{
3167 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3168}
3169
8e2f1a63
DB
3170static inline bool
3171netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3172{
3173 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3174}
3175
53511453
ED
3176/**
3177 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3178 * @dev_queue: pointer to transmit queue
3179 *
3180 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3181 * to give appropriate hint to the CPU.
53511453
ED
3182 */
3183static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3184{
3185#ifdef CONFIG_BQL
3186 prefetchw(&dev_queue->dql.num_queued);
3187#endif
3188}
3189
3190/**
3191 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3192 * @dev_queue: pointer to transmit queue
3193 *
3194 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3195 * to give appropriate hint to the CPU.
53511453
ED
3196 */
3197static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3198{
3199#ifdef CONFIG_BQL
3200 prefetchw(&dev_queue->dql.limit);
3201#endif
3202}
3203
c5d67bd7
TH
3204static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3205 unsigned int bytes)
3206{
114cf580
TH
3207#ifdef CONFIG_BQL
3208 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3209
3210 if (likely(dql_avail(&dev_queue->dql) >= 0))
3211 return;
3212
3213 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3214
3215 /*
3216 * The XOFF flag must be set before checking the dql_avail below,
3217 * because in netdev_tx_completed_queue we update the dql_completed
3218 * before checking the XOFF flag.
3219 */
3220 smp_mb();
3221
3222 /* check again in case another CPU has just made room avail */
3223 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3224 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3225#endif
c5d67bd7
TH
3226}
3227
3e59020a
ED
3228/* Variant of netdev_tx_sent_queue() for drivers that are aware
3229 * that they should not test BQL status themselves.
3230 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3231 * skb of a batch.
3232 * Returns true if the doorbell must be used to kick the NIC.
3233 */
3234static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3235 unsigned int bytes,
3236 bool xmit_more)
3237{
3238 if (xmit_more) {
3239#ifdef CONFIG_BQL
3240 dql_queued(&dev_queue->dql, bytes);
3241#endif
3242 return netif_tx_queue_stopped(dev_queue);
3243 }
3244 netdev_tx_sent_queue(dev_queue, bytes);
3245 return true;
3246}
3247
0042d0c8
FF
3248/**
3249 * netdev_sent_queue - report the number of bytes queued to hardware
3250 * @dev: network device
3251 * @bytes: number of bytes queued to the hardware device queue
3252 *
3253 * Report the number of bytes queued for sending/completion to the network
3254 * device hardware queue. @bytes should be a good approximation and should
3255 * exactly match netdev_completed_queue() @bytes
3256 */
c5d67bd7
TH
3257static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3258{
3259 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3260}
3261
620344c4
HK
3262static inline bool __netdev_sent_queue(struct net_device *dev,
3263 unsigned int bytes,
3264 bool xmit_more)
3265{
3266 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3267 xmit_more);
3268}
3269
c5d67bd7 3270static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3271 unsigned int pkts, unsigned int bytes)
c5d67bd7 3272{
114cf580 3273#ifdef CONFIG_BQL
b37c0fbe
AD
3274 if (unlikely(!bytes))
3275 return;
3276
3277 dql_completed(&dev_queue->dql, bytes);
3278
3279 /*
3280 * Without the memory barrier there is a small possiblity that
3281 * netdev_tx_sent_queue will miss the update and cause the queue to
3282 * be stopped forever
3283 */
3284 smp_mb();
3285
f3acd33d 3286 if (unlikely(dql_avail(&dev_queue->dql) < 0))
b37c0fbe
AD
3287 return;
3288
3289 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3290 netif_schedule_queue(dev_queue);
114cf580 3291#endif
c5d67bd7
TH
3292}
3293
0042d0c8
FF
3294/**
3295 * netdev_completed_queue - report bytes and packets completed by device
3296 * @dev: network device
3297 * @pkts: actual number of packets sent over the medium
3298 * @bytes: actual number of bytes sent over the medium
3299 *
3300 * Report the number of bytes and packets transmitted by the network device
3301 * hardware queue over the physical medium, @bytes must exactly match the
3302 * @bytes amount passed to netdev_sent_queue()
3303 */
c5d67bd7 3304static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3305 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3306{
3307 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3308}
3309
3310static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3311{
114cf580 3312#ifdef CONFIG_BQL
5c490354 3313 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3314 dql_reset(&q->dql);
3315#endif
c5d67bd7
TH
3316}
3317
0042d0c8
FF
3318/**
3319 * netdev_reset_queue - reset the packets and bytes count of a network device
3320 * @dev_queue: network device
3321 *
3322 * Reset the bytes and packet count of a network device and clear the
3323 * software flow control OFF bit for this network device
3324 */
c5d67bd7
TH
3325static inline void netdev_reset_queue(struct net_device *dev_queue)
3326{
3327 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3328}
3329
b9507bda
DB
3330/**
3331 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3332 * @dev: network device
3333 * @queue_index: given tx queue index
3334 *
3335 * Returns 0 if given tx queue index >= number of device tx queues,
3336 * otherwise returns the originally passed tx queue index.
3337 */
3338static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3339{
3340 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3341 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3342 dev->name, queue_index,
3343 dev->real_num_tx_queues);
3344 return 0;
3345 }
3346
3347 return queue_index;
3348}
3349
bea3348e
SH
3350/**
3351 * netif_running - test if up
3352 * @dev: network device
3353 *
3354 * Test if the device has been brought up.
3355 */
4d29515f 3356static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3357{
3358 return test_bit(__LINK_STATE_START, &dev->state);
3359}
3360
f25f4e44 3361/*
5e82b4b2 3362 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3363 * stop, and a check if it's stopped. All other device management is
3364 * done at the overall netdevice level.
3365 * Also test the device if we're multiqueue.
3366 */
bea3348e
SH
3367
3368/**
3369 * netif_start_subqueue - allow sending packets on subqueue
3370 * @dev: network device
3371 * @queue_index: sub queue index
3372 *
3373 * Start individual transmit queue of a device with multiple transmit queues.
3374 */
f25f4e44
PWJ
3375static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3376{
fd2ea0a7 3377 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3378
3379 netif_tx_start_queue(txq);
f25f4e44
PWJ
3380}
3381
bea3348e
SH
3382/**
3383 * netif_stop_subqueue - stop sending packets on subqueue
3384 * @dev: network device
3385 * @queue_index: sub queue index
3386 *
3387 * Stop individual transmit queue of a device with multiple transmit queues.
3388 */
f25f4e44
PWJ
3389static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3390{
fd2ea0a7 3391 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3392 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3393}
3394
bea3348e
SH
3395/**
3396 * netif_subqueue_stopped - test status of subqueue
3397 * @dev: network device
3398 * @queue_index: sub queue index
3399 *
3400 * Check individual transmit queue of a device with multiple transmit queues.
3401 */
4d29515f
DM
3402static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3403 u16 queue_index)
f25f4e44 3404{
fd2ea0a7 3405 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3406
3407 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3408}
3409
4d29515f
DM
3410static inline bool netif_subqueue_stopped(const struct net_device *dev,
3411 struct sk_buff *skb)
668f895a
PE
3412{
3413 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3414}
bea3348e 3415
738b35cc
FF
3416/**
3417 * netif_wake_subqueue - allow sending packets on subqueue
3418 * @dev: network device
3419 * @queue_index: sub queue index
3420 *
3421 * Resume individual transmit queue of a device with multiple transmit queues.
3422 */
3423static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3424{
3425 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3426
3427 netif_tx_wake_queue(txq);
3428}
f25f4e44 3429
537c00de 3430#ifdef CONFIG_XPS
53af53ae 3431int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3432 u16 index);
80d19669
AN
3433int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
3434 u16 index, bool is_rxqs_map);
3435
3436/**
3437 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3438 * @j: CPU/Rx queue index
3439 * @mask: bitmask of all cpus/rx queues
3440 * @nr_bits: number of bits in the bitmask
3441 *
3442 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3443 */
3444static inline bool netif_attr_test_mask(unsigned long j,
3445 const unsigned long *mask,
3446 unsigned int nr_bits)
3447{
3448 cpu_max_bits_warn(j, nr_bits);
3449 return test_bit(j, mask);
3450}
3451
3452/**
3453 * netif_attr_test_online - Test for online CPU/Rx queue
3454 * @j: CPU/Rx queue index
3455 * @online_mask: bitmask for CPUs/Rx queues that are online
3456 * @nr_bits: number of bits in the bitmask
3457 *
3458 * Returns true if a CPU/Rx queue is online.
3459 */
3460static inline bool netif_attr_test_online(unsigned long j,
3461 const unsigned long *online_mask,
3462 unsigned int nr_bits)
3463{
3464 cpu_max_bits_warn(j, nr_bits);
3465
3466 if (online_mask)
3467 return test_bit(j, online_mask);
3468
3469 return (j < nr_bits);
3470}
3471
3472/**
3473 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3474 * @n: CPU/Rx queue index
3475 * @srcp: the cpumask/Rx queue mask pointer
3476 * @nr_bits: number of bits in the bitmask
3477 *
3478 * Returns >= nr_bits if no further CPUs/Rx queues set.
3479 */
3480static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3481 unsigned int nr_bits)
3482{
3483 /* -1 is a legal arg here. */
3484 if (n != -1)
3485 cpu_max_bits_warn(n, nr_bits);
3486
3487 if (srcp)
3488 return find_next_bit(srcp, nr_bits, n + 1);
3489
3490 return n + 1;
3491}
3492
3493/**
3494 * netif_attrmask_next_and - get the next CPU/Rx queue in *src1p & *src2p
3495 * @n: CPU/Rx queue index
3496 * @src1p: the first CPUs/Rx queues mask pointer
3497 * @src2p: the second CPUs/Rx queues mask pointer
3498 * @nr_bits: number of bits in the bitmask
3499 *
3500 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3501 */
3502static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3503 const unsigned long *src2p,
3504 unsigned int nr_bits)
3505{
3506 /* -1 is a legal arg here. */
3507 if (n != -1)
3508 cpu_max_bits_warn(n, nr_bits);
3509
3510 if (src1p && src2p)
3511 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3512 else if (src1p)
3513 return find_next_bit(src1p, nr_bits, n + 1);
3514 else if (src2p)
3515 return find_next_bit(src2p, nr_bits, n + 1);
3516
3517 return n + 1;
3518}
537c00de
AD
3519#else
3520static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3521 const struct cpumask *mask,
537c00de
AD
3522 u16 index)
3523{
3524 return 0;
3525}
c9fbb2d2
KK
3526
3527static inline int __netif_set_xps_queue(struct net_device *dev,
3528 const unsigned long *mask,
3529 u16 index, bool is_rxqs_map)
3530{
3531 return 0;
3532}
537c00de
AD
3533#endif
3534
bea3348e
SH
3535/**
3536 * netif_is_multiqueue - test if device has multiple transmit queues
3537 * @dev: network device
3538 *
3539 * Check if device has multiple transmit queues
bea3348e 3540 */
4d29515f 3541static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3542{
a02cec21 3543 return dev->num_tx_queues > 1;
f25f4e44 3544}
1da177e4 3545
f629d208 3546int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3547
a953be53 3548#ifdef CONFIG_SYSFS
f629d208 3549int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3550#else
3551static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3552 unsigned int rxqs)
62fe0b40 3553{
c29c2ebd 3554 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3555 return 0;
3556}
3557#endif
3558
65073a67
DB
3559static inline struct netdev_rx_queue *
3560__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
3561{
3562 return dev->_rx + rxq;
3563}
3564
a953be53
MD
3565#ifdef CONFIG_SYSFS
3566static inline unsigned int get_netdev_rx_queue_index(
3567 struct netdev_rx_queue *queue)
3568{
3569 struct net_device *dev = queue->dev;
3570 int index = queue - dev->_rx;
3571
3572 BUG_ON(index >= dev->num_rx_queues);
3573 return index;
3574}
3575#endif
3576
16917b87 3577#define DEFAULT_MAX_NUM_RSS_QUEUES (8)
f629d208 3578int netif_get_num_default_rss_queues(void);
16917b87 3579
e6247027
ED
3580enum skb_free_reason {
3581 SKB_REASON_CONSUMED,
3582 SKB_REASON_DROPPED,
3583};
3584
3585void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3586void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3587
e6247027
ED
3588/*
3589 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3590 * interrupt context or with hardware interrupts being disabled.
3591 * (in_irq() || irqs_disabled())
3592 *
3593 * We provide four helpers that can be used in following contexts :
3594 *
3595 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3596 * replacing kfree_skb(skb)
3597 *
3598 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3599 * Typically used in place of consume_skb(skb) in TX completion path
3600 *
3601 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3602 * replacing kfree_skb(skb)
3603 *
3604 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3605 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3606 */
e6247027
ED
3607static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3608{
3609 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3610}
3611
3612static inline void dev_consume_skb_irq(struct sk_buff *skb)
3613{
3614 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3615}
3616
3617static inline void dev_kfree_skb_any(struct sk_buff *skb)
3618{
3619 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3620}
3621
3622static inline void dev_consume_skb_any(struct sk_buff *skb)
3623{
3624 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3625}
1da177e4 3626
7c497478
JW
3627void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3628int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208
JP
3629int netif_rx(struct sk_buff *skb);
3630int netif_rx_ni(struct sk_buff *skb);
04eb4489 3631int netif_receive_skb(struct sk_buff *skb);
1c601d82 3632int netif_receive_skb_core(struct sk_buff *skb);
f6ad8c1b 3633void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3634gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3635void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3636struct sk_buff *napi_get_frags(struct napi_struct *napi);
3637gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3638struct packet_offload *gro_find_receive_by_type(__be16 type);
3639struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3640
3641static inline void napi_free_frags(struct napi_struct *napi)
3642{
3643 kfree_skb(napi->skb);
3644 napi->skb = NULL;
3645}
3646
24b27fc4 3647bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
3648int netdev_rx_handler_register(struct net_device *dev,
3649 rx_handler_func_t *rx_handler,
3650 void *rx_handler_data);
3651void netdev_rx_handler_unregister(struct net_device *dev);
3652
3653bool dev_valid_name(const char *name);
44c02a2c
AV
3654int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
3655 bool *need_copyout);
36fd633e 3656int dev_ifconf(struct net *net, struct ifconf *, int);
f629d208
JP
3657int dev_ethtool(struct net *net, struct ifreq *);
3658unsigned int dev_get_flags(const struct net_device *);
6d040321
PM
3659int __dev_change_flags(struct net_device *dev, unsigned int flags,
3660 struct netlink_ext_ack *extack);
567c5e13
PM
3661int dev_change_flags(struct net_device *dev, unsigned int flags,
3662 struct netlink_ext_ack *extack);
cb178190
DM
3663void __dev_notify_flags(struct net_device *, unsigned int old_flags,
3664 unsigned int gchanges);
f629d208
JP
3665int dev_change_name(struct net_device *, const char *);
3666int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 3667int dev_get_alias(const struct net_device *, char *, size_t);
f629d208 3668int dev_change_net_namespace(struct net_device *, struct net *, const char *);
f51048c3 3669int __dev_set_mtu(struct net_device *, int);
7a4c53be
SH
3670int dev_set_mtu_ext(struct net_device *dev, int mtu,
3671 struct netlink_ext_ack *extack);
f629d208 3672int dev_set_mtu(struct net_device *, int);
6a643ddb 3673int dev_change_tx_queue_len(struct net_device *, unsigned long);
f629d208 3674void dev_set_group(struct net_device *, int);
d59cdf94
PM
3675int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
3676 struct netlink_ext_ack *extack);
3a37a963
PM
3677int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
3678 struct netlink_ext_ack *extack);
f629d208
JP
3679int dev_change_carrier(struct net_device *, bool new_carrier);
3680int dev_get_phys_port_id(struct net_device *dev,
02637fce 3681 struct netdev_phys_item_id *ppid);
db24a904
DA
3682int dev_get_phys_port_name(struct net_device *dev,
3683 char *name, size_t len);
d6abc596
FF
3684int dev_get_port_parent_id(struct net_device *dev,
3685 struct netdev_phys_item_id *ppid, bool recurse);
3686bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
d746d707 3687int dev_change_proto_down(struct net_device *dev, bool proto_down);
b5899679 3688int dev_change_proto_down_generic(struct net_device *dev, bool proto_down);
f53c7239 3689struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
3690struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3691 struct netdev_queue *txq, int *ret);
d67b9cd2 3692
f4e63525 3693typedef int (*bpf_op_t)(struct net_device *dev, struct netdev_bpf *bpf);
d67b9cd2
DB
3694int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack,
3695 int fd, u32 flags);
a25717d2
JK
3696u32 __dev_xdp_query(struct net_device *dev, bpf_op_t xdp_op,
3697 enum bpf_netdev_command cmd);
84c6b868 3698int xdp_umem_query(struct net_device *dev, u16 queue_id);
d67b9cd2 3699
a0265d28 3700int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 3701int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
3702bool is_skb_forwardable(const struct net_device *dev,
3703 const struct sk_buff *skb);
1da177e4 3704
4e3264d2
MKL
3705static __always_inline int ____dev_forward_skb(struct net_device *dev,
3706 struct sk_buff *skb)
3707{
3708 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
3709 unlikely(!is_skb_forwardable(dev, skb))) {
3710 atomic_long_inc(&dev->rx_dropped);
3711 kfree_skb(skb);
3712 return NET_RX_DROP;
3713 }
3714
3715 skb_scrub_packet(skb, true);
3716 skb->priority = 0;
3717 return 0;
3718}
3719
9f9a742d 3720bool dev_nit_active(struct net_device *dev);
74b20582
DA
3721void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
3722
20380731 3723extern int netdev_budget;
7acf8a1e 3724extern unsigned int netdev_budget_usecs;
1da177e4
LT
3725
3726/* Called by rtnetlink.c:rtnl_unlock() */
f629d208 3727void netdev_run_todo(void);
1da177e4 3728
bea3348e
SH
3729/**
3730 * dev_put - release reference to device
3731 * @dev: network device
3732 *
9ef4429b 3733 * Release reference to device to allow it to be freed.
bea3348e 3734 */
1da177e4
LT
3735static inline void dev_put(struct net_device *dev)
3736{
933393f5 3737 this_cpu_dec(*dev->pcpu_refcnt);
1da177e4
LT
3738}
3739
bea3348e
SH
3740/**
3741 * dev_hold - get reference to device
3742 * @dev: network device
3743 *
9ef4429b 3744 * Hold reference to device to keep it from being freed.
bea3348e 3745 */
15333061
SH
3746static inline void dev_hold(struct net_device *dev)
3747{
933393f5 3748 this_cpu_inc(*dev->pcpu_refcnt);
15333061 3749}
1da177e4
LT
3750
3751/* Carrier loss detection, dial on demand. The functions netif_carrier_on
3752 * and _off may be called from IRQ context, but it is caller
3753 * who is responsible for serialization of these calls.
b00055aa
SR
3754 *
3755 * The name carrier is inappropriate, these functions should really be
3756 * called netif_lowerlayer_*() because they represent the state of any
3757 * kind of lower layer not just hardware media.
1da177e4
LT
3758 */
3759
f629d208
JP
3760void linkwatch_init_dev(struct net_device *dev);
3761void linkwatch_fire_event(struct net_device *dev);
3762void linkwatch_forget_dev(struct net_device *dev);
1da177e4 3763
bea3348e
SH
3764/**
3765 * netif_carrier_ok - test if carrier present
3766 * @dev: network device
3767 *
3768 * Check if carrier is present on device
3769 */
4d29515f 3770static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
3771{
3772 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
3773}
3774
f629d208 3775unsigned long dev_trans_start(struct net_device *dev);
9d21493b 3776
f629d208 3777void __netdev_watchdog_up(struct net_device *dev);
1da177e4 3778
f629d208 3779void netif_carrier_on(struct net_device *dev);
1da177e4 3780
f629d208 3781void netif_carrier_off(struct net_device *dev);
1da177e4 3782
bea3348e
SH
3783/**
3784 * netif_dormant_on - mark device as dormant.
3785 * @dev: network device
3786 *
3787 * Mark device as dormant (as per RFC2863).
3788 *
3789 * The dormant state indicates that the relevant interface is not
3790 * actually in a condition to pass packets (i.e., it is not 'up') but is
3791 * in a "pending" state, waiting for some external event. For "on-
3792 * demand" interfaces, this new state identifies the situation where the
3793 * interface is waiting for events to place it in the up state.
bea3348e 3794 */
b00055aa
SR
3795static inline void netif_dormant_on(struct net_device *dev)
3796{
3797 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
3798 linkwatch_fire_event(dev);
3799}
3800
bea3348e
SH
3801/**
3802 * netif_dormant_off - set device as not dormant.
3803 * @dev: network device
3804 *
3805 * Device is not in dormant state.
3806 */
b00055aa
SR
3807static inline void netif_dormant_off(struct net_device *dev)
3808{
3809 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
3810 linkwatch_fire_event(dev);
3811}
3812
bea3348e 3813/**
8ecbc40a 3814 * netif_dormant - test if device is dormant
bea3348e
SH
3815 * @dev: network device
3816 *
8ecbc40a 3817 * Check if device is dormant.
bea3348e 3818 */
4d29515f 3819static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
3820{
3821 return test_bit(__LINK_STATE_DORMANT, &dev->state);
3822}
3823
3824
bea3348e
SH
3825/**
3826 * netif_oper_up - test if device is operational
3827 * @dev: network device
3828 *
3829 * Check if carrier is operational
3830 */
4d29515f 3831static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 3832{
b00055aa
SR
3833 return (dev->operstate == IF_OPER_UP ||
3834 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
3835}
3836
bea3348e
SH
3837/**
3838 * netif_device_present - is device available or removed
3839 * @dev: network device
3840 *
3841 * Check if device has not been removed from system.
3842 */
4d29515f 3843static inline bool netif_device_present(struct net_device *dev)
1da177e4
LT
3844{
3845 return test_bit(__LINK_STATE_PRESENT, &dev->state);
3846}
3847
f629d208 3848void netif_device_detach(struct net_device *dev);
1da177e4 3849
f629d208 3850void netif_device_attach(struct net_device *dev);
1da177e4
LT
3851
3852/*
3853 * Network interface message level settings
3854 */
1da177e4
LT
3855
3856enum {
3857 NETIF_MSG_DRV = 0x0001,
3858 NETIF_MSG_PROBE = 0x0002,
3859 NETIF_MSG_LINK = 0x0004,
3860 NETIF_MSG_TIMER = 0x0008,
3861 NETIF_MSG_IFDOWN = 0x0010,
3862 NETIF_MSG_IFUP = 0x0020,
3863 NETIF_MSG_RX_ERR = 0x0040,
3864 NETIF_MSG_TX_ERR = 0x0080,
3865 NETIF_MSG_TX_QUEUED = 0x0100,
3866 NETIF_MSG_INTR = 0x0200,
3867 NETIF_MSG_TX_DONE = 0x0400,
3868 NETIF_MSG_RX_STATUS = 0x0800,
3869 NETIF_MSG_PKTDATA = 0x1000,
3870 NETIF_MSG_HW = 0x2000,
3871 NETIF_MSG_WOL = 0x4000,
3872};
3873
3874#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
3875#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
3876#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
3877#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
3878#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
3879#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
3880#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
3881#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
3882#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
3883#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
3884#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
3885#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
3886#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
3887#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
3888#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
3889
3890static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
3891{
3892 /* use default */
3893 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
3894 return default_msg_enable_bits;
3895 if (debug_value == 0) /* no output */
3896 return 0;
3897 /* set low N bits */
f4d7b3e2 3898 return (1U << debug_value) - 1;
1da177e4
LT
3899}
3900
c773e847 3901static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 3902{
c773e847
DM
3903 spin_lock(&txq->_xmit_lock);
3904 txq->xmit_lock_owner = cpu;
22dd7495
JHS
3905}
3906
5a717f4f
MT
3907static inline bool __netif_tx_acquire(struct netdev_queue *txq)
3908{
3909 __acquire(&txq->_xmit_lock);
3910 return true;
3911}
3912
3913static inline void __netif_tx_release(struct netdev_queue *txq)
3914{
3915 __release(&txq->_xmit_lock);
3916}
3917
fd2ea0a7
DM
3918static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
3919{
3920 spin_lock_bh(&txq->_xmit_lock);
3921 txq->xmit_lock_owner = smp_processor_id();
3922}
3923
4d29515f 3924static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 3925{
4d29515f 3926 bool ok = spin_trylock(&txq->_xmit_lock);
c3f26a26
DM
3927 if (likely(ok))
3928 txq->xmit_lock_owner = smp_processor_id();
3929 return ok;
3930}
3931
3932static inline void __netif_tx_unlock(struct netdev_queue *txq)
3933{
3934 txq->xmit_lock_owner = -1;
3935 spin_unlock(&txq->_xmit_lock);
3936}
3937
3938static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
3939{
3940 txq->xmit_lock_owner = -1;
3941 spin_unlock_bh(&txq->_xmit_lock);
3942}
3943
08baf561
ED
3944static inline void txq_trans_update(struct netdev_queue *txq)
3945{
3946 if (txq->xmit_lock_owner != -1)
3947 txq->trans_start = jiffies;
3948}
3949
ba162f8e
FW
3950/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
3951static inline void netif_trans_update(struct net_device *dev)
3952{
9b36627a
FW
3953 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
3954
3955 if (txq->trans_start != jiffies)
3956 txq->trans_start = jiffies;
ba162f8e
FW
3957}
3958
d29f749e
DJ
3959/**
3960 * netif_tx_lock - grab network device transmit lock
3961 * @dev: network device
d29f749e
DJ
3962 *
3963 * Get network device transmit lock
3964 */
22dd7495
JHS
3965static inline void netif_tx_lock(struct net_device *dev)
3966{
e8a0464c 3967 unsigned int i;
c3f26a26 3968 int cpu;
c773e847 3969
c3f26a26
DM
3970 spin_lock(&dev->tx_global_lock);
3971 cpu = smp_processor_id();
e8a0464c
DM
3972 for (i = 0; i < dev->num_tx_queues; i++) {
3973 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
3974
3975 /* We are the only thread of execution doing a
3976 * freeze, but we have to grab the _xmit_lock in
3977 * order to synchronize with threads which are in
3978 * the ->hard_start_xmit() handler and already
3979 * checked the frozen bit.
3980 */
e8a0464c 3981 __netif_tx_lock(txq, cpu);
c3f26a26
DM
3982 set_bit(__QUEUE_STATE_FROZEN, &txq->state);
3983 __netif_tx_unlock(txq);
e8a0464c 3984 }
932ff279
HX
3985}
3986
3987static inline void netif_tx_lock_bh(struct net_device *dev)
3988{
e8a0464c
DM
3989 local_bh_disable();
3990 netif_tx_lock(dev);
932ff279
HX
3991}
3992
932ff279
HX
3993static inline void netif_tx_unlock(struct net_device *dev)
3994{
e8a0464c
DM
3995 unsigned int i;
3996
3997 for (i = 0; i < dev->num_tx_queues; i++) {
3998 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c773e847 3999
c3f26a26
DM
4000 /* No need to grab the _xmit_lock here. If the
4001 * queue is not stopped for another reason, we
4002 * force a schedule.
4003 */
4004 clear_bit(__QUEUE_STATE_FROZEN, &txq->state);
7b3d3e4f 4005 netif_schedule_queue(txq);
c3f26a26
DM
4006 }
4007 spin_unlock(&dev->tx_global_lock);
932ff279
HX
4008}
4009
4010static inline void netif_tx_unlock_bh(struct net_device *dev)
4011{
e8a0464c
DM
4012 netif_tx_unlock(dev);
4013 local_bh_enable();
932ff279
HX
4014}
4015
c773e847 4016#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 4017 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4018 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
4019 } else { \
4020 __netif_tx_acquire(txq); \
22dd7495
JHS
4021 } \
4022}
4023
5efeac44
EB
4024#define HARD_TX_TRYLOCK(dev, txq) \
4025 (((dev->features & NETIF_F_LLTX) == 0) ? \
4026 __netif_tx_trylock(txq) : \
5a717f4f 4027 __netif_tx_acquire(txq))
5efeac44 4028
c773e847 4029#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 4030 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4031 __netif_tx_unlock(txq); \
5a717f4f
MT
4032 } else { \
4033 __netif_tx_release(txq); \
22dd7495
JHS
4034 } \
4035}
4036
1da177e4
LT
4037static inline void netif_tx_disable(struct net_device *dev)
4038{
fd2ea0a7 4039 unsigned int i;
c3f26a26 4040 int cpu;
fd2ea0a7 4041
c3f26a26
DM
4042 local_bh_disable();
4043 cpu = smp_processor_id();
fd2ea0a7
DM
4044 for (i = 0; i < dev->num_tx_queues; i++) {
4045 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4046
4047 __netif_tx_lock(txq, cpu);
fd2ea0a7 4048 netif_tx_stop_queue(txq);
c3f26a26 4049 __netif_tx_unlock(txq);
fd2ea0a7 4050 }
c3f26a26 4051 local_bh_enable();
1da177e4
LT
4052}
4053
e308a5d8
DM
4054static inline void netif_addr_lock(struct net_device *dev)
4055{
4056 spin_lock(&dev->addr_list_lock);
4057}
4058
2429f7ac
JP
4059static inline void netif_addr_lock_nested(struct net_device *dev)
4060{
25175ba5
VY
4061 int subclass = SINGLE_DEPTH_NESTING;
4062
4063 if (dev->netdev_ops->ndo_get_lock_subclass)
4064 subclass = dev->netdev_ops->ndo_get_lock_subclass(dev);
4065
4066 spin_lock_nested(&dev->addr_list_lock, subclass);
2429f7ac
JP
4067}
4068
e308a5d8
DM
4069static inline void netif_addr_lock_bh(struct net_device *dev)
4070{
4071 spin_lock_bh(&dev->addr_list_lock);
4072}
4073
4074static inline void netif_addr_unlock(struct net_device *dev)
4075{
4076 spin_unlock(&dev->addr_list_lock);
4077}
4078
4079static inline void netif_addr_unlock_bh(struct net_device *dev)
4080{
4081 spin_unlock_bh(&dev->addr_list_lock);
4082}
4083
f001fde5 4084/*
31278e71 4085 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4086 * rcu_read_lock held.
4087 */
4088#define for_each_dev_addr(dev, ha) \
31278e71 4089 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4090
1da177e4
LT
4091/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4092
f629d208 4093void ether_setup(struct net_device *dev);
1da177e4
LT
4094
4095/* Support for loadable net-drivers */
f629d208 4096struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4097 unsigned char name_assign_type,
f629d208
JP
4098 void (*setup)(struct net_device *),
4099 unsigned int txqs, unsigned int rxqs);
0ad646c8
CW
4100int dev_get_valid_name(struct net *net, struct net_device *dev,
4101 const char *name);
4102
c835a677
TG
4103#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4104 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4105
c835a677
TG
4106#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4107 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4108 count)
36909ea4 4109
f629d208
JP
4110int register_netdev(struct net_device *dev);
4111void unregister_netdev(struct net_device *dev);
f001fde5 4112
22bedad3 4113/* General hardware address lists handling functions */
f629d208
JP
4114int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4115 struct netdev_hw_addr_list *from_list, int addr_len);
4116void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4117 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4118int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4119 struct net_device *dev,
4120 int (*sync)(struct net_device *, const unsigned char *),
4121 int (*unsync)(struct net_device *,
4122 const unsigned char *));
e7946760
IK
4123int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4124 struct net_device *dev,
4125 int (*sync)(struct net_device *,
4126 const unsigned char *, int),
4127 int (*unsync)(struct net_device *,
4128 const unsigned char *, int));
4129void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4130 struct net_device *dev,
4131 int (*unsync)(struct net_device *,
4132 const unsigned char *, int));
670e5b8e
AD
4133void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4134 struct net_device *dev,
4135 int (*unsync)(struct net_device *,
4136 const unsigned char *));
f629d208 4137void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4138
f001fde5 4139/* Functions used for device addresses handling */
f629d208
JP
4140int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4141 unsigned char addr_type);
4142int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4143 unsigned char addr_type);
f629d208
JP
4144void dev_addr_flush(struct net_device *dev);
4145int dev_addr_init(struct net_device *dev);
a748ee24
JP
4146
4147/* Functions used for unicast addresses handling */
f629d208
JP
4148int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4149int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4150int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4151int dev_uc_sync(struct net_device *to, struct net_device *from);
4152int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4153void dev_uc_unsync(struct net_device *to, struct net_device *from);
4154void dev_uc_flush(struct net_device *dev);
4155void dev_uc_init(struct net_device *dev);
f001fde5 4156
670e5b8e
AD
4157/**
4158 * __dev_uc_sync - Synchonize device's unicast list
4159 * @dev: device to sync
4160 * @sync: function to call if address should be added
4161 * @unsync: function to call if address should be removed
4162 *
4163 * Add newly added addresses to the interface, and release
4164 * addresses that have been deleted.
5e82b4b2 4165 */
670e5b8e
AD
4166static inline int __dev_uc_sync(struct net_device *dev,
4167 int (*sync)(struct net_device *,
4168 const unsigned char *),
4169 int (*unsync)(struct net_device *,
4170 const unsigned char *))
4171{
4172 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4173}
4174
4175/**
e793c0f7 4176 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4177 * @dev: device to sync
4178 * @unsync: function to call if address should be removed
4179 *
4180 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4181 */
670e5b8e
AD
4182static inline void __dev_uc_unsync(struct net_device *dev,
4183 int (*unsync)(struct net_device *,
4184 const unsigned char *))
4185{
4186 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4187}
4188
22bedad3 4189/* Functions used for multicast addresses handling */
f629d208
JP
4190int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4191int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4192int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4193int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4194int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4195int dev_mc_sync(struct net_device *to, struct net_device *from);
4196int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4197void dev_mc_unsync(struct net_device *to, struct net_device *from);
4198void dev_mc_flush(struct net_device *dev);
4199void dev_mc_init(struct net_device *dev);
f001fde5 4200
670e5b8e
AD
4201/**
4202 * __dev_mc_sync - Synchonize device's multicast list
4203 * @dev: device to sync
4204 * @sync: function to call if address should be added
4205 * @unsync: function to call if address should be removed
4206 *
4207 * Add newly added addresses to the interface, and release
4208 * addresses that have been deleted.
5e82b4b2 4209 */
670e5b8e
AD
4210static inline int __dev_mc_sync(struct net_device *dev,
4211 int (*sync)(struct net_device *,
4212 const unsigned char *),
4213 int (*unsync)(struct net_device *,
4214 const unsigned char *))
4215{
4216 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4217}
4218
4219/**
e793c0f7 4220 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4221 * @dev: device to sync
4222 * @unsync: function to call if address should be removed
4223 *
4224 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4225 */
670e5b8e
AD
4226static inline void __dev_mc_unsync(struct net_device *dev,
4227 int (*unsync)(struct net_device *,
4228 const unsigned char *))
4229{
4230 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4231}
4232
4417da66 4233/* Functions used for secondary unicast and multicast support */
f629d208
JP
4234void dev_set_rx_mode(struct net_device *dev);
4235void __dev_set_rx_mode(struct net_device *dev);
4236int dev_set_promiscuity(struct net_device *dev, int inc);
4237int dev_set_allmulti(struct net_device *dev, int inc);
4238void netdev_state_change(struct net_device *dev);
4239void netdev_notify_peers(struct net_device *dev);
4240void netdev_features_change(struct net_device *dev);
1da177e4 4241/* Load a device via the kmod */
f629d208
JP
4242void dev_load(struct net *net, const char *name);
4243struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4244 struct rtnl_link_stats64 *storage);
4245void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4246 const struct net_device_stats *netdev_stats);
eeda3fd6 4247
1da177e4 4248extern int netdev_max_backlog;
3b098e2d 4249extern int netdev_tstamp_prequeue;
1da177e4 4250extern int weight_p;
3d48b53f
MT
4251extern int dev_weight_rx_bias;
4252extern int dev_weight_tx_bias;
4253extern int dev_rx_weight;
4254extern int dev_tx_weight;
323ebb61 4255extern int gro_normal_batch;
9ff162a8 4256
f629d208 4257bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4258struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4259 struct list_head **iter);
f629d208
JP
4260struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
4261 struct list_head **iter);
8b5be856 4262
44a40855
VY
4263/* iterate through upper list, must be called under RCU read lock */
4264#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4265 for (iter = &(dev)->adj_list.upper, \
4266 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4267 updev; \
4268 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4269
1a3f060c
DA
4270int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4271 int (*fn)(struct net_device *upper_dev,
4272 void *data),
4273 void *data);
4274
4275bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4276 struct net_device *upper_dev);
4277
25cc72a3
IS
4278bool netdev_has_any_upper_dev(struct net_device *dev);
4279
f629d208
JP
4280void *netdev_lower_get_next_private(struct net_device *dev,
4281 struct list_head **iter);
4282void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4283 struct list_head **iter);
31088a11
VF
4284
4285#define netdev_for_each_lower_private(dev, priv, iter) \
4286 for (iter = (dev)->adj_list.lower.next, \
4287 priv = netdev_lower_get_next_private(dev, &(iter)); \
4288 priv; \
4289 priv = netdev_lower_get_next_private(dev, &(iter)))
4290
4291#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4292 for (iter = &(dev)->adj_list.lower, \
4293 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4294 priv; \
4295 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4296
4085ebe8
VY
4297void *netdev_lower_get_next(struct net_device *dev,
4298 struct list_head **iter);
7ce856aa 4299
4085ebe8 4300#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4301 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4302 ldev = netdev_lower_get_next(dev, &(iter)); \
4303 ldev; \
4304 ldev = netdev_lower_get_next(dev, &(iter)))
4305
7ce856aa
JP
4306struct net_device *netdev_all_lower_get_next(struct net_device *dev,
4307 struct list_head **iter);
4308struct net_device *netdev_all_lower_get_next_rcu(struct net_device *dev,
4309 struct list_head **iter);
4310
1a3f060c
DA
4311int netdev_walk_all_lower_dev(struct net_device *dev,
4312 int (*fn)(struct net_device *lower_dev,
4313 void *data),
4314 void *data);
4315int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4316 int (*fn)(struct net_device *lower_dev,
4317 void *data),
4318 void *data);
4319
f629d208 4320void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4321void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4322struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4323struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4324int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4325 struct netlink_ext_ack *extack);
f629d208 4326int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4327 struct net_device *upper_dev,
42ab19ee
DA
4328 void *upper_priv, void *upper_info,
4329 struct netlink_ext_ack *extack);
f629d208
JP
4330void netdev_upper_dev_unlink(struct net_device *dev,
4331 struct net_device *upper_dev);
5bb025fa 4332void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4333void *netdev_lower_dev_get_private(struct net_device *dev,
4334 struct net_device *lower_dev);
04d48266
JP
4335void netdev_lower_state_changed(struct net_device *lower_dev,
4336 void *lower_state_info);
960fb622
ED
4337
4338/* RSS keys are 40 or 52 bytes long */
4339#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4340extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4341void netdev_rss_key_fill(void *buffer, size_t len);
4342
952fcfd0 4343int dev_get_nest_level(struct net_device *dev);
f629d208 4344int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4345int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4346int skb_csum_hwoffload_help(struct sk_buff *skb,
4347 const netdev_features_t features);
4348
f629d208
JP
4349struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
4350 netdev_features_t features, bool tx_path);
4351struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
4352 netdev_features_t features);
12b0004d 4353
61bd3857
MS
4354struct netdev_bonding_info {
4355 ifslave slave;
4356 ifbond master;
4357};
4358
4359struct netdev_notifier_bonding_info {
4360 struct netdev_notifier_info info; /* must be first */
4361 struct netdev_bonding_info bonding_info;
4362};
4363
4364void netdev_bonding_info_change(struct net_device *dev,
4365 struct netdev_bonding_info *bonding_info);
4366
12b0004d
CW
4367static inline
4368struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
4369{
4370 return __skb_gso_segment(skb, features, true);
4371}
53d6471c 4372__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4373
4374static inline bool can_checksum_protocol(netdev_features_t features,
4375 __be16 protocol)
4376{
c8cd0989
TH
4377 if (protocol == htons(ETH_P_FCOE))
4378 return !!(features & NETIF_F_FCOE_CRC);
4379
4380 /* Assume this is an IP checksum (not SCTP CRC) */
4381
4382 if (features & NETIF_F_HW_CSUM) {
4383 /* Can checksum everything */
4384 return true;
4385 }
4386
4387 switch (protocol) {
4388 case htons(ETH_P_IP):
4389 return !!(features & NETIF_F_IP_CSUM);
4390 case htons(ETH_P_IPV6):
4391 return !!(features & NETIF_F_IPV6_CSUM);
4392 default:
4393 return false;
4394 }
ec5f0615 4395}
12b0004d 4396
fb286bb2 4397#ifdef CONFIG_BUG
7fe50ac8 4398void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4399#else
7fe50ac8
CW
4400static inline void netdev_rx_csum_fault(struct net_device *dev,
4401 struct sk_buff *skb)
fb286bb2
HX
4402{
4403}
4404#endif
1da177e4 4405/* rx skb timestamps */
f629d208
JP
4406void net_enable_timestamp(void);
4407void net_disable_timestamp(void);
1da177e4 4408
20380731 4409#ifdef CONFIG_PROC_FS
f629d208 4410int __init dev_proc_init(void);
900ff8c6
CW
4411#else
4412#define dev_proc_init() 0
20380731
ACM
4413#endif
4414
4798248e 4415static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4416 struct sk_buff *skb, struct net_device *dev,
4417 bool more)
4798248e 4418{
6b16f9ee 4419 __this_cpu_write(softnet_data.xmit.more, more);
0b725a2c 4420 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4421}
4422
97cdcf37
FW
4423static inline bool netdev_xmit_more(void)
4424{
4425 return __this_cpu_read(softnet_data.xmit.more);
4426}
4427
10b3ad8c 4428static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4429 struct netdev_queue *txq, bool more)
4798248e
DM
4430{
4431 const struct net_device_ops *ops = dev->netdev_ops;
2183435c 4432 netdev_tx_t rc;
4798248e 4433
fa2dbdc2 4434 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4435 if (rc == NETDEV_TX_OK)
4436 txq_trans_update(txq);
4437
4438 return rc;
4798248e
DM
4439}
4440
b793dc5c 4441int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4442 const void *ns);
b793dc5c 4443void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4444 const void *ns);
58292cbe 4445
b793dc5c 4446static inline int netdev_class_create_file(const struct class_attribute *class_attr)
58292cbe
TH
4447{
4448 return netdev_class_create_file_ns(class_attr, NULL);
4449}
4450
b793dc5c 4451static inline void netdev_class_remove_file(const struct class_attribute *class_attr)
58292cbe
TH
4452{
4453 netdev_class_remove_file_ns(class_attr, NULL);
4454}
b8a9787e 4455
737aec57 4456extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4457
f629d208 4458const char *netdev_drivername(const struct net_device *dev);
6579e57b 4459
f629d208 4460void linkwatch_run_queue(void);
20380731 4461
da08143b
MK
4462static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4463 netdev_features_t f2)
4464{
c8cd0989
TH
4465 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4466 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4467 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4468 else
b6a0e72a 4469 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4470 }
da08143b 4471
c8cd0989 4472 return f1 & f2;
da08143b
MK
4473}
4474
c8f44aff
MM
4475static inline netdev_features_t netdev_get_wanted_features(
4476 struct net_device *dev)
5455c699
MM
4477{
4478 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4479}
c8f44aff
MM
4480netdev_features_t netdev_increment_features(netdev_features_t all,
4481 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4482
4483/* Allow TSO being used on stacked device :
4484 * Performing the GSO segmentation before last device
4485 * is a performance improvement.
4486 */
4487static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4488 netdev_features_t mask)
4489{
4490 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4491}
4492
6cb6a27c 4493int __netdev_update_features(struct net_device *dev);
5455c699 4494void netdev_update_features(struct net_device *dev);
afe12cc8 4495void netdev_change_features(struct net_device *dev);
7f353bf2 4496
fc4a7489
PM
4497void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4498 struct net_device *dev);
4499
e38f3025
TM
4500netdev_features_t passthru_features_check(struct sk_buff *skb,
4501 struct net_device *dev,
4502 netdev_features_t features);
c1e756bf 4503netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 4504
4d29515f 4505static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 4506{
7b748340 4507 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
4508
4509 /* check flags correspondence */
4510 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
4511 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
4512 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 4513 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
4514 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
4515 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4516 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
4517 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
4518 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
4519 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4520 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
4521 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 4522 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 4523 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 4524 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 4525 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 4526 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 4527 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
0345e186 4528
d6b4991a 4529 return (features & feature) == feature;
576a30eb
HX
4530}
4531
4d29515f 4532static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 4533{
278b2513 4534 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 4535 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
4536}
4537
8b86a61d 4538static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 4539 netdev_features_t features)
7967168c 4540{
fc741216 4541 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
4542 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
4543 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
4544}
4545
82cc1a7a
PWJ
4546static inline void netif_set_gso_max_size(struct net_device *dev,
4547 unsigned int size)
4548{
4549 dev->gso_max_size = size;
4550}
4551
7a7ffbab
WCC
4552static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
4553 int pulled_hlen, u16 mac_offset,
4554 int mac_len)
4555{
4556 skb->protocol = protocol;
4557 skb->encapsulation = 1;
4558 skb_push(skb, pulled_hlen);
4559 skb_reset_transport_header(skb);
4560 skb->mac_header = mac_offset;
4561 skb->network_header = skb->mac_header + mac_len;
4562 skb->mac_len = mac_len;
4563}
4564
3c175784
SD
4565static inline bool netif_is_macsec(const struct net_device *dev)
4566{
4567 return dev->priv_flags & IFF_MACSEC;
4568}
4569
b618aaa9 4570static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
4571{
4572 return dev->priv_flags & IFF_MACVLAN;
4573}
4574
b618aaa9 4575static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
4576{
4577 return dev->priv_flags & IFF_MACVLAN_PORT;
4578}
4579
b618aaa9 4580static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 4581{
4582 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
4583}
4584
b618aaa9 4585static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
4586{
4587 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
4588}
4589
3bdc0eba
BG
4590static inline bool netif_supports_nofcs(struct net_device *dev)
4591{
4592 return dev->priv_flags & IFF_SUPP_NOFCS;
4593}
4594
d5256083
DB
4595static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
4596{
4597 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
4598}
4599
007979ea 4600static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 4601{
007979ea 4602 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
4603}
4604
fee6d4c7
DA
4605static inline bool netif_is_l3_slave(const struct net_device *dev)
4606{
4607 return dev->priv_flags & IFF_L3MDEV_SLAVE;
4608}
4609
0894ae3f
JP
4610static inline bool netif_is_bridge_master(const struct net_device *dev)
4611{
4612 return dev->priv_flags & IFF_EBRIDGE;
4613}
4614
28f9ee22
VY
4615static inline bool netif_is_bridge_port(const struct net_device *dev)
4616{
4617 return dev->priv_flags & IFF_BRIDGE_PORT;
4618}
4619
35d4e172
JP
4620static inline bool netif_is_ovs_master(const struct net_device *dev)
4621{
4622 return dev->priv_flags & IFF_OPENVSWITCH;
4623}
4624
5be66141
JP
4625static inline bool netif_is_ovs_port(const struct net_device *dev)
4626{
4627 return dev->priv_flags & IFF_OVS_DATAPATH;
4628}
4629
b618aaa9 4630static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
4631{
4632 return dev->priv_flags & IFF_TEAM;
4633}
4634
b618aaa9 4635static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
4636{
4637 return dev->priv_flags & IFF_TEAM_PORT;
4638}
4639
b618aaa9 4640static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
4641{
4642 return netif_is_bond_master(dev) || netif_is_team_master(dev);
4643}
4644
b618aaa9 4645static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
4646{
4647 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
4648}
4649
d4ab4286
KJ
4650static inline bool netif_is_rxfh_configured(const struct net_device *dev)
4651{
4652 return dev->priv_flags & IFF_RXFH_CONFIGURED;
4653}
4654
30c8bd5a
SS
4655static inline bool netif_is_failover(const struct net_device *dev)
4656{
4657 return dev->priv_flags & IFF_FAILOVER;
4658}
4659
4660static inline bool netif_is_failover_slave(const struct net_device *dev)
4661{
4662 return dev->priv_flags & IFF_FAILOVER_SLAVE;
4663}
4664
02875878
ED
4665/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
4666static inline void netif_keep_dst(struct net_device *dev)
4667{
4668 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
4669}
4670
18d3df3e
PA
4671/* return true if dev can't cope with mtu frames that need vlan tag insertion */
4672static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
4673{
4674 /* TODO: reserve and use an additional IFF bit, if we get more users */
4675 return dev->priv_flags & IFF_MACSEC;
4676}
4677
505d4f73 4678extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 4679
571ba423
JP
4680/* Logging, debugging and troubleshooting/diagnostic helpers. */
4681
4682/* netdev_printk helpers, similar to dev_printk */
4683
4684static inline const char *netdev_name(const struct net_device *dev)
4685{
c6f854d5
VF
4686 if (!dev->name[0] || strchr(dev->name, '%'))
4687 return "(unnamed net_device)";
571ba423
JP
4688 return dev->name;
4689}
4690
8397ed36
DA
4691static inline bool netdev_unregistering(const struct net_device *dev)
4692{
4693 return dev->reg_state == NETREG_UNREGISTERING;
4694}
4695
ccc7f496
VF
4696static inline const char *netdev_reg_state(const struct net_device *dev)
4697{
4698 switch (dev->reg_state) {
4699 case NETREG_UNINITIALIZED: return " (uninitialized)";
4700 case NETREG_REGISTERED: return "";
4701 case NETREG_UNREGISTERING: return " (unregistering)";
4702 case NETREG_UNREGISTERED: return " (unregistered)";
4703 case NETREG_RELEASED: return " (released)";
4704 case NETREG_DUMMY: return " (dummy)";
4705 }
4706
4707 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
4708 return " (unknown)";
4709}
4710
ce3fdb69 4711__printf(3, 4) __cold
6ea754eb
JP
4712void netdev_printk(const char *level, const struct net_device *dev,
4713 const char *format, ...);
ce3fdb69 4714__printf(2, 3) __cold
6ea754eb 4715void netdev_emerg(const struct net_device *dev, const char *format, ...);
ce3fdb69 4716__printf(2, 3) __cold
6ea754eb 4717void netdev_alert(const struct net_device *dev, const char *format, ...);
ce3fdb69 4718__printf(2, 3) __cold
6ea754eb 4719void netdev_crit(const struct net_device *dev, const char *format, ...);
ce3fdb69 4720__printf(2, 3) __cold
6ea754eb 4721void netdev_err(const struct net_device *dev, const char *format, ...);
ce3fdb69 4722__printf(2, 3) __cold
6ea754eb 4723void netdev_warn(const struct net_device *dev, const char *format, ...);
ce3fdb69 4724__printf(2, 3) __cold
6ea754eb 4725void netdev_notice(const struct net_device *dev, const char *format, ...);
ce3fdb69 4726__printf(2, 3) __cold
6ea754eb 4727void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 4728
375ef2b1
GP
4729#define netdev_level_once(level, dev, fmt, ...) \
4730do { \
4731 static bool __print_once __read_mostly; \
4732 \
4733 if (!__print_once) { \
4734 __print_once = true; \
4735 netdev_printk(level, dev, fmt, ##__VA_ARGS__); \
4736 } \
4737} while (0)
4738
4739#define netdev_emerg_once(dev, fmt, ...) \
4740 netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
4741#define netdev_alert_once(dev, fmt, ...) \
4742 netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
4743#define netdev_crit_once(dev, fmt, ...) \
4744 netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
4745#define netdev_err_once(dev, fmt, ...) \
4746 netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
4747#define netdev_warn_once(dev, fmt, ...) \
4748 netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
4749#define netdev_notice_once(dev, fmt, ...) \
4750 netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
4751#define netdev_info_once(dev, fmt, ...) \
4752 netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
4753
8909c9ad
VK
4754#define MODULE_ALIAS_NETDEV(device) \
4755 MODULE_ALIAS("netdev-" device)
4756
b558c96f 4757#if defined(CONFIG_DYNAMIC_DEBUG)
571ba423
JP
4758#define netdev_dbg(__dev, format, args...) \
4759do { \
ffa10cb4 4760 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 4761} while (0)
b558c96f
JC
4762#elif defined(DEBUG)
4763#define netdev_dbg(__dev, format, args...) \
4764 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
4765#else
4766#define netdev_dbg(__dev, format, args...) \
4767({ \
4768 if (0) \
4769 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
4770})
4771#endif
4772
4773#if defined(VERBOSE_DEBUG)
4774#define netdev_vdbg netdev_dbg
4775#else
4776
4777#define netdev_vdbg(dev, format, args...) \
4778({ \
4779 if (0) \
4780 netdev_printk(KERN_DEBUG, dev, format, ##args); \
4781 0; \
4782})
4783#endif
4784
4785/*
4786 * netdev_WARN() acts like dev_printk(), but with the key difference
4787 * of using a WARN/WARN_ON to get the message out, including the
4788 * file/line information and a backtrace.
4789 */
4790#define netdev_WARN(dev, format, args...) \
e1cfe3d0 4791 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 4792 netdev_reg_state(dev), ##args)
571ba423 4793
72dd831e 4794#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 4795 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
4796 netdev_reg_state(dev), ##args)
4797
b3d95c5c
JP
4798/* netif printk helpers, similar to netdev_printk */
4799
4800#define netif_printk(priv, type, level, dev, fmt, args...) \
4801do { \
4802 if (netif_msg_##type(priv)) \
4803 netdev_printk(level, (dev), fmt, ##args); \
4804} while (0)
4805
f45f4321
JP
4806#define netif_level(level, priv, type, dev, fmt, args...) \
4807do { \
4808 if (netif_msg_##type(priv)) \
4809 netdev_##level(dev, fmt, ##args); \
4810} while (0)
4811
b3d95c5c 4812#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 4813 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 4814#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 4815 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 4816#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 4817 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 4818#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 4819 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 4820#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 4821 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 4822#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 4823 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 4824#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 4825 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 4826
0053ea9c 4827#if defined(CONFIG_DYNAMIC_DEBUG)
b3d95c5c
JP
4828#define netif_dbg(priv, type, netdev, format, args...) \
4829do { \
4830 if (netif_msg_##type(priv)) \
b5fb0a03 4831 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 4832} while (0)
0053ea9c
JP
4833#elif defined(DEBUG)
4834#define netif_dbg(priv, type, dev, format, args...) \
4835 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
4836#else
4837#define netif_dbg(priv, type, dev, format, args...) \
4838({ \
4839 if (0) \
4840 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
4841 0; \
4842})
4843#endif
4844
f617f276
EC
4845/* if @cond then downgrade to debug, else print at @level */
4846#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...) \
4847 do { \
4848 if (cond) \
4849 netif_dbg(priv, type, netdev, fmt, ##args); \
4850 else \
4851 netif_ ## level(priv, type, netdev, fmt, ##args); \
4852 } while (0)
4853
b3d95c5c 4854#if defined(VERBOSE_DEBUG)
bcfcc450 4855#define netif_vdbg netif_dbg
b3d95c5c
JP
4856#else
4857#define netif_vdbg(priv, type, dev, format, args...) \
4858({ \
4859 if (0) \
a4ed89cb 4860 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
4861 0; \
4862})
4863#endif
571ba423 4864
900ff8c6
CW
4865/*
4866 * The list of packet types we will receive (as opposed to discard)
4867 * and the routines to invoke.
4868 *
4869 * Why 16. Because with 16 the only overlap we get on a hash of the
4870 * low nibble of the protocol value is RARP/SNAP/X.25.
4871 *
900ff8c6 4872 * 0800 IP
900ff8c6
CW
4873 * 0001 802.3
4874 * 0002 AX.25
4875 * 0004 802.2
4876 * 8035 RARP
4877 * 0005 SNAP
4878 * 0805 X.25
4879 * 0806 ARP
4880 * 8137 IPX
4881 * 0009 Localtalk
4882 * 86DD IPv6
4883 */
4884#define PTYPE_HASH_SIZE (16)
4885#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
4886
4de83b88
MB
4887extern struct net_device *blackhole_netdev;
4888
385a154c 4889#endif /* _LINUX_NETDEVICE_H */